2 ctdb system specific code to manage raw sockets on linux
4 Copyright (C) Ronnie Sahlberg 2007
5 Copyright (C) Andrew Tridgell 2007
6 Copyright (C) Marc Dequènes (Duck) 2009
8 This program is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 3 of the License, or
11 (at your option) any later version.
13 This program is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with this program; if not, see <http://www.gnu.org/licenses/>.
22 This file is a copy of 'common/system_linux.c' adapted for Hurd needs,
23 and inspired by 'common/system_aix.c' for the pcap usage.
27 #include "system/network.h"
28 #include "system/filesys.h"
29 #include "system/wait.h"
31 #include "lib/util/debug.h"
32 #include "lib/util/blocking.h"
34 #include "protocol/protocol.h"
36 #include <net/ethernet.h>
37 #include <netinet/ip6.h>
38 #include <net/if_arp.h>
41 #include "common/logging.h"
42 #include "common/system.h"
45 #define ETHERTYPE_IP6 0x86dd
49 calculate the tcp checksum for tcp over ipv6
51 static uint16_t tcp_checksum6(uint16_t *data
, size_t n
, struct ip6_hdr
*ip6
)
57 sum
+= uint16_checksum((uint16_t *)(void *)&ip6
->ip6_src
, 16);
58 sum
+= uint16_checksum((uint16_t *)(void *)&ip6
->ip6_dst
, 16);
61 phdr
[1] = htonl(ip6
->ip6_nxt
);
62 sum
+= uint16_checksum((uint16_t *)phdr
, 8);
64 sum
+= uint16_checksum(data
, n
);
66 sum
= (sum
& 0xFFFF) + (sum
>> 16);
67 sum
= (sum
& 0xFFFF) + (sum
>> 16);
77 send gratuitous arp reply after we have taken over an ip address
79 saddr is the address we are trying to claim
80 iface is the interface name we will be using to claim the address
82 int ctdb_sys_send_arp(const ctdb_sock_addr
*addr
, const char *iface
)
84 /* FIXME GNU/Hurd: We don't do gratuitous arp yet */
90 simple TCP checksum - assumes data is multiple of 2 bytes long
92 static uint16_t tcp_checksum(uint16_t *data
, size_t n
, struct iphdr
*ip
)
94 uint32_t sum
= uint16_checksum(data
, n
);
96 sum
+= uint16_checksum((uint16_t *)(void *)&ip
->saddr
,
98 sum
+= uint16_checksum((uint16_t *)(void *)&ip
->daddr
,
100 sum
+= ip
->protocol
+ n
;
101 sum
= (sum
& 0xFFFF) + (sum
>> 16);
102 sum
= (sum
& 0xFFFF) + (sum
>> 16);
112 Send tcp segment from the specified IP/port to the specified
115 This is used to trigger the receiving host into sending its own ACK,
116 which should trigger early detection of TCP reset by the client
119 This can also be used to send RST segments (if rst is true) and also
120 if correct seq and ack numbers are provided.
122 int ctdb_sys_send_tcp(const ctdb_sock_addr
*dest
,
123 const ctdb_sock_addr
*src
,
124 uint32_t seq
, uint32_t ack
, int rst
)
130 ctdb_sock_addr
*tmpdest
;
141 switch (src
->ip
.sin_family
) {
144 ip4pkt
.ip
.version
= 4;
145 ip4pkt
.ip
.ihl
= sizeof(ip4pkt
.ip
)/4;
146 ip4pkt
.ip
.tot_len
= htons(sizeof(ip4pkt
));
148 ip4pkt
.ip
.protocol
= IPPROTO_TCP
;
149 ip4pkt
.ip
.saddr
= src
->ip
.sin_addr
.s_addr
;
150 ip4pkt
.ip
.daddr
= dest
->ip
.sin_addr
.s_addr
;
153 ip4pkt
.tcp
.source
= src
->ip
.sin_port
;
154 ip4pkt
.tcp
.dest
= dest
->ip
.sin_port
;
155 ip4pkt
.tcp
.seq
= seq
;
156 ip4pkt
.tcp
.ack_seq
= ack
;
161 ip4pkt
.tcp
.doff
= sizeof(ip4pkt
.tcp
)/4;
162 /* this makes it easier to spot in a sniffer */
163 ip4pkt
.tcp
.window
= htons(1234);
164 ip4pkt
.tcp
.check
= tcp_checksum((uint16_t *)&ip4pkt
.tcp
, sizeof(ip4pkt
.tcp
), &ip4pkt
.ip
);
166 /* open a raw socket to send this segment from */
167 s
= socket(AF_INET
, SOCK_RAW
, IPPROTO_RAW
);
169 DEBUG(DEBUG_CRIT
,(__location__
" failed to open raw socket (%s)\n",
174 ret
= setsockopt(s
, IPPROTO_IP
, IP_HDRINCL
, &one
, sizeof(one
));
176 DEBUG(DEBUG_CRIT
,(__location__
" failed to setup IP headers (%s)\n",
182 ret
= sendto(s
, &ip4pkt
, sizeof(ip4pkt
), 0,
183 &dest
->ip
, sizeof(dest
->ip
));
186 if (ret
!= sizeof(ip4pkt
)) {
188 ("Failed sendto (%s)\n", strerror(saved_errno
)));
194 ip6pkt
.ip6
.ip6_vfc
= 0x60;
195 ip6pkt
.ip6
.ip6_plen
= htons(20);
196 ip6pkt
.ip6
.ip6_nxt
= IPPROTO_TCP
;
197 ip6pkt
.ip6
.ip6_hlim
= 64;
198 ip6pkt
.ip6
.ip6_src
= src
->ip6
.sin6_addr
;
199 ip6pkt
.ip6
.ip6_dst
= dest
->ip6
.sin6_addr
;
201 ip6pkt
.tcp
.source
= src
->ip6
.sin6_port
;
202 ip6pkt
.tcp
.dest
= dest
->ip6
.sin6_port
;
203 ip6pkt
.tcp
.seq
= seq
;
204 ip6pkt
.tcp
.ack_seq
= ack
;
209 ip6pkt
.tcp
.doff
= sizeof(ip6pkt
.tcp
)/4;
210 /* this makes it easier to spot in a sniffer */
211 ip6pkt
.tcp
.window
= htons(1234);
212 ip6pkt
.tcp
.check
= tcp_checksum6((uint16_t *)&ip6pkt
.tcp
, sizeof(ip6pkt
.tcp
), &ip6pkt
.ip6
);
214 s
= socket(AF_INET6
, SOCK_RAW
, IPPROTO_RAW
);
216 DEBUG(DEBUG_CRIT
, (__location__
" Failed to open sending socket\n"));
220 /* sendto() don't like if the port is set and the socket is
223 tmpdest
= discard_const(dest
);
224 tmpport
= tmpdest
->ip6
.sin6_port
;
226 tmpdest
->ip6
.sin6_port
= 0;
227 ret
= sendto(s
, &ip6pkt
, sizeof(ip6pkt
), 0,
228 &dest
->ip6
, sizeof(dest
->ip6
));
230 tmpdest
->ip6
.sin6_port
= tmpport
;
233 if (ret
!= sizeof(ip6pkt
)) {
235 ("Failed sendto (%s)\n", strerror(saved_errno
)));
241 DEBUG(DEBUG_CRIT
,(__location__
" not an ipv4/v6 address\n"));
249 This function is used to open a raw socket to capture from
251 int ctdb_sys_open_capture_socket(const char *iface
, void **private_data
)
255 pt
=pcap_open_live(iface
, 100, 0, 0, NULL
);
257 DEBUG(DEBUG_CRIT
,("Failed to open capture device %s\n", iface
));
260 *((pcap_t
**)private_data
) = pt
;
262 return pcap_fileno(pt
);
265 /* This function is used to close the capture socket
267 int ctdb_sys_close_capture_socket(void *private_data
)
269 pcap_t
*pt
= (pcap_t
*)private_data
;
276 called when the raw socket becomes readable
278 int ctdb_sys_read_tcp_packet(int s
, void *private_data
,
279 ctdb_sock_addr
*src
, ctdb_sock_addr
*dst
,
280 uint32_t *ack_seq
, uint32_t *seq
,
281 int *rst
, uint16_t *window
)
284 #define RCVPKTSIZE 100
285 char pkt
[RCVPKTSIZE
];
286 struct ether_header
*eth
;
291 ret
= recv(s
, pkt
, RCVPKTSIZE
, MSG_TRUNC
);
292 if (ret
< sizeof(*eth
)+sizeof(*ip
)) {
297 eth
= (struct ether_header
*)pkt
;
299 /* we want either IPv4 or IPv6 */
300 if (ntohs(eth
->ether_type
) == ETHERTYPE_IP
) {
302 ip
= (struct iphdr
*)(eth
+1);
304 /* We only want IPv4 packets */
305 if (ip
->version
!= 4) {
308 /* Dont look at fragments */
309 if ((ntohs(ip
->frag_off
)&0x1fff) != 0) {
312 /* we only want TCP */
313 if (ip
->protocol
!= IPPROTO_TCP
) {
317 /* make sure its not a short packet */
318 if (offsetof(struct tcphdr
, ack_seq
) + 4 +
319 (ip
->ihl
*4) + sizeof(*eth
) > ret
) {
323 tcp
= (struct tcphdr
*)((ip
->ihl
*4) + (char *)ip
);
325 /* tell the caller which one we've found */
326 src
->ip
.sin_family
= AF_INET
;
327 src
->ip
.sin_addr
.s_addr
= ip
->saddr
;
328 src
->ip
.sin_port
= tcp
->source
;
329 dst
->ip
.sin_family
= AF_INET
;
330 dst
->ip
.sin_addr
.s_addr
= ip
->daddr
;
331 dst
->ip
.sin_port
= tcp
->dest
;
332 *ack_seq
= tcp
->ack_seq
;
334 if (window
!= NULL
) {
335 *window
= tcp
->window
;
342 } else if (ntohs(eth
->ether_type
) == ETHERTYPE_IP6
) {
344 ip6
= (struct ip6_hdr
*)(eth
+1);
346 /* we only want TCP */
347 if (ip6
->ip6_nxt
!= IPPROTO_TCP
) {
352 tcp
= (struct tcphdr
*)(ip6
+1);
354 /* tell the caller which one we've found */
355 src
->ip6
.sin6_family
= AF_INET6
;
356 src
->ip6
.sin6_port
= tcp
->source
;
357 src
->ip6
.sin6_addr
= ip6
->ip6_src
;
359 dst
->ip6
.sin6_family
= AF_INET6
;
360 dst
->ip6
.sin6_port
= tcp
->dest
;
361 dst
->ip6
.sin6_addr
= ip6
->ip6_dst
;
363 *ack_seq
= tcp
->ack_seq
;
365 if (window
!= NULL
) {
366 *window
= tcp
->window
;
378 bool ctdb_sys_check_iface_exists(const char *iface
)
380 /* FIXME GNU/Hurd: Interface always considered present */
384 int ctdb_get_peer_pid(const int fd
, pid_t
*peer_pid
)
386 /* FIXME GNU/Hurd: get_peer_pid not implemented */