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.
6 * The options processing module for ip.c
8 * Authors: A.N.Kuznetsov
12 #include <linux/capability.h>
13 #include <linux/module.h>
14 #include <linux/slab.h>
15 #include <linux/types.h>
16 #include <asm/uaccess.h>
17 #include <asm/unaligned.h>
18 #include <linux/skbuff.h>
20 #include <linux/icmp.h>
21 #include <linux/netdevice.h>
22 #include <linux/rtnetlink.h>
26 #include <net/route.h>
27 #include <net/cipso_ipv4.h>
30 * Write options to IP header, record destination address to
31 * source route option, address of outgoing interface
32 * (we should already know it, so that this function is allowed be
33 * called only after routing decision) and timestamp,
34 * if we originate this datagram.
36 * daddr is real destination address, next hop is recorded in IP header.
37 * saddr is address of outgoing interface.
40 void ip_options_build(struct sk_buff
*skb
, struct ip_options
*opt
,
41 __be32 daddr
, struct rtable
*rt
, int is_frag
)
43 unsigned char *iph
= skb_network_header(skb
);
45 memcpy(&(IPCB(skb
)->opt
), opt
, sizeof(struct ip_options
));
46 memcpy(iph
+sizeof(struct iphdr
), opt
->__data
, opt
->optlen
);
47 opt
= &(IPCB(skb
)->opt
);
50 memcpy(iph
+opt
->srr
+iph
[opt
->srr
+1]-4, &daddr
, 4);
54 ip_rt_get_source(iph
+opt
->rr
+iph
[opt
->rr
+2]-5, skb
, rt
);
56 ip_rt_get_source(iph
+opt
->ts
+iph
[opt
->ts
+2]-9, skb
, rt
);
57 if (opt
->ts_needtime
) {
61 midtime
= htonl((tv
.tv_sec
% 86400) * MSEC_PER_SEC
+ tv
.tv_nsec
/ NSEC_PER_MSEC
);
62 memcpy(iph
+opt
->ts
+iph
[opt
->ts
+2]-5, &midtime
, 4);
67 memset(iph
+opt
->rr
, IPOPT_NOP
, iph
[opt
->rr
+1]);
72 memset(iph
+opt
->ts
, IPOPT_NOP
, iph
[opt
->ts
+1]);
74 opt
->ts_needaddr
= opt
->ts_needtime
= 0;
79 * Provided (sopt, skb) points to received options,
80 * build in dopt compiled option set appropriate for answering.
81 * i.e. invert SRR option, copy anothers,
82 * and grab room in RR/TS options.
84 * NOTE: dopt cannot point to skb.
87 int ip_options_echo(struct ip_options
*dopt
, struct sk_buff
*skb
)
89 const struct ip_options
*sopt
;
90 unsigned char *sptr
, *dptr
;
95 memset(dopt
, 0, sizeof(struct ip_options
));
97 sopt
= &(IPCB(skb
)->opt
);
99 if (sopt
->optlen
== 0)
102 sptr
= skb_network_header(skb
);
105 daddr
= skb_rtable(skb
)->rt_spec_dst
;
108 optlen
= sptr
[sopt
->rr
+1];
109 soffset
= sptr
[sopt
->rr
+2];
110 dopt
->rr
= dopt
->optlen
+ sizeof(struct iphdr
);
111 memcpy(dptr
, sptr
+sopt
->rr
, optlen
);
112 if (sopt
->rr_needaddr
&& soffset
<= optlen
) {
113 if (soffset
+ 3 > optlen
)
115 dptr
[2] = soffset
+ 4;
116 dopt
->rr_needaddr
= 1;
119 dopt
->optlen
+= optlen
;
122 optlen
= sptr
[sopt
->ts
+1];
123 soffset
= sptr
[sopt
->ts
+2];
124 dopt
->ts
= dopt
->optlen
+ sizeof(struct iphdr
);
125 memcpy(dptr
, sptr
+sopt
->ts
, optlen
);
126 if (soffset
<= optlen
) {
127 if (sopt
->ts_needaddr
) {
128 if (soffset
+ 3 > optlen
)
130 dopt
->ts_needaddr
= 1;
133 if (sopt
->ts_needtime
) {
134 if (soffset
+ 3 > optlen
)
136 if ((dptr
[3]&0xF) != IPOPT_TS_PRESPEC
) {
137 dopt
->ts_needtime
= 1;
140 dopt
->ts_needtime
= 0;
142 if (soffset
+ 7 <= optlen
) {
145 memcpy(&addr
, dptr
+soffset
-1, 4);
146 if (inet_addr_type(dev_net(skb_dst(skb
)->dev
), addr
) != RTN_UNICAST
) {
147 dopt
->ts_needtime
= 1;
156 dopt
->optlen
+= optlen
;
159 unsigned char *start
= sptr
+sopt
->srr
;
165 if (soffset
> optlen
)
166 soffset
= optlen
+ 1;
169 memcpy(&faddr
, &start
[soffset
-1], 4);
170 for (soffset
-=4, doffset
=4; soffset
> 3; soffset
-=4, doffset
+=4)
171 memcpy(&dptr
[doffset
-1], &start
[soffset
-1], 4);
173 * RFC1812 requires to fix illegal source routes.
175 if (memcmp(&ip_hdr(skb
)->saddr
,
176 &start
[soffset
+ 3], 4) == 0)
180 memcpy(&start
[doffset
-1], &daddr
, 4);
186 dopt
->srr
= dopt
->optlen
+ sizeof(struct iphdr
);
187 dopt
->optlen
+= doffset
+3;
188 dopt
->is_strictroute
= sopt
->is_strictroute
;
192 optlen
= sptr
[sopt
->cipso
+1];
193 dopt
->cipso
= dopt
->optlen
+sizeof(struct iphdr
);
194 memcpy(dptr
, sptr
+sopt
->cipso
, optlen
);
196 dopt
->optlen
+= optlen
;
198 while (dopt
->optlen
& 3) {
206 * Options "fragmenting", just fill options not
207 * allowed in fragments with NOOPs.
208 * Simple and stupid 8), but the most efficient way.
211 void ip_options_fragment(struct sk_buff
* skb
)
213 unsigned char *optptr
= skb_network_header(skb
) + sizeof(struct iphdr
);
214 struct ip_options
* opt
= &(IPCB(skb
)->opt
);
228 if (optlen
<2 || optlen
>l
)
230 if (!IPOPT_COPIED(*optptr
))
231 memset(optptr
, IPOPT_NOOP
, optlen
);
237 opt
->rr_needaddr
= 0;
238 opt
->ts_needaddr
= 0;
239 opt
->ts_needtime
= 0;
243 * Verify options and fill pointers in struct options.
244 * Caller should clear *opt, and set opt->data.
245 * If opt == NULL, then skb->data should point to IP header.
248 int ip_options_compile(struct net
*net
,
249 struct ip_options
* opt
, struct sk_buff
* skb
)
253 unsigned char * optptr
;
255 unsigned char * pp_ptr
= NULL
;
256 struct rtable
*rt
= NULL
;
259 rt
= skb_rtable(skb
);
260 optptr
= (unsigned char *)&(ip_hdr(skb
)[1]);
262 optptr
= opt
->__data
;
263 iph
= optptr
- sizeof(struct iphdr
);
265 for (l
= opt
->optlen
; l
> 0; ) {
268 for (optptr
++, l
--; l
>0; optptr
++, l
--) {
269 if (*optptr
!= IPOPT_END
) {
281 if (optlen
<2 || optlen
>l
) {
296 /* NB: cf RFC-1812 5.2.4.1 */
302 if (optptr
[2] != 4 || optlen
< 7 || ((optlen
-3) & 3)) {
306 memcpy(&opt
->faddr
, &optptr
[3], 4);
308 memmove(&optptr
[3], &optptr
[7], optlen
-7);
310 opt
->is_strictroute
= (optptr
[0] == IPOPT_SSRR
);
311 opt
->srr
= optptr
- iph
;
326 if (optptr
[2] <= optlen
) {
327 if (optptr
[2]+3 > optlen
) {
332 memcpy(&optptr
[optptr
[2]-1], &rt
->rt_spec_dst
, 4);
336 opt
->rr_needaddr
= 1;
338 opt
->rr
= optptr
- iph
;
340 case IPOPT_TIMESTAMP
:
353 if (optptr
[2] <= optlen
) {
354 unsigned char *timeptr
= NULL
;
355 if (optptr
[2]+3 > optptr
[1]) {
359 switch (optptr
[3]&0xF) {
360 case IPOPT_TS_TSONLY
:
361 opt
->ts
= optptr
- iph
;
363 timeptr
= &optptr
[optptr
[2]-1];
364 opt
->ts_needtime
= 1;
367 case IPOPT_TS_TSANDADDR
:
368 if (optptr
[2]+7 > optptr
[1]) {
372 opt
->ts
= optptr
- iph
;
374 memcpy(&optptr
[optptr
[2]-1], &rt
->rt_spec_dst
, 4);
375 timeptr
= &optptr
[optptr
[2]+3];
377 opt
->ts_needaddr
= 1;
378 opt
->ts_needtime
= 1;
381 case IPOPT_TS_PRESPEC
:
382 if (optptr
[2]+7 > optptr
[1]) {
386 opt
->ts
= optptr
- iph
;
389 memcpy(&addr
, &optptr
[optptr
[2]-1], 4);
390 if (inet_addr_type(net
, addr
) == RTN_UNICAST
)
393 timeptr
= &optptr
[optptr
[2]+3];
395 opt
->ts_needtime
= 1;
399 if (!skb
&& !capable(CAP_NET_RAW
)) {
409 midtime
= (tv
.tv_sec
% 86400) * MSEC_PER_SEC
+ tv
.tv_nsec
/ NSEC_PER_MSEC
;
410 put_unaligned_be32(midtime
, timeptr
);
414 unsigned overflow
= optptr
[3]>>4;
415 if (overflow
== 15) {
419 opt
->ts
= optptr
- iph
;
421 optptr
[3] = (optptr
[3]&0xF)|((overflow
+1)<<4);
431 if (optptr
[2] == 0 && optptr
[3] == 0)
432 opt
->router_alert
= optptr
- iph
;
435 if ((!skb
&& !capable(CAP_NET_RAW
)) || opt
->cipso
) {
439 opt
->cipso
= optptr
- iph
;
440 if (cipso_v4_validate(skb
, &optptr
)) {
448 if (!skb
&& !capable(CAP_NET_RAW
)) {
464 icmp_send(skb
, ICMP_PARAMETERPROB
, 0, htonl((pp_ptr
-iph
)<<24));
468 EXPORT_SYMBOL(ip_options_compile
);
471 * Undo all the changes done by ip_options_compile().
474 void ip_options_undo(struct ip_options
* opt
)
477 unsigned char * optptr
= opt
->__data
+opt
->srr
-sizeof(struct iphdr
);
478 memmove(optptr
+7, optptr
+3, optptr
[1]-7);
479 memcpy(optptr
+3, &opt
->faddr
, 4);
481 if (opt
->rr_needaddr
) {
482 unsigned char * optptr
= opt
->__data
+opt
->rr
-sizeof(struct iphdr
);
484 memset(&optptr
[optptr
[2]-1], 0, 4);
487 unsigned char * optptr
= opt
->__data
+opt
->ts
-sizeof(struct iphdr
);
488 if (opt
->ts_needtime
) {
490 memset(&optptr
[optptr
[2]-1], 0, 4);
491 if ((optptr
[3]&0xF) == IPOPT_TS_PRESPEC
)
494 if (opt
->ts_needaddr
) {
496 memset(&optptr
[optptr
[2]-1], 0, 4);
501 static struct ip_options_rcu
*ip_options_get_alloc(const int optlen
)
503 return kzalloc(sizeof(struct ip_options_rcu
) + ((optlen
+ 3) & ~3),
507 static int ip_options_get_finish(struct net
*net
, struct ip_options_rcu
**optp
,
508 struct ip_options_rcu
*opt
, int optlen
)
511 opt
->opt
.__data
[optlen
++] = IPOPT_END
;
512 opt
->opt
.optlen
= optlen
;
513 if (optlen
&& ip_options_compile(net
, &opt
->opt
, NULL
)) {
522 int ip_options_get_from_user(struct net
*net
, struct ip_options_rcu
**optp
,
523 unsigned char __user
*data
, int optlen
)
525 struct ip_options_rcu
*opt
= ip_options_get_alloc(optlen
);
529 if (optlen
&& copy_from_user(opt
->opt
.__data
, data
, optlen
)) {
533 return ip_options_get_finish(net
, optp
, opt
, optlen
);
536 int ip_options_get(struct net
*net
, struct ip_options_rcu
**optp
,
537 unsigned char *data
, int optlen
)
539 struct ip_options_rcu
*opt
= ip_options_get_alloc(optlen
);
544 memcpy(opt
->opt
.__data
, data
, optlen
);
545 return ip_options_get_finish(net
, optp
, opt
, optlen
);
548 void ip_forward_options(struct sk_buff
*skb
)
550 struct ip_options
* opt
= &(IPCB(skb
)->opt
);
551 unsigned char * optptr
;
552 struct rtable
*rt
= skb_rtable(skb
);
553 unsigned char *raw
= skb_network_header(skb
);
555 if (opt
->rr_needaddr
) {
556 optptr
= (unsigned char *)raw
+ opt
->rr
;
557 ip_rt_get_source(&optptr
[optptr
[2]-5], skb
, rt
);
560 if (opt
->srr_is_hit
) {
561 int srrptr
, srrspace
;
563 optptr
= raw
+ opt
->srr
;
565 for ( srrptr
=optptr
[2], srrspace
= optptr
[1];
569 if (srrptr
+ 3 > srrspace
)
571 if (memcmp(&ip_hdr(skb
)->daddr
, &optptr
[srrptr
-1], 4) == 0)
574 if (srrptr
+ 3 <= srrspace
) {
576 ip_rt_get_source(&optptr
[srrptr
-1], skb
, rt
);
577 optptr
[2] = srrptr
+4;
578 } else if (net_ratelimit())
579 printk(KERN_CRIT
"ip_forward(): Argh! Destination lost!\n");
580 if (opt
->ts_needaddr
) {
581 optptr
= raw
+ opt
->ts
;
582 ip_rt_get_source(&optptr
[optptr
[2]-9], skb
, rt
);
586 if (opt
->is_changed
) {
588 ip_send_check(ip_hdr(skb
));
592 int ip_options_rcv_srr(struct sk_buff
*skb
)
594 struct ip_options
*opt
= &(IPCB(skb
)->opt
);
595 int srrspace
, srrptr
;
597 struct iphdr
*iph
= ip_hdr(skb
);
598 unsigned char *optptr
= skb_network_header(skb
) + opt
->srr
;
599 struct rtable
*rt
= skb_rtable(skb
);
601 unsigned long orefdst
;
607 if (skb
->pkt_type
!= PACKET_HOST
)
609 if (rt
->rt_type
== RTN_UNICAST
) {
610 if (!opt
->is_strictroute
)
612 icmp_send(skb
, ICMP_PARAMETERPROB
, 0, htonl(16<<24));
615 if (rt
->rt_type
!= RTN_LOCAL
)
618 for (srrptr
=optptr
[2], srrspace
= optptr
[1]; srrptr
<= srrspace
; srrptr
+= 4) {
619 if (srrptr
+ 3 > srrspace
) {
620 icmp_send(skb
, ICMP_PARAMETERPROB
, 0, htonl((opt
->srr
+2)<<24));
623 memcpy(&nexthop
, &optptr
[srrptr
-1], 4);
625 orefdst
= skb
->_skb_refdst
;
626 skb_dst_set(skb
, NULL
);
627 err
= ip_route_input(skb
, nexthop
, iph
->saddr
, iph
->tos
, skb
->dev
);
628 rt2
= skb_rtable(skb
);
629 if (err
|| (rt2
->rt_type
!= RTN_UNICAST
&& rt2
->rt_type
!= RTN_LOCAL
)) {
631 skb
->_skb_refdst
= orefdst
;
634 refdst_drop(orefdst
);
635 if (rt2
->rt_type
!= RTN_LOCAL
)
637 /* Superfast 8) loopback forward */
638 iph
->daddr
= nexthop
;
641 if (srrptr
<= srrspace
) {
647 EXPORT_SYMBOL(ip_options_rcv_srr
);