aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
Diffstat
-rw-r--r--include/net/netns/ipv4.h2
-rw-r--r--include/net/ping.h57
-rw-r--r--net/ipv4/Makefile2
-rw-r--r--net/ipv4/af_inet.c22
-rw-r--r--net/ipv4/icmp.c12
-rw-r--r--net/ipv4/ping.c937
-rw-r--r--net/ipv4/sysctl_net_ipv4.c80
7 files changed, 1110 insertions, 2 deletions
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 542195d9469e..d786b4fc02a4 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -54,6 +54,8 @@ struct netns_ipv4 {
54 int sysctl_rt_cache_rebuild_count; 54 int sysctl_rt_cache_rebuild_count;
55 int current_rt_cache_rebuild_count; 55 int current_rt_cache_rebuild_count;
56 56
57 unsigned int sysctl_ping_group_range[2];
58
57 atomic_t rt_genid; 59 atomic_t rt_genid;
58 atomic_t dev_addr_genid; 60 atomic_t dev_addr_genid;
59 61
diff --git a/include/net/ping.h b/include/net/ping.h
new file mode 100644
index 000000000000..23062c369c62
--- /dev/null
+++ b/include/net/ping.h
@@ -0,0 +1,57 @@
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * Definitions for the "ping" module.
7 *
8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 */
13#ifndef _PING_H
14#define _PING_H
15
16#include <net/netns/hash.h>
17
18/* PING_HTABLE_SIZE must be power of 2 */
19#define PING_HTABLE_SIZE 64
20#define PING_HTABLE_MASK (PING_HTABLE_SIZE-1)
21
22#define ping_portaddr_for_each_entry(__sk, node, list) \
23 hlist_nulls_for_each_entry(__sk, node, list, sk_nulls_node)
24
25/*
26 * gid_t is either uint or ushort. We want to pass it to
27 * proc_dointvec_minmax(), so it must not be larger than MAX_INT
28 */
29#define GID_T_MAX (((gid_t)~0U) >> 1)
30
31struct ping_table {
32 struct hlist_nulls_head hash[PING_HTABLE_SIZE];
33 rwlock_t lock;
34};
35
36struct ping_iter_state {
37 struct seq_net_private p;
38 int bucket;
39};
40
41extern struct proto ping_prot;
42
43
44extern void ping_rcv(struct sk_buff *);
45extern void ping_err(struct sk_buff *, u32 info);
46
47extern void inet_get_ping_group_range_net(struct net *net, unsigned int *low, unsigned int *high);
48
49#ifdef CONFIG_PROC_FS
50extern int __init ping_proc_init(void);
51extern void ping_proc_exit(void);
52#endif
53
54void __init ping_init(void);
55
56
57#endif /* _PING_H */
diff --git a/net/ipv4/Makefile b/net/ipv4/Makefile
index 0dc772d0d125..f2dc69cffb57 100644
--- a/net/ipv4/Makefile
+++ b/net/ipv4/Makefile
@@ -11,7 +11,7 @@ obj-y := route.o inetpeer.o protocol.o \
11 datagram.o raw.o udp.o udplite.o \ 11 datagram.o raw.o udp.o udplite.o \
12 arp.o icmp.o devinet.o af_inet.o igmp.o \ 12 arp.o icmp.o devinet.o af_inet.o igmp.o \
13 fib_frontend.o fib_semantics.o fib_trie.o \ 13 fib_frontend.o fib_semantics.o fib_trie.o \
14 inet_fragment.o 14 inet_fragment.o ping.o
15 15
16obj-$(CONFIG_SYSCTL) += sysctl_net_ipv4.o 16obj-$(CONFIG_SYSCTL) += sysctl_net_ipv4.o
17obj-$(CONFIG_PROC_FS) += proc.o 17obj-$(CONFIG_PROC_FS) += proc.o
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 851aa056854b..cc1463156cd0 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -105,6 +105,7 @@
105#include <net/tcp.h> 105#include <net/tcp.h>
106#include <net/udp.h> 106#include <net/udp.h>
107#include <net/udplite.h> 107#include <net/udplite.h>
108#include <net/ping.h>
108#include <linux/skbuff.h> 109#include <linux/skbuff.h>
109#include <net/sock.h> 110#include <net/sock.h>
110#include <net/raw.h> 111#include <net/raw.h>
@@ -1008,6 +1009,14 @@ static struct inet_protosw inetsw_array[] =
1008 .flags = INET_PROTOSW_PERMANENT, 1009 .flags = INET_PROTOSW_PERMANENT,
1009 }, 1010 },
1010 1011
1012 {
1013 .type = SOCK_DGRAM,
1014 .protocol = IPPROTO_ICMP,
1015 .prot = &ping_prot,
1016 .ops = &inet_dgram_ops,
1017 .no_check = UDP_CSUM_DEFAULT,
1018 .flags = INET_PROTOSW_REUSE,
1019 },
1011 1020
1012 { 1021 {
1013 .type = SOCK_RAW, 1022 .type = SOCK_RAW,
@@ -1527,6 +1536,7 @@ static const struct net_protocol udp_protocol = {
1527 1536
1528static const struct net_protocol icmp_protocol = { 1537static const struct net_protocol icmp_protocol = {
1529 .handler = icmp_rcv, 1538 .handler = icmp_rcv,
1539 .err_handler = ping_err,
1530 .no_policy = 1, 1540 .no_policy = 1,
1531 .netns_ok = 1, 1541 .netns_ok = 1,
1532}; 1542};
@@ -1642,6 +1652,10 @@ static int __init inet_init(void)
1642 if (rc) 1652 if (rc)
1643 goto out_unregister_udp_proto; 1653 goto out_unregister_udp_proto;
1644 1654
1655 rc = proto_register(&ping_prot, 1);
1656 if (rc)
1657 goto out_unregister_raw_proto;
1658
1645 /* 1659 /*
1646 * Tell SOCKET that we are alive... 1660 * Tell SOCKET that we are alive...
1647 */ 1661 */
@@ -1697,6 +1711,8 @@ static int __init inet_init(void)
1697 /* Add UDP-Lite (RFC 3828) */ 1711 /* Add UDP-Lite (RFC 3828) */
1698 udplite4_register(); 1712 udplite4_register();
1699 1713
1714 ping_init();
1715
1700 /* 1716 /*
1701 * Set the ICMP layer up 1717 * Set the ICMP layer up
1702 */ 1718 */
@@ -1727,6 +1743,8 @@ static int __init inet_init(void)
1727 rc = 0; 1743 rc = 0;
1728out: 1744out:
1729 return rc; 1745 return rc;
1746out_unregister_raw_proto:
1747 proto_unregister(&raw_prot);
1730out_unregister_udp_proto: 1748out_unregister_udp_proto:
1731 proto_unregister(&udp_prot); 1749 proto_unregister(&udp_prot);
1732out_unregister_tcp_proto: 1750out_unregister_tcp_proto:
@@ -1751,11 +1769,15 @@ static int __init ipv4_proc_init(void)
1751 goto out_tcp; 1769 goto out_tcp;
1752 if (udp4_proc_init()) 1770 if (udp4_proc_init())
1753 goto out_udp; 1771 goto out_udp;
1772 if (ping_proc_init())
1773 goto out_ping;
1754 if (ip_misc_proc_init()) 1774 if (ip_misc_proc_init())
1755 goto out_misc; 1775 goto out_misc;
1756out: 1776out:
1757 return rc; 1777 return rc;
1758out_misc: 1778out_misc:
1779 ping_proc_exit();
1780out_ping:
1759 udp4_proc_exit(); 1781 udp4_proc_exit();
1760out_udp: 1782out_udp:
1761 tcp4_proc_exit(); 1783 tcp4_proc_exit();
diff --git a/net/ipv4/icmp.c b/net/ipv4/icmp.c
index 3314394f0aab..3f47585aad68 100644
--- a/net/ipv4/icmp.c
+++ b/net/ipv4/icmp.c
@@ -83,6 +83,7 @@
83#include <net/tcp.h> 83#include <net/tcp.h>
84#include <net/udp.h> 84#include <net/udp.h>
85#include <net/raw.h> 85#include <net/raw.h>
86#include <net/ping.h>
86#include <linux/skbuff.h> 87#include <linux/skbuff.h>
87#include <net/sock.h> 88#include <net/sock.h>
88#include <linux/errno.h> 89#include <linux/errno.h>
@@ -781,6 +782,15 @@ static void icmp_redirect(struct sk_buff *skb)
781 iph->saddr, skb->dev); 782 iph->saddr, skb->dev);
782 break; 783 break;
783 } 784 }
785
786 /* Ping wants to see redirects.
787 * Let's pretend they are errors of sorts... */
788 if (iph->protocol == IPPROTO_ICMP &&
789 iph->ihl >= 5 &&
790 pskb_may_pull(skb, (iph->ihl<<2)+8)) {
791 ping_err(skb, icmp_hdr(skb)->un.gateway);
792 }
793
784out: 794out:
785 return; 795 return;
786out_err: 796out_err:
@@ -1041,7 +1051,7 @@ error:
1041 */ 1051 */
1042static const struct icmp_control icmp_pointers[NR_ICMP_TYPES + 1] = { 1052static const struct icmp_control icmp_pointers[NR_ICMP_TYPES + 1] = {
1043 [ICMP_ECHOREPLY] = { 1053 [ICMP_ECHOREPLY] = {
1044 .handler = icmp_discard, 1054 .handler = ping_rcv,
1045 }, 1055 },
1046 [1] = { 1056 [1] = {
1047 .handler = icmp_discard, 1057 .handler = icmp_discard,
diff --git a/net/ipv4/ping.c b/net/ipv4/ping.c
new file mode 100644
index 000000000000..a77e2d788dac
--- /dev/null
+++ b/net/ipv4/ping.c
@@ -0,0 +1,937 @@
1/*
2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
5 *
6 * "Ping" sockets
7 *
8 * This program is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU General Public License
10 * as published by the Free Software Foundation; either version
11 * 2 of the License, or (at your option) any later version.
12 *
13 * Based on ipv4/udp.c code.
14 *
15 * Authors: Vasiliy Kulikov / Openwall (for Linux 2.6),
16 * Pavel Kankovsky (for Linux 2.4.32)
17 *
18 * Pavel gave all rights to bugs to Vasiliy,
19 * none of the bugs are Pavel's now.
20 *
21 */
22
23#include <asm/system.h>
24#include <linux/uaccess.h>
25#include <asm/ioctls.h>
26#include <linux/types.h>
27#include <linux/fcntl.h>
28#include <linux/socket.h>
29#include <linux/sockios.h>
30#include <linux/in.h>
31#include <linux/errno.h>
32#include <linux/timer.h>
33#include <linux/mm.h>
34#include <linux/inet.h>
35#include <linux/netdevice.h>
36#include <net/snmp.h>
37#include <net/ip.h>
38#include <net/ipv6.h>
39#include <net/icmp.h>
40#include <net/protocol.h>
41#include <linux/skbuff.h>
42#include <linux/proc_fs.h>
43#include <net/sock.h>
44#include <net/ping.h>
45#include <net/icmp.h>
46#include <net/udp.h>
47#include <net/route.h>
48#include <net/inet_common.h>
49#include <net/checksum.h>
50
51
52struct ping_table ping_table __read_mostly;
53
54u16 ping_port_rover;
55
56static inline int ping_hashfn(struct net *net, unsigned num, unsigned mask)
57{
58 int res = (num + net_hash_mix(net)) & mask;
59 pr_debug("hash(%d) = %d\n", num, res);
60 return res;
61}
62
63static inline struct hlist_nulls_head *ping_hashslot(struct ping_table *table,
64 struct net *net, unsigned num)
65{
66 return &table->hash[ping_hashfn(net, num, PING_HTABLE_MASK)];
67}
68
69static int ping_v4_get_port(struct sock *sk, unsigned short ident)
70{
71 struct hlist_nulls_node *node;
72 struct hlist_nulls_head *hlist;
73 struct inet_sock *isk, *isk2;
74 struct sock *sk2 = NULL;
75
76 isk = inet_sk(sk);
77 write_lock_bh(&ping_table.lock);
78 if (ident == 0) {
79 u32 i;
80 u16 result = ping_port_rover + 1;
81
82 for (i = 0; i < (1L << 16); i++, result++) {
83 if (!result)
84 result++; /* avoid zero */
85 hlist = ping_hashslot(&ping_table, sock_net(sk),
86 result);
87 ping_portaddr_for_each_entry(sk2, node, hlist) {
88 isk2 = inet_sk(sk2);
89
90 if (isk2->inet_num == result)
91 goto next_port;
92 }
93
94 /* found */
95 ping_port_rover = ident = result;
96 break;
97next_port:
98 ;
99 }
100 if (i >= (1L << 16))
101 goto fail;
102 } else {
103 hlist = ping_hashslot(&ping_table, sock_net(sk), ident);
104 ping_portaddr_for_each_entry(sk2, node, hlist) {
105 isk2 = inet_sk(sk2);
106
107 if ((isk2->inet_num == ident) &&
108 (sk2 != sk) &&
109 (!sk2->sk_reuse || !sk->sk_reuse))
110 goto fail;
111 }
112 }
113
114 pr_debug("found port/ident = %d\n", ident);
115 isk->inet_num = ident;
116 if (sk_unhashed(sk)) {
117 pr_debug("was not hashed\n");
118 sock_hold(sk);
119 hlist_nulls_add_head(&sk->sk_nulls_node, hlist);
120 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
121 }
122 write_unlock_bh(&ping_table.lock);
123 return 0;
124
125fail:
126 write_unlock_bh(&ping_table.lock);
127 return 1;
128}
129
130static void ping_v4_hash(struct sock *sk)
131{
132 pr_debug("ping_v4_hash(sk->port=%u)\n", inet_sk(sk)->inet_num);
133 BUG(); /* "Please do not press this button again." */
134}
135
136static void ping_v4_unhash(struct sock *sk)
137{
138 struct inet_sock *isk = inet_sk(sk);
139 pr_debug("ping_v4_unhash(isk=%p,isk->num=%u)\n", isk, isk->inet_num);
140 if (sk_hashed(sk)) {
141 struct hlist_nulls_head *hslot;
142
143 hslot = ping_hashslot(&ping_table, sock_net(sk), isk->inet_num);
144 write_lock_bh(&ping_table.lock);
145 hlist_nulls_del(&sk->sk_nulls_node);
146 sock_put(sk);
147 isk->inet_num = isk->inet_sport = 0;
148 sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
149 write_unlock_bh(&ping_table.lock);
150 }
151}
152
153struct sock *ping_v4_lookup(struct net *net, u32 saddr, u32 daddr,
154 u16 ident, int dif)
155{
156 struct hlist_nulls_head *hslot = ping_hashslot(&ping_table, net, ident);
157 struct sock *sk = NULL;
158 struct inet_sock *isk;
159 struct hlist_nulls_node *hnode;
160
161 pr_debug("try to find: num = %d, daddr = %ld, dif = %d\n",
162 (int)ident, (unsigned long)daddr, dif);
163 read_lock_bh(&ping_table.lock);
164
165 ping_portaddr_for_each_entry(sk, hnode, hslot) {
166 isk = inet_sk(sk);
167
168 pr_debug("found: %p: num = %d, daddr = %ld, dif = %d\n", sk,
169 (int)isk->inet_num, (unsigned long)isk->inet_rcv_saddr,
170 sk->sk_bound_dev_if);
171
172 pr_debug("iterate\n");
173 if (isk->inet_num != ident)
174 continue;
175 if (isk->inet_rcv_saddr && isk->inet_rcv_saddr != daddr)
176 continue;
177 if (sk->sk_bound_dev_if && sk->sk_bound_dev_if != dif)
178 continue;
179
180 sock_hold(sk);
181 goto exit;
182 }
183
184 sk = NULL;
185exit:
186 read_unlock_bh(&ping_table.lock);
187
188 return sk;
189}
190
191static int ping_init_sock(struct sock *sk)
192{
193 struct net *net = sock_net(sk);
194 gid_t group = current_egid();
195 gid_t range[2];
196 struct group_info *group_info = get_current_groups();
197 int i, j, count = group_info->ngroups;
198
199 inet_get_ping_group_range_net(net, range, range+1);
200 if (range[0] <= group && group <= range[1])
201 return 0;
202
203 for (i = 0; i < group_info->nblocks; i++) {
204 int cp_count = min_t(int, NGROUPS_PER_BLOCK, count);
205
206 for (j = 0; j < cp_count; j++) {
207 group = group_info->blocks[i][j];
208 if (range[0] <= group && group <= range[1])
209 return 0;
210 }
211
212 count -= cp_count;
213 }
214
215 return -EACCES;
216}
217
218static void ping_close(struct sock *sk, long timeout)
219{
220 pr_debug("ping_close(sk=%p,sk->num=%u)\n",
221 inet_sk(sk), inet_sk(sk)->inet_num);
222 pr_debug("isk->refcnt = %d\n", sk->sk_refcnt.counter);
223
224 sk_common_release(sk);
225}
226
227/*
228 * We need our own bind because there are no privileged id's == local ports.
229 * Moreover, we don't allow binding to multi- and broadcast addresses.
230 */
231
232static int ping_bind(struct sock *sk, struct sockaddr *uaddr, int addr_len)
233{
234 struct sockaddr_in *addr = (struct sockaddr_in *)uaddr;
235 struct inet_sock *isk = inet_sk(sk);
236 unsigned short snum;
237 int chk_addr_ret;
238 int err;
239
240 if (addr_len < sizeof(struct sockaddr_in))
241 return -EINVAL;
242
243 pr_debug("ping_v4_bind(sk=%p,sa_addr=%08x,sa_port=%d)\n",
244 sk, addr->sin_addr.s_addr, ntohs(addr->sin_port));
245
246 chk_addr_ret = inet_addr_type(sock_net(sk), addr->sin_addr.s_addr);
247 if (addr->sin_addr.s_addr == INADDR_ANY)
248 chk_addr_ret = RTN_LOCAL;
249
250 if ((sysctl_ip_nonlocal_bind == 0 &&
251 isk->freebind == 0 && isk->transparent == 0 &&
252 chk_addr_ret != RTN_LOCAL) ||
253 chk_addr_ret == RTN_MULTICAST ||
254 chk_addr_ret == RTN_BROADCAST)
255 return -EADDRNOTAVAIL;
256
257 lock_sock(sk);
258
259 err = -EINVAL;
260 if (isk->inet_num != 0)
261 goto out;
262
263 err = -EADDRINUSE;
264 isk->inet_rcv_saddr = isk->inet_saddr = addr->sin_addr.s_addr;
265 snum = ntohs(addr->sin_port);
266 if (ping_v4_get_port(sk, snum) != 0) {
267 isk->inet_saddr = isk->inet_rcv_saddr = 0;
268 goto out;
269 }
270
271 pr_debug("after bind(): num = %d, daddr = %ld, dif = %d\n",
272 (int)isk->inet_num,
273 (unsigned long) isk->inet_rcv_saddr,
274 (int)sk->sk_bound_dev_if);
275
276 err = 0;
277 if (isk->inet_rcv_saddr)
278 sk->sk_userlocks |= SOCK_BINDADDR_LOCK;
279 if (snum)
280 sk->sk_userlocks |= SOCK_BINDPORT_LOCK;
281 isk->inet_sport = htons(isk->inet_num);
282 isk->inet_daddr = 0;
283 isk->inet_dport = 0;
284 sk_dst_reset(sk);
285out:
286 release_sock(sk);
287 pr_debug("ping_v4_bind -> %d\n", err);
288 return err;
289}
290
291/*
292 * Is this a supported type of ICMP message?
293 */
294
295static inline int ping_supported(int type, int code)
296{
297 if (type == ICMP_ECHO && code == 0)
298 return 1;
299 return 0;
300}
301
302/*
303 * This routine is called by the ICMP module when it gets some
304 * sort of error condition.
305 */
306
307static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb);
308
309void ping_err(struct sk_buff *skb, u32 info)
310{
311 struct iphdr *iph = (struct iphdr *)skb->data;
312 struct icmphdr *icmph = (struct icmphdr *)(skb->data+(iph->ihl<<2));
313 struct inet_sock *inet_sock;
314 int type = icmph->type;
315 int code = icmph->code;
316 struct net *net = dev_net(skb->dev);
317 struct sock *sk;
318 int harderr;
319 int err;
320
321 /* We assume the packet has already been checked by icmp_unreach */
322
323 if (!ping_supported(icmph->type, icmph->code))
324 return;
325
326 pr_debug("ping_err(type=%04x,code=%04x,id=%04x,seq=%04x)\n", type,
327 code, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
328
329 sk = ping_v4_lookup(net, iph->daddr, iph->saddr,
330 ntohs(icmph->un.echo.id), skb->dev->ifindex);
331 if (sk == NULL) {
332 ICMP_INC_STATS_BH(net, ICMP_MIB_INERRORS);
333 pr_debug("no socket, dropping\n");
334 return; /* No socket for error */
335 }
336 pr_debug("err on socket %p\n", sk);
337
338 err = 0;
339 harderr = 0;
340 inet_sock = inet_sk(sk);
341
342 switch (type) {
343 default:
344 case ICMP_TIME_EXCEEDED:
345 err = EHOSTUNREACH;
346 break;
347 case ICMP_SOURCE_QUENCH:
348 /* This is not a real error but ping wants to see it.
349 * Report it with some fake errno. */
350 err = EREMOTEIO;
351 break;
352 case ICMP_PARAMETERPROB:
353 err = EPROTO;
354 harderr = 1;
355 break;
356 case ICMP_DEST_UNREACH:
357 if (code == ICMP_FRAG_NEEDED) { /* Path MTU discovery */
358 if (inet_sock->pmtudisc != IP_PMTUDISC_DONT) {
359 err = EMSGSIZE;
360 harderr = 1;
361 break;
362 }
363 goto out;
364 }
365 err = EHOSTUNREACH;
366 if (code <= NR_ICMP_UNREACH) {
367 harderr = icmp_err_convert[code].fatal;
368 err = icmp_err_convert[code].errno;
369 }
370 break;
371 case ICMP_REDIRECT:
372 /* See ICMP_SOURCE_QUENCH */
373 err = EREMOTEIO;
374 break;
375 }
376
377 /*
378 * RFC1122: OK. Passes ICMP errors back to application, as per
379 * 4.1.3.3.
380 */
381 if (!inet_sock->recverr) {
382 if (!harderr || sk->sk_state != TCP_ESTABLISHED)
383 goto out;
384 } else {
385 ip_icmp_error(sk, skb, err, 0 /* no remote port */,
386 info, (u8 *)icmph);
387 }
388 sk->sk_err = err;
389 sk->sk_error_report(sk);
390out:
391 sock_put(sk);
392}
393
394/*
395 * Copy and checksum an ICMP Echo packet from user space into a buffer.
396 */
397
398struct pingfakehdr {
399 struct icmphdr icmph;
400 struct iovec *iov;
401 u32 wcheck;
402};
403
404static int ping_getfrag(void *from, char * to,
405 int offset, int fraglen, int odd, struct sk_buff *skb)
406{
407 struct pingfakehdr *pfh = (struct pingfakehdr *)from;
408
409 if (offset == 0) {
410 if (fraglen < sizeof(struct icmphdr))
411 BUG();
412 if (csum_partial_copy_fromiovecend(to + sizeof(struct icmphdr),
413 pfh->iov, 0, fraglen - sizeof(struct icmphdr),
414 &pfh->wcheck))
415 return -EFAULT;
416
417 return 0;
418 }
419 if (offset < sizeof(struct icmphdr))
420 BUG();
421 if (csum_partial_copy_fromiovecend
422 (to, pfh->iov, offset - sizeof(struct icmphdr),
423 fraglen, &pfh->wcheck))
424 return -EFAULT;
425 return 0;
426}
427
428static int ping_push_pending_frames(struct sock *sk, struct pingfakehdr *pfh, struct flowi4 *fl4)
429{
430 struct sk_buff *skb = skb_peek(&sk->sk_write_queue);
431
432 pfh->wcheck = csum_partial((char *)&pfh->icmph,
433 sizeof(struct icmphdr), pfh->wcheck);
434 pfh->icmph.checksum = csum_fold(pfh->wcheck);
435 memcpy(icmp_hdr(skb), &pfh->icmph, sizeof(struct icmphdr));
436 skb->ip_summed = CHECKSUM_NONE;
437 return ip_push_pending_frames(sk, fl4);
438}
439
440int ping_sendmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
441 size_t len)
442{
443 struct net *net = sock_net(sk);
444 struct flowi4 fl4;
445 struct inet_sock *inet = inet_sk(sk);
446 struct ipcm_cookie ipc;
447 struct icmphdr user_icmph;
448 struct pingfakehdr pfh;
449 struct rtable *rt = NULL;
450 struct ip_options_data opt_copy;
451 int free = 0;
452 u32 saddr, daddr, faddr;
453 u8 tos;
454 int err;
455
456 pr_debug("ping_sendmsg(sk=%p,sk->num=%u)\n", inet, inet->inet_num);
457
458
459 if (len > 0xFFFF)
460 return -EMSGSIZE;
461
462 /*
463 * Check the flags.
464 */
465
466 /* Mirror BSD error message compatibility */
467 if (msg->msg_flags & MSG_OOB)
468 return -EOPNOTSUPP;
469
470 /*
471 * Fetch the ICMP header provided by the userland.
472 * iovec is modified!
473 */
474
475 if (memcpy_fromiovec((u8 *)&user_icmph, msg->msg_iov,
476 sizeof(struct icmphdr)))
477 return -EFAULT;
478 if (!ping_supported(user_icmph.type, user_icmph.code))
479 return -EINVAL;
480
481 /*
482 * Get and verify the address.
483 */
484
485 if (msg->msg_name) {
486 struct sockaddr_in *usin = (struct sockaddr_in *)msg->msg_name;
487 if (msg->msg_namelen < sizeof(*usin))
488 return -EINVAL;
489 if (usin->sin_family != AF_INET)
490 return -EINVAL;
491 daddr = usin->sin_addr.s_addr;
492 /* no remote port */
493 } else {
494 if (sk->sk_state != TCP_ESTABLISHED)
495 return -EDESTADDRREQ;
496 daddr = inet->inet_daddr;
497 /* no remote port */
498 }
499
500 ipc.addr = inet->inet_saddr;
501 ipc.opt = NULL;
502 ipc.oif = sk->sk_bound_dev_if;
503 ipc.tx_flags = 0;
504 err = sock_tx_timestamp(sk, &ipc.tx_flags);
505 if (err)
506 return err;
507
508 if (msg->msg_controllen) {
509 err = ip_cmsg_send(sock_net(sk), msg, &ipc);
510 if (err)
511 return err;
512 if (ipc.opt)
513 free = 1;
514 }
515 if (!ipc.opt) {
516 struct ip_options_rcu *inet_opt;
517
518 rcu_read_lock();
519 inet_opt = rcu_dereference(inet->inet_opt);
520 if (inet_opt) {
521 memcpy(&opt_copy, inet_opt,
522 sizeof(*inet_opt) + inet_opt->opt.optlen);
523 ipc.opt = &opt_copy.opt;
524 }
525 rcu_read_unlock();
526 }
527
528 saddr = ipc.addr;
529 ipc.addr = faddr = daddr;
530
531 if (ipc.opt && ipc.opt->opt.srr) {
532 if (!daddr)
533 return -EINVAL;
534 faddr = ipc.opt->opt.faddr;
535 }
536 tos = RT_TOS(inet->tos);
537 if (sock_flag(sk, SOCK_LOCALROUTE) ||
538 (msg->msg_flags & MSG_DONTROUTE) ||
539 (ipc.opt && ipc.opt->opt.is_strictroute)) {
540 tos |= RTO_ONLINK;
541 }
542
543 if (ipv4_is_multicast(daddr)) {
544 if (!ipc.oif)
545 ipc.oif = inet->mc_index;
546 if (!saddr)
547 saddr = inet->mc_addr;
548 }
549
550 flowi4_init_output(&fl4, ipc.oif, sk->sk_mark, tos,
551 RT_SCOPE_UNIVERSE, sk->sk_protocol,
552 inet_sk_flowi_flags(sk), faddr, saddr, 0, 0);
553
554 security_sk_classify_flow(sk, flowi4_to_flowi(&fl4));
555 rt = ip_route_output_flow(net, &fl4, sk);
556 if (IS_ERR(rt)) {
557 err = PTR_ERR(rt);
558 rt = NULL;
559 if (err == -ENETUNREACH)
560 IP_INC_STATS_BH(net, IPSTATS_MIB_OUTNOROUTES);
561 goto out;
562 }
563
564 err = -EACCES;
565 if ((rt->rt_flags & RTCF_BROADCAST) &&
566 !sock_flag(sk, SOCK_BROADCAST))
567 goto out;
568
569 if (msg->msg_flags & MSG_CONFIRM)
570 goto do_confirm;
571back_from_confirm:
572
573 if (!ipc.addr)
574 ipc.addr = fl4.daddr;
575
576 lock_sock(sk);
577
578 pfh.icmph.type = user_icmph.type; /* already checked */
579 pfh.icmph.code = user_icmph.code; /* ditto */
580 pfh.icmph.checksum = 0;
581 pfh.icmph.un.echo.id = inet->inet_sport;
582 pfh.icmph.un.echo.sequence = user_icmph.un.echo.sequence;
583 pfh.iov = msg->msg_iov;
584 pfh.wcheck = 0;
585
586 err = ip_append_data(sk, &fl4, ping_getfrag, &pfh, len,
587 0, &ipc, &rt, msg->msg_flags);
588 if (err)
589 ip_flush_pending_frames(sk);
590 else
591 err = ping_push_pending_frames(sk, &pfh, &fl4);
592 release_sock(sk);
593
594out:
595 ip_rt_put(rt);
596 if (free)
597 kfree(ipc.opt);
598 if (!err) {
599 icmp_out_count(sock_net(sk), user_icmph.type);
600 return len;
601 }
602 return err;
603
604do_confirm:
605 dst_confirm(&rt->dst);
606 if (!(msg->msg_flags & MSG_PROBE) || len)
607 goto back_from_confirm;
608 err = 0;
609 goto out;
610}
611
612/*
613 * IOCTL requests applicable to the UDP^H^H^HICMP protocol
614 */
615
616int ping_ioctl(struct sock *sk, int cmd, unsigned long arg)
617{
618 pr_debug("ping_ioctl(sk=%p,sk->num=%u,cmd=%d,arg=%lu)\n",
619 inet_sk(sk), inet_sk(sk)->inet_num, cmd, arg);
620 switch (cmd) {
621 case SIOCOUTQ:
622 case SIOCINQ:
623 return udp_ioctl(sk, cmd, arg);
624 default:
625 return -ENOIOCTLCMD;
626 }
627}
628
629int ping_recvmsg(struct kiocb *iocb, struct sock *sk, struct msghdr *msg,
630 size_t len, int noblock, int flags, int *addr_len)
631{
632 struct inet_sock *isk = inet_sk(sk);
633 struct sockaddr_in *sin = (struct sockaddr_in *)msg->msg_name;
634 struct sk_buff *skb;
635 int copied, err;
636
637 pr_debug("ping_recvmsg(sk=%p,sk->num=%u)\n", isk, isk->inet_num);
638
639 if (flags & MSG_OOB)
640 goto out;
641
642 if (addr_len)
643 *addr_len = sizeof(*sin);
644
645 if (flags & MSG_ERRQUEUE)
646 return ip_recv_error(sk, msg, len);
647
648 skb = skb_recv_datagram(sk, flags, noblock, &err);
649 if (!skb)
650 goto out;
651
652 copied = skb->len;
653 if (copied > len) {
654 msg->msg_flags |= MSG_TRUNC;
655 copied = len;
656 }
657
658 /* Don't bother checking the checksum */
659 err = skb_copy_datagram_iovec(skb, 0, msg->msg_iov, copied);
660 if (err)
661 goto done;
662
663 sock_recv_timestamp(msg, sk, skb);
664
665 /* Copy the address. */
666 if (sin) {
667 sin->sin_family = AF_INET;
668 sin->sin_port = 0 /* skb->h.uh->source */;
669 sin->sin_addr.s_addr = ip_hdr(skb)->saddr;
670 memset(sin->sin_zero, 0, sizeof(sin->sin_zero));
671 }
672 if (isk->cmsg_flags)
673 ip_cmsg_recv(msg, skb);
674 err = copied;
675
676done:
677 skb_free_datagram(sk, skb);
678out:
679 pr_debug("ping_recvmsg -> %d\n", err);
680 return err;
681}
682
683static int ping_queue_rcv_skb(struct sock *sk, struct sk_buff *skb)
684{
685 pr_debug("ping_queue_rcv_skb(sk=%p,sk->num=%d,skb=%p)\n",
686 inet_sk(sk), inet_sk(sk)->inet_num, skb);
687 if (sock_queue_rcv_skb(sk, skb) < 0) {
688 ICMP_INC_STATS_BH(sock_net(sk), ICMP_MIB_INERRORS);
689 kfree_skb(skb);
690 pr_debug("ping_queue_rcv_skb -> failed\n");
691 return -1;
692 }
693 return 0;
694}
695
696
697/*
698 * All we need to do is get the socket.
699 */
700
701void ping_rcv(struct sk_buff *skb)
702{
703 struct sock *sk;
704 struct net *net = dev_net(skb->dev);
705 struct iphdr *iph = ip_hdr(skb);
706 struct icmphdr *icmph = icmp_hdr(skb);
707 u32 saddr = iph->saddr;
708 u32 daddr = iph->daddr;
709
710 /* We assume the packet has already been checked by icmp_rcv */
711
712 pr_debug("ping_rcv(skb=%p,id=%04x,seq=%04x)\n",
713 skb, ntohs(icmph->un.echo.id), ntohs(icmph->un.echo.sequence));
714
715 /* Push ICMP header back */
716 skb_push(skb, skb->data - (u8 *)icmph);
717
718 sk = ping_v4_lookup(net, saddr, daddr, ntohs(icmph->un.echo.id),
719 skb->dev->ifindex);
720 if (sk != NULL) {
721 pr_debug("rcv on socket %p\n", sk);
722 ping_queue_rcv_skb(sk, skb_get(skb));
723 sock_put(sk);
724 return;
725 }
726 pr_debug("no socket, dropping\n");
727
728 /* We're called from icmp_rcv(). kfree_skb() is done there. */
729}
730
731struct proto ping_prot = {
732 .name = "PING",
733 .owner = THIS_MODULE,
734 .init = ping_init_sock,
735 .close = ping_close,
736 .connect = ip4_datagram_connect,
737 .disconnect = udp_disconnect,
738 .ioctl = ping_ioctl,
739 .setsockopt = ip_setsockopt,
740 .getsockopt = ip_getsockopt,
741 .sendmsg = ping_sendmsg,
742 .recvmsg = ping_recvmsg,
743 .bind = ping_bind,
744 .backlog_rcv = ping_queue_rcv_skb,
745 .hash = ping_v4_hash,
746 .unhash = ping_v4_unhash,
747 .get_port = ping_v4_get_port,
748 .obj_size = sizeof(struct inet_sock),
749};
750EXPORT_SYMBOL(ping_prot);
751
752#ifdef CONFIG_PROC_FS
753
754static struct sock *ping_get_first(struct seq_file *seq, int start)
755{
756 struct sock *sk;
757 struct ping_iter_state *state = seq->private;
758 struct net *net = seq_file_net(seq);
759
760 for (state->bucket = start; state->bucket < PING_HTABLE_SIZE;
761 ++state->bucket) {
762 struct hlist_nulls_node *node;
763 struct hlist_nulls_head *hslot = &ping_table.hash[state->bucket];
764
765 if (hlist_nulls_empty(hslot))
766 continue;
767
768 sk_nulls_for_each(sk, node, hslot) {
769 if (net_eq(sock_net(sk), net))
770 goto found;
771 }
772 }
773 sk = NULL;
774found:
775 return sk;
776}
777
778static struct sock *ping_get_next(struct seq_file *seq, struct sock *sk)
779{
780 struct ping_iter_state *state = seq->private;
781 struct net *net = seq_file_net(seq);
782
783 do {
784 sk = sk_nulls_next(sk);
785 } while (sk && (!net_eq(sock_net(sk), net)));
786
787 if (!sk)
788 return ping_get_first(seq, state->bucket + 1);
789 return sk;
790}
791
792static struct sock *ping_get_idx(struct seq_file *seq, loff_t pos)
793{
794 struct sock *sk = ping_get_first(seq, 0);
795
796 if (sk)
797 while (pos && (sk = ping_get_next(seq, sk)) != NULL)
798 --pos;
799 return pos ? NULL : sk;
800}
801
802static void *ping_seq_start(struct seq_file *seq, loff_t *pos)
803{
804 struct ping_iter_state *state = seq->private;
805 state->bucket = 0;
806
807 read_lock_bh(&ping_table.lock);
808
809 return *pos ? ping_get_idx(seq, *pos-1) : SEQ_START_TOKEN;
810}
811
812static void *ping_seq_next(struct seq_file *seq, void *v, loff_t *pos)
813{
814 struct sock *sk;
815
816 if (v == SEQ_START_TOKEN)
817 sk = ping_get_idx(seq, 0);
818 else
819 sk = ping_get_next(seq, v);
820
821 ++*pos;
822 return sk;
823}
824
825static void ping_seq_stop(struct seq_file *seq, void *v)
826{
827 read_unlock_bh(&ping_table.lock);
828}
829
830static void ping_format_sock(struct sock *sp, struct seq_file *f,
831 int bucket, int *len)
832{
833 struct inet_sock *inet = inet_sk(sp);
834 __be32 dest = inet->inet_daddr;
835 __be32 src = inet->inet_rcv_saddr;
836 __u16 destp = ntohs(inet->inet_dport);
837 __u16 srcp = ntohs(inet->inet_sport);
838
839 seq_printf(f, "%5d: %08X:%04X %08X:%04X"
840 " %02X %08X:%08X %02X:%08lX %08X %5d %8d %lu %d %pK %d%n",
841 bucket, src, srcp, dest, destp, sp->sk_state,
842 sk_wmem_alloc_get(sp),
843 sk_rmem_alloc_get(sp),
844 0, 0L, 0, sock_i_uid(sp), 0, sock_i_ino(sp),
845 atomic_read(&sp->sk_refcnt), sp,
846 atomic_read(&sp->sk_drops), len);
847}
848
849static int ping_seq_show(struct seq_file *seq, void *v)
850{
851 if (v == SEQ_START_TOKEN)
852 seq_printf(seq, "%-127s\n",
853 " sl local_address rem_address st tx_queue "
854 "rx_queue tr tm->when retrnsmt uid timeout "
855 "inode ref pointer drops");
856 else {
857 struct ping_iter_state *state = seq->private;
858 int len;
859
860 ping_format_sock(v, seq, state->bucket, &len);
861 seq_printf(seq, "%*s\n", 127 - len, "");
862 }
863 return 0;
864}
865
866static const struct seq_operations ping_seq_ops = {
867 .show = ping_seq_show,
868 .start = ping_seq_start,
869 .next = ping_seq_next,
870 .stop = ping_seq_stop,
871};
872
873static int ping_seq_open(struct inode *inode, struct file *file)
874{
875 return seq_open_net(inode, file, &ping_seq_ops,
876 sizeof(struct ping_iter_state));
877}
878
879static const struct file_operations ping_seq_fops = {
880 .open = ping_seq_open,
881 .read = seq_read,
882 .llseek = seq_lseek,
883 .release = seq_release_net,
884};
885
886static int ping_proc_register(struct net *net)
887{
888 struct proc_dir_entry *p;
889 int rc = 0;
890
891 p = proc_net_fops_create(net, "icmp", S_IRUGO, &ping_seq_fops);
892 if (!p)
893 rc = -ENOMEM;
894 return rc;
895}
896
897static void ping_proc_unregister(struct net *net)
898{
899 proc_net_remove(net, "icmp");
900}
901
902
903static int __net_init ping_proc_init_net(struct net *net)
904{
905 return ping_proc_register(net);
906}
907
908static void __net_exit ping_proc_exit_net(struct net *net)
909{
910 ping_proc_unregister(net);
911}
912
913static struct pernet_operations ping_net_ops = {
914 .init = ping_proc_init_net,
915 .exit = ping_proc_exit_net,
916};
917
918int __init ping_proc_init(void)
919{
920 return register_pernet_subsys(&ping_net_ops);
921}
922
923void ping_proc_exit(void)
924{
925 unregister_pernet_subsys(&ping_net_ops);
926}
927
928#endif
929
930void __init ping_init(void)
931{
932 int i;
933
934 for (i = 0; i < PING_HTABLE_SIZE; i++)
935 INIT_HLIST_NULLS_HEAD(&ping_table.hash[i], i);
936 rwlock_init(&ping_table.lock);
937}
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 321e6e84dbcc..28e8273bbef8 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -13,6 +13,7 @@
13#include <linux/seqlock.h> 13#include <linux/seqlock.h>
14#include <linux/init.h> 14#include <linux/init.h>
15#include <linux/slab.h> 15#include <linux/slab.h>
16#include <linux/nsproxy.h>
16#include <net/snmp.h> 17#include <net/snmp.h>
17#include <net/icmp.h> 18#include <net/icmp.h>
18#include <net/ip.h> 19#include <net/ip.h>
@@ -21,6 +22,7 @@
21#include <net/udp.h> 22#include <net/udp.h>
22#include <net/cipso_ipv4.h> 23#include <net/cipso_ipv4.h>
23#include <net/inet_frag.h> 24#include <net/inet_frag.h>
25#include <net/ping.h>
24 26
25static int zero; 27static int zero;
26static int tcp_retr1_max = 255; 28static int tcp_retr1_max = 255;
@@ -30,6 +32,8 @@ static int tcp_adv_win_scale_min = -31;
30static int tcp_adv_win_scale_max = 31; 32static int tcp_adv_win_scale_max = 31;
31static int ip_ttl_min = 1; 33static int ip_ttl_min = 1;
32static int ip_ttl_max = 255; 34static int ip_ttl_max = 255;
35static int ip_ping_group_range_min[] = { 0, 0 };
36static int ip_ping_group_range_max[] = { GID_T_MAX, GID_T_MAX };
33 37
34/* Update system visible IP port range */ 38/* Update system visible IP port range */
35static void set_local_port_range(int range[2]) 39static void set_local_port_range(int range[2])
@@ -68,6 +72,65 @@ static int ipv4_local_port_range(ctl_table *table, int write,
68 return ret; 72 return ret;
69} 73}
70 74
75
76void inet_get_ping_group_range_net(struct net *net, gid_t *low, gid_t *high)
77{
78 gid_t *data = net->ipv4.sysctl_ping_group_range;
79 unsigned seq;
80 do {
81 seq = read_seqbegin(&sysctl_local_ports.lock);
82
83 *low = data[0];
84 *high = data[1];
85 } while (read_seqretry(&sysctl_local_ports.lock, seq));
86}
87
88void inet_get_ping_group_range_table(struct ctl_table *table, gid_t *low, gid_t *high)
89{
90 gid_t *data = table->data;
91 unsigned seq;
92 do {
93 seq = read_seqbegin(&sysctl_local_ports.lock);
94
95 *low = data[0];
96 *high = data[1];
97 } while (read_seqretry(&sysctl_local_ports.lock, seq));
98}
99
100/* Update system visible IP port range */
101static void set_ping_group_range(struct ctl_table *table, int range[2])
102{
103 gid_t *data = table->data;
104 write_seqlock(&sysctl_local_ports.lock);
105 data[0] = range[0];
106 data[1] = range[1];
107 write_sequnlock(&sysctl_local_ports.lock);
108}
109
110/* Validate changes from /proc interface. */
111static int ipv4_ping_group_range(ctl_table *table, int write,
112 void __user *buffer,
113 size_t *lenp, loff_t *ppos)
114{
115 int ret;
116 gid_t range[2];
117 ctl_table tmp = {
118 .data = &range,
119 .maxlen = sizeof(range),
120 .mode = table->mode,
121 .extra1 = &ip_ping_group_range_min,
122 .extra2 = &ip_ping_group_range_max,
123 };
124
125 inet_get_ping_group_range_table(table, range, range + 1);
126 ret = proc_dointvec_minmax(&tmp, write, buffer, lenp, ppos);
127
128 if (write && ret == 0)
129 set_ping_group_range(table, range);
130
131 return ret;
132}
133
71static int proc_tcp_congestion_control(ctl_table *ctl, int write, 134static int proc_tcp_congestion_control(ctl_table *ctl, int write,
72 void __user *buffer, size_t *lenp, loff_t *ppos) 135 void __user *buffer, size_t *lenp, loff_t *ppos)
73{ 136{
@@ -677,6 +740,13 @@ static struct ctl_table ipv4_net_table[] = {
677 .mode = 0644, 740 .mode = 0644,
678 .proc_handler = proc_dointvec 741 .proc_handler = proc_dointvec
679 }, 742 },
743 {
744 .procname = "ping_group_range",
745 .data = &init_net.ipv4.sysctl_ping_group_range,
746 .maxlen = sizeof(init_net.ipv4.sysctl_ping_group_range),
747 .mode = 0644,
748 .proc_handler = ipv4_ping_group_range,
749 },
680 { } 750 { }
681}; 751};
682 752
@@ -711,8 +781,18 @@ static __net_init int ipv4_sysctl_init_net(struct net *net)
711 &net->ipv4.sysctl_icmp_ratemask; 781 &net->ipv4.sysctl_icmp_ratemask;
712 table[6].data = 782 table[6].data =
713 &net->ipv4.sysctl_rt_cache_rebuild_count; 783 &net->ipv4.sysctl_rt_cache_rebuild_count;
784 table[7].data =
785 &net->ipv4.sysctl_ping_group_range;
786
714 } 787 }
715 788
789 /*
790 * Sane defaults - nobody may create ping sockets.
791 * Boot scripts should set this to distro-specific group.
792 */
793 net->ipv4.sysctl_ping_group_range[0] = 1;
794 net->ipv4.sysctl_ping_group_range[1] = 0;
795
716 net->ipv4.sysctl_rt_cache_rebuild_count = 4; 796 net->ipv4.sysctl_rt_cache_rebuild_count = 4;
717 797
718 net->ipv4.ipv4_hdr = register_net_sysctl_table(net, 798 net->ipv4.ipv4_hdr = register_net_sysctl_table(net,
generated by cgit v1.2.2 (git 2.25.0) at 2024-11-22 13:15:29 -0500