aboutsummaryrefslogtreecommitdiffstats
path: root/net
diff options
context:
space:
mode:
authorJeff Garzik <jgarzik@pobox.com>2005-09-14 08:19:08 -0400
committerJeff Garzik <jgarzik@pobox.com>2005-09-14 08:19:08 -0400
commit905ec87e93bc9e01b15c60035cd6a50c636cbaef (patch)
tree46fd7618d6511611ffc19eb0dd4d7bc6b90a41c2 /net
parent1d6ae775d7a948c9575658eb41184fd2e506c0df (diff)
parent2f4ba45a75d6383b4a1201169a808ffea416ffa0 (diff)
Merge /spare/repo/linux-2.6/
Diffstat (limited to 'net')
-rw-r--r--net/atm/mpc.c2
-rw-r--r--net/ax25/af_ax25.c46
-rw-r--r--net/ax25/ax25_addr.c27
-rw-r--r--net/ax25/ax25_ip.c6
-rw-r--r--net/compat.c44
-rw-r--r--net/core/dst.c3
-rw-r--r--net/core/netpoll.c4
-rw-r--r--net/core/pktgen.c15
-rw-r--r--net/dccp/ccids/ccid3.c321
-rw-r--r--net/dccp/ccids/ccid3.h16
-rw-r--r--net/dccp/ccids/lib/packet_history.h3
-rw-r--r--net/dccp/dccp.h16
-rw-r--r--net/dccp/input.c4
-rw-r--r--net/dccp/ipv4.c5
-rw-r--r--net/dccp/minisocks.c1
-rw-r--r--net/dccp/options.c90
-rw-r--r--net/dccp/output.c3
-rw-r--r--net/decnet/dn_route.c3
-rw-r--r--net/ieee80211/Kconfig1
-rw-r--r--net/ipv4/af_inet.c13
-rw-r--r--net/ipv4/fib_trie.c804
-rw-r--r--net/ipv4/inetpeer.c3
-rw-r--r--net/ipv4/ipconfig.c6
-rw-r--r--net/ipv4/netfilter/ip_conntrack_netbios_ns.c24
-rw-r--r--net/ipv4/netfilter/ipt_REJECT.c5
-rw-r--r--net/ipv4/netfilter/ipt_owner.c1
-rw-r--r--net/ipv4/route.c29
-rw-r--r--net/ipv4/tcp_output.c7
-rw-r--r--net/ipv4/udp.c2
-rw-r--r--net/ipv6/addrconf.c3
-rw-r--r--net/ipv6/datagram.c139
-rw-r--r--net/ipv6/exthdrs.c119
-rw-r--r--net/ipv6/icmp.c20
-rw-r--r--net/ipv6/ip6_fib.c2
-rw-r--r--net/ipv6/ip6_flowlabel.c16
-rw-r--r--net/ipv6/ip6_output.c24
-rw-r--r--net/ipv6/ip6_tunnel.c7
-rw-r--r--net/ipv6/ipv6_sockglue.c186
-rw-r--r--net/ipv6/ndisc.c16
-rw-r--r--net/ipv6/netfilter/ip6t_REJECT.c5
-rw-r--r--net/ipv6/netfilter/ip6t_owner.c1
-rw-r--r--net/ipv6/netfilter/ip6t_rt.c14
-rw-r--r--net/ipv6/raw.c21
-rw-r--r--net/ipv6/reassembly.c9
-rw-r--r--net/ipv6/tcp_ipv6.c36
-rw-r--r--net/ipv6/udp.c25
-rw-r--r--net/irda/ircomm/ircomm_tty.c9
-rw-r--r--net/netrom/af_netrom.c28
-rw-r--r--net/netrom/nr_dev.c54
-rw-r--r--net/netrom/nr_in.c15
-rw-r--r--net/netrom/nr_loopback.c2
-rw-r--r--net/netrom/nr_subr.c7
-rw-r--r--net/netrom/sysctl_net_netrom.c12
-rw-r--r--net/rose/af_rose.c8
-rw-r--r--net/rose/rose_dev.c2
-rw-r--r--net/rose/rose_subr.c4
-rw-r--r--net/sched/sch_api.c2
-rw-r--r--net/socket.c3
-rw-r--r--net/sunrpc/svcsock.c3
-rw-r--r--net/xfrm/xfrm_policy.c8
60 files changed, 1333 insertions, 971 deletions
diff --git a/net/atm/mpc.c b/net/atm/mpc.c
index 17a81ebe7e6e..526d9531411f 100644
--- a/net/atm/mpc.c
+++ b/net/atm/mpc.c
@@ -105,7 +105,7 @@ extern void mpc_proc_clean(void);
105 105
106struct mpoa_client *mpcs = NULL; /* FIXME */ 106struct mpoa_client *mpcs = NULL; /* FIXME */
107static struct atm_mpoa_qos *qos_head = NULL; 107static struct atm_mpoa_qos *qos_head = NULL;
108static struct timer_list mpc_timer = TIMER_INITIALIZER(NULL, 0, 0); 108static DEFINE_TIMER(mpc_timer, NULL, 0, 0);
109 109
110 110
111static struct mpoa_client *find_mpc_by_itfnum(int itf) 111static struct mpoa_client *find_mpc_by_itfnum(int itf)
diff --git a/net/ax25/af_ax25.c b/net/ax25/af_ax25.c
index ed705ddad56b..8e37e71e34ff 100644
--- a/net/ax25/af_ax25.c
+++ b/net/ax25/af_ax25.c
@@ -1695,16 +1695,12 @@ static int ax25_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
1695 /* These two are safe on a single CPU system as only user tasks fiddle here */ 1695 /* These two are safe on a single CPU system as only user tasks fiddle here */
1696 if ((skb = skb_peek(&sk->sk_receive_queue)) != NULL) 1696 if ((skb = skb_peek(&sk->sk_receive_queue)) != NULL)
1697 amount = skb->len; 1697 amount = skb->len;
1698 res = put_user(amount, (int __user *)argp); 1698 res = put_user(amount, (int __user *) argp);
1699 break; 1699 break;
1700 } 1700 }
1701 1701
1702 case SIOCGSTAMP: 1702 case SIOCGSTAMP:
1703 if (sk != NULL) { 1703 res = sock_get_timestamp(sk, argp);
1704 res = sock_get_timestamp(sk, argp);
1705 break;
1706 }
1707 res = -EINVAL;
1708 break; 1704 break;
1709 1705
1710 case SIOCAX25ADDUID: /* Add a uid to the uid/call map table */ 1706 case SIOCAX25ADDUID: /* Add a uid to the uid/call map table */
@@ -1951,24 +1947,24 @@ static struct net_proto_family ax25_family_ops = {
1951}; 1947};
1952 1948
1953static struct proto_ops ax25_proto_ops = { 1949static struct proto_ops ax25_proto_ops = {
1954 .family = PF_AX25, 1950 .family = PF_AX25,
1955 .owner = THIS_MODULE, 1951 .owner = THIS_MODULE,
1956 .release = ax25_release, 1952 .release = ax25_release,
1957 .bind = ax25_bind, 1953 .bind = ax25_bind,
1958 .connect = ax25_connect, 1954 .connect = ax25_connect,
1959 .socketpair = sock_no_socketpair, 1955 .socketpair = sock_no_socketpair,
1960 .accept = ax25_accept, 1956 .accept = ax25_accept,
1961 .getname = ax25_getname, 1957 .getname = ax25_getname,
1962 .poll = datagram_poll, 1958 .poll = datagram_poll,
1963 .ioctl = ax25_ioctl, 1959 .ioctl = ax25_ioctl,
1964 .listen = ax25_listen, 1960 .listen = ax25_listen,
1965 .shutdown = ax25_shutdown, 1961 .shutdown = ax25_shutdown,
1966 .setsockopt = ax25_setsockopt, 1962 .setsockopt = ax25_setsockopt,
1967 .getsockopt = ax25_getsockopt, 1963 .getsockopt = ax25_getsockopt,
1968 .sendmsg = ax25_sendmsg, 1964 .sendmsg = ax25_sendmsg,
1969 .recvmsg = ax25_recvmsg, 1965 .recvmsg = ax25_recvmsg,
1970 .mmap = sock_no_mmap, 1966 .mmap = sock_no_mmap,
1971 .sendpage = sock_no_sendpage, 1967 .sendpage = sock_no_sendpage,
1972}; 1968};
1973 1969
1974/* 1970/*
@@ -1984,7 +1980,7 @@ static struct notifier_block ax25_dev_notifier = {
1984 .notifier_call =ax25_device_event, 1980 .notifier_call =ax25_device_event,
1985}; 1981};
1986 1982
1987EXPORT_SYMBOL(ax25_encapsulate); 1983EXPORT_SYMBOL(ax25_hard_header);
1988EXPORT_SYMBOL(ax25_rebuild_header); 1984EXPORT_SYMBOL(ax25_rebuild_header);
1989EXPORT_SYMBOL(ax25_findbyuid); 1985EXPORT_SYMBOL(ax25_findbyuid);
1990EXPORT_SYMBOL(ax25_find_cb); 1986EXPORT_SYMBOL(ax25_find_cb);
diff --git a/net/ax25/ax25_addr.c b/net/ax25/ax25_addr.c
index dca179daf415..0164a155b8c4 100644
--- a/net/ax25/ax25_addr.c
+++ b/net/ax25/ax25_addr.c
@@ -67,37 +67,34 @@ char *ax2asc(char *buf, ax25_address *a)
67/* 67/*
68 * ascii -> ax25 conversion 68 * ascii -> ax25 conversion
69 */ 69 */
70ax25_address *asc2ax(char *callsign) 70void asc2ax(ax25_address *addr, char *callsign)
71{ 71{
72 static ax25_address addr;
73 char *s; 72 char *s;
74 int n; 73 int n;
75 74
76 for (s = callsign, n = 0; n < 6; n++) { 75 for (s = callsign, n = 0; n < 6; n++) {
77 if (*s != '\0' && *s != '-') 76 if (*s != '\0' && *s != '-')
78 addr.ax25_call[n] = *s++; 77 addr->ax25_call[n] = *s++;
79 else 78 else
80 addr.ax25_call[n] = ' '; 79 addr->ax25_call[n] = ' ';
81 addr.ax25_call[n] <<= 1; 80 addr->ax25_call[n] <<= 1;
82 addr.ax25_call[n] &= 0xFE; 81 addr->ax25_call[n] &= 0xFE;
83 } 82 }
84 83
85 if (*s++ == '\0') { 84 if (*s++ == '\0') {
86 addr.ax25_call[6] = 0x00; 85 addr->ax25_call[6] = 0x00;
87 return &addr; 86 return;
88 } 87 }
89 88
90 addr.ax25_call[6] = *s++ - '0'; 89 addr->ax25_call[6] = *s++ - '0';
91 90
92 if (*s != '\0') { 91 if (*s != '\0') {
93 addr.ax25_call[6] *= 10; 92 addr->ax25_call[6] *= 10;
94 addr.ax25_call[6] += *s++ - '0'; 93 addr->ax25_call[6] += *s++ - '0';
95 } 94 }
96 95
97 addr.ax25_call[6] <<= 1; 96 addr->ax25_call[6] <<= 1;
98 addr.ax25_call[6] &= 0x1E; 97 addr->ax25_call[6] &= 0x1E;
99
100 return &addr;
101} 98}
102 99
103/* 100/*
diff --git a/net/ax25/ax25_ip.c b/net/ax25/ax25_ip.c
index bba0173e2d65..d643dac3eccc 100644
--- a/net/ax25/ax25_ip.c
+++ b/net/ax25/ax25_ip.c
@@ -47,7 +47,7 @@
47 47
48#ifdef CONFIG_INET 48#ifdef CONFIG_INET
49 49
50int ax25_encapsulate(struct sk_buff *skb, struct net_device *dev, unsigned short type, void *daddr, void *saddr, unsigned len) 50int ax25_hard_header(struct sk_buff *skb, struct net_device *dev, unsigned short type, void *daddr, void *saddr, unsigned len)
51{ 51{
52 unsigned char *buff; 52 unsigned char *buff;
53 53
@@ -88,7 +88,7 @@ int ax25_encapsulate(struct sk_buff *skb, struct net_device *dev, unsigned short
88 *buff++ = AX25_P_ARP; 88 *buff++ = AX25_P_ARP;
89 break; 89 break;
90 default: 90 default:
91 printk(KERN_ERR "AX.25: ax25_encapsulate - wrong protocol type 0x%2.2x\n", type); 91 printk(KERN_ERR "AX.25: ax25_hard_header - wrong protocol type 0x%2.2x\n", type);
92 *buff++ = 0; 92 *buff++ = 0;
93 break; 93 break;
94 } 94 }
@@ -209,7 +209,7 @@ put:
209 209
210#else /* INET */ 210#else /* INET */
211 211
212int ax25_encapsulate(struct sk_buff *skb, struct net_device *dev, unsigned short type, void *daddr, void *saddr, unsigned len) 212int ax25_hard_header(struct sk_buff *skb, struct net_device *dev, unsigned short type, void *daddr, void *saddr, unsigned len)
213{ 213{
214 return -AX25_HEADER_LEN; 214 return -AX25_HEADER_LEN;
215} 215}
diff --git a/net/compat.c b/net/compat.c
index d99ab9695893..e593dace2fdb 100644
--- a/net/compat.c
+++ b/net/compat.c
@@ -135,13 +135,14 @@ static inline struct compat_cmsghdr __user *cmsg_compat_nxthdr(struct msghdr *ms
135 * thus placement) of cmsg headers and length are different for 135 * thus placement) of cmsg headers and length are different for
136 * 32-bit apps. -DaveM 136 * 32-bit apps. -DaveM
137 */ 137 */
138int cmsghdr_from_user_compat_to_kern(struct msghdr *kmsg, 138int cmsghdr_from_user_compat_to_kern(struct msghdr *kmsg, struct sock *sk,
139 unsigned char *stackbuf, int stackbuf_size) 139 unsigned char *stackbuf, int stackbuf_size)
140{ 140{
141 struct compat_cmsghdr __user *ucmsg; 141 struct compat_cmsghdr __user *ucmsg;
142 struct cmsghdr *kcmsg, *kcmsg_base; 142 struct cmsghdr *kcmsg, *kcmsg_base;
143 compat_size_t ucmlen; 143 compat_size_t ucmlen;
144 __kernel_size_t kcmlen, tmp; 144 __kernel_size_t kcmlen, tmp;
145 int err = -EFAULT;
145 146
146 kcmlen = 0; 147 kcmlen = 0;
147 kcmsg_base = kcmsg = (struct cmsghdr *)stackbuf; 148 kcmsg_base = kcmsg = (struct cmsghdr *)stackbuf;
@@ -156,6 +157,7 @@ int cmsghdr_from_user_compat_to_kern(struct msghdr *kmsg,
156 157
157 tmp = ((ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg))) + 158 tmp = ((ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg))) +
158 CMSG_ALIGN(sizeof(struct cmsghdr))); 159 CMSG_ALIGN(sizeof(struct cmsghdr)));
160 tmp = CMSG_ALIGN(tmp);
159 kcmlen += tmp; 161 kcmlen += tmp;
160 ucmsg = cmsg_compat_nxthdr(kmsg, ucmsg, ucmlen); 162 ucmsg = cmsg_compat_nxthdr(kmsg, ucmsg, ucmlen);
161 } 163 }
@@ -167,30 +169,34 @@ int cmsghdr_from_user_compat_to_kern(struct msghdr *kmsg,
167 * until we have successfully copied over all of the data 169 * until we have successfully copied over all of the data
168 * from the user. 170 * from the user.
169 */ 171 */
170 if(kcmlen > stackbuf_size) 172 if (kcmlen > stackbuf_size)
171 kcmsg_base = kcmsg = kmalloc(kcmlen, GFP_KERNEL); 173 kcmsg_base = kcmsg = sock_kmalloc(sk, kcmlen, GFP_KERNEL);
172 if(kcmsg == NULL) 174 if (kcmsg == NULL)
173 return -ENOBUFS; 175 return -ENOBUFS;
174 176
175 /* Now copy them over neatly. */ 177 /* Now copy them over neatly. */
176 memset(kcmsg, 0, kcmlen); 178 memset(kcmsg, 0, kcmlen);
177 ucmsg = CMSG_COMPAT_FIRSTHDR(kmsg); 179 ucmsg = CMSG_COMPAT_FIRSTHDR(kmsg);
178 while(ucmsg != NULL) { 180 while(ucmsg != NULL) {
179 __get_user(ucmlen, &ucmsg->cmsg_len); 181 if (__get_user(ucmlen, &ucmsg->cmsg_len))
182 goto Efault;
183 if (!CMSG_COMPAT_OK(ucmlen, ucmsg, kmsg))
184 goto Einval;
180 tmp = ((ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg))) + 185 tmp = ((ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg))) +
181 CMSG_ALIGN(sizeof(struct cmsghdr))); 186 CMSG_ALIGN(sizeof(struct cmsghdr)));
187 if ((char *)kcmsg_base + kcmlen - (char *)kcmsg < CMSG_ALIGN(tmp))
188 goto Einval;
182 kcmsg->cmsg_len = tmp; 189 kcmsg->cmsg_len = tmp;
183 __get_user(kcmsg->cmsg_level, &ucmsg->cmsg_level); 190 tmp = CMSG_ALIGN(tmp);
184 __get_user(kcmsg->cmsg_type, &ucmsg->cmsg_type); 191 if (__get_user(kcmsg->cmsg_level, &ucmsg->cmsg_level) ||
185 192 __get_user(kcmsg->cmsg_type, &ucmsg->cmsg_type) ||
186 /* Copy over the data. */ 193 copy_from_user(CMSG_DATA(kcmsg),
187 if(copy_from_user(CMSG_DATA(kcmsg), 194 CMSG_COMPAT_DATA(ucmsg),
188 CMSG_COMPAT_DATA(ucmsg), 195 (ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg)))))
189 (ucmlen - CMSG_COMPAT_ALIGN(sizeof(*ucmsg))))) 196 goto Efault;
190 goto out_free_efault;
191 197
192 /* Advance. */ 198 /* Advance. */
193 kcmsg = (struct cmsghdr *)((char *)kcmsg + CMSG_ALIGN(tmp)); 199 kcmsg = (struct cmsghdr *)((char *)kcmsg + tmp);
194 ucmsg = cmsg_compat_nxthdr(kmsg, ucmsg, ucmlen); 200 ucmsg = cmsg_compat_nxthdr(kmsg, ucmsg, ucmlen);
195 } 201 }
196 202
@@ -199,10 +205,12 @@ int cmsghdr_from_user_compat_to_kern(struct msghdr *kmsg,
199 kmsg->msg_controllen = kcmlen; 205 kmsg->msg_controllen = kcmlen;
200 return 0; 206 return 0;
201 207
202out_free_efault: 208Einval:
203 if(kcmsg_base != (struct cmsghdr *)stackbuf) 209 err = -EINVAL;
204 kfree(kcmsg_base); 210Efault:
205 return -EFAULT; 211 if (kcmsg_base != (struct cmsghdr *)stackbuf)
212 sock_kfree_s(sk, kcmsg_base, kcmlen);
213 return err;
206} 214}
207 215
208int put_cmsg_compat(struct msghdr *kmsg, int level, int type, int len, void *data) 216int put_cmsg_compat(struct msghdr *kmsg, int level, int type, int len, void *data)
diff --git a/net/core/dst.c b/net/core/dst.c
index 334790da9f16..470c05bc4cb2 100644
--- a/net/core/dst.c
+++ b/net/core/dst.c
@@ -39,8 +39,7 @@ static unsigned long dst_gc_timer_inc = DST_GC_MAX;
39static void dst_run_gc(unsigned long); 39static void dst_run_gc(unsigned long);
40static void ___dst_free(struct dst_entry * dst); 40static void ___dst_free(struct dst_entry * dst);
41 41
42static struct timer_list dst_gc_timer = 42static DEFINE_TIMER(dst_gc_timer, dst_run_gc, DST_GC_MIN, 0);
43 TIMER_INITIALIZER(dst_run_gc, DST_GC_MIN, 0);
44 43
45static void dst_run_gc(unsigned long dummy) 44static void dst_run_gc(unsigned long dummy)
46{ 45{
diff --git a/net/core/netpoll.c b/net/core/netpoll.c
index a1a9a7abff50..5265dfd69928 100644
--- a/net/core/netpoll.c
+++ b/net/core/netpoll.c
@@ -645,10 +645,10 @@ int netpoll_setup(struct netpoll *np)
645 645
646 npinfo->rx_flags = 0; 646 npinfo->rx_flags = 0;
647 npinfo->rx_np = NULL; 647 npinfo->rx_np = NULL;
648 npinfo->poll_lock = SPIN_LOCK_UNLOCKED; 648 spin_lock_init(&npinfo->poll_lock);
649 npinfo->poll_owner = -1; 649 npinfo->poll_owner = -1;
650 npinfo->tries = MAX_RETRIES; 650 npinfo->tries = MAX_RETRIES;
651 npinfo->rx_lock = SPIN_LOCK_UNLOCKED; 651 spin_lock_init(&npinfo->rx_lock);
652 } else 652 } else
653 npinfo = ndev->npinfo; 653 npinfo = ndev->npinfo;
654 654
diff --git a/net/core/pktgen.c b/net/core/pktgen.c
index 8eb083b6041a..ef430b1e8e42 100644
--- a/net/core/pktgen.c
+++ b/net/core/pktgen.c
@@ -503,7 +503,7 @@ static int pg_delay_d = 0;
503static int pg_clone_skb_d = 0; 503static int pg_clone_skb_d = 0;
504static int debug = 0; 504static int debug = 0;
505 505
506static spinlock_t _thread_lock = SPIN_LOCK_UNLOCKED; 506static DEFINE_SPINLOCK(_thread_lock);
507static struct pktgen_thread *pktgen_threads = NULL; 507static struct pktgen_thread *pktgen_threads = NULL;
508 508
509static char module_fname[128]; 509static char module_fname[128];
@@ -1452,8 +1452,7 @@ static int proc_thread_write(struct file *file, const char __user *user_buffer,
1452 thread_lock(); 1452 thread_lock();
1453 t->control |= T_REMDEV; 1453 t->control |= T_REMDEV;
1454 thread_unlock(); 1454 thread_unlock();
1455 current->state = TASK_INTERRUPTIBLE; 1455 schedule_timeout_interruptible(msecs_to_jiffies(125)); /* Propagate thread->control */
1456 schedule_timeout(HZ/8); /* Propagate thread->control */
1457 ret = count; 1456 ret = count;
1458 sprintf(pg_result, "OK: rem_device_all"); 1457 sprintf(pg_result, "OK: rem_device_all");
1459 goto out; 1458 goto out;
@@ -1716,10 +1715,9 @@ static void spin(struct pktgen_dev *pkt_dev, __u64 spin_until_us)
1716 printk(KERN_INFO "sleeping for %d\n", (int)(spin_until_us - now)); 1715 printk(KERN_INFO "sleeping for %d\n", (int)(spin_until_us - now));
1717 while (now < spin_until_us) { 1716 while (now < spin_until_us) {
1718 /* TODO: optimise sleeping behavior */ 1717 /* TODO: optimise sleeping behavior */
1719 if (spin_until_us - now > (1000000/HZ)+1) { 1718 if (spin_until_us - now > jiffies_to_usecs(1)+1)
1720 current->state = TASK_INTERRUPTIBLE; 1719 schedule_timeout_interruptible(1);
1721 schedule_timeout(1); 1720 else if (spin_until_us - now > 100) {
1722 } else if (spin_until_us - now > 100) {
1723 do_softirq(); 1721 do_softirq();
1724 if (!pkt_dev->running) 1722 if (!pkt_dev->running)
1725 return; 1723 return;
@@ -2449,8 +2447,7 @@ static void pktgen_run_all_threads(void)
2449 } 2447 }
2450 thread_unlock(); 2448 thread_unlock();
2451 2449
2452 current->state = TASK_INTERRUPTIBLE; 2450 schedule_timeout_interruptible(msecs_to_jiffies(125)); /* Propagate thread->control */
2453 schedule_timeout(HZ/8); /* Propagate thread->control */
2454 2451
2455 pktgen_wait_all_threads_run(); 2452 pktgen_wait_all_threads_run();
2456} 2453}
diff --git a/net/dccp/ccids/ccid3.c b/net/dccp/ccids/ccid3.c
index 7bf3b3a91e97..38aa84986118 100644
--- a/net/dccp/ccids/ccid3.c
+++ b/net/dccp/ccids/ccid3.c
@@ -43,12 +43,22 @@
43#include "ccid3.h" 43#include "ccid3.h"
44 44
45/* 45/*
46 * Reason for maths with 10 here is to avoid 32 bit overflow when a is big. 46 * Reason for maths here is to avoid 32 bit overflow when a is big.
47 * With this we get close to the limit.
47 */ 48 */
48static inline u32 usecs_div(const u32 a, const u32 b) 49static inline u32 usecs_div(const u32 a, const u32 b)
49{ 50{
50 const u32 tmp = a * (USEC_PER_SEC / 10); 51 const u32 div = a < (UINT_MAX / (USEC_PER_SEC / 10)) ? 10 :
51 return b > 20 ? tmp / (b / 10) : tmp; 52 a < (UINT_MAX / (USEC_PER_SEC / 50)) ? 50 :
53 a < (UINT_MAX / (USEC_PER_SEC / 100)) ? 100 :
54 a < (UINT_MAX / (USEC_PER_SEC / 500)) ? 500 :
55 a < (UINT_MAX / (USEC_PER_SEC / 1000)) ? 1000 :
56 a < (UINT_MAX / (USEC_PER_SEC / 5000)) ? 5000 :
57 a < (UINT_MAX / (USEC_PER_SEC / 10000)) ? 10000 :
58 a < (UINT_MAX / (USEC_PER_SEC / 50000)) ? 50000 :
59 100000;
60 const u32 tmp = a * (USEC_PER_SEC / div);
61 return (b >= 2 * div) ? tmp / (b / div) : tmp;
52} 62}
53 63
54static int ccid3_debug; 64static int ccid3_debug;
@@ -68,13 +78,11 @@ static struct dccp_li_hist *ccid3_li_hist;
68 78
69static int ccid3_init(struct sock *sk) 79static int ccid3_init(struct sock *sk)
70{ 80{
71 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk);
72 return 0; 81 return 0;
73} 82}
74 83
75static void ccid3_exit(struct sock *sk) 84static void ccid3_exit(struct sock *sk)
76{ 85{
77 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk);
78} 86}
79 87
80/* TFRC sender states */ 88/* TFRC sender states */
@@ -102,8 +110,7 @@ static const char *ccid3_tx_state_name(enum ccid3_hc_tx_states state)
102static inline void ccid3_hc_tx_set_state(struct sock *sk, 110static inline void ccid3_hc_tx_set_state(struct sock *sk,
103 enum ccid3_hc_tx_states state) 111 enum ccid3_hc_tx_states state)
104{ 112{
105 struct dccp_sock *dp = dccp_sk(sk); 113 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
106 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private;
107 enum ccid3_hc_tx_states oldstate = hctx->ccid3hctx_state; 114 enum ccid3_hc_tx_states oldstate = hctx->ccid3hctx_state;
108 115
109 ccid3_pr_debug("%s(%p) %-8.8s -> %s\n", 116 ccid3_pr_debug("%s(%p) %-8.8s -> %s\n",
@@ -144,8 +151,7 @@ static inline void ccid3_calc_new_delta(struct ccid3_hc_tx_sock *hctx)
144 */ 151 */
145static void ccid3_hc_tx_update_x(struct sock *sk) 152static void ccid3_hc_tx_update_x(struct sock *sk)
146{ 153{
147 struct dccp_sock *dp = dccp_sk(sk); 154 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
148 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private;
149 155
150 /* To avoid large error in calcX */ 156 /* To avoid large error in calcX */
151 if (hctx->ccid3hctx_p >= TFRC_SMALLEST_P) { 157 if (hctx->ccid3hctx_p >= TFRC_SMALLEST_P) {
@@ -159,7 +165,7 @@ static void ccid3_hc_tx_update_x(struct sock *sk)
159 } else { 165 } else {
160 struct timeval now; 166 struct timeval now;
161 167
162 do_gettimeofday(&now); 168 dccp_timestamp(sk, &now);
163 if (timeval_delta(&now, &hctx->ccid3hctx_t_ld) >= 169 if (timeval_delta(&now, &hctx->ccid3hctx_t_ld) >=
164 hctx->ccid3hctx_rtt) { 170 hctx->ccid3hctx_rtt) {
165 hctx->ccid3hctx_x = max_t(u32, min_t(u32, hctx->ccid3hctx_x_recv, 171 hctx->ccid3hctx_x = max_t(u32, min_t(u32, hctx->ccid3hctx_x_recv,
@@ -174,9 +180,8 @@ static void ccid3_hc_tx_update_x(struct sock *sk)
174static void ccid3_hc_tx_no_feedback_timer(unsigned long data) 180static void ccid3_hc_tx_no_feedback_timer(unsigned long data)
175{ 181{
176 struct sock *sk = (struct sock *)data; 182 struct sock *sk = (struct sock *)data;
177 struct dccp_sock *dp = dccp_sk(sk);
178 unsigned long next_tmout = 0; 183 unsigned long next_tmout = 0;
179 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 184 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
180 185
181 bh_lock_sock(sk); 186 bh_lock_sock(sk);
182 if (sock_owned_by_user(sk)) { 187 if (sock_owned_by_user(sk)) {
@@ -274,20 +279,20 @@ static int ccid3_hc_tx_send_packet(struct sock *sk,
274 struct sk_buff *skb, int len) 279 struct sk_buff *skb, int len)
275{ 280{
276 struct dccp_sock *dp = dccp_sk(sk); 281 struct dccp_sock *dp = dccp_sk(sk);
277 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 282 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
278 struct dccp_tx_hist_entry *new_packet; 283 struct dccp_tx_hist_entry *new_packet;
279 struct timeval now; 284 struct timeval now;
280 long delay; 285 long delay;
281 int rc = -ENOTCONN; 286 int rc = -ENOTCONN;
282 287
283 /* Check if pure ACK or Terminating*/ 288 BUG_ON(hctx == NULL || hctx->ccid3hctx_state == TFRC_SSTATE_TERM);
284 289
290 /* Check if pure ACK or Terminating*/
285 /* 291 /*
286 * XXX: We only call this function for DATA and DATAACK, on, these 292 * XXX: We only call this function for DATA and DATAACK, on, these
287 * packets can have zero length, but why the comment about "pure ACK"? 293 * packets can have zero length, but why the comment about "pure ACK"?
288 */ 294 */
289 if (hctx == NULL || len == 0 || 295 if (unlikely(len == 0))
290 hctx->ccid3hctx_state == TFRC_SSTATE_TERM)
291 goto out; 296 goto out;
292 297
293 /* See if last packet allocated was not sent */ 298 /* See if last packet allocated was not sent */
@@ -297,23 +302,20 @@ static int ccid3_hc_tx_send_packet(struct sock *sk,
297 SLAB_ATOMIC); 302 SLAB_ATOMIC);
298 303
299 rc = -ENOBUFS; 304 rc = -ENOBUFS;
300 if (new_packet == NULL) { 305 if (unlikely(new_packet == NULL)) {
301 ccid3_pr_debug("%s, sk=%p, not enough mem to add " 306 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, not enough "
302 "to history, send refused\n", 307 "mem to add to history, send refused\n",
303 dccp_role(sk), sk); 308 __FUNCTION__, dccp_role(sk), sk);
304 goto out; 309 goto out;
305 } 310 }
306 311
307 dccp_tx_hist_add_entry(&hctx->ccid3hctx_hist, new_packet); 312 dccp_tx_hist_add_entry(&hctx->ccid3hctx_hist, new_packet);
308 } 313 }
309 314
310 do_gettimeofday(&now); 315 dccp_timestamp(sk, &now);
311 316
312 switch (hctx->ccid3hctx_state) { 317 switch (hctx->ccid3hctx_state) {
313 case TFRC_SSTATE_NO_SENT: 318 case TFRC_SSTATE_NO_SENT:
314 ccid3_pr_debug("%s, sk=%p, first packet(%llu)\n",
315 dccp_role(sk), sk, dp->dccps_gss);
316
317 hctx->ccid3hctx_no_feedback_timer.function = ccid3_hc_tx_no_feedback_timer; 319 hctx->ccid3hctx_no_feedback_timer.function = ccid3_hc_tx_no_feedback_timer;
318 hctx->ccid3hctx_no_feedback_timer.data = (unsigned long)sk; 320 hctx->ccid3hctx_no_feedback_timer.data = (unsigned long)sk;
319 sk_reset_timer(sk, &hctx->ccid3hctx_no_feedback_timer, 321 sk_reset_timer(sk, &hctx->ccid3hctx_no_feedback_timer,
@@ -321,7 +323,7 @@ static int ccid3_hc_tx_send_packet(struct sock *sk,
321 hctx->ccid3hctx_last_win_count = 0; 323 hctx->ccid3hctx_last_win_count = 0;
322 hctx->ccid3hctx_t_last_win_count = now; 324 hctx->ccid3hctx_t_last_win_count = now;
323 ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK); 325 ccid3_hc_tx_set_state(sk, TFRC_SSTATE_NO_FBACK);
324 hctx->ccid3hctx_t_ipi = TFRC_INITIAL_TIMEOUT; 326 hctx->ccid3hctx_t_ipi = TFRC_INITIAL_IPI;
325 327
326 /* Set nominal send time for initial packet */ 328 /* Set nominal send time for initial packet */
327 hctx->ccid3hctx_t_nom = now; 329 hctx->ccid3hctx_t_nom = now;
@@ -334,7 +336,6 @@ static int ccid3_hc_tx_send_packet(struct sock *sk,
334 case TFRC_SSTATE_FBACK: 336 case TFRC_SSTATE_FBACK:
335 delay = (timeval_delta(&now, &hctx->ccid3hctx_t_nom) - 337 delay = (timeval_delta(&now, &hctx->ccid3hctx_t_nom) -
336 hctx->ccid3hctx_delta); 338 hctx->ccid3hctx_delta);
337 ccid3_pr_debug("send_packet delay=%ld\n", delay);
338 delay /= -1000; 339 delay /= -1000;
339 /* divide by -1000 is to convert to ms and get sign right */ 340 /* divide by -1000 is to convert to ms and get sign right */
340 rc = delay > 0 ? delay : 0; 341 rc = delay > 0 ? delay : 0;
@@ -348,29 +349,25 @@ static int ccid3_hc_tx_send_packet(struct sock *sk,
348 } 349 }
349 350
350 /* Can we send? if so add options and add to packet history */ 351 /* Can we send? if so add options and add to packet history */
351 if (rc == 0) 352 if (rc == 0) {
353 dp->dccps_hc_tx_insert_options = 1;
352 new_packet->dccphtx_ccval = 354 new_packet->dccphtx_ccval =
353 DCCP_SKB_CB(skb)->dccpd_ccval = 355 DCCP_SKB_CB(skb)->dccpd_ccval =
354 hctx->ccid3hctx_last_win_count; 356 hctx->ccid3hctx_last_win_count;
357 }
355out: 358out:
356 return rc; 359 return rc;
357} 360}
358 361
359static void ccid3_hc_tx_packet_sent(struct sock *sk, int more, int len) 362static void ccid3_hc_tx_packet_sent(struct sock *sk, int more, int len)
360{ 363{
361 struct dccp_sock *dp = dccp_sk(sk); 364 const struct dccp_sock *dp = dccp_sk(sk);
362 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 365 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
363 struct timeval now; 366 struct timeval now;
364 367
365 BUG_ON(hctx == NULL); 368 BUG_ON(hctx == NULL || hctx->ccid3hctx_state == TFRC_SSTATE_TERM);
366
367 if (hctx->ccid3hctx_state == TFRC_SSTATE_TERM) {
368 ccid3_pr_debug("%s, sk=%p, while state is TFRC_SSTATE_TERM!\n",
369 dccp_role(sk), sk);
370 return;
371 }
372 369
373 do_gettimeofday(&now); 370 dccp_timestamp(sk, &now);
374 371
375 /* check if we have sent a data packet */ 372 /* check if we have sent a data packet */
376 if (len > 0) { 373 if (len > 0) {
@@ -378,14 +375,14 @@ static void ccid3_hc_tx_packet_sent(struct sock *sk, int more, int len)
378 struct dccp_tx_hist_entry *packet; 375 struct dccp_tx_hist_entry *packet;
379 376
380 packet = dccp_tx_hist_head(&hctx->ccid3hctx_hist); 377 packet = dccp_tx_hist_head(&hctx->ccid3hctx_hist);
381 if (packet == NULL) { 378 if (unlikely(packet == NULL)) {
382 printk(KERN_CRIT "%s: packet doesn't exists in " 379 LIMIT_NETDEBUG(KERN_WARNING "%s: packet doesn't "
383 "history!\n", __FUNCTION__); 380 "exists in history!\n", __FUNCTION__);
384 return; 381 return;
385 } 382 }
386 if (packet->dccphtx_sent) { 383 if (unlikely(packet->dccphtx_sent)) {
387 printk(KERN_CRIT "%s: no unsent packet in history!\n", 384 LIMIT_NETDEBUG(KERN_WARNING "%s: no unsent packet in "
388 __FUNCTION__); 385 "history!\n", __FUNCTION__);
389 return; 386 return;
390 } 387 }
391 packet->dccphtx_tstamp = now; 388 packet->dccphtx_tstamp = now;
@@ -445,24 +442,18 @@ static void ccid3_hc_tx_packet_sent(struct sock *sk, int more, int len)
445 442
446static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb) 443static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
447{ 444{
448 struct dccp_sock *dp = dccp_sk(sk); 445 const struct dccp_sock *dp = dccp_sk(sk);
449 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 446 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
450 struct ccid3_options_received *opt_recv; 447 struct ccid3_options_received *opt_recv;
451 struct dccp_tx_hist_entry *packet; 448 struct dccp_tx_hist_entry *packet;
449 struct timeval now;
452 unsigned long next_tmout; 450 unsigned long next_tmout;
453 u32 t_elapsed; 451 u32 t_elapsed;
454 u32 pinv; 452 u32 pinv;
455 u32 x_recv; 453 u32 x_recv;
456 u32 r_sample; 454 u32 r_sample;
457 455
458 if (hctx == NULL) 456 BUG_ON(hctx == NULL || hctx->ccid3hctx_state == TFRC_SSTATE_TERM);
459 return;
460
461 if (hctx->ccid3hctx_state == TFRC_SSTATE_TERM) {
462 ccid3_pr_debug("%s, sk=%p, received a packet when "
463 "terminating!\n", dccp_role(sk), sk);
464 return;
465 }
466 457
467 /* we are only interested in ACKs */ 458 /* we are only interested in ACKs */
468 if (!(DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_ACK || 459 if (!(DCCP_SKB_CB(skb)->dccpd_type == DCCP_PKT_ACK ||
@@ -471,7 +462,7 @@ static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
471 462
472 opt_recv = &hctx->ccid3hctx_options_received; 463 opt_recv = &hctx->ccid3hctx_options_received;
473 464
474 t_elapsed = dp->dccps_options_received.dccpor_elapsed_time; 465 t_elapsed = dp->dccps_options_received.dccpor_elapsed_time * 10;
475 x_recv = opt_recv->ccid3or_receive_rate; 466 x_recv = opt_recv->ccid3or_receive_rate;
476 pinv = opt_recv->ccid3or_loss_event_rate; 467 pinv = opt_recv->ccid3or_loss_event_rate;
477 468
@@ -486,19 +477,24 @@ static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
486 /* get t_recvdata from history */ 477 /* get t_recvdata from history */
487 packet = dccp_tx_hist_find_entry(&hctx->ccid3hctx_hist, 478 packet = dccp_tx_hist_find_entry(&hctx->ccid3hctx_hist,
488 DCCP_SKB_CB(skb)->dccpd_ack_seq); 479 DCCP_SKB_CB(skb)->dccpd_ack_seq);
489 if (packet == NULL) { 480 if (unlikely(packet == NULL)) {
490 ccid3_pr_debug("%s, sk=%p, seqno %llu(%s) does't " 481 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, seqno "
491 "exist in history!\n", 482 "%llu(%s) does't exist in history!\n",
492 dccp_role(sk), sk, 483 __FUNCTION__, dccp_role(sk), sk,
493 DCCP_SKB_CB(skb)->dccpd_ack_seq, 484 (unsigned long long)DCCP_SKB_CB(skb)->dccpd_ack_seq,
494 dccp_packet_name(DCCP_SKB_CB(skb)->dccpd_type)); 485 dccp_packet_name(DCCP_SKB_CB(skb)->dccpd_type));
495 return; 486 return;
496 } 487 }
497 488
498 /* Update RTT */ 489 /* Update RTT */
499 r_sample = timeval_now_delta(&packet->dccphtx_tstamp); 490 dccp_timestamp(sk, &now);
500 /* FIXME: */ 491 r_sample = timeval_delta(&now, &packet->dccphtx_tstamp);
501 // r_sample -= usecs_to_jiffies(t_elapsed * 10); 492 if (unlikely(r_sample <= t_elapsed))
493 LIMIT_NETDEBUG(KERN_WARNING "%s: r_sample=%uus, "
494 "t_elapsed=%uus\n",
495 __FUNCTION__, r_sample, t_elapsed);
496 else
497 r_sample -= t_elapsed;
502 498
503 /* Update RTT estimate by 499 /* Update RTT estimate by
504 * If (No feedback recv) 500 * If (No feedback recv)
@@ -591,11 +587,11 @@ static void ccid3_hc_tx_packet_recv(struct sock *sk, struct sk_buff *skb)
591 587
592static void ccid3_hc_tx_insert_options(struct sock *sk, struct sk_buff *skb) 588static void ccid3_hc_tx_insert_options(struct sock *sk, struct sk_buff *skb)
593{ 589{
594 const struct dccp_sock *dp = dccp_sk(sk); 590 const struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
595 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 591
592 BUG_ON(hctx == NULL);
596 593
597 if (hctx == NULL || !(sk->sk_state == DCCP_OPEN || 594 if (!(sk->sk_state == DCCP_OPEN || sk->sk_state == DCCP_PARTOPEN))
598 sk->sk_state == DCCP_PARTOPEN))
599 return; 595 return;
600 596
601 DCCP_SKB_CB(skb)->dccpd_ccval = hctx->ccid3hctx_last_win_count; 597 DCCP_SKB_CB(skb)->dccpd_ccval = hctx->ccid3hctx_last_win_count;
@@ -606,12 +602,11 @@ static int ccid3_hc_tx_parse_options(struct sock *sk, unsigned char option,
606 unsigned char *value) 602 unsigned char *value)
607{ 603{
608 int rc = 0; 604 int rc = 0;
609 struct dccp_sock *dp = dccp_sk(sk); 605 const struct dccp_sock *dp = dccp_sk(sk);
610 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 606 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
611 struct ccid3_options_received *opt_recv; 607 struct ccid3_options_received *opt_recv;
612 608
613 if (hctx == NULL) 609 BUG_ON(hctx == NULL);
614 return 0;
615 610
616 opt_recv = &hctx->ccid3hctx_options_received; 611 opt_recv = &hctx->ccid3hctx_options_received;
617 612
@@ -625,10 +620,10 @@ static int ccid3_hc_tx_parse_options(struct sock *sk, unsigned char option,
625 620
626 switch (option) { 621 switch (option) {
627 case TFRC_OPT_LOSS_EVENT_RATE: 622 case TFRC_OPT_LOSS_EVENT_RATE:
628 if (len != 4) { 623 if (unlikely(len != 4)) {
629 ccid3_pr_debug("%s, sk=%p, invalid len for " 624 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, invalid "
630 "TFRC_OPT_LOSS_EVENT_RATE\n", 625 "len for TFRC_OPT_LOSS_EVENT_RATE\n",
631 dccp_role(sk), sk); 626 __FUNCTION__, dccp_role(sk), sk);
632 rc = -EINVAL; 627 rc = -EINVAL;
633 } else { 628 } else {
634 opt_recv->ccid3or_loss_event_rate = ntohl(*(u32 *)value); 629 opt_recv->ccid3or_loss_event_rate = ntohl(*(u32 *)value);
@@ -646,10 +641,10 @@ static int ccid3_hc_tx_parse_options(struct sock *sk, unsigned char option,
646 opt_recv->ccid3or_loss_intervals_len); 641 opt_recv->ccid3or_loss_intervals_len);
647 break; 642 break;
648 case TFRC_OPT_RECEIVE_RATE: 643 case TFRC_OPT_RECEIVE_RATE:
649 if (len != 4) { 644 if (unlikely(len != 4)) {
650 ccid3_pr_debug("%s, sk=%p, invalid len for " 645 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, invalid "
651 "TFRC_OPT_RECEIVE_RATE\n", 646 "len for TFRC_OPT_RECEIVE_RATE\n",
652 dccp_role(sk), sk); 647 __FUNCTION__, dccp_role(sk), sk);
653 rc = -EINVAL; 648 rc = -EINVAL;
654 } else { 649 } else {
655 opt_recv->ccid3or_receive_rate = ntohl(*(u32 *)value); 650 opt_recv->ccid3or_receive_rate = ntohl(*(u32 *)value);
@@ -668,13 +663,11 @@ static int ccid3_hc_tx_init(struct sock *sk)
668 struct dccp_sock *dp = dccp_sk(sk); 663 struct dccp_sock *dp = dccp_sk(sk);
669 struct ccid3_hc_tx_sock *hctx; 664 struct ccid3_hc_tx_sock *hctx;
670 665
671 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk); 666 dp->dccps_hc_tx_ccid_private = kmalloc(sizeof(*hctx), gfp_any());
672 667 if (dp->dccps_hc_tx_ccid_private == NULL)
673 hctx = dp->dccps_hc_tx_ccid_private = kmalloc(sizeof(*hctx),
674 gfp_any());
675 if (hctx == NULL)
676 return -ENOMEM; 668 return -ENOMEM;
677 669
670 hctx = ccid3_hc_tx_sk(sk);
678 memset(hctx, 0, sizeof(*hctx)); 671 memset(hctx, 0, sizeof(*hctx));
679 672
680 if (dp->dccps_packet_size >= TFRC_MIN_PACKET_SIZE && 673 if (dp->dccps_packet_size >= TFRC_MIN_PACKET_SIZE &&
@@ -696,9 +689,8 @@ static int ccid3_hc_tx_init(struct sock *sk)
696static void ccid3_hc_tx_exit(struct sock *sk) 689static void ccid3_hc_tx_exit(struct sock *sk)
697{ 690{
698 struct dccp_sock *dp = dccp_sk(sk); 691 struct dccp_sock *dp = dccp_sk(sk);
699 struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private; 692 struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
700 693
701 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk);
702 BUG_ON(hctx == NULL); 694 BUG_ON(hctx == NULL);
703 695
704 ccid3_hc_tx_set_state(sk, TFRC_SSTATE_TERM); 696 ccid3_hc_tx_set_state(sk, TFRC_SSTATE_TERM);
@@ -738,8 +730,7 @@ static const char *ccid3_rx_state_name(enum ccid3_hc_rx_states state)
738static inline void ccid3_hc_rx_set_state(struct sock *sk, 730static inline void ccid3_hc_rx_set_state(struct sock *sk,
739 enum ccid3_hc_rx_states state) 731 enum ccid3_hc_rx_states state)
740{ 732{
741 struct dccp_sock *dp = dccp_sk(sk); 733 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
742 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
743 enum ccid3_hc_rx_states oldstate = hcrx->ccid3hcrx_state; 734 enum ccid3_hc_rx_states oldstate = hcrx->ccid3hcrx_state;
744 735
745 ccid3_pr_debug("%s(%p) %-8.8s -> %s\n", 736 ccid3_pr_debug("%s(%p) %-8.8s -> %s\n",
@@ -751,14 +742,14 @@ static inline void ccid3_hc_rx_set_state(struct sock *sk,
751 742
752static void ccid3_hc_rx_send_feedback(struct sock *sk) 743static void ccid3_hc_rx_send_feedback(struct sock *sk)
753{ 744{
745 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
754 struct dccp_sock *dp = dccp_sk(sk); 746 struct dccp_sock *dp = dccp_sk(sk);
755 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
756 struct dccp_rx_hist_entry *packet; 747 struct dccp_rx_hist_entry *packet;
757 struct timeval now; 748 struct timeval now;
758 749
759 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk); 750 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk);
760 751
761 do_gettimeofday(&now); 752 dccp_timestamp(sk, &now);
762 753
763 switch (hcrx->ccid3hcrx_state) { 754 switch (hcrx->ccid3hcrx_state) {
764 case TFRC_RSTATE_NO_DATA: 755 case TFRC_RSTATE_NO_DATA:
@@ -767,11 +758,8 @@ static void ccid3_hc_rx_send_feedback(struct sock *sk)
767 case TFRC_RSTATE_DATA: { 758 case TFRC_RSTATE_DATA: {
768 const u32 delta = timeval_delta(&now, 759 const u32 delta = timeval_delta(&now,
769 &hcrx->ccid3hcrx_tstamp_last_feedback); 760 &hcrx->ccid3hcrx_tstamp_last_feedback);
770 761 hcrx->ccid3hcrx_x_recv = usecs_div(hcrx->ccid3hcrx_bytes_recv,
771 hcrx->ccid3hcrx_x_recv = (hcrx->ccid3hcrx_bytes_recv * 762 delta);
772 USEC_PER_SEC);
773 if (likely(delta > 1))
774 hcrx->ccid3hcrx_x_recv /= delta;
775 } 763 }
776 break; 764 break;
777 default: 765 default:
@@ -782,10 +770,10 @@ static void ccid3_hc_rx_send_feedback(struct sock *sk)
782 } 770 }
783 771
784 packet = dccp_rx_hist_find_data_packet(&hcrx->ccid3hcrx_hist); 772 packet = dccp_rx_hist_find_data_packet(&hcrx->ccid3hcrx_hist);
785 if (packet == NULL) { 773 if (unlikely(packet == NULL)) {
786 printk(KERN_CRIT "%s: %s, sk=%p, no data packet in history!\n", 774 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, no data packet "
787 __FUNCTION__, dccp_role(sk), sk); 775 "in history!\n",
788 dump_stack(); 776 __FUNCTION__, dccp_role(sk), sk);
789 return; 777 return;
790 } 778 }
791 779
@@ -801,17 +789,18 @@ static void ccid3_hc_rx_send_feedback(struct sock *sk)
801 hcrx->ccid3hcrx_pinv = ~0; 789 hcrx->ccid3hcrx_pinv = ~0;
802 else 790 else
803 hcrx->ccid3hcrx_pinv = 1000000 / hcrx->ccid3hcrx_p; 791 hcrx->ccid3hcrx_pinv = 1000000 / hcrx->ccid3hcrx_p;
792 dp->dccps_hc_rx_insert_options = 1;
804 dccp_send_ack(sk); 793 dccp_send_ack(sk);
805} 794}
806 795
807static void ccid3_hc_rx_insert_options(struct sock *sk, struct sk_buff *skb) 796static void ccid3_hc_rx_insert_options(struct sock *sk, struct sk_buff *skb)
808{ 797{
809 const struct dccp_sock *dp = dccp_sk(sk); 798 const struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
810 u32 x_recv, pinv; 799 u32 x_recv, pinv;
811 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
812 800
813 if (hcrx == NULL || !(sk->sk_state == DCCP_OPEN || 801 BUG_ON(hcrx == NULL);
814 sk->sk_state == DCCP_PARTOPEN)) 802
803 if (!(sk->sk_state == DCCP_OPEN || sk->sk_state == DCCP_PARTOPEN))
815 return; 804 return;
816 805
817 DCCP_SKB_CB(skb)->dccpd_ccval = hcrx->ccid3hcrx_last_counter; 806 DCCP_SKB_CB(skb)->dccpd_ccval = hcrx->ccid3hcrx_last_counter;
@@ -837,8 +826,7 @@ static void ccid3_hc_rx_insert_options(struct sock *sk, struct sk_buff *skb)
837 826
838static u32 ccid3_hc_rx_calc_first_li(struct sock *sk) 827static u32 ccid3_hc_rx_calc_first_li(struct sock *sk)
839{ 828{
840 struct dccp_sock *dp = dccp_sk(sk); 829 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
841 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
842 struct dccp_rx_hist_entry *entry, *next, *tail = NULL; 830 struct dccp_rx_hist_entry *entry, *next, *tail = NULL;
843 u32 rtt, delta, x_recv, fval, p, tmp2; 831 u32 rtt, delta, x_recv, fval, p, tmp2;
844 struct timeval tstamp = { 0, }; 832 struct timeval tstamp = { 0, };
@@ -869,17 +857,17 @@ static u32 ccid3_hc_rx_calc_first_li(struct sock *sk)
869 } 857 }
870 } 858 }
871 859
872 if (step == 0) { 860 if (unlikely(step == 0)) {
873 printk(KERN_CRIT "%s: %s, sk=%p, packet history contains no " 861 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, packet history "
874 "data packets!\n", 862 "contains no data packets!\n",
875 __FUNCTION__, dccp_role(sk), sk); 863 __FUNCTION__, dccp_role(sk), sk);
876 return ~0; 864 return ~0;
877 } 865 }
878 866
879 if (interval == 0) { 867 if (unlikely(interval == 0)) {
880 ccid3_pr_debug("%s, sk=%p, Could not find a win_count " 868 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, Could not find a "
881 "interval > 0. Defaulting to 1\n", 869 "win_count interval > 0. Defaulting to 1\n",
882 dccp_role(sk), sk); 870 __FUNCTION__, dccp_role(sk), sk);
883 interval = 1; 871 interval = 1;
884 } 872 }
885found: 873found:
@@ -889,10 +877,9 @@ found:
889 if (rtt == 0) 877 if (rtt == 0)
890 rtt = 1; 878 rtt = 1;
891 879
892 delta = timeval_now_delta(&hcrx->ccid3hcrx_tstamp_last_feedback); 880 dccp_timestamp(sk, &tstamp);
893 x_recv = hcrx->ccid3hcrx_bytes_recv * USEC_PER_SEC; 881 delta = timeval_delta(&tstamp, &hcrx->ccid3hcrx_tstamp_last_feedback);
894 if (likely(delta > 1)) 882 x_recv = usecs_div(hcrx->ccid3hcrx_bytes_recv, delta);
895 x_recv /= delta;
896 883
897 tmp1 = (u64)x_recv * (u64)rtt; 884 tmp1 = (u64)x_recv * (u64)rtt;
898 do_div(tmp1,10000000); 885 do_div(tmp1,10000000);
@@ -911,8 +898,7 @@ found:
911 898
912static void ccid3_hc_rx_update_li(struct sock *sk, u64 seq_loss, u8 win_loss) 899static void ccid3_hc_rx_update_li(struct sock *sk, u64 seq_loss, u8 win_loss)
913{ 900{
914 struct dccp_sock *dp = dccp_sk(sk); 901 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
915 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
916 902
917 if (seq_loss != DCCP_MAX_SEQNO + 1 && 903 if (seq_loss != DCCP_MAX_SEQNO + 1 &&
918 list_empty(&hcrx->ccid3hcrx_li_hist)) { 904 list_empty(&hcrx->ccid3hcrx_li_hist)) {
@@ -924,14 +910,14 @@ static void ccid3_hc_rx_update_li(struct sock *sk, u64 seq_loss, u8 win_loss)
924 if (li_tail == NULL) 910 if (li_tail == NULL)
925 return; 911 return;
926 li_tail->dccplih_interval = ccid3_hc_rx_calc_first_li(sk); 912 li_tail->dccplih_interval = ccid3_hc_rx_calc_first_li(sk);
927 } 913 } else
928 /* FIXME: find end of interval */ 914 LIMIT_NETDEBUG(KERN_WARNING "%s: FIXME: find end of "
915 "interval\n", __FUNCTION__);
929} 916}
930 917
931static void ccid3_hc_rx_detect_loss(struct sock *sk) 918static void ccid3_hc_rx_detect_loss(struct sock *sk)
932{ 919{
933 struct dccp_sock *dp = dccp_sk(sk); 920 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
934 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
935 u8 win_loss; 921 u8 win_loss;
936 const u64 seq_loss = dccp_rx_hist_detect_loss(&hcrx->ccid3hcrx_hist, 922 const u64 seq_loss = dccp_rx_hist_detect_loss(&hcrx->ccid3hcrx_hist,
937 &hcrx->ccid3hcrx_li_hist, 923 &hcrx->ccid3hcrx_li_hist,
@@ -942,22 +928,19 @@ static void ccid3_hc_rx_detect_loss(struct sock *sk)
942 928
943static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb) 929static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
944{ 930{
945 struct dccp_sock *dp = dccp_sk(sk); 931 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
946 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
947 const struct dccp_options_received *opt_recv; 932 const struct dccp_options_received *opt_recv;
948 struct dccp_rx_hist_entry *packet; 933 struct dccp_rx_hist_entry *packet;
949 struct timeval now; 934 struct timeval now;
950 u8 win_count; 935 u8 win_count;
951 u32 p_prev; 936 u32 p_prev, r_sample, t_elapsed;
952 int ins; 937 int ins;
953 938
954 if (hcrx == NULL) 939 BUG_ON(hcrx == NULL ||
955 return; 940 !(hcrx->ccid3hcrx_state == TFRC_RSTATE_NO_DATA ||
956
957 BUG_ON(!(hcrx->ccid3hcrx_state == TFRC_RSTATE_NO_DATA ||
958 hcrx->ccid3hcrx_state == TFRC_RSTATE_DATA)); 941 hcrx->ccid3hcrx_state == TFRC_RSTATE_DATA));
959 942
960 opt_recv = &dp->dccps_options_received; 943 opt_recv = &dccp_sk(sk)->dccps_options_received;
961 944
962 switch (DCCP_SKB_CB(skb)->dccpd_type) { 945 switch (DCCP_SKB_CB(skb)->dccpd_type) {
963 case DCCP_PKT_ACK: 946 case DCCP_PKT_ACK:
@@ -967,10 +950,24 @@ static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
967 if (opt_recv->dccpor_timestamp_echo == 0) 950 if (opt_recv->dccpor_timestamp_echo == 0)
968 break; 951 break;
969 p_prev = hcrx->ccid3hcrx_rtt; 952 p_prev = hcrx->ccid3hcrx_rtt;
970 do_gettimeofday(&now); 953 dccp_timestamp(sk, &now);
971 hcrx->ccid3hcrx_rtt = timeval_usecs(&now) - 954 timeval_sub_usecs(&now, opt_recv->dccpor_timestamp_echo * 10);
972 (opt_recv->dccpor_timestamp_echo - 955 r_sample = timeval_usecs(&now);
973 opt_recv->dccpor_elapsed_time) * 10; 956 t_elapsed = opt_recv->dccpor_elapsed_time * 10;
957
958 if (unlikely(r_sample <= t_elapsed))
959 LIMIT_NETDEBUG(KERN_WARNING "%s: r_sample=%uus, "
960 "t_elapsed=%uus\n",
961 __FUNCTION__, r_sample, t_elapsed);
962 else
963 r_sample -= t_elapsed;
964
965 if (hcrx->ccid3hcrx_state == TFRC_RSTATE_NO_DATA)
966 hcrx->ccid3hcrx_rtt = r_sample;
967 else
968 hcrx->ccid3hcrx_rtt = (hcrx->ccid3hcrx_rtt * 9) / 10 +
969 r_sample / 10;
970
974 if (p_prev != hcrx->ccid3hcrx_rtt) 971 if (p_prev != hcrx->ccid3hcrx_rtt)
975 ccid3_pr_debug("%s, New RTT=%luus, elapsed time=%u\n", 972 ccid3_pr_debug("%s, New RTT=%luus, elapsed time=%u\n",
976 dccp_role(sk), hcrx->ccid3hcrx_rtt, 973 dccp_role(sk), hcrx->ccid3hcrx_rtt,
@@ -978,19 +975,16 @@ static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
978 break; 975 break;
979 case DCCP_PKT_DATA: 976 case DCCP_PKT_DATA:
980 break; 977 break;
981 default: 978 default: /* We're not interested in other packet types, move along */
982 ccid3_pr_debug("%s, sk=%p, not DATA/DATAACK/ACK packet(%s)\n",
983 dccp_role(sk), sk,
984 dccp_packet_name(DCCP_SKB_CB(skb)->dccpd_type));
985 return; 979 return;
986 } 980 }
987 981
988 packet = dccp_rx_hist_entry_new(ccid3_rx_hist, opt_recv->dccpor_ndp, 982 packet = dccp_rx_hist_entry_new(ccid3_rx_hist, sk, opt_recv->dccpor_ndp,
989 skb, SLAB_ATOMIC); 983 skb, SLAB_ATOMIC);
990 if (packet == NULL) { 984 if (unlikely(packet == NULL)) {
991 ccid3_pr_debug("%s, sk=%p, Not enough mem to add rx packet " 985 LIMIT_NETDEBUG(KERN_WARNING "%s: %s, sk=%p, Not enough mem to "
992 "to history (consider it lost)!", 986 "add rx packet to history, consider it lost!\n",
993 dccp_role(sk), sk); 987 __FUNCTION__, dccp_role(sk), sk);
994 return; 988 return;
995 } 989 }
996 990
@@ -1017,7 +1011,7 @@ static void ccid3_hc_rx_packet_recv(struct sock *sk, struct sk_buff *skb)
1017 if (ins != 0) 1011 if (ins != 0)
1018 break; 1012 break;
1019 1013
1020 do_gettimeofday(&now); 1014 dccp_timestamp(sk, &now);
1021 if (timeval_delta(&now, &hcrx->ccid3hcrx_tstamp_last_ack) >= 1015 if (timeval_delta(&now, &hcrx->ccid3hcrx_tstamp_last_ack) >=
1022 hcrx->ccid3hcrx_rtt) { 1016 hcrx->ccid3hcrx_rtt) {
1023 hcrx->ccid3hcrx_tstamp_last_ack = now; 1017 hcrx->ccid3hcrx_tstamp_last_ack = now;
@@ -1056,11 +1050,11 @@ static int ccid3_hc_rx_init(struct sock *sk)
1056 1050
1057 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk); 1051 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk);
1058 1052
1059 hcrx = dp->dccps_hc_rx_ccid_private = kmalloc(sizeof(*hcrx), 1053 dp->dccps_hc_rx_ccid_private = kmalloc(sizeof(*hcrx), gfp_any());
1060 gfp_any()); 1054 if (dp->dccps_hc_rx_ccid_private == NULL)
1061 if (hcrx == NULL)
1062 return -ENOMEM; 1055 return -ENOMEM;
1063 1056
1057 hcrx = ccid3_hc_rx_sk(sk);
1064 memset(hcrx, 0, sizeof(*hcrx)); 1058 memset(hcrx, 0, sizeof(*hcrx));
1065 1059
1066 if (dp->dccps_packet_size >= TFRC_MIN_PACKET_SIZE && 1060 if (dp->dccps_packet_size >= TFRC_MIN_PACKET_SIZE &&
@@ -1072,23 +1066,18 @@ static int ccid3_hc_rx_init(struct sock *sk)
1072 hcrx->ccid3hcrx_state = TFRC_RSTATE_NO_DATA; 1066 hcrx->ccid3hcrx_state = TFRC_RSTATE_NO_DATA;
1073 INIT_LIST_HEAD(&hcrx->ccid3hcrx_hist); 1067 INIT_LIST_HEAD(&hcrx->ccid3hcrx_hist);
1074 INIT_LIST_HEAD(&hcrx->ccid3hcrx_li_hist); 1068 INIT_LIST_HEAD(&hcrx->ccid3hcrx_li_hist);
1075 /* 1069 dccp_timestamp(sk, &hcrx->ccid3hcrx_tstamp_last_ack);
1076 * XXX this seems to be paranoid, need to think more about this, for 1070 hcrx->ccid3hcrx_tstamp_last_feedback = hcrx->ccid3hcrx_tstamp_last_ack;
1077 * now start with something different than zero. -acme 1071 hcrx->ccid3hcrx_rtt = 5000; /* XXX 5ms for now... */
1078 */
1079 hcrx->ccid3hcrx_rtt = USEC_PER_SEC / 5;
1080 return 0; 1072 return 0;
1081} 1073}
1082 1074
1083static void ccid3_hc_rx_exit(struct sock *sk) 1075static void ccid3_hc_rx_exit(struct sock *sk)
1084{ 1076{
1077 struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
1085 struct dccp_sock *dp = dccp_sk(sk); 1078 struct dccp_sock *dp = dccp_sk(sk);
1086 struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
1087 1079
1088 ccid3_pr_debug("%s, sk=%p\n", dccp_role(sk), sk); 1080 BUG_ON(hcrx == NULL);
1089
1090 if (hcrx == NULL)
1091 return;
1092 1081
1093 ccid3_hc_rx_set_state(sk, TFRC_RSTATE_TERM); 1082 ccid3_hc_rx_set_state(sk, TFRC_RSTATE_TERM);
1094 1083
@@ -1104,12 +1093,14 @@ static void ccid3_hc_rx_exit(struct sock *sk)
1104 1093
1105static void ccid3_hc_rx_get_info(struct sock *sk, struct tcp_info *info) 1094static void ccid3_hc_rx_get_info(struct sock *sk, struct tcp_info *info)
1106{ 1095{
1107 const struct dccp_sock *dp = dccp_sk(sk); 1096 const struct ccid3_hc_rx_sock *hcrx = ccid3_hc_rx_sk(sk);
1108 const struct ccid3_hc_rx_sock *hcrx = dp->dccps_hc_rx_ccid_private;
1109 1097
1110 if (hcrx == NULL) 1098 /* Listen socks doesn't have a private CCID block */
1099 if (sk->sk_state == DCCP_LISTEN)
1111 return; 1100 return;
1112 1101
1102 BUG_ON(hcrx == NULL);
1103
1113 info->tcpi_ca_state = hcrx->ccid3hcrx_state; 1104 info->tcpi_ca_state = hcrx->ccid3hcrx_state;
1114 info->tcpi_options |= TCPI_OPT_TIMESTAMPS; 1105 info->tcpi_options |= TCPI_OPT_TIMESTAMPS;
1115 info->tcpi_rcv_rtt = hcrx->ccid3hcrx_rtt; 1106 info->tcpi_rcv_rtt = hcrx->ccid3hcrx_rtt;
@@ -1117,12 +1108,14 @@ static void ccid3_hc_rx_get_info(struct sock *sk, struct tcp_info *info)
1117 1108
1118static void ccid3_hc_tx_get_info(struct sock *sk, struct tcp_info *info) 1109static void ccid3_hc_tx_get_info(struct sock *sk, struct tcp_info *info)
1119{ 1110{
1120 const struct dccp_sock *dp = dccp_sk(sk); 1111 const struct ccid3_hc_tx_sock *hctx = ccid3_hc_tx_sk(sk);
1121 const struct ccid3_hc_tx_sock *hctx = dp->dccps_hc_tx_ccid_private;
1122 1112
1123 if (hctx == NULL) 1113 /* Listen socks doesn't have a private CCID block */
1114 if (sk->sk_state == DCCP_LISTEN)
1124 return; 1115 return;
1125 1116
1117 BUG_ON(hctx == NULL);
1118
1126 info->tcpi_rto = hctx->ccid3hctx_t_rto; 1119 info->tcpi_rto = hctx->ccid3hctx_t_rto;
1127 info->tcpi_rtt = hctx->ccid3hctx_rtt; 1120 info->tcpi_rtt = hctx->ccid3hctx_rtt;
1128} 1121}
diff --git a/net/dccp/ccids/ccid3.h b/net/dccp/ccids/ccid3.h
index ee8cbace6630..eb248778eea3 100644
--- a/net/dccp/ccids/ccid3.h
+++ b/net/dccp/ccids/ccid3.h
@@ -48,6 +48,8 @@
48/* Two seconds as per CCID3 spec */ 48/* Two seconds as per CCID3 spec */
49#define TFRC_INITIAL_TIMEOUT (2 * USEC_PER_SEC) 49#define TFRC_INITIAL_TIMEOUT (2 * USEC_PER_SEC)
50 50
51#define TFRC_INITIAL_IPI (USEC_PER_SEC / 4)
52
51/* In usecs - half the scheduling granularity as per RFC3448 4.6 */ 53/* In usecs - half the scheduling granularity as per RFC3448 4.6 */
52#define TFRC_OPSYS_HALF_TIME_GRAN (USEC_PER_SEC / (2 * HZ)) 54#define TFRC_OPSYS_HALF_TIME_GRAN (USEC_PER_SEC / (2 * HZ))
53 55
@@ -115,7 +117,7 @@ struct ccid3_hc_rx_sock {
115 u64 ccid3hcrx_seqno_last_counter:48, 117 u64 ccid3hcrx_seqno_last_counter:48,
116 ccid3hcrx_state:8, 118 ccid3hcrx_state:8,
117 ccid3hcrx_last_counter:4; 119 ccid3hcrx_last_counter:4;
118 unsigned long ccid3hcrx_rtt; 120 u32 ccid3hcrx_rtt;
119 u32 ccid3hcrx_p; 121 u32 ccid3hcrx_p;
120 u32 ccid3hcrx_bytes_recv; 122 u32 ccid3hcrx_bytes_recv;
121 struct timeval ccid3hcrx_tstamp_last_feedback; 123 struct timeval ccid3hcrx_tstamp_last_feedback;
@@ -128,10 +130,14 @@ struct ccid3_hc_rx_sock {
128 u32 ccid3hcrx_x_recv; 130 u32 ccid3hcrx_x_recv;
129}; 131};
130 132
131#define ccid3_hc_tx_field(s,field) (s->dccps_hc_tx_ccid_private == NULL ? 0 : \ 133static inline struct ccid3_hc_tx_sock *ccid3_hc_tx_sk(const struct sock *sk)
132 ((struct ccid3_hc_tx_sock *)s->dccps_hc_tx_ccid_private)->ccid3hctx_##field) 134{
135 return dccp_sk(sk)->dccps_hc_tx_ccid_private;
136}
133 137
134#define ccid3_hc_rx_field(s,field) (s->dccps_hc_rx_ccid_private == NULL ? 0 : \ 138static inline struct ccid3_hc_rx_sock *ccid3_hc_rx_sk(const struct sock *sk)
135 ((struct ccid3_hc_rx_sock *)s->dccps_hc_rx_ccid_private)->ccid3hcrx_##field) 139{
140 return dccp_sk(sk)->dccps_hc_rx_ccid_private;
141}
136 142
137#endif /* _DCCP_CCID3_H_ */ 143#endif /* _DCCP_CCID3_H_ */
diff --git a/net/dccp/ccids/lib/packet_history.h b/net/dccp/ccids/lib/packet_history.h
index fb90a91aa93d..b375ebdb7dcf 100644
--- a/net/dccp/ccids/lib/packet_history.h
+++ b/net/dccp/ccids/lib/packet_history.h
@@ -134,6 +134,7 @@ static inline struct dccp_tx_hist_entry *
134 134
135static inline struct dccp_rx_hist_entry * 135static inline struct dccp_rx_hist_entry *
136 dccp_rx_hist_entry_new(struct dccp_rx_hist *hist, 136 dccp_rx_hist_entry_new(struct dccp_rx_hist *hist,
137 const struct sock *sk,
137 const u32 ndp, 138 const u32 ndp,
138 const struct sk_buff *skb, 139 const struct sk_buff *skb,
139 const unsigned int __nocast prio) 140 const unsigned int __nocast prio)
@@ -148,7 +149,7 @@ static inline struct dccp_rx_hist_entry *
148 entry->dccphrx_ccval = dh->dccph_ccval; 149 entry->dccphrx_ccval = dh->dccph_ccval;
149 entry->dccphrx_type = dh->dccph_type; 150 entry->dccphrx_type = dh->dccph_type;
150 entry->dccphrx_ndp = ndp; 151 entry->dccphrx_ndp = ndp;
151 do_gettimeofday(&(entry->dccphrx_tstamp)); 152 dccp_timestamp(sk, &entry->dccphrx_tstamp);
152 } 153 }
153 154
154 return entry; 155 return entry;
diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h
index 33456c0d5937..95c4630b3b18 100644
--- a/net/dccp/dccp.h
+++ b/net/dccp/dccp.h
@@ -426,10 +426,13 @@ extern struct dccp_ackpkts *
426 dccp_ackpkts_alloc(unsigned int len, 426 dccp_ackpkts_alloc(unsigned int len,
427 const unsigned int __nocast priority); 427 const unsigned int __nocast priority);
428extern void dccp_ackpkts_free(struct dccp_ackpkts *ap); 428extern void dccp_ackpkts_free(struct dccp_ackpkts *ap);
429extern int dccp_ackpkts_add(struct dccp_ackpkts *ap, u64 ackno, u8 state); 429extern int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk,
430 u64 ackno, u8 state);
430extern void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap, 431extern void dccp_ackpkts_check_rcv_ackno(struct dccp_ackpkts *ap,
431 struct sock *sk, u64 ackno); 432 struct sock *sk, u64 ackno);
432 433
434extern void dccp_timestamp(const struct sock *sk, struct timeval *tv);
435
433static inline suseconds_t timeval_usecs(const struct timeval *tv) 436static inline suseconds_t timeval_usecs(const struct timeval *tv)
434{ 437{
435 return tv->tv_sec * USEC_PER_SEC + tv->tv_usec; 438 return tv->tv_sec * USEC_PER_SEC + tv->tv_usec;
@@ -468,17 +471,6 @@ static inline void timeval_sub_usecs(struct timeval *tv,
468 } 471 }
469} 472}
470 473
471/*
472 * Returns the difference in usecs between timeval
473 * passed in and current time
474 */
475static inline suseconds_t timeval_now_delta(const struct timeval *tv)
476{
477 struct timeval now;
478 do_gettimeofday(&now);
479 return timeval_delta(&now, tv);
480}
481
482#ifdef CONFIG_IP_DCCP_DEBUG 474#ifdef CONFIG_IP_DCCP_DEBUG
483extern void dccp_ackvector_print(const u64 ackno, 475extern void dccp_ackvector_print(const u64 ackno,
484 const unsigned char *vector, int len); 476 const unsigned char *vector, int len);
diff --git a/net/dccp/input.c b/net/dccp/input.c
index ef29cef1dafe..c60bc3433f5e 100644
--- a/net/dccp/input.c
+++ b/net/dccp/input.c
@@ -170,7 +170,7 @@ int dccp_rcv_established(struct sock *sk, struct sk_buff *skb,
170 if (dp->dccps_options.dccpo_send_ack_vector) { 170 if (dp->dccps_options.dccpo_send_ack_vector) {
171 struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts; 171 struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
172 172
173 if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, 173 if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk,
174 DCCP_SKB_CB(skb)->dccpd_seq, 174 DCCP_SKB_CB(skb)->dccpd_seq,
175 DCCP_ACKPKTS_STATE_RECEIVED)) { 175 DCCP_ACKPKTS_STATE_RECEIVED)) {
176 LIMIT_NETDEBUG(KERN_WARNING "DCCP: acknowledgeable " 176 LIMIT_NETDEBUG(KERN_WARNING "DCCP: acknowledgeable "
@@ -498,7 +498,7 @@ int dccp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
498 * DCCP_ACKPKTS_STATE_ECN_MARKED 498 * DCCP_ACKPKTS_STATE_ECN_MARKED
499 */ 499 */
500 if (dp->dccps_options.dccpo_send_ack_vector) { 500 if (dp->dccps_options.dccpo_send_ack_vector) {
501 if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, 501 if (dccp_ackpkts_add(dp->dccps_hc_rx_ackpkts, sk,
502 DCCP_SKB_CB(skb)->dccpd_seq, 502 DCCP_SKB_CB(skb)->dccpd_seq,
503 DCCP_ACKPKTS_STATE_RECEIVED)) 503 DCCP_ACKPKTS_STATE_RECEIVED))
504 goto discard; 504 goto discard;
diff --git a/net/dccp/ipv4.c b/net/dccp/ipv4.c
index 3fc75dbee4b8..2afaa464e7f0 100644
--- a/net/dccp/ipv4.c
+++ b/net/dccp/ipv4.c
@@ -641,16 +641,12 @@ int dccp_v4_send_reset(struct sock *sk, enum dccp_reset_codes code)
641 641
642 skb = dccp_make_reset(sk, sk->sk_dst_cache, code); 642 skb = dccp_make_reset(sk, sk->sk_dst_cache, code);
643 if (skb != NULL) { 643 if (skb != NULL) {
644 const struct dccp_sock *dp = dccp_sk(sk);
645 const struct inet_sock *inet = inet_sk(sk); 644 const struct inet_sock *inet = inet_sk(sk);
646 645
647 err = ip_build_and_send_pkt(skb, sk, 646 err = ip_build_and_send_pkt(skb, sk,
648 inet->saddr, inet->daddr, NULL); 647 inet->saddr, inet->daddr, NULL);
649 if (err == NET_XMIT_CN) 648 if (err == NET_XMIT_CN)
650 err = 0; 649 err = 0;
651
652 ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk);
653 ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk);
654 } 650 }
655 651
656 return err; 652 return err;
@@ -1243,6 +1239,7 @@ static int dccp_v4_init_sock(struct sock *sk)
1243 static int dccp_ctl_socket_init = 1; 1239 static int dccp_ctl_socket_init = 1;
1244 1240
1245 dccp_options_init(&dp->dccps_options); 1241 dccp_options_init(&dp->dccps_options);
1242 do_gettimeofday(&dp->dccps_epoch);
1246 1243
1247 if (dp->dccps_options.dccpo_send_ack_vector) { 1244 if (dp->dccps_options.dccpo_send_ack_vector) {
1248 dp->dccps_hc_rx_ackpkts = 1245 dp->dccps_hc_rx_ackpkts =
diff --git a/net/dccp/minisocks.c b/net/dccp/minisocks.c
index ce5dff4ac22e..18461bc04cbe 100644
--- a/net/dccp/minisocks.c
+++ b/net/dccp/minisocks.c
@@ -96,6 +96,7 @@ struct sock *dccp_create_openreq_child(struct sock *sk,
96 newdp->dccps_hc_rx_ackpkts = NULL; 96 newdp->dccps_hc_rx_ackpkts = NULL;
97 newdp->dccps_role = DCCP_ROLE_SERVER; 97 newdp->dccps_role = DCCP_ROLE_SERVER;
98 newicsk->icsk_rto = DCCP_TIMEOUT_INIT; 98 newicsk->icsk_rto = DCCP_TIMEOUT_INIT;
99 do_gettimeofday(&newdp->dccps_epoch);
99 100
100 if (newdp->dccps_options.dccpo_send_ack_vector) { 101 if (newdp->dccps_options.dccpo_send_ack_vector) {
101 newdp->dccps_hc_rx_ackpkts = 102 newdp->dccps_hc_rx_ackpkts =
diff --git a/net/dccp/options.c b/net/dccp/options.c
index 382c5894acb2..d4c4242d8dd7 100644
--- a/net/dccp/options.c
+++ b/net/dccp/options.c
@@ -72,6 +72,7 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
72 struct dccp_options_received *opt_recv = &dp->dccps_options_received; 72 struct dccp_options_received *opt_recv = &dp->dccps_options_received;
73 unsigned char opt, len; 73 unsigned char opt, len;
74 unsigned char *value; 74 unsigned char *value;
75 u32 elapsed_time;
75 76
76 memset(opt_recv, 0, sizeof(*opt_recv)); 77 memset(opt_recv, 0, sizeof(*opt_recv));
77 78
@@ -139,7 +140,7 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
139 opt_recv->dccpor_timestamp = ntohl(*(u32 *)value); 140 opt_recv->dccpor_timestamp = ntohl(*(u32 *)value);
140 141
141 dp->dccps_timestamp_echo = opt_recv->dccpor_timestamp; 142 dp->dccps_timestamp_echo = opt_recv->dccpor_timestamp;
142 do_gettimeofday(&dp->dccps_timestamp_time); 143 dccp_timestamp(sk, &dp->dccps_timestamp_time);
143 144
144 dccp_pr_debug("%sTIMESTAMP=%u, ackno=%llu\n", 145 dccp_pr_debug("%sTIMESTAMP=%u, ackno=%llu\n",
145 debug_prefix, opt_recv->dccpor_timestamp, 146 debug_prefix, opt_recv->dccpor_timestamp,
@@ -159,18 +160,18 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
159 (unsigned long long) 160 (unsigned long long)
160 DCCP_SKB_CB(skb)->dccpd_ack_seq); 161 DCCP_SKB_CB(skb)->dccpd_ack_seq);
161 162
162 if (len > 4) {
163 if (len == 6)
164 opt_recv->dccpor_elapsed_time =
165 ntohs(*(u16 *)(value + 4));
166 else
167 opt_recv->dccpor_elapsed_time =
168 ntohl(*(u32 *)(value + 4));
169 163
170 dccp_pr_debug("%sTIMESTAMP_ECHO ELAPSED_TIME=%d\n", 164 if (len == 4)
171 debug_prefix, 165 break;
172 opt_recv->dccpor_elapsed_time); 166
173 } 167 if (len == 6)
168 elapsed_time = ntohs(*(u16 *)(value + 4));
169 else
170 elapsed_time = ntohl(*(u32 *)(value + 4));
171
172 /* Give precedence to the biggest ELAPSED_TIME */
173 if (elapsed_time > opt_recv->dccpor_elapsed_time)
174 opt_recv->dccpor_elapsed_time = elapsed_time;
174 break; 175 break;
175 case DCCPO_ELAPSED_TIME: 176 case DCCPO_ELAPSED_TIME:
176 if (len != 2 && len != 4) 177 if (len != 2 && len != 4)
@@ -180,14 +181,15 @@ int dccp_parse_options(struct sock *sk, struct sk_buff *skb)
180 continue; 181 continue;
181 182
182 if (len == 2) 183 if (len == 2)
183 opt_recv->dccpor_elapsed_time = 184 elapsed_time = ntohs(*(u16 *)value);
184 ntohs(*(u16 *)value);
185 else 185 else
186 opt_recv->dccpor_elapsed_time = 186 elapsed_time = ntohl(*(u32 *)value);
187 ntohl(*(u32 *)value); 187
188 if (elapsed_time > opt_recv->dccpor_elapsed_time)
189 opt_recv->dccpor_elapsed_time = elapsed_time;
188 190
189 dccp_pr_debug("%sELAPSED_TIME=%d\n", debug_prefix, 191 dccp_pr_debug("%sELAPSED_TIME=%d\n", debug_prefix,
190 opt_recv->dccpor_elapsed_time); 192 elapsed_time);
191 break; 193 break;
192 /* 194 /*
193 * From draft-ietf-dccp-spec-11.txt: 195 * From draft-ietf-dccp-spec-11.txt:
@@ -359,9 +361,13 @@ static void dccp_insert_option_ack_vector(struct sock *sk, struct sk_buff *skb)
359#endif 361#endif
360 struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts; 362 struct dccp_ackpkts *ap = dp->dccps_hc_rx_ackpkts;
361 int len = ap->dccpap_buf_vector_len + 2; 363 int len = ap->dccpap_buf_vector_len + 2;
362 const u32 elapsed_time = timeval_now_delta(&ap->dccpap_time) / 10; 364 struct timeval now;
365 u32 elapsed_time;
363 unsigned char *to, *from; 366 unsigned char *to, *from;
364 367
368 dccp_timestamp(sk, &now);
369 elapsed_time = timeval_delta(&now, &ap->dccpap_time) / 10;
370
365 if (elapsed_time != 0) 371 if (elapsed_time != 0)
366 dccp_insert_option_elapsed_time(sk, skb, elapsed_time); 372 dccp_insert_option_elapsed_time(sk, skb, elapsed_time);
367 373
@@ -426,13 +432,29 @@ static void dccp_insert_option_ack_vector(struct sock *sk, struct sk_buff *skb)
426 (unsigned long long) ap->dccpap_ack_ackno); 432 (unsigned long long) ap->dccpap_ack_ackno);
427} 433}
428 434
435void dccp_timestamp(const struct sock *sk, struct timeval *tv)
436{
437 const struct dccp_sock *dp = dccp_sk(sk);
438
439 do_gettimeofday(tv);
440 tv->tv_sec -= dp->dccps_epoch.tv_sec;
441 tv->tv_usec -= dp->dccps_epoch.tv_usec;
442
443 while (tv->tv_usec < 0) {
444 tv->tv_sec--;
445 tv->tv_usec += USEC_PER_SEC;
446 }
447}
448
449EXPORT_SYMBOL_GPL(dccp_timestamp);
450
429void dccp_insert_option_timestamp(struct sock *sk, struct sk_buff *skb) 451void dccp_insert_option_timestamp(struct sock *sk, struct sk_buff *skb)
430{ 452{
431 struct timeval tv; 453 struct timeval tv;
432 u32 now; 454 u32 now;
433 455
434 do_gettimeofday(&tv); 456 dccp_timestamp(sk, &tv);
435 now = (tv.tv_sec * USEC_PER_SEC + tv.tv_usec) / 10; 457 now = timeval_usecs(&tv) / 10;
436 /* yes this will overflow but that is the point as we want a 458 /* yes this will overflow but that is the point as we want a
437 * 10 usec 32 bit timer which mean it wraps every 11.9 hours */ 459 * 10 usec 32 bit timer which mean it wraps every 11.9 hours */
438 460
@@ -450,13 +472,17 @@ static void dccp_insert_option_timestamp_echo(struct sock *sk,
450 const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ? 472 const char *debug_prefix = dp->dccps_role == DCCP_ROLE_CLIENT ?
451 "CLIENT TX opt: " : "server TX opt: "; 473 "CLIENT TX opt: " : "server TX opt: ";
452#endif 474#endif
475 struct timeval now;
453 u32 tstamp_echo; 476 u32 tstamp_echo;
454 const u32 elapsed_time = 477 u32 elapsed_time;
455 timeval_now_delta(&dp->dccps_timestamp_time) / 10; 478 int len, elapsed_time_len;
456 const int elapsed_time_len = dccp_elapsed_time_len(elapsed_time);
457 const int len = 6 + elapsed_time_len;
458 unsigned char *to; 479 unsigned char *to;
459 480
481 dccp_timestamp(sk, &now);
482 elapsed_time = timeval_delta(&now, &dp->dccps_timestamp_time) / 10;
483 elapsed_time_len = dccp_elapsed_time_len(elapsed_time);
484 len = 6 + elapsed_time_len;
485
460 if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) { 486 if (DCCP_SKB_CB(skb)->dccpd_opt_len + len > DCCP_MAX_OPT_LEN) {
461 LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to insert " 487 LIMIT_NETDEBUG(KERN_INFO "DCCP: packet too small to insert "
462 "timestamp echo!\n"); 488 "timestamp echo!\n");
@@ -505,13 +531,18 @@ void dccp_insert_options(struct sock *sk, struct sk_buff *skb)
505 (dp->dccps_hc_rx_ackpkts->dccpap_buf_ackno != 531 (dp->dccps_hc_rx_ackpkts->dccpap_buf_ackno !=
506 DCCP_MAX_SEQNO + 1)) 532 DCCP_MAX_SEQNO + 1))
507 dccp_insert_option_ack_vector(sk, skb); 533 dccp_insert_option_ack_vector(sk, skb);
508
509 if (dp->dccps_timestamp_echo != 0) 534 if (dp->dccps_timestamp_echo != 0)
510 dccp_insert_option_timestamp_echo(sk, skb); 535 dccp_insert_option_timestamp_echo(sk, skb);
511 } 536 }
512 537
513 ccid_hc_rx_insert_options(dp->dccps_hc_rx_ccid, sk, skb); 538 if (dp->dccps_hc_rx_insert_options) {
514 ccid_hc_tx_insert_options(dp->dccps_hc_tx_ccid, sk, skb); 539 ccid_hc_rx_insert_options(dp->dccps_hc_rx_ccid, sk, skb);
540 dp->dccps_hc_rx_insert_options = 0;
541 }
542 if (dp->dccps_hc_tx_insert_options) {
543 ccid_hc_tx_insert_options(dp->dccps_hc_tx_ccid, sk, skb);
544 dp->dccps_hc_tx_insert_options = 0;
545 }
515 546
516 /* XXX: insert other options when appropriate */ 547 /* XXX: insert other options when appropriate */
517 548
@@ -616,7 +647,8 @@ static inline int dccp_ackpkts_set_buf_head_state(struct dccp_ackpkts *ap,
616/* 647/*
617 * Implements the draft-ietf-dccp-spec-11.txt Appendix A 648 * Implements the draft-ietf-dccp-spec-11.txt Appendix A
618 */ 649 */
619int dccp_ackpkts_add(struct dccp_ackpkts *ap, u64 ackno, u8 state) 650int dccp_ackpkts_add(struct dccp_ackpkts *ap, const struct sock *sk,
651 u64 ackno, u8 state)
620{ 652{
621 /* 653 /*
622 * Check at the right places if the buffer is full, if it is, tell the 654 * Check at the right places if the buffer is full, if it is, tell the
@@ -697,7 +729,7 @@ int dccp_ackpkts_add(struct dccp_ackpkts *ap, u64 ackno, u8 state)
697 } 729 }
698 730
699 ap->dccpap_buf_ackno = ackno; 731 ap->dccpap_buf_ackno = ackno;
700 do_gettimeofday(&ap->dccpap_time); 732 dccp_timestamp(sk, &ap->dccpap_time);
701out: 733out:
702 dccp_pr_debug(""); 734 dccp_pr_debug("");
703 dccp_ackpkts_print(ap); 735 dccp_ackpkts_print(ap);
diff --git a/net/dccp/output.c b/net/dccp/output.c
index 28de157a4326..ea6d0e91e511 100644
--- a/net/dccp/output.c
+++ b/net/dccp/output.c
@@ -522,7 +522,4 @@ void dccp_send_close(struct sock *sk, const int active)
522 dccp_transmit_skb(sk, skb_clone(skb, prio)); 522 dccp_transmit_skb(sk, skb_clone(skb, prio));
523 } else 523 } else
524 dccp_transmit_skb(sk, skb); 524 dccp_transmit_skb(sk, skb);
525
526 ccid_hc_rx_exit(dp->dccps_hc_rx_ccid, sk);
527 ccid_hc_tx_exit(dp->dccps_hc_tx_ccid, sk);
528} 525}
diff --git a/net/decnet/dn_route.c b/net/decnet/dn_route.c
index 2c915f305be3..3407f190afe8 100644
--- a/net/decnet/dn_route.c
+++ b/net/decnet/dn_route.c
@@ -117,8 +117,7 @@ static struct dn_rt_hash_bucket *dn_rt_hash_table;
117static unsigned dn_rt_hash_mask; 117static unsigned dn_rt_hash_mask;
118 118
119static struct timer_list dn_route_timer; 119static struct timer_list dn_route_timer;
120static struct timer_list dn_rt_flush_timer = 120static DEFINE_TIMER(dn_rt_flush_timer, dn_run_flush, 0, 0);
121 TIMER_INITIALIZER(dn_run_flush, 0, 0);
122int decnet_dst_gc_interval = 2; 121int decnet_dst_gc_interval = 2;
123 122
124static struct dst_ops dn_dst_ops = { 123static struct dst_ops dn_dst_ops = {
diff --git a/net/ieee80211/Kconfig b/net/ieee80211/Kconfig
index 58ed4319e693..91b16fbf91f0 100644
--- a/net/ieee80211/Kconfig
+++ b/net/ieee80211/Kconfig
@@ -1,6 +1,5 @@
1config IEEE80211 1config IEEE80211
2 tristate "Generic IEEE 802.11 Networking Stack" 2 tristate "Generic IEEE 802.11 Networking Stack"
3 select NET_RADIO
4 ---help--- 3 ---help---
5 This option enables the hardware independent IEEE 802.11 4 This option enables the hardware independent IEEE 802.11
6 networking stack. 5 networking stack.
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index bf147f8db399..a9d84f93442c 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1248,11 +1248,6 @@ module_init(inet_init);
1248/* ------------------------------------------------------------------------ */ 1248/* ------------------------------------------------------------------------ */
1249 1249
1250#ifdef CONFIG_PROC_FS 1250#ifdef CONFIG_PROC_FS
1251#ifdef CONFIG_IP_FIB_TRIE
1252extern int fib_stat_proc_init(void);
1253extern void fib_stat_proc_exit(void);
1254#endif
1255
1256static int __init ipv4_proc_init(void) 1251static int __init ipv4_proc_init(void)
1257{ 1252{
1258 int rc = 0; 1253 int rc = 0;
@@ -1265,19 +1260,11 @@ static int __init ipv4_proc_init(void)
1265 goto out_udp; 1260 goto out_udp;
1266 if (fib_proc_init()) 1261 if (fib_proc_init())
1267 goto out_fib; 1262 goto out_fib;
1268#ifdef CONFIG_IP_FIB_TRIE
1269 if (fib_stat_proc_init())
1270 goto out_fib_stat;
1271#endif
1272 if (ip_misc_proc_init()) 1263 if (ip_misc_proc_init())
1273 goto out_misc; 1264 goto out_misc;
1274out: 1265out:
1275 return rc; 1266 return rc;
1276out_misc: 1267out_misc:
1277#ifdef CONFIG_IP_FIB_TRIE
1278 fib_stat_proc_exit();
1279out_fib_stat:
1280#endif
1281 fib_proc_exit(); 1268 fib_proc_exit();
1282out_fib: 1269out_fib:
1283 udp4_proc_exit(); 1270 udp4_proc_exit();
diff --git a/net/ipv4/fib_trie.c b/net/ipv4/fib_trie.c
index b2dea4e5da77..1b63b4824164 100644
--- a/net/ipv4/fib_trie.c
+++ b/net/ipv4/fib_trie.c
@@ -43,7 +43,7 @@
43 * 2 of the License, or (at your option) any later version. 43 * 2 of the License, or (at your option) any later version.
44 */ 44 */
45 45
46#define VERSION "0.402" 46#define VERSION "0.403"
47 47
48#include <linux/config.h> 48#include <linux/config.h>
49#include <asm/uaccess.h> 49#include <asm/uaccess.h>
@@ -164,7 +164,6 @@ static struct node *resize(struct trie *t, struct tnode *tn);
164static struct tnode *inflate(struct trie *t, struct tnode *tn); 164static struct tnode *inflate(struct trie *t, struct tnode *tn);
165static struct tnode *halve(struct trie *t, struct tnode *tn); 165static struct tnode *halve(struct trie *t, struct tnode *tn);
166static void tnode_free(struct tnode *tn); 166static void tnode_free(struct tnode *tn);
167static void trie_dump_seq(struct seq_file *seq, struct trie *t);
168 167
169static kmem_cache_t *fn_alias_kmem __read_mostly; 168static kmem_cache_t *fn_alias_kmem __read_mostly;
170static struct trie *trie_local = NULL, *trie_main = NULL; 169static struct trie *trie_local = NULL, *trie_main = NULL;
@@ -1971,558 +1970,525 @@ struct fib_table * __init fib_hash_init(int id)
1971 return tb; 1970 return tb;
1972} 1971}
1973 1972
1974/* Trie dump functions */ 1973#ifdef CONFIG_PROC_FS
1974/* Depth first Trie walk iterator */
1975struct fib_trie_iter {
1976 struct tnode *tnode;
1977 struct trie *trie;
1978 unsigned index;
1979 unsigned depth;
1980};
1975 1981
1976static void putspace_seq(struct seq_file *seq, int n) 1982static struct node *fib_trie_get_next(struct fib_trie_iter *iter)
1977{ 1983{
1978 while (n--) 1984 struct tnode *tn = iter->tnode;
1979 seq_printf(seq, " "); 1985 unsigned cindex = iter->index;
1980} 1986 struct tnode *p;
1981 1987
1982static void printbin_seq(struct seq_file *seq, unsigned int v, int bits) 1988 pr_debug("get_next iter={node=%p index=%d depth=%d}\n",
1983{ 1989 iter->tnode, iter->index, iter->depth);
1984 while (bits--) 1990rescan:
1985 seq_printf(seq, "%s", (v & (1<<bits))?"1":"0"); 1991 while (cindex < (1<<tn->bits)) {
1986} 1992 struct node *n = tnode_get_child(tn, cindex);
1987 1993
1988static void printnode_seq(struct seq_file *seq, int indent, struct node *n, 1994 if (n) {
1989 int pend, int cindex, int bits) 1995 if (IS_LEAF(n)) {
1990{ 1996 iter->tnode = tn;
1991 putspace_seq(seq, indent); 1997 iter->index = cindex + 1;
1992 if (IS_LEAF(n)) 1998 } else {
1993 seq_printf(seq, "|"); 1999 /* push down one level */
1994 else 2000 iter->tnode = (struct tnode *) n;
1995 seq_printf(seq, "+"); 2001 iter->index = 0;
1996 if (bits) { 2002 ++iter->depth;
1997 seq_printf(seq, "%d/", cindex); 2003 }
1998 printbin_seq(seq, cindex, bits); 2004 return n;
1999 seq_printf(seq, ": "); 2005 }
2000 } else
2001 seq_printf(seq, "<root>: ");
2002 seq_printf(seq, "%s:%p ", IS_LEAF(n)?"Leaf":"Internal node", n);
2003 2006
2004 if (IS_LEAF(n)) { 2007 ++cindex;
2005 struct leaf *l = (struct leaf *)n; 2008 }
2006 struct fib_alias *fa;
2007 int i;
2008 2009
2009 seq_printf(seq, "key=%d.%d.%d.%d\n", 2010 /* Current node exhausted, pop back up */
2010 n->key >> 24, (n->key >> 16) % 256, (n->key >> 8) % 256, n->key % 256); 2011 p = NODE_PARENT(tn);
2011 2012 if (p) {
2012 for (i = 32; i >= 0; i--) 2013 cindex = tkey_extract_bits(tn->key, p->pos, p->bits)+1;
2013 if (find_leaf_info(&l->list, i)) { 2014 tn = p;
2014 struct list_head *fa_head = get_fa_head(l, i); 2015 --iter->depth;
2015 2016 goto rescan;
2016 if (!fa_head)
2017 continue;
2018
2019 if (list_empty(fa_head))
2020 continue;
2021
2022 putspace_seq(seq, indent+2);
2023 seq_printf(seq, "{/%d...dumping}\n", i);
2024
2025 list_for_each_entry_rcu(fa, fa_head, fa_list) {
2026 putspace_seq(seq, indent+2);
2027 if (fa->fa_info == NULL) {
2028 seq_printf(seq, "Error fa_info=NULL\n");
2029 continue;
2030 }
2031 if (fa->fa_info->fib_nh == NULL) {
2032 seq_printf(seq, "Error _fib_nh=NULL\n");
2033 continue;
2034 }
2035
2036 seq_printf(seq, "{type=%d scope=%d TOS=%d}\n",
2037 fa->fa_type,
2038 fa->fa_scope,
2039 fa->fa_tos);
2040 }
2041 }
2042 } else {
2043 struct tnode *tn = (struct tnode *)n;
2044 int plen = ((struct tnode *)n)->pos;
2045 t_key prf = MASK_PFX(n->key, plen);
2046
2047 seq_printf(seq, "key=%d.%d.%d.%d/%d\n",
2048 prf >> 24, (prf >> 16) % 256, (prf >> 8) % 256, prf % 256, plen);
2049
2050 putspace_seq(seq, indent); seq_printf(seq, "| ");
2051 seq_printf(seq, "{key prefix=%08x/", tn->key & TKEY_GET_MASK(0, tn->pos));
2052 printbin_seq(seq, tkey_extract_bits(tn->key, 0, tn->pos), tn->pos);
2053 seq_printf(seq, "}\n");
2054 putspace_seq(seq, indent); seq_printf(seq, "| ");
2055 seq_printf(seq, "{pos=%d", tn->pos);
2056 seq_printf(seq, " (skip=%d bits)", tn->pos - pend);
2057 seq_printf(seq, " bits=%d (%u children)}\n", tn->bits, (1 << tn->bits));
2058 putspace_seq(seq, indent); seq_printf(seq, "| ");
2059 seq_printf(seq, "{empty=%d full=%d}\n", tn->empty_children, tn->full_children);
2060 } 2017 }
2018
2019 /* got root? */
2020 return NULL;
2061} 2021}
2062 2022
2063static void trie_dump_seq(struct seq_file *seq, struct trie *t) 2023static struct node *fib_trie_get_first(struct fib_trie_iter *iter,
2024 struct trie *t)
2064{ 2025{
2065 struct node *n; 2026 struct node *n = rcu_dereference(t->trie);
2066 int cindex = 0;
2067 int indent = 1;
2068 int pend = 0;
2069 int depth = 0;
2070 struct tnode *tn;
2071
2072 rcu_read_lock();
2073 n = rcu_dereference(t->trie);
2074 seq_printf(seq, "------ trie_dump of t=%p ------\n", t);
2075 2027
2076 if (!n) { 2028 if (n && IS_TNODE(n)) {
2077 seq_printf(seq, "------ trie is empty\n"); 2029 iter->tnode = (struct tnode *) n;
2078 2030 iter->trie = t;
2079 rcu_read_unlock(); 2031 iter->index = 0;
2080 return; 2032 iter->depth = 0;
2033 return n;
2081 } 2034 }
2035 return NULL;
2036}
2082 2037
2083 printnode_seq(seq, indent, n, pend, cindex, 0); 2038static void trie_collect_stats(struct trie *t, struct trie_stat *s)
2084 2039{
2085 if (!IS_TNODE(n)) { 2040 struct node *n;
2086 rcu_read_unlock(); 2041 struct fib_trie_iter iter;
2087 return;
2088 }
2089
2090 tn = (struct tnode *)n;
2091 pend = tn->pos+tn->bits;
2092 putspace_seq(seq, indent); seq_printf(seq, "\\--\n");
2093 indent += 3;
2094 depth++;
2095
2096 while (tn && cindex < (1 << tn->bits)) {
2097 struct node *child = rcu_dereference(tn->child[cindex]);
2098 if (!child)
2099 cindex++;
2100 else {
2101 /* Got a child */
2102 printnode_seq(seq, indent, child, pend,
2103 cindex, tn->bits);
2104
2105 if (IS_LEAF(child))
2106 cindex++;
2107
2108 else {
2109 /*
2110 * New tnode. Decend one level
2111 */
2112
2113 depth++;
2114 n = child;
2115 tn = (struct tnode *)n;
2116 pend = tn->pos+tn->bits;
2117 putspace_seq(seq, indent);
2118 seq_printf(seq, "\\--\n");
2119 indent += 3;
2120 cindex = 0;
2121 }
2122 }
2123
2124 /*
2125 * Test if we are done
2126 */
2127
2128 while (cindex >= (1 << tn->bits)) {
2129 /*
2130 * Move upwards and test for root
2131 * pop off all traversed nodes
2132 */
2133 2042
2134 if (NODE_PARENT(tn) == NULL) { 2043 memset(s, 0, sizeof(*s));
2135 tn = NULL;
2136 break;
2137 }
2138 2044
2139 cindex = tkey_extract_bits(tn->key, NODE_PARENT(tn)->pos, NODE_PARENT(tn)->bits); 2045 rcu_read_lock();
2140 cindex++; 2046 for (n = fib_trie_get_first(&iter, t); n;
2141 tn = NODE_PARENT(tn); 2047 n = fib_trie_get_next(&iter)) {
2142 pend = tn->pos + tn->bits; 2048 if (IS_LEAF(n)) {
2143 indent -= 3; 2049 s->leaves++;
2144 depth--; 2050 s->totdepth += iter.depth;
2051 if (iter.depth > s->maxdepth)
2052 s->maxdepth = iter.depth;
2053 } else {
2054 const struct tnode *tn = (const struct tnode *) n;
2055 int i;
2056
2057 s->tnodes++;
2058 s->nodesizes[tn->bits]++;
2059 for (i = 0; i < (1<<tn->bits); i++)
2060 if (!tn->child[i])
2061 s->nullpointers++;
2145 } 2062 }
2146 } 2063 }
2147 rcu_read_unlock(); 2064 rcu_read_unlock();
2148} 2065}
2149 2066
2150static struct trie_stat *trie_stat_new(void) 2067/*
2068 * This outputs /proc/net/fib_triestats
2069 */
2070static void trie_show_stats(struct seq_file *seq, struct trie_stat *stat)
2151{ 2071{
2152 struct trie_stat *s; 2072 unsigned i, max, pointers, bytes, avdepth;
2153 int i;
2154 2073
2155 s = kmalloc(sizeof(struct trie_stat), GFP_KERNEL); 2074 if (stat->leaves)
2156 if (!s) 2075 avdepth = stat->totdepth*100 / stat->leaves;
2157 return NULL; 2076 else
2077 avdepth = 0;
2158 2078
2159 s->totdepth = 0; 2079 seq_printf(seq, "\tAver depth: %d.%02d\n", avdepth / 100, avdepth % 100 );
2160 s->maxdepth = 0; 2080 seq_printf(seq, "\tMax depth: %u\n", stat->maxdepth);
2161 s->tnodes = 0;
2162 s->leaves = 0;
2163 s->nullpointers = 0;
2164 2081
2165 for (i = 0; i < MAX_CHILDS; i++) 2082 seq_printf(seq, "\tLeaves: %u\n", stat->leaves);
2166 s->nodesizes[i] = 0;
2167 2083
2168 return s; 2084 bytes = sizeof(struct leaf) * stat->leaves;
2169} 2085 seq_printf(seq, "\tInternal nodes: %d\n\t", stat->tnodes);
2086 bytes += sizeof(struct tnode) * stat->tnodes;
2170 2087
2171static struct trie_stat *trie_collect_stats(struct trie *t) 2088 max = MAX_CHILDS-1;
2172{ 2089 while (max >= 0 && stat->nodesizes[max] == 0)
2173 struct node *n; 2090 max--;
2174 struct trie_stat *s = trie_stat_new();
2175 int cindex = 0;
2176 int pend = 0;
2177 int depth = 0;
2178 2091
2179 if (!s) 2092 pointers = 0;
2180 return NULL; 2093 for (i = 1; i <= max; i++)
2094 if (stat->nodesizes[i] != 0) {
2095 seq_printf(seq, " %d: %d", i, stat->nodesizes[i]);
2096 pointers += (1<<i) * stat->nodesizes[i];
2097 }
2098 seq_putc(seq, '\n');
2099 seq_printf(seq, "\tPointers: %d\n", pointers);
2181 2100
2182 rcu_read_lock(); 2101 bytes += sizeof(struct node *) * pointers;
2183 n = rcu_dereference(t->trie); 2102 seq_printf(seq, "Null ptrs: %d\n", stat->nullpointers);
2103 seq_printf(seq, "Total size: %d kB\n", (bytes + 1023) / 1024);
2184 2104
2185 if (!n) 2105#ifdef CONFIG_IP_FIB_TRIE_STATS
2186 return s; 2106 seq_printf(seq, "Counters:\n---------\n");
2107 seq_printf(seq,"gets = %d\n", t->stats.gets);
2108 seq_printf(seq,"backtracks = %d\n", t->stats.backtrack);
2109 seq_printf(seq,"semantic match passed = %d\n", t->stats.semantic_match_passed);
2110 seq_printf(seq,"semantic match miss = %d\n", t->stats.semantic_match_miss);
2111 seq_printf(seq,"null node hit= %d\n", t->stats.null_node_hit);
2112 seq_printf(seq,"skipped node resize = %d\n", t->stats.resize_node_skipped);
2113#ifdef CLEAR_STATS
2114 memset(&(t->stats), 0, sizeof(t->stats));
2115#endif
2116#endif /* CONFIG_IP_FIB_TRIE_STATS */
2117}
2187 2118
2188 if (IS_TNODE(n)) { 2119static int fib_triestat_seq_show(struct seq_file *seq, void *v)
2189 struct tnode *tn = (struct tnode *)n; 2120{
2190 pend = tn->pos+tn->bits; 2121 struct trie_stat *stat;
2191 s->nodesizes[tn->bits]++;
2192 depth++;
2193
2194 while (tn && cindex < (1 << tn->bits)) {
2195 struct node *ch = rcu_dereference(tn->child[cindex]);
2196 if (ch) {
2197
2198 /* Got a child */
2199
2200 if (IS_LEAF(tn->child[cindex])) {
2201 cindex++;
2202
2203 /* stats */
2204 if (depth > s->maxdepth)
2205 s->maxdepth = depth;
2206 s->totdepth += depth;
2207 s->leaves++;
2208 } else {
2209 /*
2210 * New tnode. Decend one level
2211 */
2212
2213 s->tnodes++;
2214 s->nodesizes[tn->bits]++;
2215 depth++;
2216
2217 n = ch;
2218 tn = (struct tnode *)n;
2219 pend = tn->pos+tn->bits;
2220
2221 cindex = 0;
2222 }
2223 } else {
2224 cindex++;
2225 s->nullpointers++;
2226 }
2227 2122
2228 /* 2123 stat = kmalloc(sizeof(*stat), GFP_KERNEL);
2229 * Test if we are done 2124 if (!stat)
2230 */ 2125 return -ENOMEM;
2231 2126
2232 while (cindex >= (1 << tn->bits)) { 2127 seq_printf(seq, "Basic info: size of leaf: %Zd bytes, size of tnode: %Zd bytes.\n",
2233 /* 2128 sizeof(struct leaf), sizeof(struct tnode));
2234 * Move upwards and test for root
2235 * pop off all traversed nodes
2236 */
2237 2129
2238 if (NODE_PARENT(tn) == NULL) { 2130 if (trie_local) {
2239 tn = NULL; 2131 seq_printf(seq, "Local:\n");
2240 n = NULL; 2132 trie_collect_stats(trie_local, stat);
2241 break; 2133 trie_show_stats(seq, stat);
2242 } 2134 }
2243 2135
2244 cindex = tkey_extract_bits(tn->key, NODE_PARENT(tn)->pos, NODE_PARENT(tn)->bits); 2136 if (trie_main) {
2245 tn = NODE_PARENT(tn); 2137 seq_printf(seq, "Main:\n");
2246 cindex++; 2138 trie_collect_stats(trie_main, stat);
2247 n = (struct node *)tn; 2139 trie_show_stats(seq, stat);
2248 pend = tn->pos+tn->bits;
2249 depth--;
2250 }
2251 }
2252 } 2140 }
2141 kfree(stat);
2253 2142
2254 rcu_read_unlock(); 2143 return 0;
2255 return s;
2256} 2144}
2257 2145
2258#ifdef CONFIG_PROC_FS 2146static int fib_triestat_seq_open(struct inode *inode, struct file *file)
2259
2260static struct fib_alias *fib_triestat_get_first(struct seq_file *seq)
2261{ 2147{
2262 return NULL; 2148 return single_open(file, fib_triestat_seq_show, NULL);
2263} 2149}
2264 2150
2265static struct fib_alias *fib_triestat_get_next(struct seq_file *seq) 2151static struct file_operations fib_triestat_fops = {
2152 .owner = THIS_MODULE,
2153 .open = fib_triestat_seq_open,
2154 .read = seq_read,
2155 .llseek = seq_lseek,
2156 .release = single_release,
2157};
2158
2159static struct node *fib_trie_get_idx(struct fib_trie_iter *iter,
2160 loff_t pos)
2266{ 2161{
2162 loff_t idx = 0;
2163 struct node *n;
2164
2165 for (n = fib_trie_get_first(iter, trie_local);
2166 n; ++idx, n = fib_trie_get_next(iter)) {
2167 if (pos == idx)
2168 return n;
2169 }
2170
2171 for (n = fib_trie_get_first(iter, trie_main);
2172 n; ++idx, n = fib_trie_get_next(iter)) {
2173 if (pos == idx)
2174 return n;
2175 }
2267 return NULL; 2176 return NULL;
2268} 2177}
2269 2178
2270static void *fib_triestat_seq_start(struct seq_file *seq, loff_t *pos) 2179static void *fib_trie_seq_start(struct seq_file *seq, loff_t *pos)
2271{ 2180{
2272 if (!ip_fib_main_table) 2181 rcu_read_lock();
2273 return NULL; 2182 if (*pos == 0)
2274
2275 if (*pos)
2276 return fib_triestat_get_next(seq);
2277 else
2278 return SEQ_START_TOKEN; 2183 return SEQ_START_TOKEN;
2184 return fib_trie_get_idx(seq->private, *pos - 1);
2279} 2185}
2280 2186
2281static void *fib_triestat_seq_next(struct seq_file *seq, void *v, loff_t *pos) 2187static void *fib_trie_seq_next(struct seq_file *seq, void *v, loff_t *pos)
2282{ 2188{
2189 struct fib_trie_iter *iter = seq->private;
2190 void *l = v;
2191
2283 ++*pos; 2192 ++*pos;
2284 if (v == SEQ_START_TOKEN) 2193 if (v == SEQ_START_TOKEN)
2285 return fib_triestat_get_first(seq); 2194 return fib_trie_get_idx(iter, 0);
2286 else
2287 return fib_triestat_get_next(seq);
2288}
2289 2195
2290static void fib_triestat_seq_stop(struct seq_file *seq, void *v) 2196 v = fib_trie_get_next(iter);
2291{ 2197 BUG_ON(v == l);
2198 if (v)
2199 return v;
2292 2200
2293} 2201 /* continue scan in next trie */
2202 if (iter->trie == trie_local)
2203 return fib_trie_get_first(iter, trie_main);
2294 2204
2295/* 2205 return NULL;
2296 * This outputs /proc/net/fib_triestats 2206}
2297 *
2298 * It always works in backward compatibility mode.
2299 * The format of the file is not supposed to be changed.
2300 */
2301 2207
2302static void collect_and_show(struct trie *t, struct seq_file *seq) 2208static void fib_trie_seq_stop(struct seq_file *seq, void *v)
2303{ 2209{
2304 int bytes = 0; /* How many bytes are used, a ref is 4 bytes */ 2210 rcu_read_unlock();
2305 int i, max, pointers; 2211}
2306 struct trie_stat *stat;
2307 int avdepth;
2308
2309 stat = trie_collect_stats(t);
2310
2311 bytes = 0;
2312 seq_printf(seq, "trie=%p\n", t);
2313
2314 if (stat) {
2315 if (stat->leaves)
2316 avdepth = stat->totdepth*100 / stat->leaves;
2317 else
2318 avdepth = 0;
2319 seq_printf(seq, "Aver depth: %d.%02d\n", avdepth / 100, avdepth % 100);
2320 seq_printf(seq, "Max depth: %4d\n", stat->maxdepth);
2321 2212
2322 seq_printf(seq, "Leaves: %d\n", stat->leaves); 2213static void seq_indent(struct seq_file *seq, int n)
2323 bytes += sizeof(struct leaf) * stat->leaves; 2214{
2324 seq_printf(seq, "Internal nodes: %d\n", stat->tnodes); 2215 while (n-- > 0) seq_puts(seq, " ");
2325 bytes += sizeof(struct tnode) * stat->tnodes; 2216}
2326 2217
2327 max = MAX_CHILDS-1; 2218static inline const char *rtn_scope(enum rt_scope_t s)
2219{
2220 static char buf[32];
2328 2221
2329 while (max >= 0 && stat->nodesizes[max] == 0) 2222 switch(s) {
2330 max--; 2223 case RT_SCOPE_UNIVERSE: return "universe";
2331 pointers = 0; 2224 case RT_SCOPE_SITE: return "site";
2225 case RT_SCOPE_LINK: return "link";
2226 case RT_SCOPE_HOST: return "host";
2227 case RT_SCOPE_NOWHERE: return "nowhere";
2228 default:
2229 snprintf(buf, sizeof(buf), "scope=%d", s);
2230 return buf;
2231 }
2232}
2332 2233
2333 for (i = 1; i <= max; i++) 2234static const char *rtn_type_names[__RTN_MAX] = {
2334 if (stat->nodesizes[i] != 0) { 2235 [RTN_UNSPEC] = "UNSPEC",
2335 seq_printf(seq, " %d: %d", i, stat->nodesizes[i]); 2236 [RTN_UNICAST] = "UNICAST",
2336 pointers += (1<<i) * stat->nodesizes[i]; 2237 [RTN_LOCAL] = "LOCAL",
2337 } 2238 [RTN_BROADCAST] = "BROADCAST",
2338 seq_printf(seq, "\n"); 2239 [RTN_ANYCAST] = "ANYCAST",
2339 seq_printf(seq, "Pointers: %d\n", pointers); 2240 [RTN_MULTICAST] = "MULTICAST",
2340 bytes += sizeof(struct node *) * pointers; 2241 [RTN_BLACKHOLE] = "BLACKHOLE",
2341 seq_printf(seq, "Null ptrs: %d\n", stat->nullpointers); 2242 [RTN_UNREACHABLE] = "UNREACHABLE",
2342 seq_printf(seq, "Total size: %d kB\n", bytes / 1024); 2243 [RTN_PROHIBIT] = "PROHIBIT",
2244 [RTN_THROW] = "THROW",
2245 [RTN_NAT] = "NAT",
2246 [RTN_XRESOLVE] = "XRESOLVE",
2247};
2343 2248
2344 kfree(stat); 2249static inline const char *rtn_type(unsigned t)
2345 } 2250{
2251 static char buf[32];
2346 2252
2347#ifdef CONFIG_IP_FIB_TRIE_STATS 2253 if (t < __RTN_MAX && rtn_type_names[t])
2348 seq_printf(seq, "Counters:\n---------\n"); 2254 return rtn_type_names[t];
2349 seq_printf(seq,"gets = %d\n", t->stats.gets); 2255 snprintf(buf, sizeof(buf), "type %d", t);
2350 seq_printf(seq,"backtracks = %d\n", t->stats.backtrack); 2256 return buf;
2351 seq_printf(seq,"semantic match passed = %d\n", t->stats.semantic_match_passed);
2352 seq_printf(seq,"semantic match miss = %d\n", t->stats.semantic_match_miss);
2353 seq_printf(seq,"null node hit= %d\n", t->stats.null_node_hit);
2354 seq_printf(seq,"skipped node resize = %d\n", t->stats.resize_node_skipped);
2355#ifdef CLEAR_STATS
2356 memset(&(t->stats), 0, sizeof(t->stats));
2357#endif
2358#endif /* CONFIG_IP_FIB_TRIE_STATS */
2359} 2257}
2360 2258
2361static int fib_triestat_seq_show(struct seq_file *seq, void *v) 2259/* Pretty print the trie */
2260static int fib_trie_seq_show(struct seq_file *seq, void *v)
2362{ 2261{
2363 char bf[128]; 2262 const struct fib_trie_iter *iter = seq->private;
2263 struct node *n = v;
2364 2264
2365 if (v == SEQ_START_TOKEN) { 2265 if (v == SEQ_START_TOKEN)
2366 seq_printf(seq, "Basic info: size of leaf: %Zd bytes, size of tnode: %Zd bytes.\n", 2266 return 0;
2367 sizeof(struct leaf), sizeof(struct tnode));
2368 if (trie_local)
2369 collect_and_show(trie_local, seq);
2370 2267
2371 if (trie_main) 2268 if (IS_TNODE(n)) {
2372 collect_and_show(trie_main, seq); 2269 struct tnode *tn = (struct tnode *) n;
2373 } else { 2270 t_key prf = ntohl(MASK_PFX(tn->key, tn->pos));
2374 snprintf(bf, sizeof(bf), "*\t%08X\t%08X", 200, 400);
2375 2271
2376 seq_printf(seq, "%-127s\n", bf); 2272 if (!NODE_PARENT(n)) {
2273 if (iter->trie == trie_local)
2274 seq_puts(seq, "<local>:\n");
2275 else
2276 seq_puts(seq, "<main>:\n");
2277 } else {
2278 seq_indent(seq, iter->depth-1);
2279 seq_printf(seq, " +-- %d.%d.%d.%d/%d\n",
2280 NIPQUAD(prf), tn->pos);
2281 }
2282 } else {
2283 struct leaf *l = (struct leaf *) n;
2284 int i;
2285 u32 val = ntohl(l->key);
2286
2287 seq_indent(seq, iter->depth);
2288 seq_printf(seq, " |-- %d.%d.%d.%d\n", NIPQUAD(val));
2289 for (i = 32; i >= 0; i--) {
2290 struct leaf_info *li = find_leaf_info(&l->list, i);
2291 if (li) {
2292 struct fib_alias *fa;
2293 list_for_each_entry_rcu(fa, &li->falh, fa_list) {
2294 seq_indent(seq, iter->depth+1);
2295 seq_printf(seq, " /%d %s %s", i,
2296 rtn_scope(fa->fa_scope),
2297 rtn_type(fa->fa_type));
2298 if (fa->fa_tos)
2299 seq_printf(seq, "tos =%d\n",
2300 fa->fa_tos);
2301 seq_putc(seq, '\n');
2302 }
2303 }
2304 }
2377 } 2305 }
2306
2378 return 0; 2307 return 0;
2379} 2308}
2380 2309
2381static struct seq_operations fib_triestat_seq_ops = { 2310static struct seq_operations fib_trie_seq_ops = {
2382 .start = fib_triestat_seq_start, 2311 .start = fib_trie_seq_start,
2383 .next = fib_triestat_seq_next, 2312 .next = fib_trie_seq_next,
2384 .stop = fib_triestat_seq_stop, 2313 .stop = fib_trie_seq_stop,
2385 .show = fib_triestat_seq_show, 2314 .show = fib_trie_seq_show,
2386}; 2315};
2387 2316
2388static int fib_triestat_seq_open(struct inode *inode, struct file *file) 2317static int fib_trie_seq_open(struct inode *inode, struct file *file)
2389{ 2318{
2390 struct seq_file *seq; 2319 struct seq_file *seq;
2391 int rc = -ENOMEM; 2320 int rc = -ENOMEM;
2321 struct fib_trie_iter *s = kmalloc(sizeof(*s), GFP_KERNEL);
2392 2322
2393 rc = seq_open(file, &fib_triestat_seq_ops); 2323 if (!s)
2324 goto out;
2325
2326 rc = seq_open(file, &fib_trie_seq_ops);
2394 if (rc) 2327 if (rc)
2395 goto out_kfree; 2328 goto out_kfree;
2396 2329
2397 seq = file->private_data; 2330 seq = file->private_data;
2331 seq->private = s;
2332 memset(s, 0, sizeof(*s));
2398out: 2333out:
2399 return rc; 2334 return rc;
2400out_kfree: 2335out_kfree:
2336 kfree(s);
2401 goto out; 2337 goto out;
2402} 2338}
2403 2339
2404static struct file_operations fib_triestat_seq_fops = { 2340static struct file_operations fib_trie_fops = {
2405 .owner = THIS_MODULE, 2341 .owner = THIS_MODULE,
2406 .open = fib_triestat_seq_open, 2342 .open = fib_trie_seq_open,
2407 .read = seq_read, 2343 .read = seq_read,
2408 .llseek = seq_lseek, 2344 .llseek = seq_lseek,
2409 .release = seq_release_private, 2345 .release = seq_release_private,
2410}; 2346};
2411 2347
2412int __init fib_stat_proc_init(void) 2348static unsigned fib_flag_trans(int type, u32 mask, const struct fib_info *fi)
2413{
2414 if (!proc_net_fops_create("fib_triestat", S_IRUGO, &fib_triestat_seq_fops))
2415 return -ENOMEM;
2416 return 0;
2417}
2418
2419void __init fib_stat_proc_exit(void)
2420{ 2349{
2421 proc_net_remove("fib_triestat"); 2350 static unsigned type2flags[RTN_MAX + 1] = {
2422} 2351 [7] = RTF_REJECT, [8] = RTF_REJECT,
2352 };
2353 unsigned flags = type2flags[type];
2423 2354
2424static struct fib_alias *fib_trie_get_first(struct seq_file *seq) 2355 if (fi && fi->fib_nh->nh_gw)
2425{ 2356 flags |= RTF_GATEWAY;
2426 return NULL; 2357 if (mask == 0xFFFFFFFF)
2358 flags |= RTF_HOST;
2359 flags |= RTF_UP;
2360 return flags;
2427} 2361}
2428 2362
2429static struct fib_alias *fib_trie_get_next(struct seq_file *seq) 2363/*
2364 * This outputs /proc/net/route.
2365 * The format of the file is not supposed to be changed
2366 * and needs to be same as fib_hash output to avoid breaking
2367 * legacy utilities
2368 */
2369static int fib_route_seq_show(struct seq_file *seq, void *v)
2430{ 2370{
2431 return NULL; 2371 struct leaf *l = v;
2432} 2372 int i;
2373 char bf[128];
2433 2374
2434static void *fib_trie_seq_start(struct seq_file *seq, loff_t *pos) 2375 if (v == SEQ_START_TOKEN) {
2435{ 2376 seq_printf(seq, "%-127s\n", "Iface\tDestination\tGateway "
2436 if (!ip_fib_main_table) 2377 "\tFlags\tRefCnt\tUse\tMetric\tMask\t\tMTU"
2437 return NULL; 2378 "\tWindow\tIRTT");
2379 return 0;
2380 }
2438 2381
2439 if (*pos) 2382 if (IS_TNODE(l))
2440 return fib_trie_get_next(seq); 2383 return 0;
2441 else
2442 return SEQ_START_TOKEN;
2443}
2444 2384
2445static void *fib_trie_seq_next(struct seq_file *seq, void *v, loff_t *pos) 2385 for (i=32; i>=0; i--) {
2446{ 2386 struct leaf_info *li = find_leaf_info(&l->list, i);
2447 ++*pos; 2387 struct fib_alias *fa;
2448 if (v == SEQ_START_TOKEN) 2388 u32 mask, prefix;
2449 return fib_trie_get_first(seq);
2450 else
2451 return fib_trie_get_next(seq);
2452 2389
2453} 2390 if (!li)
2391 continue;
2454 2392
2455static void fib_trie_seq_stop(struct seq_file *seq, void *v) 2393 mask = inet_make_mask(li->plen);
2456{ 2394 prefix = htonl(l->key);
2457}
2458 2395
2459/* 2396 list_for_each_entry_rcu(fa, &li->falh, fa_list) {
2460 * This outputs /proc/net/fib_trie. 2397 const struct fib_info *fi = rcu_dereference(fa->fa_info);
2461 * 2398 unsigned flags = fib_flag_trans(fa->fa_type, mask, fi);
2462 * It always works in backward compatibility mode.
2463 * The format of the file is not supposed to be changed.
2464 */
2465 2399
2466static int fib_trie_seq_show(struct seq_file *seq, void *v) 2400 if (fa->fa_type == RTN_BROADCAST
2467{ 2401 || fa->fa_type == RTN_MULTICAST)
2468 char bf[128]; 2402 continue;
2469 2403
2470 if (v == SEQ_START_TOKEN) { 2404 if (fi)
2471 if (trie_local) 2405 snprintf(bf, sizeof(bf),
2472 trie_dump_seq(seq, trie_local); 2406 "%s\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u",
2407 fi->fib_dev ? fi->fib_dev->name : "*",
2408 prefix,
2409 fi->fib_nh->nh_gw, flags, 0, 0,
2410 fi->fib_priority,
2411 mask,
2412 (fi->fib_advmss ? fi->fib_advmss + 40 : 0),
2413 fi->fib_window,
2414 fi->fib_rtt >> 3);
2415 else
2416 snprintf(bf, sizeof(bf),
2417 "*\t%08X\t%08X\t%04X\t%d\t%u\t%d\t%08X\t%d\t%u\t%u",
2418 prefix, 0, flags, 0, 0, 0,
2419 mask, 0, 0, 0);
2473 2420
2474 if (trie_main) 2421 seq_printf(seq, "%-127s\n", bf);
2475 trie_dump_seq(seq, trie_main); 2422 }
2476 } else {
2477 snprintf(bf, sizeof(bf),
2478 "*\t%08X\t%08X", 200, 400);
2479 seq_printf(seq, "%-127s\n", bf);
2480 } 2423 }
2481 2424
2482 return 0; 2425 return 0;
2483} 2426}
2484 2427
2485static struct seq_operations fib_trie_seq_ops = { 2428static struct seq_operations fib_route_seq_ops = {
2486 .start = fib_trie_seq_start, 2429 .start = fib_trie_seq_start,
2487 .next = fib_trie_seq_next, 2430 .next = fib_trie_seq_next,
2488 .stop = fib_trie_seq_stop, 2431 .stop = fib_trie_seq_stop,
2489 .show = fib_trie_seq_show, 2432 .show = fib_route_seq_show,
2490}; 2433};
2491 2434
2492static int fib_trie_seq_open(struct inode *inode, struct file *file) 2435static int fib_route_seq_open(struct inode *inode, struct file *file)
2493{ 2436{
2494 struct seq_file *seq; 2437 struct seq_file *seq;
2495 int rc = -ENOMEM; 2438 int rc = -ENOMEM;
2439 struct fib_trie_iter *s = kmalloc(sizeof(*s), GFP_KERNEL);
2496 2440
2497 rc = seq_open(file, &fib_trie_seq_ops); 2441 if (!s)
2442 goto out;
2443
2444 rc = seq_open(file, &fib_route_seq_ops);
2498 if (rc) 2445 if (rc)
2499 goto out_kfree; 2446 goto out_kfree;
2500 2447
2501 seq = file->private_data; 2448 seq = file->private_data;
2449 seq->private = s;
2450 memset(s, 0, sizeof(*s));
2502out: 2451out:
2503 return rc; 2452 return rc;
2504out_kfree: 2453out_kfree:
2454 kfree(s);
2505 goto out; 2455 goto out;
2506} 2456}
2507 2457
2508static struct file_operations fib_trie_seq_fops = { 2458static struct file_operations fib_route_fops = {
2509 .owner = THIS_MODULE, 2459 .owner = THIS_MODULE,
2510 .open = fib_trie_seq_open, 2460 .open = fib_route_seq_open,
2511 .read = seq_read, 2461 .read = seq_read,
2512 .llseek = seq_lseek, 2462 .llseek = seq_lseek,
2513 .release= seq_release_private, 2463 .release = seq_release_private,
2514}; 2464};
2515 2465
2516int __init fib_proc_init(void) 2466int __init fib_proc_init(void)
2517{ 2467{
2518 if (!proc_net_fops_create("fib_trie", S_IRUGO, &fib_trie_seq_fops)) 2468 if (!proc_net_fops_create("fib_trie", S_IRUGO, &fib_trie_fops))
2519 return -ENOMEM; 2469 goto out1;
2470
2471 if (!proc_net_fops_create("fib_triestat", S_IRUGO, &fib_triestat_fops))
2472 goto out2;
2473
2474 if (!proc_net_fops_create("route", S_IRUGO, &fib_route_fops))
2475 goto out3;
2476
2520 return 0; 2477 return 0;
2478
2479out3:
2480 proc_net_remove("fib_triestat");
2481out2:
2482 proc_net_remove("fib_trie");
2483out1:
2484 return -ENOMEM;
2521} 2485}
2522 2486
2523void __init fib_proc_exit(void) 2487void __init fib_proc_exit(void)
2524{ 2488{
2525 proc_net_remove("fib_trie"); 2489 proc_net_remove("fib_trie");
2490 proc_net_remove("fib_triestat");
2491 proc_net_remove("route");
2526} 2492}
2527 2493
2528#endif /* CONFIG_PROC_FS */ 2494#endif /* CONFIG_PROC_FS */
diff --git a/net/ipv4/inetpeer.c b/net/ipv4/inetpeer.c
index f84ba9c96551..2fc3fd38924f 100644
--- a/net/ipv4/inetpeer.c
+++ b/net/ipv4/inetpeer.c
@@ -100,8 +100,7 @@ DEFINE_SPINLOCK(inet_peer_unused_lock);
100#define PEER_MAX_CLEANUP_WORK 30 100#define PEER_MAX_CLEANUP_WORK 30
101 101
102static void peer_check_expire(unsigned long dummy); 102static void peer_check_expire(unsigned long dummy);
103static struct timer_list peer_periodic_timer = 103static DEFINE_TIMER(peer_periodic_timer, peer_check_expire, 0, 0);
104 TIMER_INITIALIZER(peer_check_expire, 0, 0);
105 104
106/* Exported for sysctl_net_ipv4. */ 105/* Exported for sysctl_net_ipv4. */
107int inet_peer_gc_mintime = 10 * HZ, 106int inet_peer_gc_mintime = 10 * HZ,
diff --git a/net/ipv4/ipconfig.c b/net/ipv4/ipconfig.c
index 953129d392d2..e8674baaa8d9 100644
--- a/net/ipv4/ipconfig.c
+++ b/net/ipv4/ipconfig.c
@@ -1103,10 +1103,8 @@ static int __init ic_dynamic(void)
1103#endif 1103#endif
1104 1104
1105 jiff = jiffies + (d->next ? CONF_INTER_TIMEOUT : timeout); 1105 jiff = jiffies + (d->next ? CONF_INTER_TIMEOUT : timeout);
1106 while (time_before(jiffies, jiff) && !ic_got_reply) { 1106 while (time_before(jiffies, jiff) && !ic_got_reply)
1107 set_current_state(TASK_UNINTERRUPTIBLE); 1107 schedule_timeout_uninterruptible(1);
1108 schedule_timeout(1);
1109 }
1110#ifdef IPCONFIG_DHCP 1108#ifdef IPCONFIG_DHCP
1111 /* DHCP isn't done until we get a DHCPACK. */ 1109 /* DHCP isn't done until we get a DHCPACK. */
1112 if ((ic_got_reply & IC_BOOTP) 1110 if ((ic_got_reply & IC_BOOTP)
diff --git a/net/ipv4/netfilter/ip_conntrack_netbios_ns.c b/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
index 2b5cf9c51309..bb7246683b74 100644
--- a/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
+++ b/net/ipv4/netfilter/ip_conntrack_netbios_ns.c
@@ -104,12 +104,28 @@ out:
104static struct ip_conntrack_helper helper = { 104static struct ip_conntrack_helper helper = {
105 .name = "netbios-ns", 105 .name = "netbios-ns",
106 .tuple = { 106 .tuple = {
107 .src.u.udp.port = __constant_htons(137), 107 .src = {
108 .dst.protonum = IPPROTO_UDP, 108 .u = {
109 .udp = {
110 .port = __constant_htons(137),
111 }
112 }
113 },
114 .dst = {
115 .protonum = IPPROTO_UDP,
116 },
109 }, 117 },
110 .mask = { 118 .mask = {
111 .src.u.udp.port = 0xFFFF, 119 .src = {
112 .dst.protonum = 0xFF, 120 .u = {
121 .udp = {
122 .port = 0xFFFF,
123 }
124 }
125 },
126 .dst = {
127 .protonum = 0xFF,
128 },
113 }, 129 },
114 .max_expected = 1, 130 .max_expected = 1,
115 .me = THIS_MODULE, 131 .me = THIS_MODULE,
diff --git a/net/ipv4/netfilter/ipt_REJECT.c b/net/ipv4/netfilter/ipt_REJECT.c
index f115a84a4ac6..f057025a719e 100644
--- a/net/ipv4/netfilter/ipt_REJECT.c
+++ b/net/ipv4/netfilter/ipt_REJECT.c
@@ -92,10 +92,7 @@ static inline struct rtable *route_reverse(struct sk_buff *skb,
92 fl.fl_ip_sport = tcph->dest; 92 fl.fl_ip_sport = tcph->dest;
93 fl.fl_ip_dport = tcph->source; 93 fl.fl_ip_dport = tcph->source;
94 94
95 if (xfrm_lookup((struct dst_entry **)&rt, &fl, NULL, 0)) { 95 xfrm_lookup((struct dst_entry **)&rt, &fl, NULL, 0);
96 dst_release(&rt->u.dst);
97 rt = NULL;
98 }
99 96
100 return rt; 97 return rt;
101} 98}
diff --git a/net/ipv4/netfilter/ipt_owner.c b/net/ipv4/netfilter/ipt_owner.c
index c1889f88262b..0cee2862ed85 100644
--- a/net/ipv4/netfilter/ipt_owner.c
+++ b/net/ipv4/netfilter/ipt_owner.c
@@ -11,6 +11,7 @@
11#include <linux/module.h> 11#include <linux/module.h>
12#include <linux/skbuff.h> 12#include <linux/skbuff.h>
13#include <linux/file.h> 13#include <linux/file.h>
14#include <linux/rcupdate.h>
14#include <net/sock.h> 15#include <net/sock.h>
15 16
16#include <linux/netfilter_ipv4/ipt_owner.h> 17#include <linux/netfilter_ipv4/ipt_owner.h>
diff --git a/net/ipv4/route.c b/net/ipv4/route.c
index 8c0b14e3beec..8549f26e2495 100644
--- a/net/ipv4/route.c
+++ b/net/ipv4/route.c
@@ -1760,6 +1760,7 @@ static inline int __mkroute_input(struct sk_buff *skb,
1760 goto cleanup; 1760 goto cleanup;
1761 } 1761 }
1762 1762
1763 atomic_set(&rth->u.dst.__refcnt, 1);
1763 rth->u.dst.flags= DST_HOST; 1764 rth->u.dst.flags= DST_HOST;
1764#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED 1765#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
1765 if (res->fi->fib_nhs > 1) 1766 if (res->fi->fib_nhs > 1)
@@ -1820,7 +1821,6 @@ static inline int ip_mkroute_input_def(struct sk_buff *skb,
1820 err = __mkroute_input(skb, res, in_dev, daddr, saddr, tos, &rth); 1821 err = __mkroute_input(skb, res, in_dev, daddr, saddr, tos, &rth);
1821 if (err) 1822 if (err)
1822 return err; 1823 return err;
1823 atomic_set(&rth->u.dst.__refcnt, 1);
1824 1824
1825 /* put it into the cache */ 1825 /* put it into the cache */
1826 hash = rt_hash_code(daddr, saddr ^ (fl->iif << 5), tos); 1826 hash = rt_hash_code(daddr, saddr ^ (fl->iif << 5), tos);
@@ -1834,8 +1834,8 @@ static inline int ip_mkroute_input(struct sk_buff *skb,
1834 u32 daddr, u32 saddr, u32 tos) 1834 u32 daddr, u32 saddr, u32 tos)
1835{ 1835{
1836#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED 1836#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
1837 struct rtable* rth = NULL; 1837 struct rtable* rth = NULL, *rtres;
1838 unsigned char hop, hopcount, lasthop; 1838 unsigned char hop, hopcount;
1839 int err = -EINVAL; 1839 int err = -EINVAL;
1840 unsigned int hash; 1840 unsigned int hash;
1841 1841
@@ -1844,8 +1844,6 @@ static inline int ip_mkroute_input(struct sk_buff *skb,
1844 else 1844 else
1845 hopcount = 1; 1845 hopcount = 1;
1846 1846
1847 lasthop = hopcount - 1;
1848
1849 /* distinguish between multipath and singlepath */ 1847 /* distinguish between multipath and singlepath */
1850 if (hopcount < 2) 1848 if (hopcount < 2)
1851 return ip_mkroute_input_def(skb, res, fl, in_dev, daddr, 1849 return ip_mkroute_input_def(skb, res, fl, in_dev, daddr,
@@ -1855,6 +1853,10 @@ static inline int ip_mkroute_input(struct sk_buff *skb,
1855 for (hop = 0; hop < hopcount; hop++) { 1853 for (hop = 0; hop < hopcount; hop++) {
1856 res->nh_sel = hop; 1854 res->nh_sel = hop;
1857 1855
1856 /* put reference to previous result */
1857 if (hop)
1858 ip_rt_put(rtres);
1859
1858 /* create a routing cache entry */ 1860 /* create a routing cache entry */
1859 err = __mkroute_input(skb, res, in_dev, daddr, saddr, tos, 1861 err = __mkroute_input(skb, res, in_dev, daddr, saddr, tos,
1860 &rth); 1862 &rth);
@@ -1863,7 +1865,7 @@ static inline int ip_mkroute_input(struct sk_buff *skb,
1863 1865
1864 /* put it into the cache */ 1866 /* put it into the cache */
1865 hash = rt_hash_code(daddr, saddr ^ (fl->iif << 5), tos); 1867 hash = rt_hash_code(daddr, saddr ^ (fl->iif << 5), tos);
1866 err = rt_intern_hash(hash, rth, (struct rtable**)&skb->dst); 1868 err = rt_intern_hash(hash, rth, &rtres);
1867 if (err) 1869 if (err)
1868 return err; 1870 return err;
1869 1871
@@ -1873,13 +1875,8 @@ static inline int ip_mkroute_input(struct sk_buff *skb,
1873 FIB_RES_NETMASK(*res), 1875 FIB_RES_NETMASK(*res),
1874 res->prefixlen, 1876 res->prefixlen,
1875 &FIB_RES_NH(*res)); 1877 &FIB_RES_NH(*res));
1876
1877 /* only for the last hop the reference count is handled
1878 * outside
1879 */
1880 if (hop == lasthop)
1881 atomic_set(&(skb->dst->__refcnt), 1);
1882 } 1878 }
1879 skb->dst = &rtres->u.dst;
1883 return err; 1880 return err;
1884#else /* CONFIG_IP_ROUTE_MULTIPATH_CACHED */ 1881#else /* CONFIG_IP_ROUTE_MULTIPATH_CACHED */
1885 return ip_mkroute_input_def(skb, res, fl, in_dev, daddr, saddr, tos); 1882 return ip_mkroute_input_def(skb, res, fl, in_dev, daddr, saddr, tos);
@@ -2208,6 +2205,7 @@ static inline int __mkroute_output(struct rtable **result,
2208 goto cleanup; 2205 goto cleanup;
2209 } 2206 }
2210 2207
2208 atomic_set(&rth->u.dst.__refcnt, 1);
2211 rth->u.dst.flags= DST_HOST; 2209 rth->u.dst.flags= DST_HOST;
2212#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED 2210#ifdef CONFIG_IP_ROUTE_MULTIPATH_CACHED
2213 if (res->fi) { 2211 if (res->fi) {
@@ -2290,8 +2288,6 @@ static inline int ip_mkroute_output_def(struct rtable **rp,
2290 if (err == 0) { 2288 if (err == 0) {
2291 u32 tos = RT_FL_TOS(oldflp); 2289 u32 tos = RT_FL_TOS(oldflp);
2292 2290
2293 atomic_set(&rth->u.dst.__refcnt, 1);
2294
2295 hash = rt_hash_code(oldflp->fl4_dst, 2291 hash = rt_hash_code(oldflp->fl4_dst,
2296 oldflp->fl4_src ^ (oldflp->oif << 5), tos); 2292 oldflp->fl4_src ^ (oldflp->oif << 5), tos);
2297 err = rt_intern_hash(hash, rth, rp); 2293 err = rt_intern_hash(hash, rth, rp);
@@ -2326,6 +2322,10 @@ static inline int ip_mkroute_output(struct rtable** rp,
2326 dev2nexthop = FIB_RES_DEV(*res); 2322 dev2nexthop = FIB_RES_DEV(*res);
2327 dev_hold(dev2nexthop); 2323 dev_hold(dev2nexthop);
2328 2324
2325 /* put reference to previous result */
2326 if (hop)
2327 ip_rt_put(*rp);
2328
2329 err = __mkroute_output(&rth, res, fl, oldflp, 2329 err = __mkroute_output(&rth, res, fl, oldflp,
2330 dev2nexthop, flags); 2330 dev2nexthop, flags);
2331 2331
@@ -2350,7 +2350,6 @@ static inline int ip_mkroute_output(struct rtable** rp,
2350 if (err != 0) 2350 if (err != 0)
2351 return err; 2351 return err;
2352 } 2352 }
2353 atomic_set(&(*rp)->u.dst.__refcnt, 1);
2354 return err; 2353 return err;
2355 } else { 2354 } else {
2356 return ip_mkroute_output_def(rp, res, fl, oldflp, dev_out, 2355 return ip_mkroute_output_def(rp, res, fl, oldflp, dev_out,
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 6094db5e11be..c10e4435e3b1 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -485,11 +485,6 @@ int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len, unsigned int mss
485 TCP_SKB_CB(buff)->when = TCP_SKB_CB(skb)->when; 485 TCP_SKB_CB(buff)->when = TCP_SKB_CB(skb)->when;
486 buff->tstamp = skb->tstamp; 486 buff->tstamp = skb->tstamp;
487 487
488 if (TCP_SKB_CB(skb)->sacked & TCPCB_LOST) {
489 tp->lost_out -= tcp_skb_pcount(skb);
490 tp->left_out -= tcp_skb_pcount(skb);
491 }
492
493 old_factor = tcp_skb_pcount(skb); 488 old_factor = tcp_skb_pcount(skb);
494 489
495 /* Fix up tso_factor for both original and new SKB. */ 490 /* Fix up tso_factor for both original and new SKB. */
@@ -499,7 +494,7 @@ int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len, unsigned int mss
499 /* If this packet has been sent out already, we must 494 /* If this packet has been sent out already, we must
500 * adjust the various packet counters. 495 * adjust the various packet counters.
501 */ 496 */
502 if (after(tp->snd_nxt, TCP_SKB_CB(buff)->end_seq)) { 497 if (!before(tp->snd_nxt, TCP_SKB_CB(buff)->end_seq)) {
503 int diff = old_factor - tcp_skb_pcount(skb) - 498 int diff = old_factor - tcp_skb_pcount(skb) -
504 tcp_skb_pcount(buff); 499 tcp_skb_pcount(buff);
505 500
diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
index e5beca7de86c..e0bd1013cb0d 100644
--- a/net/ipv4/udp.c
+++ b/net/ipv4/udp.c
@@ -1141,7 +1141,7 @@ int udp_rcv(struct sk_buff *skb)
1141 if (ulen > len || ulen < sizeof(*uh)) 1141 if (ulen > len || ulen < sizeof(*uh))
1142 goto short_packet; 1142 goto short_packet;
1143 1143
1144 if (pskb_trim(skb, ulen)) 1144 if (pskb_trim_rcsum(skb, ulen))
1145 goto short_packet; 1145 goto short_packet;
1146 1146
1147 if (udp_checksum_init(skb, uh, ulen, saddr, daddr) < 0) 1147 if (udp_checksum_init(skb, uh, ulen, saddr, daddr) < 0)
diff --git a/net/ipv6/addrconf.c b/net/ipv6/addrconf.c
index 6d6fb74f3b52..2fea3f4402a0 100644
--- a/net/ipv6/addrconf.c
+++ b/net/ipv6/addrconf.c
@@ -123,8 +123,7 @@ DEFINE_RWLOCK(addrconf_lock);
123 123
124static void addrconf_verify(unsigned long); 124static void addrconf_verify(unsigned long);
125 125
126static struct timer_list addr_chk_timer = 126static DEFINE_TIMER(addr_chk_timer, addrconf_verify, 0, 0);
127 TIMER_INITIALIZER(addrconf_verify, 0, 0);
128static DEFINE_SPINLOCK(addrconf_verify_lock); 127static DEFINE_SPINLOCK(addrconf_verify_lock);
129 128
130static void addrconf_join_anycast(struct inet6_ifaddr *ifp); 129static void addrconf_join_anycast(struct inet6_ifaddr *ifp);
diff --git a/net/ipv6/datagram.c b/net/ipv6/datagram.c
index 01468fab3d3d..cc518405b3e1 100644
--- a/net/ipv6/datagram.c
+++ b/net/ipv6/datagram.c
@@ -175,10 +175,8 @@ ipv4_connected:
175 if (final_p) 175 if (final_p)
176 ipv6_addr_copy(&fl.fl6_dst, final_p); 176 ipv6_addr_copy(&fl.fl6_dst, final_p);
177 177
178 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) { 178 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0)
179 dst_release(dst);
180 goto out; 179 goto out;
181 }
182 180
183 /* source address lookup done in ip6_dst_lookup */ 181 /* source address lookup done in ip6_dst_lookup */
184 182
@@ -390,32 +388,101 @@ int datagram_recv_ctl(struct sock *sk, struct msghdr *msg, struct sk_buff *skb)
390 put_cmsg(msg, SOL_IPV6, IPV6_HOPLIMIT, sizeof(hlim), &hlim); 388 put_cmsg(msg, SOL_IPV6, IPV6_HOPLIMIT, sizeof(hlim), &hlim);
391 } 389 }
392 390
391 if (np->rxopt.bits.rxtclass) {
392 int tclass = (ntohl(*(u32 *)skb->nh.ipv6h) >> 20) & 0xff;
393 put_cmsg(msg, SOL_IPV6, IPV6_TCLASS, sizeof(tclass), &tclass);
394 }
395
393 if (np->rxopt.bits.rxflow && (*(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK)) { 396 if (np->rxopt.bits.rxflow && (*(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK)) {
394 u32 flowinfo = *(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK; 397 u32 flowinfo = *(u32*)skb->nh.raw & IPV6_FLOWINFO_MASK;
395 put_cmsg(msg, SOL_IPV6, IPV6_FLOWINFO, sizeof(flowinfo), &flowinfo); 398 put_cmsg(msg, SOL_IPV6, IPV6_FLOWINFO, sizeof(flowinfo), &flowinfo);
396 } 399 }
400
401 /* HbH is allowed only once */
397 if (np->rxopt.bits.hopopts && opt->hop) { 402 if (np->rxopt.bits.hopopts && opt->hop) {
398 u8 *ptr = skb->nh.raw + opt->hop; 403 u8 *ptr = skb->nh.raw + opt->hop;
399 put_cmsg(msg, SOL_IPV6, IPV6_HOPOPTS, (ptr[1]+1)<<3, ptr); 404 put_cmsg(msg, SOL_IPV6, IPV6_HOPOPTS, (ptr[1]+1)<<3, ptr);
400 } 405 }
401 if (np->rxopt.bits.dstopts && opt->dst0) { 406
407 if (opt->lastopt &&
408 (np->rxopt.bits.dstopts || np->rxopt.bits.srcrt)) {
409 /*
410 * Silly enough, but we need to reparse in order to
411 * report extension headers (except for HbH)
412 * in order.
413 *
414 * Also note that IPV6_RECVRTHDRDSTOPTS is NOT
415 * (and WILL NOT be) defined because
416 * IPV6_RECVDSTOPTS is more generic. --yoshfuji
417 */
418 unsigned int off = sizeof(struct ipv6hdr);
419 u8 nexthdr = skb->nh.ipv6h->nexthdr;
420
421 while (off <= opt->lastopt) {
422 unsigned len;
423 u8 *ptr = skb->nh.raw + off;
424
425 switch(nexthdr) {
426 case IPPROTO_DSTOPTS:
427 nexthdr = ptr[0];
428 len = (ptr[1] + 1) << 3;
429 if (np->rxopt.bits.dstopts)
430 put_cmsg(msg, SOL_IPV6, IPV6_DSTOPTS, len, ptr);
431 break;
432 case IPPROTO_ROUTING:
433 nexthdr = ptr[0];
434 len = (ptr[1] + 1) << 3;
435 if (np->rxopt.bits.srcrt)
436 put_cmsg(msg, SOL_IPV6, IPV6_RTHDR, len, ptr);
437 break;
438 case IPPROTO_AH:
439 nexthdr = ptr[0];
440 len = (ptr[1] + 1) << 2;
441 break;
442 default:
443 nexthdr = ptr[0];
444 len = (ptr[1] + 1) << 3;
445 break;
446 }
447
448 off += len;
449 }
450 }
451
452 /* socket options in old style */
453 if (np->rxopt.bits.rxoinfo) {
454 struct in6_pktinfo src_info;
455
456 src_info.ipi6_ifindex = opt->iif;
457 ipv6_addr_copy(&src_info.ipi6_addr, &skb->nh.ipv6h->daddr);
458 put_cmsg(msg, SOL_IPV6, IPV6_2292PKTINFO, sizeof(src_info), &src_info);
459 }
460 if (np->rxopt.bits.rxohlim) {
461 int hlim = skb->nh.ipv6h->hop_limit;
462 put_cmsg(msg, SOL_IPV6, IPV6_2292HOPLIMIT, sizeof(hlim), &hlim);
463 }
464 if (np->rxopt.bits.ohopopts && opt->hop) {
465 u8 *ptr = skb->nh.raw + opt->hop;
466 put_cmsg(msg, SOL_IPV6, IPV6_2292HOPOPTS, (ptr[1]+1)<<3, ptr);
467 }
468 if (np->rxopt.bits.odstopts && opt->dst0) {
402 u8 *ptr = skb->nh.raw + opt->dst0; 469 u8 *ptr = skb->nh.raw + opt->dst0;
403 put_cmsg(msg, SOL_IPV6, IPV6_DSTOPTS, (ptr[1]+1)<<3, ptr); 470 put_cmsg(msg, SOL_IPV6, IPV6_2292DSTOPTS, (ptr[1]+1)<<3, ptr);
404 } 471 }
405 if (np->rxopt.bits.srcrt && opt->srcrt) { 472 if (np->rxopt.bits.osrcrt && opt->srcrt) {
406 struct ipv6_rt_hdr *rthdr = (struct ipv6_rt_hdr *)(skb->nh.raw + opt->srcrt); 473 struct ipv6_rt_hdr *rthdr = (struct ipv6_rt_hdr *)(skb->nh.raw + opt->srcrt);
407 put_cmsg(msg, SOL_IPV6, IPV6_RTHDR, (rthdr->hdrlen+1) << 3, rthdr); 474 put_cmsg(msg, SOL_IPV6, IPV6_2292RTHDR, (rthdr->hdrlen+1) << 3, rthdr);
408 } 475 }
409 if (np->rxopt.bits.dstopts && opt->dst1) { 476 if (np->rxopt.bits.odstopts && opt->dst1) {
410 u8 *ptr = skb->nh.raw + opt->dst1; 477 u8 *ptr = skb->nh.raw + opt->dst1;
411 put_cmsg(msg, SOL_IPV6, IPV6_DSTOPTS, (ptr[1]+1)<<3, ptr); 478 put_cmsg(msg, SOL_IPV6, IPV6_2292DSTOPTS, (ptr[1]+1)<<3, ptr);
412 } 479 }
413 return 0; 480 return 0;
414} 481}
415 482
416int datagram_send_ctl(struct msghdr *msg, struct flowi *fl, 483int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
417 struct ipv6_txoptions *opt, 484 struct ipv6_txoptions *opt,
418 int *hlimit) 485 int *hlimit, int *tclass)
419{ 486{
420 struct in6_pktinfo *src_info; 487 struct in6_pktinfo *src_info;
421 struct cmsghdr *cmsg; 488 struct cmsghdr *cmsg;
@@ -438,6 +505,7 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
438 505
439 switch (cmsg->cmsg_type) { 506 switch (cmsg->cmsg_type) {
440 case IPV6_PKTINFO: 507 case IPV6_PKTINFO:
508 case IPV6_2292PKTINFO:
441 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct in6_pktinfo))) { 509 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct in6_pktinfo))) {
442 err = -EINVAL; 510 err = -EINVAL;
443 goto exit_f; 511 goto exit_f;
@@ -492,6 +560,7 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
492 fl->fl6_flowlabel = IPV6_FLOWINFO_MASK & *(u32 *)CMSG_DATA(cmsg); 560 fl->fl6_flowlabel = IPV6_FLOWINFO_MASK & *(u32 *)CMSG_DATA(cmsg);
493 break; 561 break;
494 562
563 case IPV6_2292HOPOPTS:
495 case IPV6_HOPOPTS: 564 case IPV6_HOPOPTS:
496 if (opt->hopopt || cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) { 565 if (opt->hopopt || cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) {
497 err = -EINVAL; 566 err = -EINVAL;
@@ -512,7 +581,7 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
512 opt->hopopt = hdr; 581 opt->hopopt = hdr;
513 break; 582 break;
514 583
515 case IPV6_DSTOPTS: 584 case IPV6_2292DSTOPTS:
516 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) { 585 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) {
517 err = -EINVAL; 586 err = -EINVAL;
518 goto exit_f; 587 goto exit_f;
@@ -536,6 +605,33 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
536 opt->dst1opt = hdr; 605 opt->dst1opt = hdr;
537 break; 606 break;
538 607
608 case IPV6_DSTOPTS:
609 case IPV6_RTHDRDSTOPTS:
610 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_opt_hdr))) {
611 err = -EINVAL;
612 goto exit_f;
613 }
614
615 hdr = (struct ipv6_opt_hdr *)CMSG_DATA(cmsg);
616 len = ((hdr->hdrlen + 1) << 3);
617 if (cmsg->cmsg_len < CMSG_LEN(len)) {
618 err = -EINVAL;
619 goto exit_f;
620 }
621 if (!capable(CAP_NET_RAW)) {
622 err = -EPERM;
623 goto exit_f;
624 }
625 if (cmsg->cmsg_type == IPV6_DSTOPTS) {
626 opt->opt_flen += len;
627 opt->dst1opt = hdr;
628 } else {
629 opt->opt_nflen += len;
630 opt->dst0opt = hdr;
631 }
632 break;
633
634 case IPV6_2292RTHDR:
539 case IPV6_RTHDR: 635 case IPV6_RTHDR:
540 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_rt_hdr))) { 636 if (cmsg->cmsg_len < CMSG_LEN(sizeof(struct ipv6_rt_hdr))) {
541 err = -EINVAL; 637 err = -EINVAL;
@@ -568,7 +664,7 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
568 opt->opt_nflen += len; 664 opt->opt_nflen += len;
569 opt->srcrt = rthdr; 665 opt->srcrt = rthdr;
570 666
571 if (opt->dst1opt) { 667 if (cmsg->cmsg_type == IPV6_2292RTHDR && opt->dst1opt) {
572 int dsthdrlen = ((opt->dst1opt->hdrlen+1)<<3); 668 int dsthdrlen = ((opt->dst1opt->hdrlen+1)<<3);
573 669
574 opt->opt_nflen += dsthdrlen; 670 opt->opt_nflen += dsthdrlen;
@@ -579,6 +675,7 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
579 675
580 break; 676 break;
581 677
678 case IPV6_2292HOPLIMIT:
582 case IPV6_HOPLIMIT: 679 case IPV6_HOPLIMIT:
583 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) { 680 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) {
584 err = -EINVAL; 681 err = -EINVAL;
@@ -588,6 +685,24 @@ int datagram_send_ctl(struct msghdr *msg, struct flowi *fl,
588 *hlimit = *(int *)CMSG_DATA(cmsg); 685 *hlimit = *(int *)CMSG_DATA(cmsg);
589 break; 686 break;
590 687
688 case IPV6_TCLASS:
689 {
690 int tc;
691
692 err = -EINVAL;
693 if (cmsg->cmsg_len != CMSG_LEN(sizeof(int))) {
694 goto exit_f;
695 }
696
697 tc = *(int *)CMSG_DATA(cmsg);
698 if (tc < 0 || tc > 0xff)
699 goto exit_f;
700
701 err = 0;
702 *tclass = tc;
703
704 break;
705 }
591 default: 706 default:
592 LIMIT_NETDEBUG(KERN_DEBUG "invalid cmsg type: %d\n", 707 LIMIT_NETDEBUG(KERN_DEBUG "invalid cmsg type: %d\n",
593 cmsg->cmsg_type); 708 cmsg->cmsg_type);
diff --git a/net/ipv6/exthdrs.c b/net/ipv6/exthdrs.c
index 5be6da2584ee..922549581abc 100644
--- a/net/ipv6/exthdrs.c
+++ b/net/ipv6/exthdrs.c
@@ -164,6 +164,7 @@ static int ipv6_destopt_rcv(struct sk_buff **skbp, unsigned int *nhoffp)
164 return -1; 164 return -1;
165 } 165 }
166 166
167 opt->lastopt = skb->h.raw - skb->nh.raw;
167 opt->dst1 = skb->h.raw - skb->nh.raw; 168 opt->dst1 = skb->h.raw - skb->nh.raw;
168 169
169 if (ip6_parse_tlv(tlvprocdestopt_lst, skb)) { 170 if (ip6_parse_tlv(tlvprocdestopt_lst, skb)) {
@@ -243,6 +244,7 @@ static int ipv6_rthdr_rcv(struct sk_buff **skbp, unsigned int *nhoffp)
243 244
244looped_back: 245looped_back:
245 if (hdr->segments_left == 0) { 246 if (hdr->segments_left == 0) {
247 opt->lastopt = skb->h.raw - skb->nh.raw;
246 opt->srcrt = skb->h.raw - skb->nh.raw; 248 opt->srcrt = skb->h.raw - skb->nh.raw;
247 skb->h.raw += (hdr->hdrlen + 1) << 3; 249 skb->h.raw += (hdr->hdrlen + 1) << 3;
248 opt->dst0 = opt->dst1; 250 opt->dst0 = opt->dst1;
@@ -404,8 +406,7 @@ ipv6_invert_rthdr(struct sock *sk, struct ipv6_rt_hdr *hdr)
404 406
405 memcpy(opt->srcrt, hdr, sizeof(*hdr)); 407 memcpy(opt->srcrt, hdr, sizeof(*hdr));
406 irthdr = (struct rt0_hdr*)opt->srcrt; 408 irthdr = (struct rt0_hdr*)opt->srcrt;
407 /* Obsolete field, MBZ, when originated by us */ 409 irthdr->reserved = 0;
408 irthdr->bitmap = 0;
409 opt->srcrt->segments_left = n; 410 opt->srcrt->segments_left = n;
410 for (i=0; i<n; i++) 411 for (i=0; i<n; i++)
411 memcpy(irthdr->addr+i, rthdr->addr+(n-1-i), 16); 412 memcpy(irthdr->addr+i, rthdr->addr+(n-1-i), 16);
@@ -459,11 +460,10 @@ static int ipv6_hop_jumbo(struct sk_buff *skb, int optoff)
459 IP6_INC_STATS_BH(IPSTATS_MIB_INTRUNCATEDPKTS); 460 IP6_INC_STATS_BH(IPSTATS_MIB_INTRUNCATEDPKTS);
460 goto drop; 461 goto drop;
461 } 462 }
462 if (pkt_len + sizeof(struct ipv6hdr) < skb->len) { 463
463 __pskb_trim(skb, pkt_len + sizeof(struct ipv6hdr)); 464 if (pskb_trim_rcsum(skb, pkt_len + sizeof(struct ipv6hdr)))
464 if (skb->ip_summed == CHECKSUM_HW) 465 goto drop;
465 skb->ip_summed = CHECKSUM_NONE; 466
466 }
467 return 1; 467 return 1;
468 468
469drop: 469drop:
@@ -539,10 +539,15 @@ void ipv6_push_nfrag_opts(struct sk_buff *skb, struct ipv6_txoptions *opt,
539 u8 *proto, 539 u8 *proto,
540 struct in6_addr **daddr) 540 struct in6_addr **daddr)
541{ 541{
542 if (opt->srcrt) 542 if (opt->srcrt) {
543 ipv6_push_rthdr(skb, proto, opt->srcrt, daddr); 543 ipv6_push_rthdr(skb, proto, opt->srcrt, daddr);
544 if (opt->dst0opt) 544 /*
545 ipv6_push_exthdr(skb, proto, NEXTHDR_DEST, opt->dst0opt); 545 * IPV6_RTHDRDSTOPTS is ignored
546 * unless IPV6_RTHDR is set (RFC3542).
547 */
548 if (opt->dst0opt)
549 ipv6_push_exthdr(skb, proto, NEXTHDR_DEST, opt->dst0opt);
550 }
546 if (opt->hopopt) 551 if (opt->hopopt)
547 ipv6_push_exthdr(skb, proto, NEXTHDR_HOP, opt->hopopt); 552 ipv6_push_exthdr(skb, proto, NEXTHDR_HOP, opt->hopopt);
548} 553}
@@ -573,3 +578,97 @@ ipv6_dup_options(struct sock *sk, struct ipv6_txoptions *opt)
573 } 578 }
574 return opt2; 579 return opt2;
575} 580}
581
582static int ipv6_renew_option(void *ohdr,
583 struct ipv6_opt_hdr __user *newopt, int newoptlen,
584 int inherit,
585 struct ipv6_opt_hdr **hdr,
586 char **p)
587{
588 if (inherit) {
589 if (ohdr) {
590 memcpy(*p, ohdr, ipv6_optlen((struct ipv6_opt_hdr *)ohdr));
591 *hdr = (struct ipv6_opt_hdr *)*p;
592 *p += CMSG_ALIGN(ipv6_optlen(*(struct ipv6_opt_hdr **)hdr));
593 }
594 } else {
595 if (newopt) {
596 if (copy_from_user(*p, newopt, newoptlen))
597 return -EFAULT;
598 *hdr = (struct ipv6_opt_hdr *)*p;
599 if (ipv6_optlen(*(struct ipv6_opt_hdr **)hdr) > newoptlen)
600 return -EINVAL;
601 *p += CMSG_ALIGN(newoptlen);
602 }
603 }
604 return 0;
605}
606
607struct ipv6_txoptions *
608ipv6_renew_options(struct sock *sk, struct ipv6_txoptions *opt,
609 int newtype,
610 struct ipv6_opt_hdr __user *newopt, int newoptlen)
611{
612 int tot_len = 0;
613 char *p;
614 struct ipv6_txoptions *opt2;
615 int err;
616
617 if (newtype != IPV6_HOPOPTS && opt->hopopt)
618 tot_len += CMSG_ALIGN(ipv6_optlen(opt->hopopt));
619 if (newtype != IPV6_RTHDRDSTOPTS && opt->dst0opt)
620 tot_len += CMSG_ALIGN(ipv6_optlen(opt->dst0opt));
621 if (newtype != IPV6_RTHDR && opt->srcrt)
622 tot_len += CMSG_ALIGN(ipv6_optlen(opt->srcrt));
623 if (newtype != IPV6_DSTOPTS && opt->dst1opt)
624 tot_len += CMSG_ALIGN(ipv6_optlen(opt->dst1opt));
625 if (newopt && newoptlen)
626 tot_len += CMSG_ALIGN(newoptlen);
627
628 if (!tot_len)
629 return NULL;
630
631 opt2 = sock_kmalloc(sk, tot_len, GFP_ATOMIC);
632 if (!opt2)
633 return ERR_PTR(-ENOBUFS);
634
635 memset(opt2, 0, tot_len);
636
637 opt2->tot_len = tot_len;
638 p = (char *)(opt2 + 1);
639
640 err = ipv6_renew_option(opt->hopopt, newopt, newoptlen,
641 newtype != IPV6_HOPOPTS,
642 &opt2->hopopt, &p);
643 if (err)
644 goto out;
645
646 err = ipv6_renew_option(opt->dst0opt, newopt, newoptlen,
647 newtype != IPV6_RTHDRDSTOPTS,
648 &opt2->dst0opt, &p);
649 if (err)
650 goto out;
651
652 err = ipv6_renew_option(opt->srcrt, newopt, newoptlen,
653 newtype != IPV6_RTHDR,
654 (struct ipv6_opt_hdr **)opt2->srcrt, &p);
655 if (err)
656 goto out;
657
658 err = ipv6_renew_option(opt->dst1opt, newopt, newoptlen,
659 newtype != IPV6_DSTOPTS,
660 &opt2->dst1opt, &p);
661 if (err)
662 goto out;
663
664 opt2->opt_nflen = (opt2->hopopt ? ipv6_optlen(opt2->hopopt) : 0) +
665 (opt2->dst0opt ? ipv6_optlen(opt2->dst0opt) : 0) +
666 (opt2->srcrt ? ipv6_optlen(opt2->srcrt) : 0);
667 opt2->opt_flen = (opt2->dst1opt ? ipv6_optlen(opt2->dst1opt) : 0);
668
669 return opt2;
670out:
671 sock_kfree_s(sk, p, tot_len);
672 return ERR_PTR(err);
673}
674
diff --git a/net/ipv6/icmp.c b/net/ipv6/icmp.c
index fa8f1bb0aa52..b7185fb3377c 100644
--- a/net/ipv6/icmp.c
+++ b/net/ipv6/icmp.c
@@ -287,7 +287,7 @@ void icmpv6_send(struct sk_buff *skb, int type, int code, __u32 info,
287 int iif = 0; 287 int iif = 0;
288 int addr_type = 0; 288 int addr_type = 0;
289 int len; 289 int len;
290 int hlimit; 290 int hlimit, tclass;
291 int err = 0; 291 int err = 0;
292 292
293 if ((u8*)hdr < skb->head || (u8*)(hdr+1) > skb->tail) 293 if ((u8*)hdr < skb->head || (u8*)(hdr+1) > skb->tail)
@@ -374,7 +374,7 @@ void icmpv6_send(struct sk_buff *skb, int type, int code, __u32 info,
374 if (err) 374 if (err)
375 goto out; 375 goto out;
376 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) 376 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0)
377 goto out_dst_release; 377 goto out;
378 378
379 if (ipv6_addr_is_multicast(&fl.fl6_dst)) 379 if (ipv6_addr_is_multicast(&fl.fl6_dst))
380 hlimit = np->mcast_hops; 380 hlimit = np->mcast_hops;
@@ -385,6 +385,10 @@ void icmpv6_send(struct sk_buff *skb, int type, int code, __u32 info,
385 if (hlimit < 0) 385 if (hlimit < 0)
386 hlimit = ipv6_get_hoplimit(dst->dev); 386 hlimit = ipv6_get_hoplimit(dst->dev);
387 387
388 tclass = np->cork.tclass;
389 if (tclass < 0)
390 tclass = 0;
391
388 msg.skb = skb; 392 msg.skb = skb;
389 msg.offset = skb->nh.raw - skb->data; 393 msg.offset = skb->nh.raw - skb->data;
390 394
@@ -400,7 +404,7 @@ void icmpv6_send(struct sk_buff *skb, int type, int code, __u32 info,
400 err = ip6_append_data(sk, icmpv6_getfrag, &msg, 404 err = ip6_append_data(sk, icmpv6_getfrag, &msg,
401 len + sizeof(struct icmp6hdr), 405 len + sizeof(struct icmp6hdr),
402 sizeof(struct icmp6hdr), 406 sizeof(struct icmp6hdr),
403 hlimit, NULL, &fl, (struct rt6_info*)dst, 407 hlimit, tclass, NULL, &fl, (struct rt6_info*)dst,
404 MSG_DONTWAIT); 408 MSG_DONTWAIT);
405 if (err) { 409 if (err) {
406 ip6_flush_pending_frames(sk); 410 ip6_flush_pending_frames(sk);
@@ -434,6 +438,7 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
434 struct dst_entry *dst; 438 struct dst_entry *dst;
435 int err = 0; 439 int err = 0;
436 int hlimit; 440 int hlimit;
441 int tclass;
437 442
438 saddr = &skb->nh.ipv6h->daddr; 443 saddr = &skb->nh.ipv6h->daddr;
439 444
@@ -464,7 +469,7 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
464 if (err) 469 if (err)
465 goto out; 470 goto out;
466 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) 471 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0)
467 goto out_dst_release; 472 goto out;
468 473
469 if (ipv6_addr_is_multicast(&fl.fl6_dst)) 474 if (ipv6_addr_is_multicast(&fl.fl6_dst))
470 hlimit = np->mcast_hops; 475 hlimit = np->mcast_hops;
@@ -475,13 +480,17 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
475 if (hlimit < 0) 480 if (hlimit < 0)
476 hlimit = ipv6_get_hoplimit(dst->dev); 481 hlimit = ipv6_get_hoplimit(dst->dev);
477 482
483 tclass = np->cork.tclass;
484 if (tclass < 0)
485 tclass = 0;
486
478 idev = in6_dev_get(skb->dev); 487 idev = in6_dev_get(skb->dev);
479 488
480 msg.skb = skb; 489 msg.skb = skb;
481 msg.offset = 0; 490 msg.offset = 0;
482 491
483 err = ip6_append_data(sk, icmpv6_getfrag, &msg, skb->len + sizeof(struct icmp6hdr), 492 err = ip6_append_data(sk, icmpv6_getfrag, &msg, skb->len + sizeof(struct icmp6hdr),
484 sizeof(struct icmp6hdr), hlimit, NULL, &fl, 493 sizeof(struct icmp6hdr), hlimit, tclass, NULL, &fl,
485 (struct rt6_info*)dst, MSG_DONTWAIT); 494 (struct rt6_info*)dst, MSG_DONTWAIT);
486 495
487 if (err) { 496 if (err) {
@@ -496,7 +505,6 @@ static void icmpv6_echo_reply(struct sk_buff *skb)
496out_put: 505out_put:
497 if (likely(idev != NULL)) 506 if (likely(idev != NULL))
498 in6_dev_put(idev); 507 in6_dev_put(idev);
499out_dst_release:
500 dst_release(dst); 508 dst_release(dst);
501out: 509out:
502 icmpv6_xmit_unlock(); 510 icmpv6_xmit_unlock();
diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 16af874c9e8f..4fcc5a7acf6e 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -92,7 +92,7 @@ static struct fib6_node * fib6_repair_tree(struct fib6_node *fn);
92 92
93static __u32 rt_sernum; 93static __u32 rt_sernum;
94 94
95static struct timer_list ip6_fib_timer = TIMER_INITIALIZER(fib6_run_gc, 0, 0); 95static DEFINE_TIMER(ip6_fib_timer, fib6_run_gc, 0, 0);
96 96
97struct fib6_walker_t fib6_walker_list = { 97struct fib6_walker_t fib6_walker_list = {
98 .prev = &fib6_walker_list, 98 .prev = &fib6_walker_list,
diff --git a/net/ipv6/ip6_flowlabel.c b/net/ipv6/ip6_flowlabel.c
index b6c73da5ff35..f841bde30c18 100644
--- a/net/ipv6/ip6_flowlabel.c
+++ b/net/ipv6/ip6_flowlabel.c
@@ -50,7 +50,7 @@ static atomic_t fl_size = ATOMIC_INIT(0);
50static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1]; 50static struct ip6_flowlabel *fl_ht[FL_HASH_MASK+1];
51 51
52static void ip6_fl_gc(unsigned long dummy); 52static void ip6_fl_gc(unsigned long dummy);
53static struct timer_list ip6_fl_gc_timer = TIMER_INITIALIZER(ip6_fl_gc, 0, 0); 53static DEFINE_TIMER(ip6_fl_gc_timer, ip6_fl_gc, 0, 0);
54 54
55/* FL hash table lock: it protects only of GC */ 55/* FL hash table lock: it protects only of GC */
56 56
@@ -225,16 +225,20 @@ struct ipv6_txoptions *fl6_merge_options(struct ipv6_txoptions * opt_space,
225 struct ip6_flowlabel * fl, 225 struct ip6_flowlabel * fl,
226 struct ipv6_txoptions * fopt) 226 struct ipv6_txoptions * fopt)
227{ 227{
228 struct ipv6_txoptions * fl_opt = fl->opt; 228 struct ipv6_txoptions * fl_opt = fl ? fl->opt : NULL;
229 229
230 if (fopt == NULL || fopt->opt_flen == 0) 230 if (fopt == NULL || fopt->opt_flen == 0) {
231 return fl_opt; 231 if (!fl_opt || !fl_opt->dst0opt || fl_opt->srcrt)
232 return fl_opt;
233 }
232 234
233 if (fl_opt != NULL) { 235 if (fl_opt != NULL) {
234 opt_space->hopopt = fl_opt->hopopt; 236 opt_space->hopopt = fl_opt->hopopt;
235 opt_space->dst0opt = fl_opt->dst0opt; 237 opt_space->dst0opt = fl_opt->srcrt ? fl_opt->dst0opt : NULL;
236 opt_space->srcrt = fl_opt->srcrt; 238 opt_space->srcrt = fl_opt->srcrt;
237 opt_space->opt_nflen = fl_opt->opt_nflen; 239 opt_space->opt_nflen = fl_opt->opt_nflen;
240 if (fl_opt->dst0opt && !fl_opt->srcrt)
241 opt_space->opt_nflen -= ipv6_optlen(fl_opt->dst0opt);
238 } else { 242 } else {
239 if (fopt->opt_nflen == 0) 243 if (fopt->opt_nflen == 0)
240 return fopt; 244 return fopt;
@@ -310,7 +314,7 @@ fl_create(struct in6_flowlabel_req *freq, char __user *optval, int optlen, int *
310 msg.msg_control = (void*)(fl->opt+1); 314 msg.msg_control = (void*)(fl->opt+1);
311 flowi.oif = 0; 315 flowi.oif = 0;
312 316
313 err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk); 317 err = datagram_send_ctl(&msg, &flowi, fl->opt, &junk, &junk);
314 if (err) 318 if (err)
315 goto done; 319 goto done;
316 err = -EINVAL; 320 err = -EINVAL;
diff --git a/net/ipv6/ip6_output.c b/net/ipv6/ip6_output.c
index 01ef94f7c7f1..2f589f24c093 100644
--- a/net/ipv6/ip6_output.c
+++ b/net/ipv6/ip6_output.c
@@ -166,7 +166,7 @@ int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
166 struct ipv6hdr *hdr; 166 struct ipv6hdr *hdr;
167 u8 proto = fl->proto; 167 u8 proto = fl->proto;
168 int seg_len = skb->len; 168 int seg_len = skb->len;
169 int hlimit; 169 int hlimit, tclass;
170 u32 mtu; 170 u32 mtu;
171 171
172 if (opt) { 172 if (opt) {
@@ -202,7 +202,6 @@ int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
202 * Fill in the IPv6 header 202 * Fill in the IPv6 header
203 */ 203 */
204 204
205 *(u32*)hdr = htonl(0x60000000) | fl->fl6_flowlabel;
206 hlimit = -1; 205 hlimit = -1;
207 if (np) 206 if (np)
208 hlimit = np->hop_limit; 207 hlimit = np->hop_limit;
@@ -211,6 +210,14 @@ int ip6_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl,
211 if (hlimit < 0) 210 if (hlimit < 0)
212 hlimit = ipv6_get_hoplimit(dst->dev); 211 hlimit = ipv6_get_hoplimit(dst->dev);
213 212
213 tclass = -1;
214 if (np)
215 tclass = np->tclass;
216 if (tclass < 0)
217 tclass = 0;
218
219 *(u32 *)hdr = htonl(0x60000000 | (tclass << 20)) | fl->fl6_flowlabel;
220
214 hdr->payload_len = htons(seg_len); 221 hdr->payload_len = htons(seg_len);
215 hdr->nexthdr = proto; 222 hdr->nexthdr = proto;
216 hdr->hop_limit = hlimit; 223 hdr->hop_limit = hlimit;
@@ -762,10 +769,11 @@ out_err_release:
762 return err; 769 return err;
763} 770}
764 771
765int ip6_append_data(struct sock *sk, int getfrag(void *from, char *to, int offset, int len, int odd, struct sk_buff *skb), 772int ip6_append_data(struct sock *sk, int getfrag(void *from, char *to,
766 void *from, int length, int transhdrlen, 773 int offset, int len, int odd, struct sk_buff *skb),
767 int hlimit, struct ipv6_txoptions *opt, struct flowi *fl, struct rt6_info *rt, 774 void *from, int length, int transhdrlen,
768 unsigned int flags) 775 int hlimit, int tclass, struct ipv6_txoptions *opt, struct flowi *fl,
776 struct rt6_info *rt, unsigned int flags)
769{ 777{
770 struct inet_sock *inet = inet_sk(sk); 778 struct inet_sock *inet = inet_sk(sk);
771 struct ipv6_pinfo *np = inet6_sk(sk); 779 struct ipv6_pinfo *np = inet6_sk(sk);
@@ -803,6 +811,7 @@ int ip6_append_data(struct sock *sk, int getfrag(void *from, char *to, int offse
803 np->cork.rt = rt; 811 np->cork.rt = rt;
804 inet->cork.fl = *fl; 812 inet->cork.fl = *fl;
805 np->cork.hop_limit = hlimit; 813 np->cork.hop_limit = hlimit;
814 np->cork.tclass = tclass;
806 inet->cork.fragsize = mtu = dst_mtu(rt->u.dst.path); 815 inet->cork.fragsize = mtu = dst_mtu(rt->u.dst.path);
807 if (dst_allfrag(rt->u.dst.path)) 816 if (dst_allfrag(rt->u.dst.path))
808 inet->cork.flags |= IPCORK_ALLFRAG; 817 inet->cork.flags |= IPCORK_ALLFRAG;
@@ -1084,7 +1093,8 @@ int ip6_push_pending_frames(struct sock *sk)
1084 1093
1085 skb->nh.ipv6h = hdr = (struct ipv6hdr*) skb_push(skb, sizeof(struct ipv6hdr)); 1094 skb->nh.ipv6h = hdr = (struct ipv6hdr*) skb_push(skb, sizeof(struct ipv6hdr));
1086 1095
1087 *(u32*)hdr = fl->fl6_flowlabel | htonl(0x60000000); 1096 *(u32*)hdr = fl->fl6_flowlabel |
1097 htonl(0x60000000 | ((int)np->cork.tclass << 20));
1088 1098
1089 if (skb->len <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN) 1099 if (skb->len <= sizeof(struct ipv6hdr) + IPV6_MAXPLEN)
1090 hdr->payload_len = htons(skb->len - sizeof(struct ipv6hdr)); 1100 hdr->payload_len = htons(skb->len - sizeof(struct ipv6hdr));
diff --git a/net/ipv6/ip6_tunnel.c b/net/ipv6/ip6_tunnel.c
index 09613729404c..cf94372d1af3 100644
--- a/net/ipv6/ip6_tunnel.c
+++ b/net/ipv6/ip6_tunnel.c
@@ -673,11 +673,12 @@ ip6ip6_tnl_xmit(struct sk_buff *skb, struct net_device *dev)
673 673
674 if ((dst = ip6_tnl_dst_check(t)) != NULL) 674 if ((dst = ip6_tnl_dst_check(t)) != NULL)
675 dst_hold(dst); 675 dst_hold(dst);
676 else 676 else {
677 dst = ip6_route_output(NULL, &fl); 677 dst = ip6_route_output(NULL, &fl);
678 678
679 if (dst->error || xfrm_lookup(&dst, &fl, NULL, 0) < 0) 679 if (dst->error || xfrm_lookup(&dst, &fl, NULL, 0) < 0)
680 goto tx_err_link_failure; 680 goto tx_err_link_failure;
681 }
681 682
682 tdev = dst->dev; 683 tdev = dst->dev;
683 684
diff --git a/net/ipv6/ipv6_sockglue.c b/net/ipv6/ipv6_sockglue.c
index 76466af8331e..8567873d0dd8 100644
--- a/net/ipv6/ipv6_sockglue.c
+++ b/net/ipv6/ipv6_sockglue.c
@@ -210,39 +210,139 @@ int ipv6_setsockopt(struct sock *sk, int level, int optname,
210 retv = 0; 210 retv = 0;
211 break; 211 break;
212 212
213 case IPV6_PKTINFO: 213 case IPV6_RECVPKTINFO:
214 np->rxopt.bits.rxinfo = valbool; 214 np->rxopt.bits.rxinfo = valbool;
215 retv = 0; 215 retv = 0;
216 break; 216 break;
217
218 case IPV6_2292PKTINFO:
219 np->rxopt.bits.rxoinfo = valbool;
220 retv = 0;
221 break;
217 222
218 case IPV6_HOPLIMIT: 223 case IPV6_RECVHOPLIMIT:
219 np->rxopt.bits.rxhlim = valbool; 224 np->rxopt.bits.rxhlim = valbool;
220 retv = 0; 225 retv = 0;
221 break; 226 break;
222 227
223 case IPV6_RTHDR: 228 case IPV6_2292HOPLIMIT:
229 np->rxopt.bits.rxohlim = valbool;
230 retv = 0;
231 break;
232
233 case IPV6_RECVRTHDR:
224 if (val < 0 || val > 2) 234 if (val < 0 || val > 2)
225 goto e_inval; 235 goto e_inval;
226 np->rxopt.bits.srcrt = val; 236 np->rxopt.bits.srcrt = val;
227 retv = 0; 237 retv = 0;
228 break; 238 break;
229 239
230 case IPV6_HOPOPTS: 240 case IPV6_2292RTHDR:
241 if (val < 0 || val > 2)
242 goto e_inval;
243 np->rxopt.bits.osrcrt = val;
244 retv = 0;
245 break;
246
247 case IPV6_RECVHOPOPTS:
231 np->rxopt.bits.hopopts = valbool; 248 np->rxopt.bits.hopopts = valbool;
232 retv = 0; 249 retv = 0;
233 break; 250 break;
234 251
235 case IPV6_DSTOPTS: 252 case IPV6_2292HOPOPTS:
253 np->rxopt.bits.ohopopts = valbool;
254 retv = 0;
255 break;
256
257 case IPV6_RECVDSTOPTS:
236 np->rxopt.bits.dstopts = valbool; 258 np->rxopt.bits.dstopts = valbool;
237 retv = 0; 259 retv = 0;
238 break; 260 break;
239 261
262 case IPV6_2292DSTOPTS:
263 np->rxopt.bits.odstopts = valbool;
264 retv = 0;
265 break;
266
267 case IPV6_TCLASS:
268 if (val < 0 || val > 0xff)
269 goto e_inval;
270 np->tclass = val;
271 retv = 0;
272 break;
273
274 case IPV6_RECVTCLASS:
275 np->rxopt.bits.rxtclass = valbool;
276 retv = 0;
277 break;
278
240 case IPV6_FLOWINFO: 279 case IPV6_FLOWINFO:
241 np->rxopt.bits.rxflow = valbool; 280 np->rxopt.bits.rxflow = valbool;
242 retv = 0; 281 retv = 0;
243 break; 282 break;
244 283
245 case IPV6_PKTOPTIONS: 284 case IPV6_HOPOPTS:
285 case IPV6_RTHDRDSTOPTS:
286 case IPV6_RTHDR:
287 case IPV6_DSTOPTS:
288 {
289 struct ipv6_txoptions *opt;
290 if (optlen == 0)
291 optval = 0;
292
293 /* hop-by-hop / destination options are privileged option */
294 retv = -EPERM;
295 if (optname != IPV6_RTHDR && !capable(CAP_NET_RAW))
296 break;
297
298 retv = -EINVAL;
299 if (optlen & 0x7 || optlen > 8 * 255)
300 break;
301
302 opt = ipv6_renew_options(sk, np->opt, optname,
303 (struct ipv6_opt_hdr __user *)optval,
304 optlen);
305 if (IS_ERR(opt)) {
306 retv = PTR_ERR(opt);
307 break;
308 }
309
310 /* routing header option needs extra check */
311 if (optname == IPV6_RTHDR && opt->srcrt) {
312 struct ipv6_rt_hdr *rthdr = opt->srcrt;
313 if (rthdr->type)
314 goto sticky_done;
315 if ((rthdr->hdrlen & 1) ||
316 (rthdr->hdrlen >> 1) != rthdr->segments_left)
317 goto sticky_done;
318 }
319
320 retv = 0;
321 if (sk->sk_type == SOCK_STREAM) {
322 if (opt) {
323 struct tcp_sock *tp = tcp_sk(sk);
324 if (!((1 << sk->sk_state) &
325 (TCPF_LISTEN | TCPF_CLOSE))
326 && inet_sk(sk)->daddr != LOOPBACK4_IPV6) {
327 tp->ext_header_len = opt->opt_flen + opt->opt_nflen;
328 tcp_sync_mss(sk, tp->pmtu_cookie);
329 }
330 }
331 opt = xchg(&np->opt, opt);
332 sk_dst_reset(sk);
333 } else {
334 write_lock(&sk->sk_dst_lock);
335 opt = xchg(&np->opt, opt);
336 write_unlock(&sk->sk_dst_lock);
337 sk_dst_reset(sk);
338 }
339sticky_done:
340 if (opt)
341 sock_kfree_s(sk, opt, opt->tot_len);
342 break;
343 }
344
345 case IPV6_2292PKTOPTIONS:
246 { 346 {
247 struct ipv6_txoptions *opt = NULL; 347 struct ipv6_txoptions *opt = NULL;
248 struct msghdr msg; 348 struct msghdr msg;
@@ -276,7 +376,7 @@ int ipv6_setsockopt(struct sock *sk, int level, int optname,
276 msg.msg_controllen = optlen; 376 msg.msg_controllen = optlen;
277 msg.msg_control = (void*)(opt+1); 377 msg.msg_control = (void*)(opt+1);
278 378
279 retv = datagram_send_ctl(&msg, &fl, opt, &junk); 379 retv = datagram_send_ctl(&msg, &fl, opt, &junk, &junk);
280 if (retv) 380 if (retv)
281 goto done; 381 goto done;
282update: 382update:
@@ -529,6 +629,17 @@ e_inval:
529 return -EINVAL; 629 return -EINVAL;
530} 630}
531 631
632int ipv6_getsockopt_sticky(struct sock *sk, struct ipv6_opt_hdr *hdr,
633 char __user *optval, int len)
634{
635 if (!hdr)
636 return 0;
637 len = min_t(int, len, ipv6_optlen(hdr));
638 if (copy_to_user(optval, hdr, ipv6_optlen(hdr)))
639 return -EFAULT;
640 return len;
641}
642
532int ipv6_getsockopt(struct sock *sk, int level, int optname, 643int ipv6_getsockopt(struct sock *sk, int level, int optname,
533 char __user *optval, int __user *optlen) 644 char __user *optval, int __user *optlen)
534{ 645{
@@ -567,7 +678,7 @@ int ipv6_getsockopt(struct sock *sk, int level, int optname,
567 return err; 678 return err;
568 } 679 }
569 680
570 case IPV6_PKTOPTIONS: 681 case IPV6_2292PKTOPTIONS:
571 { 682 {
572 struct msghdr msg; 683 struct msghdr msg;
573 struct sk_buff *skb; 684 struct sk_buff *skb;
@@ -601,6 +712,16 @@ int ipv6_getsockopt(struct sock *sk, int level, int optname,
601 int hlim = np->mcast_hops; 712 int hlim = np->mcast_hops;
602 put_cmsg(&msg, SOL_IPV6, IPV6_HOPLIMIT, sizeof(hlim), &hlim); 713 put_cmsg(&msg, SOL_IPV6, IPV6_HOPLIMIT, sizeof(hlim), &hlim);
603 } 714 }
715 if (np->rxopt.bits.rxoinfo) {
716 struct in6_pktinfo src_info;
717 src_info.ipi6_ifindex = np->mcast_oif;
718 ipv6_addr_copy(&src_info.ipi6_addr, &np->daddr);
719 put_cmsg(&msg, SOL_IPV6, IPV6_2292PKTINFO, sizeof(src_info), &src_info);
720 }
721 if (np->rxopt.bits.rxohlim) {
722 int hlim = np->mcast_hops;
723 put_cmsg(&msg, SOL_IPV6, IPV6_2292HOPLIMIT, sizeof(hlim), &hlim);
724 }
604 } 725 }
605 len -= msg.msg_controllen; 726 len -= msg.msg_controllen;
606 return put_user(len, optlen); 727 return put_user(len, optlen);
@@ -625,26 +746,67 @@ int ipv6_getsockopt(struct sock *sk, int level, int optname,
625 val = np->ipv6only; 746 val = np->ipv6only;
626 break; 747 break;
627 748
628 case IPV6_PKTINFO: 749 case IPV6_RECVPKTINFO:
629 val = np->rxopt.bits.rxinfo; 750 val = np->rxopt.bits.rxinfo;
630 break; 751 break;
631 752
632 case IPV6_HOPLIMIT: 753 case IPV6_2292PKTINFO:
754 val = np->rxopt.bits.rxoinfo;
755 break;
756
757 case IPV6_RECVHOPLIMIT:
633 val = np->rxopt.bits.rxhlim; 758 val = np->rxopt.bits.rxhlim;
634 break; 759 break;
635 760
636 case IPV6_RTHDR: 761 case IPV6_2292HOPLIMIT:
762 val = np->rxopt.bits.rxohlim;
763 break;
764
765 case IPV6_RECVRTHDR:
637 val = np->rxopt.bits.srcrt; 766 val = np->rxopt.bits.srcrt;
638 break; 767 break;
639 768
769 case IPV6_2292RTHDR:
770 val = np->rxopt.bits.osrcrt;
771 break;
772
640 case IPV6_HOPOPTS: 773 case IPV6_HOPOPTS:
774 case IPV6_RTHDRDSTOPTS:
775 case IPV6_RTHDR:
776 case IPV6_DSTOPTS:
777 {
778
779 lock_sock(sk);
780 len = ipv6_getsockopt_sticky(sk, np->opt->hopopt,
781 optval, len);
782 release_sock(sk);
783 return put_user(len, optlen);
784 }
785
786 case IPV6_RECVHOPOPTS:
641 val = np->rxopt.bits.hopopts; 787 val = np->rxopt.bits.hopopts;
642 break; 788 break;
643 789
644 case IPV6_DSTOPTS: 790 case IPV6_2292HOPOPTS:
791 val = np->rxopt.bits.ohopopts;
792 break;
793
794 case IPV6_RECVDSTOPTS:
645 val = np->rxopt.bits.dstopts; 795 val = np->rxopt.bits.dstopts;
646 break; 796 break;
647 797
798 case IPV6_2292DSTOPTS:
799 val = np->rxopt.bits.odstopts;
800 break;
801
802 case IPV6_TCLASS:
803 val = np->tclass;
804 break;
805
806 case IPV6_RECVTCLASS:
807 val = np->rxopt.bits.rxtclass;
808 break;
809
648 case IPV6_FLOWINFO: 810 case IPV6_FLOWINFO:
649 val = np->rxopt.bits.rxflow; 811 val = np->rxopt.bits.rxflow;
650 break; 812 break;
diff --git a/net/ipv6/ndisc.c b/net/ipv6/ndisc.c
index a7eae30f4554..555a31347eda 100644
--- a/net/ipv6/ndisc.c
+++ b/net/ipv6/ndisc.c
@@ -447,10 +447,8 @@ static void ndisc_send_na(struct net_device *dev, struct neighbour *neigh,
447 return; 447 return;
448 448
449 err = xfrm_lookup(&dst, &fl, NULL, 0); 449 err = xfrm_lookup(&dst, &fl, NULL, 0);
450 if (err < 0) { 450 if (err < 0)
451 dst_release(dst);
452 return; 451 return;
453 }
454 452
455 if (inc_opt) { 453 if (inc_opt) {
456 if (dev->addr_len) 454 if (dev->addr_len)
@@ -539,10 +537,8 @@ void ndisc_send_ns(struct net_device *dev, struct neighbour *neigh,
539 return; 537 return;
540 538
541 err = xfrm_lookup(&dst, &fl, NULL, 0); 539 err = xfrm_lookup(&dst, &fl, NULL, 0);
542 if (err < 0) { 540 if (err < 0)
543 dst_release(dst);
544 return; 541 return;
545 }
546 542
547 len = sizeof(struct icmp6hdr) + sizeof(struct in6_addr); 543 len = sizeof(struct icmp6hdr) + sizeof(struct in6_addr);
548 send_llinfo = dev->addr_len && !ipv6_addr_any(saddr); 544 send_llinfo = dev->addr_len && !ipv6_addr_any(saddr);
@@ -616,10 +612,8 @@ void ndisc_send_rs(struct net_device *dev, struct in6_addr *saddr,
616 return; 612 return;
617 613
618 err = xfrm_lookup(&dst, &fl, NULL, 0); 614 err = xfrm_lookup(&dst, &fl, NULL, 0);
619 if (err < 0) { 615 if (err < 0)
620 dst_release(dst);
621 return; 616 return;
622 }
623 617
624 len = sizeof(struct icmp6hdr); 618 len = sizeof(struct icmp6hdr);
625 if (dev->addr_len) 619 if (dev->addr_len)
@@ -1353,10 +1347,8 @@ void ndisc_send_redirect(struct sk_buff *skb, struct neighbour *neigh,
1353 return; 1347 return;
1354 1348
1355 err = xfrm_lookup(&dst, &fl, NULL, 0); 1349 err = xfrm_lookup(&dst, &fl, NULL, 0);
1356 if (err) { 1350 if (err)
1357 dst_release(dst);
1358 return; 1351 return;
1359 }
1360 1352
1361 rt = (struct rt6_info *) dst; 1353 rt = (struct rt6_info *) dst;
1362 1354
diff --git a/net/ipv6/netfilter/ip6t_REJECT.c b/net/ipv6/netfilter/ip6t_REJECT.c
index 14316c3ebde4..b03e87adca93 100644
--- a/net/ipv6/netfilter/ip6t_REJECT.c
+++ b/net/ipv6/netfilter/ip6t_REJECT.c
@@ -100,11 +100,8 @@ static void send_reset(struct sk_buff *oldskb)
100 dst = ip6_route_output(NULL, &fl); 100 dst = ip6_route_output(NULL, &fl);
101 if (dst == NULL) 101 if (dst == NULL)
102 return; 102 return;
103 if (dst->error || 103 if (dst->error || xfrm_lookup(&dst, &fl, NULL, 0))
104 xfrm_lookup(&dst, &fl, NULL, 0)) {
105 dst_release(dst);
106 return; 104 return;
107 }
108 105
109 hh_len = (dst->dev->hard_header_len + 15)&~15; 106 hh_len = (dst->dev->hard_header_len + 15)&~15;
110 nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr) 107 nskb = alloc_skb(hh_len + 15 + dst->header_len + sizeof(struct ipv6hdr)
diff --git a/net/ipv6/netfilter/ip6t_owner.c b/net/ipv6/netfilter/ip6t_owner.c
index 9b91decbfddb..4de4cdad4b7d 100644
--- a/net/ipv6/netfilter/ip6t_owner.c
+++ b/net/ipv6/netfilter/ip6t_owner.c
@@ -11,6 +11,7 @@
11#include <linux/module.h> 11#include <linux/module.h>
12#include <linux/skbuff.h> 12#include <linux/skbuff.h>
13#include <linux/file.h> 13#include <linux/file.h>
14#include <linux/rcupdate.h>
14#include <net/sock.h> 15#include <net/sock.h>
15 16
16#include <linux/netfilter_ipv6/ip6t_owner.h> 17#include <linux/netfilter_ipv6/ip6t_owner.h>
diff --git a/net/ipv6/netfilter/ip6t_rt.c b/net/ipv6/netfilter/ip6t_rt.c
index a9526b773d28..2bb670037df3 100644
--- a/net/ipv6/netfilter/ip6t_rt.c
+++ b/net/ipv6/netfilter/ip6t_rt.c
@@ -161,8 +161,8 @@ match(const struct sk_buff *skb,
161 ((rtinfo->hdrlen == hdrlen) ^ 161 ((rtinfo->hdrlen == hdrlen) ^
162 !!(rtinfo->invflags & IP6T_RT_INV_LEN)))); 162 !!(rtinfo->invflags & IP6T_RT_INV_LEN))));
163 DEBUGP("res %02X %02X %02X ", 163 DEBUGP("res %02X %02X %02X ",
164 (rtinfo->flags & IP6T_RT_RES), ((struct rt0_hdr *)rh)->bitmap, 164 (rtinfo->flags & IP6T_RT_RES), ((struct rt0_hdr *)rh)->reserved,
165 !((rtinfo->flags & IP6T_RT_RES) && (((struct rt0_hdr *)rh)->bitmap))); 165 !((rtinfo->flags & IP6T_RT_RES) && (((struct rt0_hdr *)rh)->reserved)));
166 166
167 ret = (rh != NULL) 167 ret = (rh != NULL)
168 && 168 &&
@@ -179,12 +179,12 @@ match(const struct sk_buff *skb,
179 !!(rtinfo->invflags & IP6T_RT_INV_TYP))); 179 !!(rtinfo->invflags & IP6T_RT_INV_TYP)));
180 180
181 if (ret && (rtinfo->flags & IP6T_RT_RES)) { 181 if (ret && (rtinfo->flags & IP6T_RT_RES)) {
182 u_int32_t *bp, _bitmap; 182 u_int32_t *rp, _reserved;
183 bp = skb_header_pointer(skb, 183 rp = skb_header_pointer(skb,
184 ptr + offsetof(struct rt0_hdr, bitmap), 184 ptr + offsetof(struct rt0_hdr, reserved),
185 sizeof(_bitmap), &_bitmap); 185 sizeof(_reserved), &_reserved);
186 186
187 ret = (*bp == 0); 187 ret = (*rp == 0);
188 } 188 }
189 189
190 DEBUGP("#%d ",rtinfo->addrnr); 190 DEBUGP("#%d ",rtinfo->addrnr);
diff --git a/net/ipv6/raw.c b/net/ipv6/raw.c
index ed3a76b30fd9..5aa3691c578d 100644
--- a/net/ipv6/raw.c
+++ b/net/ipv6/raw.c
@@ -655,6 +655,7 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
655 struct flowi fl; 655 struct flowi fl;
656 int addr_len = msg->msg_namelen; 656 int addr_len = msg->msg_namelen;
657 int hlimit = -1; 657 int hlimit = -1;
658 int tclass = -1;
658 u16 proto; 659 u16 proto;
659 int err; 660 int err;
660 661
@@ -740,7 +741,7 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
740 memset(opt, 0, sizeof(struct ipv6_txoptions)); 741 memset(opt, 0, sizeof(struct ipv6_txoptions));
741 opt->tot_len = sizeof(struct ipv6_txoptions); 742 opt->tot_len = sizeof(struct ipv6_txoptions);
742 743
743 err = datagram_send_ctl(msg, &fl, opt, &hlimit); 744 err = datagram_send_ctl(msg, &fl, opt, &hlimit, &tclass);
744 if (err < 0) { 745 if (err < 0) {
745 fl6_sock_release(flowlabel); 746 fl6_sock_release(flowlabel);
746 return err; 747 return err;
@@ -755,8 +756,7 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
755 } 756 }
756 if (opt == NULL) 757 if (opt == NULL)
757 opt = np->opt; 758 opt = np->opt;
758 if (flowlabel) 759 opt = fl6_merge_options(&opt_space, flowlabel, opt);
759 opt = fl6_merge_options(&opt_space, flowlabel, opt);
760 760
761 fl.proto = proto; 761 fl.proto = proto;
762 rawv6_probe_proto_opt(&fl, msg); 762 rawv6_probe_proto_opt(&fl, msg);
@@ -782,10 +782,8 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
782 if (final_p) 782 if (final_p)
783 ipv6_addr_copy(&fl.fl6_dst, final_p); 783 ipv6_addr_copy(&fl.fl6_dst, final_p);
784 784
785 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) { 785 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0)
786 dst_release(dst);
787 goto out; 786 goto out;
788 }
789 787
790 if (hlimit < 0) { 788 if (hlimit < 0) {
791 if (ipv6_addr_is_multicast(&fl.fl6_dst)) 789 if (ipv6_addr_is_multicast(&fl.fl6_dst))
@@ -798,6 +796,12 @@ static int rawv6_sendmsg(struct kiocb *iocb, struct sock *sk,
798 hlimit = ipv6_get_hoplimit(dst->dev); 796 hlimit = ipv6_get_hoplimit(dst->dev);
799 } 797 }
800 798
799 if (tclass < 0) {
800 tclass = np->cork.tclass;
801 if (tclass < 0)
802 tclass = 0;
803 }
804
801 if (msg->msg_flags&MSG_CONFIRM) 805 if (msg->msg_flags&MSG_CONFIRM)
802 goto do_confirm; 806 goto do_confirm;
803 807
@@ -806,8 +810,9 @@ back_from_confirm:
806 err = rawv6_send_hdrinc(sk, msg->msg_iov, len, &fl, (struct rt6_info*)dst, msg->msg_flags); 810 err = rawv6_send_hdrinc(sk, msg->msg_iov, len, &fl, (struct rt6_info*)dst, msg->msg_flags);
807 } else { 811 } else {
808 lock_sock(sk); 812 lock_sock(sk);
809 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov, len, 0, 813 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov,
810 hlimit, opt, &fl, (struct rt6_info*)dst, msg->msg_flags); 814 len, 0, hlimit, tclass, opt, &fl, (struct rt6_info*)dst,
815 msg->msg_flags);
811 816
812 if (err) 817 if (err)
813 ip6_flush_pending_frames(sk); 818 ip6_flush_pending_frames(sk);
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 9d9e04344c77..e4fe9ee484dd 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c
@@ -479,12 +479,9 @@ static void ip6_frag_queue(struct frag_queue *fq, struct sk_buff *skb,
479 /* Point into the IP datagram 'data' part. */ 479 /* Point into the IP datagram 'data' part. */
480 if (!pskb_pull(skb, (u8 *) (fhdr + 1) - skb->data)) 480 if (!pskb_pull(skb, (u8 *) (fhdr + 1) - skb->data))
481 goto err; 481 goto err;
482 if (end-offset < skb->len) { 482
483 if (pskb_trim(skb, end - offset)) 483 if (pskb_trim_rcsum(skb, end - offset))
484 goto err; 484 goto err;
485 if (skb->ip_summed != CHECKSUM_UNNECESSARY)
486 skb->ip_summed = CHECKSUM_NONE;
487 }
488 485
489 /* Find out which fragments are in front and at the back of us 486 /* Find out which fragments are in front and at the back of us
490 * in the chain of fragments so far. We must know where to put 487 * in the chain of fragments so far. We must know where to put
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 794734f1d230..80643e6b346b 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -632,10 +632,8 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
632 if (final_p) 632 if (final_p)
633 ipv6_addr_copy(&fl.fl6_dst, final_p); 633 ipv6_addr_copy(&fl.fl6_dst, final_p);
634 634
635 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) { 635 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0)
636 dst_release(dst);
637 goto failure; 636 goto failure;
638 }
639 637
640 if (saddr == NULL) { 638 if (saddr == NULL) {
641 saddr = &fl.fl6_src; 639 saddr = &fl.fl6_src;
@@ -849,7 +847,7 @@ static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req,
849 if (dst == NULL) { 847 if (dst == NULL) {
850 opt = np->opt; 848 opt = np->opt;
851 if (opt == NULL && 849 if (opt == NULL &&
852 np->rxopt.bits.srcrt == 2 && 850 np->rxopt.bits.osrcrt == 2 &&
853 treq->pktopts) { 851 treq->pktopts) {
854 struct sk_buff *pktopts = treq->pktopts; 852 struct sk_buff *pktopts = treq->pktopts;
855 struct inet6_skb_parm *rxopt = IP6CB(pktopts); 853 struct inet6_skb_parm *rxopt = IP6CB(pktopts);
@@ -888,7 +886,6 @@ static int tcp_v6_send_synack(struct sock *sk, struct request_sock *req,
888 } 886 }
889 887
890done: 888done:
891 dst_release(dst);
892 if (opt && opt != np->opt) 889 if (opt && opt != np->opt)
893 sock_kfree_s(sk, opt, opt->tot_len); 890 sock_kfree_s(sk, opt, opt->tot_len);
894 return err; 891 return err;
@@ -915,11 +912,10 @@ static int ipv6_opt_accepted(struct sock *sk, struct sk_buff *skb)
915 struct inet6_skb_parm *opt = IP6CB(skb); 912 struct inet6_skb_parm *opt = IP6CB(skb);
916 913
917 if (np->rxopt.all) { 914 if (np->rxopt.all) {
918 if ((opt->hop && np->rxopt.bits.hopopts) || 915 if ((opt->hop && (np->rxopt.bits.hopopts || np->rxopt.bits.ohopopts)) ||
919 ((IPV6_FLOWINFO_MASK&*(u32*)skb->nh.raw) && 916 ((IPV6_FLOWINFO_MASK & *(u32*)skb->nh.raw) && np->rxopt.bits.rxflow) ||
920 np->rxopt.bits.rxflow) || 917 (opt->srcrt && (np->rxopt.bits.srcrt || np->rxopt.bits.osrcrt)) ||
921 (opt->srcrt && np->rxopt.bits.srcrt) || 918 ((opt->dst1 || opt->dst0) && (np->rxopt.bits.dstopts || np->rxopt.bits.odstopts)))
922 ((opt->dst1 || opt->dst0) && np->rxopt.bits.dstopts))
923 return 1; 919 return 1;
924 } 920 }
925 return 0; 921 return 0;
@@ -1001,10 +997,8 @@ static void tcp_v6_send_reset(struct sk_buff *skb)
1001 /* sk = NULL, but it is safe for now. RST socket required. */ 997 /* sk = NULL, but it is safe for now. RST socket required. */
1002 if (!ip6_dst_lookup(NULL, &buff->dst, &fl)) { 998 if (!ip6_dst_lookup(NULL, &buff->dst, &fl)) {
1003 999
1004 if ((xfrm_lookup(&buff->dst, &fl, NULL, 0)) < 0) { 1000 if ((xfrm_lookup(&buff->dst, &fl, NULL, 0)) < 0)
1005 dst_release(buff->dst);
1006 return; 1001 return;
1007 }
1008 1002
1009 ip6_xmit(NULL, buff, &fl, NULL, 0); 1003 ip6_xmit(NULL, buff, &fl, NULL, 0);
1010 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS); 1004 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS);
@@ -1068,10 +1062,8 @@ static void tcp_v6_send_ack(struct sk_buff *skb, u32 seq, u32 ack, u32 win, u32
1068 fl.fl_ip_sport = t1->source; 1062 fl.fl_ip_sport = t1->source;
1069 1063
1070 if (!ip6_dst_lookup(NULL, &buff->dst, &fl)) { 1064 if (!ip6_dst_lookup(NULL, &buff->dst, &fl)) {
1071 if ((xfrm_lookup(&buff->dst, &fl, NULL, 0)) < 0) { 1065 if ((xfrm_lookup(&buff->dst, &fl, NULL, 0)) < 0)
1072 dst_release(buff->dst);
1073 return; 1066 return;
1074 }
1075 ip6_xmit(NULL, buff, &fl, NULL, 0); 1067 ip6_xmit(NULL, buff, &fl, NULL, 0);
1076 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS); 1068 TCP_INC_STATS_BH(TCP_MIB_OUTSEGS);
1077 return; 1069 return;
@@ -1190,8 +1182,8 @@ static int tcp_v6_conn_request(struct sock *sk, struct sk_buff *skb)
1190 TCP_ECN_create_request(req, skb->h.th); 1182 TCP_ECN_create_request(req, skb->h.th);
1191 treq->pktopts = NULL; 1183 treq->pktopts = NULL;
1192 if (ipv6_opt_accepted(sk, skb) || 1184 if (ipv6_opt_accepted(sk, skb) ||
1193 np->rxopt.bits.rxinfo || 1185 np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo ||
1194 np->rxopt.bits.rxhlim) { 1186 np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim) {
1195 atomic_inc(&skb->users); 1187 atomic_inc(&skb->users);
1196 treq->pktopts = skb; 1188 treq->pktopts = skb;
1197 } 1189 }
@@ -1288,7 +1280,7 @@ static struct sock * tcp_v6_syn_recv_sock(struct sock *sk, struct sk_buff *skb,
1288 if (sk_acceptq_is_full(sk)) 1280 if (sk_acceptq_is_full(sk))
1289 goto out_overflow; 1281 goto out_overflow;
1290 1282
1291 if (np->rxopt.bits.srcrt == 2 && 1283 if (np->rxopt.bits.osrcrt == 2 &&
1292 opt == NULL && treq->pktopts) { 1284 opt == NULL && treq->pktopts) {
1293 struct inet6_skb_parm *rxopt = IP6CB(treq->pktopts); 1285 struct inet6_skb_parm *rxopt = IP6CB(treq->pktopts);
1294 if (rxopt->srcrt) 1286 if (rxopt->srcrt)
@@ -1544,9 +1536,9 @@ ipv6_pktoptions:
1544 tp = tcp_sk(sk); 1536 tp = tcp_sk(sk);
1545 if (TCP_SKB_CB(opt_skb)->end_seq == tp->rcv_nxt && 1537 if (TCP_SKB_CB(opt_skb)->end_seq == tp->rcv_nxt &&
1546 !((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) { 1538 !((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))) {
1547 if (np->rxopt.bits.rxinfo) 1539 if (np->rxopt.bits.rxinfo || np->rxopt.bits.rxoinfo)
1548 np->mcast_oif = inet6_iif(opt_skb); 1540 np->mcast_oif = inet6_iif(opt_skb);
1549 if (np->rxopt.bits.rxhlim) 1541 if (np->rxopt.bits.rxhlim || np->rxopt.bits.rxohlim)
1550 np->mcast_hops = opt_skb->nh.ipv6h->hop_limit; 1542 np->mcast_hops = opt_skb->nh.ipv6h->hop_limit;
1551 if (ipv6_opt_accepted(sk, opt_skb)) { 1543 if (ipv6_opt_accepted(sk, opt_skb)) {
1552 skb_set_owner_r(opt_skb, sk); 1544 skb_set_owner_r(opt_skb, sk);
@@ -1734,7 +1726,6 @@ static int tcp_v6_rebuild_header(struct sock *sk)
1734 1726
1735 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) { 1727 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) {
1736 sk->sk_err_soft = -err; 1728 sk->sk_err_soft = -err;
1737 dst_release(dst);
1738 return err; 1729 return err;
1739 } 1730 }
1740 1731
@@ -1787,7 +1778,6 @@ static int tcp_v6_xmit(struct sk_buff *skb, int ipfragok)
1787 1778
1788 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) { 1779 if ((err = xfrm_lookup(&dst, &fl, sk, 0)) < 0) {
1789 sk->sk_route_caps = 0; 1780 sk->sk_route_caps = 0;
1790 dst_release(dst);
1791 return err; 1781 return err;
1792 } 1782 }
1793 1783
diff --git a/net/ipv6/udp.c b/net/ipv6/udp.c
index 390d750449ce..69b146843a20 100644
--- a/net/ipv6/udp.c
+++ b/net/ipv6/udp.c
@@ -483,7 +483,7 @@ static int udpv6_rcv(struct sk_buff **pskb, unsigned int *nhoffp)
483 } 483 }
484 484
485 if (ulen < skb->len) { 485 if (ulen < skb->len) {
486 if (__pskb_trim(skb, ulen)) 486 if (pskb_trim_rcsum(skb, ulen))
487 goto discard; 487 goto discard;
488 saddr = &skb->nh.ipv6h->saddr; 488 saddr = &skb->nh.ipv6h->saddr;
489 daddr = &skb->nh.ipv6h->daddr; 489 daddr = &skb->nh.ipv6h->daddr;
@@ -637,6 +637,7 @@ static int udpv6_sendmsg(struct kiocb *iocb, struct sock *sk,
637 int addr_len = msg->msg_namelen; 637 int addr_len = msg->msg_namelen;
638 int ulen = len; 638 int ulen = len;
639 int hlimit = -1; 639 int hlimit = -1;
640 int tclass = -1;
640 int corkreq = up->corkflag || msg->msg_flags&MSG_MORE; 641 int corkreq = up->corkflag || msg->msg_flags&MSG_MORE;
641 int err; 642 int err;
642 643
@@ -758,7 +759,7 @@ do_udp_sendmsg:
758 memset(opt, 0, sizeof(struct ipv6_txoptions)); 759 memset(opt, 0, sizeof(struct ipv6_txoptions));
759 opt->tot_len = sizeof(*opt); 760 opt->tot_len = sizeof(*opt);
760 761
761 err = datagram_send_ctl(msg, fl, opt, &hlimit); 762 err = datagram_send_ctl(msg, fl, opt, &hlimit, &tclass);
762 if (err < 0) { 763 if (err < 0) {
763 fl6_sock_release(flowlabel); 764 fl6_sock_release(flowlabel);
764 return err; 765 return err;
@@ -773,8 +774,7 @@ do_udp_sendmsg:
773 } 774 }
774 if (opt == NULL) 775 if (opt == NULL)
775 opt = np->opt; 776 opt = np->opt;
776 if (flowlabel) 777 opt = fl6_merge_options(&opt_space, flowlabel, opt);
777 opt = fl6_merge_options(&opt_space, flowlabel, opt);
778 778
779 fl->proto = IPPROTO_UDP; 779 fl->proto = IPPROTO_UDP;
780 ipv6_addr_copy(&fl->fl6_dst, daddr); 780 ipv6_addr_copy(&fl->fl6_dst, daddr);
@@ -799,10 +799,8 @@ do_udp_sendmsg:
799 if (final_p) 799 if (final_p)
800 ipv6_addr_copy(&fl->fl6_dst, final_p); 800 ipv6_addr_copy(&fl->fl6_dst, final_p);
801 801
802 if ((err = xfrm_lookup(&dst, fl, sk, 0)) < 0) { 802 if ((err = xfrm_lookup(&dst, fl, sk, 0)) < 0)
803 dst_release(dst);
804 goto out; 803 goto out;
805 }
806 804
807 if (hlimit < 0) { 805 if (hlimit < 0) {
808 if (ipv6_addr_is_multicast(&fl->fl6_dst)) 806 if (ipv6_addr_is_multicast(&fl->fl6_dst))
@@ -815,6 +813,12 @@ do_udp_sendmsg:
815 hlimit = ipv6_get_hoplimit(dst->dev); 813 hlimit = ipv6_get_hoplimit(dst->dev);
816 } 814 }
817 815
816 if (tclass < 0) {
817 tclass = np->tclass;
818 if (tclass < 0)
819 tclass = 0;
820 }
821
818 if (msg->msg_flags&MSG_CONFIRM) 822 if (msg->msg_flags&MSG_CONFIRM)
819 goto do_confirm; 823 goto do_confirm;
820back_from_confirm: 824back_from_confirm:
@@ -834,9 +838,10 @@ back_from_confirm:
834 838
835do_append_data: 839do_append_data:
836 up->len += ulen; 840 up->len += ulen;
837 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov, ulen, sizeof(struct udphdr), 841 err = ip6_append_data(sk, ip_generic_getfrag, msg->msg_iov, ulen,
838 hlimit, opt, fl, (struct rt6_info*)dst, 842 sizeof(struct udphdr), hlimit, tclass, opt, fl,
839 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags); 843 (struct rt6_info*)dst,
844 corkreq ? msg->msg_flags|MSG_MORE : msg->msg_flags);
840 if (err) 845 if (err)
841 udp_v6_flush_pending_frames(sk); 846 udp_v6_flush_pending_frames(sk);
842 else if (!corkreq) 847 else if (!corkreq)
diff --git a/net/irda/ircomm/ircomm_tty.c b/net/irda/ircomm/ircomm_tty.c
index 5d1e61168eb7..6f20b4206e08 100644
--- a/net/irda/ircomm/ircomm_tty.c
+++ b/net/irda/ircomm/ircomm_tty.c
@@ -567,10 +567,8 @@ static void ircomm_tty_close(struct tty_struct *tty, struct file *filp)
567 self->tty = NULL; 567 self->tty = NULL;
568 568
569 if (self->blocked_open) { 569 if (self->blocked_open) {
570 if (self->close_delay) { 570 if (self->close_delay)
571 current->state = TASK_INTERRUPTIBLE; 571 schedule_timeout_interruptible(self->close_delay);
572 schedule_timeout(self->close_delay);
573 }
574 wake_up_interruptible(&self->open_wait); 572 wake_up_interruptible(&self->open_wait);
575 } 573 }
576 574
@@ -863,8 +861,7 @@ static void ircomm_tty_wait_until_sent(struct tty_struct *tty, int timeout)
863 spin_lock_irqsave(&self->spinlock, flags); 861 spin_lock_irqsave(&self->spinlock, flags);
864 while (self->tx_skb && self->tx_skb->len) { 862 while (self->tx_skb && self->tx_skb->len) {
865 spin_unlock_irqrestore(&self->spinlock, flags); 863 spin_unlock_irqrestore(&self->spinlock, flags);
866 current->state = TASK_INTERRUPTIBLE; 864 schedule_timeout_interruptible(poll_time);
867 schedule_timeout(poll_time);
868 spin_lock_irqsave(&self->spinlock, flags); 865 spin_lock_irqsave(&self->spinlock, flags);
869 if (signal_pending(current)) 866 if (signal_pending(current))
870 break; 867 break;
diff --git a/net/netrom/af_netrom.c b/net/netrom/af_netrom.c
index f4578c759ffc..e5d82d711cae 100644
--- a/net/netrom/af_netrom.c
+++ b/net/netrom/af_netrom.c
@@ -56,6 +56,7 @@ int sysctl_netrom_transport_requested_window_size = NR_DEFAULT_WINDOW;
56int sysctl_netrom_transport_no_activity_timeout = NR_DEFAULT_IDLE; 56int sysctl_netrom_transport_no_activity_timeout = NR_DEFAULT_IDLE;
57int sysctl_netrom_routing_control = NR_DEFAULT_ROUTING; 57int sysctl_netrom_routing_control = NR_DEFAULT_ROUTING;
58int sysctl_netrom_link_fails_count = NR_DEFAULT_FAILS; 58int sysctl_netrom_link_fails_count = NR_DEFAULT_FAILS;
59int sysctl_netrom_reset_circuit = NR_DEFAULT_RESET;
59 60
60static unsigned short circuit = 0x101; 61static unsigned short circuit = 0x101;
61 62
@@ -908,17 +909,17 @@ int nr_rx_frame(struct sk_buff *skb, struct net_device *dev)
908 if (frametype != NR_CONNREQ) { 909 if (frametype != NR_CONNREQ) {
909 /* 910 /*
910 * Here it would be nice to be able to send a reset but 911 * Here it would be nice to be able to send a reset but
911 * NET/ROM doesn't have one. The following hack would 912 * NET/ROM doesn't have one. We've tried to extend the protocol
912 * have been a way to extend the protocol but apparently 913 * by sending NR_CONNACK | NR_CHOKE_FLAGS replies but that
913 * it kills BPQ boxes... :-( 914 * apparently kills BPQ boxes... :-(
915 * So now we try to follow the established behaviour of
916 * G8PZT's Xrouter which is sending packets with command type 7
917 * as an extension of the protocol.
914 */ 918 */
915#if 0 919 if (sysctl_netrom_reset_circuit &&
916 /* 920 (frametype != NR_RESET || flags != 0))
917 * Never reply to a CONNACK/CHOKE. 921 nr_transmit_reset(skb, 1);
918 */ 922
919 if (frametype != NR_CONNACK || flags != NR_CHOKE_FLAG)
920 nr_transmit_refusal(skb, 1);
921#endif
922 return 0; 923 return 0;
923 } 924 }
924 925
@@ -1187,9 +1188,7 @@ static int nr_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
1187 } 1188 }
1188 1189
1189 case SIOCGSTAMP: 1190 case SIOCGSTAMP:
1190 ret = -EINVAL; 1191 ret = sock_get_timestamp(sk, argp);
1191 if (sk != NULL)
1192 ret = sock_get_timestamp(sk, argp);
1193 release_sock(sk); 1192 release_sock(sk);
1194 return ret; 1193 return ret;
1195 1194
@@ -1393,8 +1392,7 @@ static int __init nr_proto_init(void)
1393 struct net_device *dev; 1392 struct net_device *dev;
1394 1393
1395 sprintf(name, "nr%d", i); 1394 sprintf(name, "nr%d", i);
1396 dev = alloc_netdev(sizeof(struct net_device_stats), name, 1395 dev = alloc_netdev(sizeof(struct nr_private), name, nr_setup);
1397 nr_setup);
1398 if (!dev) { 1396 if (!dev) {
1399 printk(KERN_ERR "NET/ROM: nr_proto_init - unable to allocate device structure\n"); 1397 printk(KERN_ERR "NET/ROM: nr_proto_init - unable to allocate device structure\n");
1400 goto fail; 1398 goto fail;
diff --git a/net/netrom/nr_dev.c b/net/netrom/nr_dev.c
index 263da4c26494..4e66eef9a034 100644
--- a/net/netrom/nr_dev.c
+++ b/net/netrom/nr_dev.c
@@ -47,7 +47,7 @@ int nr_rx_ip(struct sk_buff *skb, struct net_device *dev)
47 struct net_device_stats *stats = netdev_priv(dev); 47 struct net_device_stats *stats = netdev_priv(dev);
48 48
49 if (!netif_running(dev)) { 49 if (!netif_running(dev)) {
50 stats->rx_errors++; 50 stats->rx_dropped++;
51 return 0; 51 return 0;
52 } 52 }
53 53
@@ -71,15 +71,10 @@ int nr_rx_ip(struct sk_buff *skb, struct net_device *dev)
71 71
72static int nr_rebuild_header(struct sk_buff *skb) 72static int nr_rebuild_header(struct sk_buff *skb)
73{ 73{
74 struct net_device *dev = skb->dev;
75 struct net_device_stats *stats = netdev_priv(dev);
76 struct sk_buff *skbn;
77 unsigned char *bp = skb->data; 74 unsigned char *bp = skb->data;
78 int len;
79 75
80 if (arp_find(bp + 7, skb)) { 76 if (arp_find(bp + 7, skb))
81 return 1; 77 return 1;
82 }
83 78
84 bp[6] &= ~AX25_CBIT; 79 bp[6] &= ~AX25_CBIT;
85 bp[6] &= ~AX25_EBIT; 80 bp[6] &= ~AX25_EBIT;
@@ -90,27 +85,7 @@ static int nr_rebuild_header(struct sk_buff *skb)
90 bp[6] |= AX25_EBIT; 85 bp[6] |= AX25_EBIT;
91 bp[6] |= AX25_SSSID_SPARE; 86 bp[6] |= AX25_SSSID_SPARE;
92 87
93 if ((skbn = skb_clone(skb, GFP_ATOMIC)) == NULL) { 88 return 0;
94 kfree_skb(skb);
95 return 1;
96 }
97
98 if (skb->sk != NULL)
99 skb_set_owner_w(skbn, skb->sk);
100
101 kfree_skb(skb);
102
103 len = skbn->len;
104
105 if (!nr_route_frame(skbn, NULL)) {
106 kfree_skb(skbn);
107 stats->tx_errors++;
108 }
109
110 stats->tx_packets++;
111 stats->tx_bytes += len;
112
113 return 1;
114} 89}
115 90
116#else 91#else
@@ -185,15 +160,27 @@ static int nr_close(struct net_device *dev)
185 160
186static int nr_xmit(struct sk_buff *skb, struct net_device *dev) 161static int nr_xmit(struct sk_buff *skb, struct net_device *dev)
187{ 162{
188 struct net_device_stats *stats = netdev_priv(dev); 163 struct nr_private *nr = netdev_priv(dev);
189 dev_kfree_skb(skb); 164 struct net_device_stats *stats = &nr->stats;
190 stats->tx_errors++; 165 unsigned int len = skb->len;
166
167 if (!nr_route_frame(skb, NULL)) {
168 kfree_skb(skb);
169 stats->tx_errors++;
170 return 0;
171 }
172
173 stats->tx_packets++;
174 stats->tx_bytes += len;
175
191 return 0; 176 return 0;
192} 177}
193 178
194static struct net_device_stats *nr_get_stats(struct net_device *dev) 179static struct net_device_stats *nr_get_stats(struct net_device *dev)
195{ 180{
196 return netdev_priv(dev); 181 struct nr_private *nr = netdev_priv(dev);
182
183 return &nr->stats;
197} 184}
198 185
199void nr_setup(struct net_device *dev) 186void nr_setup(struct net_device *dev)
@@ -208,12 +195,11 @@ void nr_setup(struct net_device *dev)
208 dev->hard_header_len = NR_NETWORK_LEN + NR_TRANSPORT_LEN; 195 dev->hard_header_len = NR_NETWORK_LEN + NR_TRANSPORT_LEN;
209 dev->addr_len = AX25_ADDR_LEN; 196 dev->addr_len = AX25_ADDR_LEN;
210 dev->type = ARPHRD_NETROM; 197 dev->type = ARPHRD_NETROM;
211 dev->tx_queue_len = 40;
212 dev->rebuild_header = nr_rebuild_header; 198 dev->rebuild_header = nr_rebuild_header;
213 dev->set_mac_address = nr_set_mac_address; 199 dev->set_mac_address = nr_set_mac_address;
214 200
215 /* New-style flags. */ 201 /* New-style flags. */
216 dev->flags = 0; 202 dev->flags = IFF_NOARP;
217 203
218 dev->get_stats = nr_get_stats; 204 dev->get_stats = nr_get_stats;
219} 205}
diff --git a/net/netrom/nr_in.c b/net/netrom/nr_in.c
index 64b81a796907..004e8599b8fe 100644
--- a/net/netrom/nr_in.c
+++ b/net/netrom/nr_in.c
@@ -98,6 +98,11 @@ static int nr_state1_machine(struct sock *sk, struct sk_buff *skb,
98 nr_disconnect(sk, ECONNREFUSED); 98 nr_disconnect(sk, ECONNREFUSED);
99 break; 99 break;
100 100
101 case NR_RESET:
102 if (sysctl_netrom_reset_circuit);
103 nr_disconnect(sk, ECONNRESET);
104 break;
105
101 default: 106 default:
102 break; 107 break;
103 } 108 }
@@ -124,6 +129,11 @@ static int nr_state2_machine(struct sock *sk, struct sk_buff *skb,
124 nr_disconnect(sk, 0); 129 nr_disconnect(sk, 0);
125 break; 130 break;
126 131
132 case NR_RESET:
133 if (sysctl_netrom_reset_circuit);
134 nr_disconnect(sk, ECONNRESET);
135 break;
136
127 default: 137 default:
128 break; 138 break;
129 } 139 }
@@ -254,6 +264,11 @@ static int nr_state3_machine(struct sock *sk, struct sk_buff *skb, int frametype
254 } 264 }
255 break; 265 break;
256 266
267 case NR_RESET:
268 if (sysctl_netrom_reset_circuit);
269 nr_disconnect(sk, ECONNRESET);
270 break;
271
257 default: 272 default:
258 break; 273 break;
259 } 274 }
diff --git a/net/netrom/nr_loopback.c b/net/netrom/nr_loopback.c
index 165b2abce110..e856ae1b360a 100644
--- a/net/netrom/nr_loopback.c
+++ b/net/netrom/nr_loopback.c
@@ -17,7 +17,7 @@
17static void nr_loopback_timer(unsigned long); 17static void nr_loopback_timer(unsigned long);
18 18
19static struct sk_buff_head loopback_queue; 19static struct sk_buff_head loopback_queue;
20static struct timer_list loopback_timer = TIMER_INITIALIZER(nr_loopback_timer, 0, 0); 20static DEFINE_TIMER(loopback_timer, nr_loopback_timer, 0, 0);
21 21
22void __init nr_loopback_init(void) 22void __init nr_loopback_init(void)
23{ 23{
diff --git a/net/netrom/nr_subr.c b/net/netrom/nr_subr.c
index 587bed2674bf..bcb9946b4f56 100644
--- a/net/netrom/nr_subr.c
+++ b/net/netrom/nr_subr.c
@@ -210,10 +210,9 @@ void nr_write_internal(struct sock *sk, int frametype)
210} 210}
211 211
212/* 212/*
213 * This routine is called when a Connect Acknowledge with the Choke Flag 213 * This routine is called to send an error reply.
214 * set is needed to refuse a connection.
215 */ 214 */
216void nr_transmit_refusal(struct sk_buff *skb, int mine) 215void __nr_transmit_reply(struct sk_buff *skb, int mine, unsigned char cmdflags)
217{ 216{
218 struct sk_buff *skbn; 217 struct sk_buff *skbn;
219 unsigned char *dptr; 218 unsigned char *dptr;
@@ -254,7 +253,7 @@ void nr_transmit_refusal(struct sk_buff *skb, int mine)
254 *dptr++ = 0; 253 *dptr++ = 0;
255 } 254 }
256 255
257 *dptr++ = NR_CONNACK | NR_CHOKE_FLAG; 256 *dptr++ = cmdflags;
258 *dptr++ = 0; 257 *dptr++ = 0;
259 258
260 if (!nr_route_frame(skbn, NULL)) 259 if (!nr_route_frame(skbn, NULL))
diff --git a/net/netrom/sysctl_net_netrom.c b/net/netrom/sysctl_net_netrom.c
index c9ed50382ea7..6bb8dda849dc 100644
--- a/net/netrom/sysctl_net_netrom.c
+++ b/net/netrom/sysctl_net_netrom.c
@@ -30,6 +30,7 @@ static int min_idle[] = {0 * HZ};
30static int max_idle[] = {65535 * HZ}; 30static int max_idle[] = {65535 * HZ};
31static int min_route[] = {0}, max_route[] = {1}; 31static int min_route[] = {0}, max_route[] = {1};
32static int min_fails[] = {1}, max_fails[] = {10}; 32static int min_fails[] = {1}, max_fails[] = {10};
33static int min_reset[] = {0}, max_reset[] = {1};
33 34
34static struct ctl_table_header *nr_table_header; 35static struct ctl_table_header *nr_table_header;
35 36
@@ -155,6 +156,17 @@ static ctl_table nr_table[] = {
155 .extra1 = &min_fails, 156 .extra1 = &min_fails,
156 .extra2 = &max_fails 157 .extra2 = &max_fails
157 }, 158 },
159 {
160 .ctl_name = NET_NETROM_RESET,
161 .procname = "reset",
162 .data = &sysctl_netrom_reset_circuit,
163 .maxlen = sizeof(int),
164 .mode = 0644,
165 .proc_handler = &proc_dointvec_minmax,
166 .strategy = &sysctl_intvec,
167 .extra1 = &min_reset,
168 .extra2 = &max_reset
169 },
158 { .ctl_name = 0 } 170 { .ctl_name = 0 }
159}; 171};
160 172
diff --git a/net/rose/af_rose.c b/net/rose/af_rose.c
index 3077878ed4f0..5acb1680524a 100644
--- a/net/rose/af_rose.c
+++ b/net/rose/af_rose.c
@@ -1243,7 +1243,7 @@ static int rose_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
1243 amount = sk->sk_sndbuf - atomic_read(&sk->sk_wmem_alloc); 1243 amount = sk->sk_sndbuf - atomic_read(&sk->sk_wmem_alloc);
1244 if (amount < 0) 1244 if (amount < 0)
1245 amount = 0; 1245 amount = 0;
1246 return put_user(amount, (unsigned int __user *)argp); 1246 return put_user(amount, (unsigned int __user *) argp);
1247 } 1247 }
1248 1248
1249 case TIOCINQ: { 1249 case TIOCINQ: {
@@ -1252,13 +1252,11 @@ static int rose_ioctl(struct socket *sock, unsigned int cmd, unsigned long arg)
1252 /* These two are safe on a single CPU system as only user tasks fiddle here */ 1252 /* These two are safe on a single CPU system as only user tasks fiddle here */
1253 if ((skb = skb_peek(&sk->sk_receive_queue)) != NULL) 1253 if ((skb = skb_peek(&sk->sk_receive_queue)) != NULL)
1254 amount = skb->len; 1254 amount = skb->len;
1255 return put_user(amount, (unsigned int __user *)argp); 1255 return put_user(amount, (unsigned int __user *) argp);
1256 } 1256 }
1257 1257
1258 case SIOCGSTAMP: 1258 case SIOCGSTAMP:
1259 if (sk != NULL) 1259 return sock_get_timestamp(sk, (struct timeval __user *) argp);
1260 return sock_get_timestamp(sk, (struct timeval __user *)argp);
1261 return -EINVAL;
1262 1260
1263 case SIOCGIFADDR: 1261 case SIOCGIFADDR:
1264 case SIOCSIFADDR: 1262 case SIOCSIFADDR:
diff --git a/net/rose/rose_dev.c b/net/rose/rose_dev.c
index a8ed9a1d09f9..d297af737d10 100644
--- a/net/rose/rose_dev.c
+++ b/net/rose/rose_dev.c
@@ -149,6 +149,6 @@ void rose_setup(struct net_device *dev)
149 dev->set_mac_address = rose_set_mac_address; 149 dev->set_mac_address = rose_set_mac_address;
150 150
151 /* New-style flags. */ 151 /* New-style flags. */
152 dev->flags = 0; 152 dev->flags = IFF_NOARP;
153 dev->get_stats = rose_get_stats; 153 dev->get_stats = rose_get_stats;
154} 154}
diff --git a/net/rose/rose_subr.c b/net/rose/rose_subr.c
index 02891ce2db37..36a77944622b 100644
--- a/net/rose/rose_subr.c
+++ b/net/rose/rose_subr.c
@@ -337,13 +337,13 @@ static int rose_parse_ccitt(unsigned char *p, struct rose_facilities_struct *fac
337 memcpy(&facilities->source_addr, p + 7, ROSE_ADDR_LEN); 337 memcpy(&facilities->source_addr, p + 7, ROSE_ADDR_LEN);
338 memcpy(callsign, p + 12, l - 10); 338 memcpy(callsign, p + 12, l - 10);
339 callsign[l - 10] = '\0'; 339 callsign[l - 10] = '\0';
340 facilities->source_call = *asc2ax(callsign); 340 asc2ax(&facilities->source_call, callsign);
341 } 341 }
342 if (*p == FAC_CCITT_SRC_NSAP) { 342 if (*p == FAC_CCITT_SRC_NSAP) {
343 memcpy(&facilities->dest_addr, p + 7, ROSE_ADDR_LEN); 343 memcpy(&facilities->dest_addr, p + 7, ROSE_ADDR_LEN);
344 memcpy(callsign, p + 12, l - 10); 344 memcpy(callsign, p + 12, l - 10);
345 callsign[l - 10] = '\0'; 345 callsign[l - 10] = '\0';
346 facilities->dest_call = *asc2ax(callsign); 346 asc2ax(&facilities->dest_call, callsign);
347 } 347 }
348 p += l + 2; 348 p += l + 2;
349 n += l + 2; 349 n += l + 2;
diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c
index 737681cb9a92..31570b9a6e9a 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -1194,7 +1194,7 @@ EXPORT_SYMBOL(psched_time_base);
1194 * with 32-bit get_cycles(). Safe up to 4GHz CPU. 1194 * with 32-bit get_cycles(). Safe up to 4GHz CPU.
1195 */ 1195 */
1196static void psched_tick(unsigned long); 1196static void psched_tick(unsigned long);
1197static struct timer_list psched_timer = TIMER_INITIALIZER(psched_tick, 0, 0); 1197static DEFINE_TIMER(psched_timer, psched_tick, 0, 0);
1198 1198
1199static void psched_tick(unsigned long dummy) 1199static void psched_tick(unsigned long dummy)
1200{ 1200{
diff --git a/net/socket.c b/net/socket.c
index e1bd5d84d7bf..c699e93c33d7 100644
--- a/net/socket.c
+++ b/net/socket.c
@@ -1745,10 +1745,11 @@ asmlinkage long sys_sendmsg(int fd, struct msghdr __user *msg, unsigned flags)
1745 goto out_freeiov; 1745 goto out_freeiov;
1746 ctl_len = msg_sys.msg_controllen; 1746 ctl_len = msg_sys.msg_controllen;
1747 if ((MSG_CMSG_COMPAT & flags) && ctl_len) { 1747 if ((MSG_CMSG_COMPAT & flags) && ctl_len) {
1748 err = cmsghdr_from_user_compat_to_kern(&msg_sys, ctl, sizeof(ctl)); 1748 err = cmsghdr_from_user_compat_to_kern(&msg_sys, sock->sk, ctl, sizeof(ctl));
1749 if (err) 1749 if (err)
1750 goto out_freeiov; 1750 goto out_freeiov;
1751 ctl_buf = msg_sys.msg_control; 1751 ctl_buf = msg_sys.msg_control;
1752 ctl_len = msg_sys.msg_controllen;
1752 } else if (ctl_len) { 1753 } else if (ctl_len) {
1753 if (ctl_len > sizeof(ctl)) 1754 if (ctl_len > sizeof(ctl))
1754 { 1755 {
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index 05fe2e735538..51885b5f744e 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1170,8 +1170,7 @@ svc_recv(struct svc_serv *serv, struct svc_rqst *rqstp, long timeout)
1170 while (rqstp->rq_arghi < pages) { 1170 while (rqstp->rq_arghi < pages) {
1171 struct page *p = alloc_page(GFP_KERNEL); 1171 struct page *p = alloc_page(GFP_KERNEL);
1172 if (!p) { 1172 if (!p) {
1173 set_current_state(TASK_UNINTERRUPTIBLE); 1173 schedule_timeout_uninterruptible(msecs_to_jiffies(500));
1174 schedule_timeout(HZ/2);
1175 continue; 1174 continue;
1176 } 1175 }
1177 rqstp->rq_argpages[rqstp->rq_arghi++] = p; 1176 rqstp->rq_argpages[rqstp->rq_arghi++] = p;
diff --git a/net/xfrm/xfrm_policy.c b/net/xfrm/xfrm_policy.c
index 83c8135e1764..fda737d77edc 100644
--- a/net/xfrm/xfrm_policy.c
+++ b/net/xfrm/xfrm_policy.c
@@ -765,8 +765,8 @@ restart:
765 switch (policy->action) { 765 switch (policy->action) {
766 case XFRM_POLICY_BLOCK: 766 case XFRM_POLICY_BLOCK:
767 /* Prohibit the flow */ 767 /* Prohibit the flow */
768 xfrm_pol_put(policy); 768 err = -EPERM;
769 return -EPERM; 769 goto error;
770 770
771 case XFRM_POLICY_ALLOW: 771 case XFRM_POLICY_ALLOW:
772 if (policy->xfrm_nr == 0) { 772 if (policy->xfrm_nr == 0) {
@@ -782,8 +782,8 @@ restart:
782 */ 782 */
783 dst = xfrm_find_bundle(fl, policy, family); 783 dst = xfrm_find_bundle(fl, policy, family);
784 if (IS_ERR(dst)) { 784 if (IS_ERR(dst)) {
785 xfrm_pol_put(policy); 785 err = PTR_ERR(dst);
786 return PTR_ERR(dst); 786 goto error;
787 } 787 }
788 788
789 if (dst) 789 if (dst)