[tcpip] Allow supported address families to be detected at runtime
[ipxe.git] / src / net / ipv6.c
1 /*
2 * Copyright (C) 2013 Michael Brown <mbrown@fensystems.co.uk>.
3 *
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU General Public License as
6 * published by the Free Software Foundation; either version 2 of the
7 * License, or any later version.
8 *
9 * This program is distributed in the hope that it will be useful, but
10 * WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
12 * General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
17 * 02110-1301, USA.
18 */
19
20 FILE_LICENCE ( GPL2_OR_LATER );
21
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <stdlib.h>
25 #include <string.h>
26 #include <errno.h>
27 #include <assert.h>
28 #include <byteswap.h>
29 #include <ipxe/iobuf.h>
30 #include <ipxe/tcpip.h>
31 #include <ipxe/if_ether.h>
32 #include <ipxe/crc32.h>
33 #include <ipxe/fragment.h>
34 #include <ipxe/ipstat.h>
35 #include <ipxe/ndp.h>
36 #include <ipxe/ipv6.h>
37
38 /** @file
39 *
40 * IPv6 protocol
41 *
42 */
43
44 /* Disambiguate the various error causes */
45 #define EINVAL_LEN __einfo_error ( EINFO_EINVAL_LEN )
46 #define EINFO_EINVAL_LEN \
47 __einfo_uniqify ( EINFO_EINVAL, 0x01, "Invalid length" )
48 #define ENOTSUP_VER __einfo_error ( EINFO_ENOTSUP_VER )
49 #define EINFO_ENOTSUP_VER \
50 __einfo_uniqify ( EINFO_ENOTSUP, 0x01, "Unsupported version" )
51 #define ENOTSUP_HDR __einfo_error ( EINFO_ENOTSUP_HDR )
52 #define EINFO_ENOTSUP_HDR \
53 __einfo_uniqify ( EINFO_ENOTSUP, 0x02, "Unsupported header type" )
54 #define ENOTSUP_OPT __einfo_error ( EINFO_ENOTSUP_OPT )
55 #define EINFO_ENOTSUP_OPT \
56 __einfo_uniqify ( EINFO_ENOTSUP, 0x03, "Unsupported option" )
57
58 /** List of IPv6 miniroutes */
59 struct list_head ipv6_miniroutes = LIST_HEAD_INIT ( ipv6_miniroutes );
60
61 /** IPv6 statistics */
62 static struct ip_statistics ipv6_stats;
63
64 /** IPv6 statistics family */
65 struct ip_statistics_family
66 ipv6_statistics_family __ip_statistics_family ( IP_STATISTICS_IPV6 ) = {
67 .version = 6,
68 .stats = &ipv6_stats,
69 };
70
71 /**
72 * Determine debugging colour for IPv6 debug messages
73 *
74 * @v in IPv6 address
75 * @ret col Debugging colour (for DBGC())
76 */
77 static uint32_t ipv6col ( struct in6_addr *in ) {
78 return crc32_le ( 0, in, sizeof ( *in ) );
79 }
80
81 /**
82 * Dump IPv6 routing table entry
83 *
84 * @v miniroute Routing table entry
85 */
86 static inline __attribute__ (( always_inline )) void
87 ipv6_dump_miniroute ( struct ipv6_miniroute *miniroute ) {
88 struct net_device *netdev = miniroute->netdev;
89
90 DBGC ( netdev, "IPv6 %s has %s %s/%d", netdev->name,
91 ( ( miniroute->flags & IPV6_HAS_ADDRESS ) ?
92 "address" : "prefix" ),
93 inet6_ntoa ( &miniroute->address ), miniroute->prefix_len );
94 if ( miniroute->flags & IPV6_HAS_ROUTER )
95 DBGC ( netdev, " router %s", inet6_ntoa ( &miniroute->router ));
96 DBGC ( netdev, "\n" );
97 }
98
99 /**
100 * Check if network device has a specific IPv6 address
101 *
102 * @v netdev Network device
103 * @v addr IPv6 address
104 * @ret has_addr Network device has this IPv6 address
105 */
106 int ipv6_has_addr ( struct net_device *netdev, struct in6_addr *addr ) {
107 struct ipv6_miniroute *miniroute;
108
109 list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
110 if ( ( miniroute->netdev == netdev ) &&
111 ( miniroute->flags & IPV6_HAS_ADDRESS ) &&
112 ( memcmp ( &miniroute->address, addr,
113 sizeof ( miniroute->address ) ) == 0 ) ) {
114 /* Found matching address */
115 return 1;
116 }
117 }
118 return 0;
119 }
120
121 /**
122 * Check if IPv6 address is within a routing table entry's local network
123 *
124 * @v miniroute Routing table entry
125 * @v address IPv6 address
126 * @ret is_on_link Address is within this entry's local network
127 */
128 static int ipv6_is_on_link ( struct ipv6_miniroute *miniroute,
129 struct in6_addr *address ) {
130 unsigned int i;
131
132 for ( i = 0 ; i < ( sizeof ( address->s6_addr32 ) /
133 sizeof ( address->s6_addr32[0] ) ) ; i++ ) {
134 if ( (( address->s6_addr32[i] ^ miniroute->address.s6_addr32[i])
135 & miniroute->prefix_mask.s6_addr32[i] ) != 0 )
136 return 0;
137 }
138 return 1;
139 }
140
141 /**
142 * Find IPv6 routing table entry for a given address
143 *
144 * @v netdev Network device
145 * @v address IPv6 address
146 * @ret miniroute Routing table entry, or NULL if not found
147 */
148 static struct ipv6_miniroute * ipv6_miniroute ( struct net_device *netdev,
149 struct in6_addr *address ) {
150 struct ipv6_miniroute *miniroute;
151
152 list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
153 if ( ( miniroute->netdev == netdev ) &&
154 ipv6_is_on_link ( miniroute, address ) ) {
155 return miniroute;
156 }
157 }
158 return NULL;
159 }
160
161 /**
162 * Add IPv6 routing table entry
163 *
164 * @v netdev Network device
165 * @v address IPv6 address (or prefix)
166 * @v prefix_len Prefix length
167 * @v flags Flags
168 * @ret miniroute Routing table entry, or NULL on failure
169 */
170 static struct ipv6_miniroute * ipv6_add_miniroute ( struct net_device *netdev,
171 struct in6_addr *address,
172 unsigned int prefix_len,
173 unsigned int flags ) {
174 struct ipv6_miniroute *miniroute;
175 uint8_t *prefix_mask;
176
177 /* Create routing table entry */
178 miniroute = zalloc ( sizeof ( *miniroute ) );
179 if ( ! miniroute )
180 return NULL;
181 miniroute->netdev = netdev_get ( netdev );
182 memcpy ( &miniroute->address, address, sizeof ( miniroute->address ) );
183 miniroute->prefix_len = prefix_len;
184 assert ( prefix_len <= ( 8 * sizeof ( miniroute->prefix_mask ) ) );
185 for ( prefix_mask = miniroute->prefix_mask.s6_addr ; prefix_len >= 8 ;
186 prefix_mask++, prefix_len -= 8 ) {
187 *prefix_mask = 0xff;
188 }
189 if ( prefix_len )
190 *prefix_mask <<= ( 8 - prefix_len );
191 miniroute->flags = flags;
192 list_add ( &miniroute->list, &ipv6_miniroutes );
193 ipv6_dump_miniroute ( miniroute );
194
195 return miniroute;
196 }
197
198 /**
199 * Define IPv6 on-link prefix
200 *
201 * @v netdev Network device
202 * @v prefix IPv6 address prefix
203 * @v prefix_len Prefix length
204 * @v router Router address (or NULL)
205 * @ret rc Return status code
206 */
207 int ipv6_set_prefix ( struct net_device *netdev, struct in6_addr *prefix,
208 unsigned int prefix_len, struct in6_addr *router ) {
209 struct ipv6_miniroute *miniroute;
210 int changed;
211
212 /* Find or create routing table entry */
213 miniroute = ipv6_miniroute ( netdev, prefix );
214 if ( ! miniroute )
215 miniroute = ipv6_add_miniroute ( netdev, prefix, prefix_len, 0);
216 if ( ! miniroute )
217 return -ENOMEM;
218
219 /* Record router and add to start or end of list as appropriate */
220 list_del ( &miniroute->list );
221 if ( router ) {
222 changed = ( ( ! ( miniroute->flags & IPV6_HAS_ROUTER ) ) ||
223 ( memcmp ( &miniroute->router, router,
224 sizeof ( miniroute->router ) ) != 0 ) );
225 miniroute->flags |= IPV6_HAS_ROUTER;
226 memcpy ( &miniroute->router, router,
227 sizeof ( miniroute->router ) );
228 list_add_tail ( &miniroute->list, &ipv6_miniroutes );
229 } else {
230 changed = ( miniroute->flags & IPV6_HAS_ROUTER );
231 miniroute->flags &= ~IPV6_HAS_ROUTER;
232 list_add ( &miniroute->list, &ipv6_miniroutes );
233 }
234 if ( changed )
235 ipv6_dump_miniroute ( miniroute );
236
237 return 0;
238 }
239
240 /**
241 * Add IPv6 on-link address
242 *
243 * @v netdev Network device
244 * @v address IPv6 address
245 * @ret rc Return status code
246 *
247 * An on-link prefix for the address must already exist.
248 */
249 int ipv6_set_address ( struct net_device *netdev, struct in6_addr *address ) {
250 struct ipv6_miniroute *miniroute;
251 int changed;
252
253 /* Find routing table entry */
254 miniroute = ipv6_miniroute ( netdev, address );
255 if ( ! miniroute )
256 return -EADDRNOTAVAIL;
257
258 /* Record address */
259 changed = ( ( ! ( miniroute->flags & IPV6_HAS_ADDRESS ) ) ||
260 ( memcmp ( &miniroute->address, address,
261 sizeof ( miniroute->address ) ) != 0 ) );
262 memcpy ( &miniroute->address, address, sizeof ( miniroute->address ) );
263 miniroute->flags |= IPV6_HAS_ADDRESS;
264 if ( changed )
265 ipv6_dump_miniroute ( miniroute );
266
267 return 0;
268 }
269
270 /**
271 * Perform IPv6 routing
272 *
273 * @v scope_id Destination address scope ID (for link-local addresses)
274 * @v dest Final destination address
275 * @ret dest Next hop destination address
276 * @ret miniroute Routing table entry to use, or NULL if no route
277 */
278 static struct ipv6_miniroute * ipv6_route ( unsigned int scope_id,
279 struct in6_addr **dest ) {
280 struct ipv6_miniroute *miniroute;
281
282 /* Find first usable route in routing table */
283 list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
284
285 /* Skip closed network devices */
286 if ( ! netdev_is_open ( miniroute->netdev ) )
287 continue;
288
289 /* Skip routing table entries with no usable source address */
290 if ( ! ( miniroute->flags & IPV6_HAS_ADDRESS ) )
291 continue;
292
293 if ( IN6_IS_ADDR_NONGLOBAL ( *dest ) ) {
294
295 /* If destination is non-global, and the scope ID
296 * matches this network device, then use this route.
297 */
298 if ( miniroute->netdev->index == scope_id )
299 return miniroute;
300
301 } else {
302
303 /* If destination is an on-link global
304 * address, then use this route.
305 */
306 if ( ipv6_is_on_link ( miniroute, *dest ) )
307 return miniroute;
308
309 /* If destination is an off-link global
310 * address, and we have a default gateway,
311 * then use this route.
312 */
313 if ( miniroute->flags & IPV6_HAS_ROUTER ) {
314 *dest = &miniroute->router;
315 return miniroute;
316 }
317 }
318 }
319
320 return NULL;
321 }
322
323 /**
324 * Determine transmitting network device
325 *
326 * @v st_dest Destination network-layer address
327 * @ret netdev Transmitting network device, or NULL
328 */
329 static struct net_device * ipv6_netdev ( struct sockaddr_tcpip *st_dest ) {
330 struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
331 struct in6_addr *dest = &sin6_dest->sin6_addr;
332 struct ipv6_miniroute *miniroute;
333
334 /* Find routing table entry */
335 miniroute = ipv6_route ( sin6_dest->sin6_scope_id, &dest );
336 if ( ! miniroute )
337 return NULL;
338
339 return miniroute->netdev;
340 }
341
342 /**
343 * Check that received options can be safely ignored
344 *
345 * @v iphdr IPv6 header
346 * @v options Options extension header
347 * @v len Maximum length of header
348 * @ret rc Return status code
349 */
350 static int ipv6_check_options ( struct ipv6_header *iphdr,
351 struct ipv6_options_header *options,
352 size_t len ) {
353 struct ipv6_option *option = options->options;
354 struct ipv6_option *end = ( ( ( void * ) options ) + len );
355
356 while ( option < end ) {
357 if ( ! IPV6_CAN_IGNORE_OPT ( option->type ) ) {
358 DBGC ( ipv6col ( &iphdr->src ), "IPv6 unrecognised "
359 "option type %#02x:\n", option->type );
360 DBGC_HDA ( ipv6col ( &iphdr->src ), 0,
361 options, len );
362 return -ENOTSUP_OPT;
363 }
364 if ( option->type == IPV6_OPT_PAD1 ) {
365 option = ( ( ( void * ) option ) + 1 );
366 } else {
367 option = ( ( ( void * ) option->value ) + option->len );
368 }
369 }
370 return 0;
371 }
372
373 /**
374 * Check if fragment matches fragment reassembly buffer
375 *
376 * @v fragment Fragment reassembly buffer
377 * @v iobuf I/O buffer
378 * @v hdrlen Length of non-fragmentable potion of I/O buffer
379 * @ret is_fragment Fragment matches this reassembly buffer
380 */
381 static int ipv6_is_fragment ( struct fragment *fragment,
382 struct io_buffer *iobuf, size_t hdrlen ) {
383 struct ipv6_header *frag_iphdr = fragment->iobuf->data;
384 struct ipv6_fragment_header *frag_fhdr =
385 ( fragment->iobuf->data + fragment->hdrlen -
386 sizeof ( *frag_fhdr ) );
387 struct ipv6_header *iphdr = iobuf->data;
388 struct ipv6_fragment_header *fhdr =
389 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
390
391 return ( ( memcmp ( &iphdr->src, &frag_iphdr->src,
392 sizeof ( iphdr->src ) ) == 0 ) &&
393 ( fhdr->ident == frag_fhdr->ident ) );
394 }
395
396 /**
397 * Get fragment offset
398 *
399 * @v iobuf I/O buffer
400 * @v hdrlen Length of non-fragmentable potion of I/O buffer
401 * @ret offset Offset
402 */
403 static size_t ipv6_fragment_offset ( struct io_buffer *iobuf, size_t hdrlen ) {
404 struct ipv6_fragment_header *fhdr =
405 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
406
407 return ( ntohs ( fhdr->offset_more ) & IPV6_MASK_OFFSET );
408 }
409
410 /**
411 * Check if more fragments exist
412 *
413 * @v iobuf I/O buffer
414 * @v hdrlen Length of non-fragmentable potion of I/O buffer
415 * @ret more_frags More fragments exist
416 */
417 static int ipv6_more_fragments ( struct io_buffer *iobuf, size_t hdrlen ) {
418 struct ipv6_fragment_header *fhdr =
419 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
420
421 return ( fhdr->offset_more & htons ( IPV6_MASK_MOREFRAGS ) );
422 }
423
424 /** Fragment reassembler */
425 static struct fragment_reassembler ipv6_reassembler = {
426 .list = LIST_HEAD_INIT ( ipv6_reassembler.list ),
427 .is_fragment = ipv6_is_fragment,
428 .fragment_offset = ipv6_fragment_offset,
429 .more_fragments = ipv6_more_fragments,
430 .stats = &ipv6_stats,
431 };
432
433 /**
434 * Calculate IPv6 pseudo-header checksum
435 *
436 * @v iphdr IPv6 header
437 * @v len Payload length
438 * @v next_header Next header type
439 * @v csum Existing checksum
440 * @ret csum Updated checksum
441 */
442 static uint16_t ipv6_pshdr_chksum ( struct ipv6_header *iphdr, size_t len,
443 int next_header, uint16_t csum ) {
444 struct ipv6_pseudo_header pshdr;
445
446 /* Build pseudo-header */
447 memcpy ( &pshdr.src, &iphdr->src, sizeof ( pshdr.src ) );
448 memcpy ( &pshdr.dest, &iphdr->dest, sizeof ( pshdr.dest ) );
449 pshdr.len = htonl ( len );
450 memset ( pshdr.zero, 0, sizeof ( pshdr.zero ) );
451 pshdr.next_header = next_header;
452
453 /* Update the checksum value */
454 return tcpip_continue_chksum ( csum, &pshdr, sizeof ( pshdr ) );
455 }
456
457 /**
458 * Transmit IPv6 packet
459 *
460 * @v iobuf I/O buffer
461 * @v tcpip Transport-layer protocol
462 * @v st_src Source network-layer address
463 * @v st_dest Destination network-layer address
464 * @v netdev Network device to use if no route found, or NULL
465 * @v trans_csum Transport-layer checksum to complete, or NULL
466 * @ret rc Status
467 *
468 * This function expects a transport-layer segment and prepends the
469 * IPv6 header
470 */
471 static int ipv6_tx ( struct io_buffer *iobuf,
472 struct tcpip_protocol *tcpip_protocol,
473 struct sockaddr_tcpip *st_src,
474 struct sockaddr_tcpip *st_dest,
475 struct net_device *netdev,
476 uint16_t *trans_csum ) {
477 struct sockaddr_in6 *sin6_src = ( ( struct sockaddr_in6 * ) st_src );
478 struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
479 struct ipv6_miniroute *miniroute;
480 struct ipv6_header *iphdr;
481 struct in6_addr *src = NULL;
482 struct in6_addr *next_hop;
483 uint8_t ll_dest_buf[MAX_LL_ADDR_LEN];
484 const void *ll_dest;
485 size_t len;
486 int rc;
487
488 /* Update statistics */
489 ipv6_stats.out_requests++;
490
491 /* Fill up the IPv6 header, except source address */
492 len = iob_len ( iobuf );
493 iphdr = iob_push ( iobuf, sizeof ( *iphdr ) );
494 memset ( iphdr, 0, sizeof ( *iphdr ) );
495 iphdr->ver_tc_label = htonl ( IPV6_VER );
496 iphdr->len = htons ( len );
497 iphdr->next_header = tcpip_protocol->tcpip_proto;
498 iphdr->hop_limit = IPV6_HOP_LIMIT;
499 memcpy ( &iphdr->dest, &sin6_dest->sin6_addr, sizeof ( iphdr->dest ) );
500
501 /* Use routing table to identify next hop and transmitting netdev */
502 next_hop = &iphdr->dest;
503 if ( ( miniroute = ipv6_route ( sin6_dest->sin6_scope_id,
504 &next_hop ) ) != NULL ) {
505 src = &miniroute->address;
506 netdev = miniroute->netdev;
507 }
508 if ( ! netdev ) {
509 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 has no route to %s\n",
510 inet6_ntoa ( &iphdr->dest ) );
511 ipv6_stats.out_no_routes++;
512 rc = -ENETUNREACH;
513 goto err;
514 }
515 if ( sin6_src && ! IN6_IS_ADDR_UNSPECIFIED ( &sin6_src->sin6_addr ) )
516 src = &sin6_src->sin6_addr;
517 if ( src )
518 memcpy ( &iphdr->src, src, sizeof ( iphdr->src ) );
519
520 /* Fix up checksums */
521 if ( trans_csum ) {
522 *trans_csum = ipv6_pshdr_chksum ( iphdr, len,
523 tcpip_protocol->tcpip_proto,
524 *trans_csum );
525 }
526
527 /* Print IPv6 header for debugging */
528 DBGC2 ( ipv6col ( &iphdr->dest ), "IPv6 TX %s->",
529 inet6_ntoa ( &iphdr->src ) );
530 DBGC2 ( ipv6col ( &iphdr->dest ), "%s len %zd next %d\n",
531 inet6_ntoa ( &iphdr->dest ), len, iphdr->next_header );
532
533 /* Calculate link-layer destination address, if possible */
534 if ( IN6_IS_ADDR_MULTICAST ( next_hop ) ) {
535 /* Multicast address */
536 ipv6_stats.out_mcast_pkts++;
537 if ( ( rc = netdev->ll_protocol->mc_hash ( AF_INET6, next_hop,
538 ll_dest_buf ) ) !=0){
539 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not hash "
540 "multicast %s: %s\n", inet6_ntoa ( next_hop ),
541 strerror ( rc ) );
542 goto err;
543 }
544 ll_dest = ll_dest_buf;
545 } else {
546 /* Unicast address */
547 ll_dest = NULL;
548 }
549
550 /* Update statistics */
551 ipv6_stats.out_transmits++;
552 ipv6_stats.out_octets += iob_len ( iobuf );
553
554 /* Hand off to link layer (via NDP if applicable) */
555 if ( ll_dest ) {
556 if ( ( rc = net_tx ( iobuf, netdev, &ipv6_protocol, ll_dest,
557 netdev->ll_addr ) ) != 0 ) {
558 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
559 "transmit packet via %s: %s\n",
560 netdev->name, strerror ( rc ) );
561 return rc;
562 }
563 } else {
564 if ( ( rc = ndp_tx ( iobuf, netdev, next_hop, &iphdr->src,
565 netdev->ll_addr ) ) != 0 ) {
566 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
567 "transmit packet via %s: %s\n",
568 netdev->name, strerror ( rc ) );
569 return rc;
570 }
571 }
572
573 return 0;
574
575 err:
576 free_iob ( iobuf );
577 return rc;
578 }
579
580 /**
581 * Process incoming IPv6 packets
582 *
583 * @v iobuf I/O buffer
584 * @v netdev Network device
585 * @v ll_dest Link-layer destination address
586 * @v ll_source Link-layer destination source
587 * @v flags Packet flags
588 * @ret rc Return status code
589 *
590 * This function expects an IPv6 network datagram. It processes the
591 * headers and sends it to the transport layer.
592 */
593 static int ipv6_rx ( struct io_buffer *iobuf, struct net_device *netdev,
594 const void *ll_dest __unused,
595 const void *ll_source __unused,
596 unsigned int flags __unused ) {
597 struct ipv6_header *iphdr = iobuf->data;
598 union ipv6_extension_header *ext;
599 union {
600 struct sockaddr_in6 sin6;
601 struct sockaddr_tcpip st;
602 } src, dest;
603 uint16_t pshdr_csum;
604 size_t len;
605 size_t hdrlen;
606 size_t extlen;
607 int this_header;
608 int next_header;
609 int rc;
610
611 /* Update statistics */
612 ipv6_stats.in_receives++;
613 ipv6_stats.in_octets += iob_len ( iobuf );
614 if ( flags & LL_BROADCAST ) {
615 ipv6_stats.in_bcast_pkts++;
616 } else if ( flags & LL_MULTICAST ) {
617 ipv6_stats.in_mcast_pkts++;
618 }
619
620 /* Sanity check the IPv6 header */
621 if ( iob_len ( iobuf ) < sizeof ( *iphdr ) ) {
622 DBGC ( ipv6col ( &iphdr->src ), "IPv6 packet too short at %zd "
623 "bytes (min %zd bytes)\n", iob_len ( iobuf ),
624 sizeof ( *iphdr ) );
625 rc = -EINVAL_LEN;
626 goto err_header;
627 }
628 if ( ( iphdr->ver_tc_label & htonl ( IPV6_MASK_VER ) ) !=
629 htonl ( IPV6_VER ) ) {
630 DBGC ( ipv6col ( &iphdr->src ), "IPv6 version %#08x not "
631 "supported\n", ntohl ( iphdr->ver_tc_label ) );
632 rc = -ENOTSUP_VER;
633 goto err_header;
634 }
635
636 /* Truncate packet to specified length */
637 len = ntohs ( iphdr->len );
638 if ( len > iob_len ( iobuf ) ) {
639 DBGC ( ipv6col ( &iphdr->src ), "IPv6 length too long at %zd "
640 "bytes (packet is %zd bytes)\n", len, iob_len ( iobuf ));
641 ipv6_stats.in_truncated_pkts++;
642 rc = -EINVAL_LEN;
643 goto err_other;
644 }
645 iob_unput ( iobuf, ( iob_len ( iobuf ) - len - sizeof ( *iphdr ) ) );
646 hdrlen = sizeof ( *iphdr );
647
648 /* Print IPv6 header for debugging */
649 DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
650 inet6_ntoa ( &iphdr->dest ) );
651 DBGC2 ( ipv6col ( &iphdr->src ), "%s len %zd next %d\n",
652 inet6_ntoa ( &iphdr->src ), len, iphdr->next_header );
653
654 /* Discard unicast packets not destined for us */
655 if ( ( ! ( flags & LL_MULTICAST ) ) &&
656 ( ! ipv6_has_addr ( netdev, &iphdr->dest ) ) ) {
657 DBGC ( ipv6col ( &iphdr->src ), "IPv6 discarding non-local "
658 "unicast packet for %s\n", inet6_ntoa ( &iphdr->dest ) );
659 ipv6_stats.in_addr_errors++;
660 rc = -EPIPE;
661 goto err_other;
662 }
663
664 /* Process any extension headers */
665 next_header = iphdr->next_header;
666 while ( 1 ) {
667
668 /* Extract extension header */
669 this_header = next_header;
670 ext = ( iobuf->data + hdrlen );
671 extlen = sizeof ( ext->pad );
672 if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
673 DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
674 "extension header type %d at %zd bytes (min "
675 "%zd bytes)\n", this_header,
676 ( iob_len ( iobuf ) - hdrlen ), extlen );
677 rc = -EINVAL_LEN;
678 goto err_header;
679 }
680
681 /* Determine size of extension header (if applicable) */
682 if ( ( this_header == IPV6_HOPBYHOP ) ||
683 ( this_header == IPV6_DESTINATION ) ||
684 ( this_header == IPV6_ROUTING ) ) {
685 /* Length field is present */
686 extlen += ext->common.len;
687 } else if ( this_header == IPV6_FRAGMENT ) {
688 /* Length field is reserved and ignored (RFC2460) */
689 } else {
690 /* Not an extension header; assume rest is payload */
691 break;
692 }
693 if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
694 DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
695 "extension header type %d at %zd bytes (min "
696 "%zd bytes)\n", this_header,
697 ( iob_len ( iobuf ) - hdrlen ), extlen );
698 rc = -EINVAL_LEN;
699 goto err_header;
700 }
701 hdrlen += extlen;
702 next_header = ext->common.next_header;
703 DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
704 inet6_ntoa ( &iphdr->dest ) );
705 DBGC2 ( ipv6col ( &iphdr->src ), "%s ext type %d len %zd next "
706 "%d\n", inet6_ntoa ( &iphdr->src ), this_header,
707 extlen, next_header );
708
709 /* Process this extension header */
710 if ( ( this_header == IPV6_HOPBYHOP ) ||
711 ( this_header == IPV6_DESTINATION ) ) {
712
713 /* Check that all options can be ignored */
714 if ( ( rc = ipv6_check_options ( iphdr, &ext->options,
715 extlen ) ) != 0 )
716 goto err_header;
717
718 } else if ( this_header == IPV6_FRAGMENT ) {
719
720 /* Reassemble fragments */
721 iobuf = fragment_reassemble ( &ipv6_reassembler, iobuf,
722 &hdrlen );
723 if ( ! iobuf )
724 return 0;
725 iphdr = iobuf->data;
726 }
727 }
728
729 /* Construct socket address, calculate pseudo-header checksum,
730 * and hand off to transport layer
731 */
732 memset ( &src, 0, sizeof ( src ) );
733 src.sin6.sin6_family = AF_INET6;
734 memcpy ( &src.sin6.sin6_addr, &iphdr->src,
735 sizeof ( src.sin6.sin6_addr ) );
736 src.sin6.sin6_scope_id = netdev->index;
737 memset ( &dest, 0, sizeof ( dest ) );
738 dest.sin6.sin6_family = AF_INET6;
739 memcpy ( &dest.sin6.sin6_addr, &iphdr->dest,
740 sizeof ( dest.sin6.sin6_addr ) );
741 dest.sin6.sin6_scope_id = netdev->index;
742 iob_pull ( iobuf, hdrlen );
743 pshdr_csum = ipv6_pshdr_chksum ( iphdr, iob_len ( iobuf ),
744 next_header, TCPIP_EMPTY_CSUM );
745 if ( ( rc = tcpip_rx ( iobuf, netdev, next_header, &src.st, &dest.st,
746 pshdr_csum, &ipv6_stats ) ) != 0 ) {
747 DBGC ( ipv6col ( &src.sin6.sin6_addr ), "IPv6 received packet "
748 "rejected by stack: %s\n", strerror ( rc ) );
749 return rc;
750 }
751
752 return 0;
753
754 err_header:
755 ipv6_stats.in_hdr_errors++;
756 err_other:
757 free_iob ( iobuf );
758 return rc;
759 }
760
761 /**
762 * Parse IPv6 address
763 *
764 * @v string IPv6 address string
765 * @ret in IPv6 address to fill in
766 * @ret rc Return status code
767 */
768 int inet6_aton ( const char *string, struct in6_addr *in ) {
769 uint16_t *word = in->s6_addr16;
770 uint16_t *end = ( word + ( sizeof ( in->s6_addr16 ) /
771 sizeof ( in->s6_addr16[0] ) ) );
772 uint16_t *pad = NULL;
773 const char *nptr = string;
774 char *endptr;
775 unsigned long value;
776 size_t pad_len;
777 size_t move_len;
778
779 /* Parse string */
780 while ( 1 ) {
781
782 /* Parse current word */
783 value = strtoul ( nptr, &endptr, 16 );
784 if ( value > 0xffff ) {
785 DBG ( "IPv6 invalid word value %#lx in \"%s\"\n",
786 value, string );
787 return -EINVAL;
788 }
789 *(word++) = htons ( value );
790
791 /* Parse separator */
792 if ( ! *endptr )
793 break;
794 if ( *endptr != ':' ) {
795 DBG ( "IPv6 invalid separator '%c' in \"%s\"\n",
796 *endptr, string );
797 return -EINVAL;
798 }
799 if ( ( endptr == nptr ) && ( nptr != string ) ) {
800 if ( pad ) {
801 DBG ( "IPv6 invalid multiple \"::\" in "
802 "\"%s\"\n", string );
803 return -EINVAL;
804 }
805 pad = word;
806 }
807 nptr = ( endptr + 1 );
808
809 /* Check for overrun */
810 if ( word == end ) {
811 DBG ( "IPv6 too many words in \"%s\"\n", string );
812 return -EINVAL;
813 }
814 }
815
816 /* Insert padding if specified */
817 if ( pad ) {
818 move_len = ( ( ( void * ) word ) - ( ( void * ) pad ) );
819 pad_len = ( ( ( void * ) end ) - ( ( void * ) word ) );
820 memmove ( ( ( ( void * ) pad ) + pad_len ), pad, move_len );
821 memset ( pad, 0, pad_len );
822 } else if ( word != end ) {
823 DBG ( "IPv6 underlength address \"%s\"\n", string );
824 return -EINVAL;
825 }
826
827 return 0;
828 }
829
830 /**
831 * Convert IPv6 address to standard notation
832 *
833 * @v in IPv6 address
834 * @ret string IPv6 address string in canonical format
835 *
836 * RFC5952 defines the canonical format for IPv6 textual representation.
837 */
838 char * inet6_ntoa ( const struct in6_addr *in ) {
839 static char buf[41]; /* ":xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */
840 char *out = buf;
841 char *longest_start = NULL;
842 char *start = NULL;
843 int longest_len = 1;
844 int len = 0;
845 char *dest;
846 unsigned int i;
847 uint16_t value;
848
849 /* Format address, keeping track of longest run of zeros */
850 for ( i = 0 ; i < ( sizeof ( in->s6_addr16 ) /
851 sizeof ( in->s6_addr16[0] ) ) ; i++ ) {
852 value = ntohs ( in->s6_addr16[i] );
853 if ( value == 0 ) {
854 if ( len++ == 0 )
855 start = out;
856 if ( len > longest_len ) {
857 longest_start = start;
858 longest_len = len;
859 }
860 } else {
861 len = 0;
862 }
863 out += sprintf ( out, ":%x", value );
864 }
865
866 /* Abbreviate longest run of zeros, if applicable */
867 if ( longest_start ) {
868 dest = strcpy ( ( longest_start + 1 ),
869 ( longest_start + ( 2 * longest_len ) ) );
870 if ( dest[0] == '\0' )
871 dest[1] = '\0';
872 dest[0] = ':';
873 }
874 return ( ( longest_start == buf ) ? buf : ( buf + 1 ) );
875 }
876
877 /**
878 * Transcribe IPv6 address
879 *
880 * @v net_addr IPv6 address
881 * @ret string IPv6 address in standard notation
882 *
883 */
884 static const char * ipv6_ntoa ( const void *net_addr ) {
885 return inet6_ntoa ( net_addr );
886 }
887
888 /**
889 * Transcribe IPv6 socket address
890 *
891 * @v sa Socket address
892 * @ret string Socket address in standard notation
893 */
894 static const char * ipv6_sock_ntoa ( struct sockaddr *sa ) {
895 static char buf[ 39 /* "xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */ +
896 1 /* "%" */ + NETDEV_NAME_LEN + 1 /* NUL */ ];
897 struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
898 struct in6_addr *in = &sin6->sin6_addr;
899 struct net_device *netdev;
900 const char *netdev_name;
901
902 /* Identify network device, if applicable */
903 if ( IN6_IS_ADDR_NONGLOBAL ( in ) ) {
904 netdev = find_netdev_by_index ( sin6->sin6_scope_id );
905 netdev_name = ( netdev ? netdev->name : "UNKNOWN" );
906 } else {
907 netdev_name = NULL;
908 }
909
910 /* Format socket address */
911 snprintf ( buf, sizeof ( buf ), "%s%s%s", inet6_ntoa ( in ),
912 ( netdev_name ? "%" : "" ),
913 ( netdev_name ? netdev_name : "" ) );
914 return buf;
915 }
916
917 /**
918 * Parse IPv6 socket address
919 *
920 * @v string Socket address string
921 * @v sa Socket address to fill in
922 * @ret rc Return status code
923 */
924 static int ipv6_sock_aton ( const char *string, struct sockaddr *sa ) {
925 struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
926 struct in6_addr in;
927 struct net_device *netdev;
928 size_t len;
929 char *tmp;
930 char *in_string;
931 char *netdev_string;
932 int rc;
933
934 /* Create modifiable copy of string */
935 tmp = strdup ( string );
936 if ( ! tmp ) {
937 rc = -ENOMEM;
938 goto err_alloc;
939 }
940 in_string = tmp;
941
942 /* Strip surrounding "[...]", if present */
943 len = strlen ( in_string );
944 if ( ( in_string[0] == '[' ) && ( in_string[ len - 1 ] == ']' ) ) {
945 in_string[ len - 1 ] = '\0';
946 in_string++;
947 }
948
949 /* Split at network device name, if present */
950 netdev_string = strchr ( in_string, '%' );
951 if ( netdev_string )
952 *(netdev_string++) = '\0';
953
954 /* Parse IPv6 address portion */
955 if ( ( rc = inet6_aton ( in_string, &in ) ) != 0 )
956 goto err_inet6_aton;
957
958 /* Parse scope ID, if applicable */
959 if ( netdev_string ) {
960
961 /* Parse explicit network device name, if present */
962 netdev = find_netdev ( netdev_string );
963 if ( ! netdev ) {
964 rc = -ENODEV;
965 goto err_find_netdev;
966 }
967 sin6->sin6_scope_id = netdev->index;
968
969 } else if ( IN6_IS_ADDR_NONGLOBAL ( &in ) ) {
970
971 /* If no network device is explicitly specified for a
972 * link-local or multicast address, default to using
973 * "netX" (if existent).
974 */
975 netdev = last_opened_netdev();
976 if ( netdev )
977 sin6->sin6_scope_id = netdev->index;
978 }
979
980 /* Copy IPv6 address portion to socket address */
981 memcpy ( &sin6->sin6_addr, &in, sizeof ( sin6->sin6_addr ) );
982
983 err_find_netdev:
984 err_inet6_aton:
985 free ( tmp );
986 err_alloc:
987 return rc;
988 }
989
990 /** IPv6 protocol */
991 struct net_protocol ipv6_protocol __net_protocol = {
992 .name = "IPv6",
993 .net_proto = htons ( ETH_P_IPV6 ),
994 .net_addr_len = sizeof ( struct in6_addr ),
995 .rx = ipv6_rx,
996 .ntoa = ipv6_ntoa,
997 };
998
999 /** IPv6 TCPIP net protocol */
1000 struct tcpip_net_protocol ipv6_tcpip_protocol __tcpip_net_protocol = {
1001 .name = "IPv6",
1002 .sa_family = AF_INET6,
1003 .header_len = sizeof ( struct ipv6_header ),
1004 .net_protocol = &ipv6_protocol,
1005 .tx = ipv6_tx,
1006 .netdev = ipv6_netdev,
1007 };
1008
1009 /** IPv6 socket address converter */
1010 struct sockaddr_converter ipv6_sockaddr_converter __sockaddr_converter = {
1011 .family = AF_INET6,
1012 .ntoa = ipv6_sock_ntoa,
1013 .aton = ipv6_sock_aton,
1014 };
1015
1016 /**
1017 * Parse IPv6 address setting value
1018 *
1019 * @v type Setting type
1020 * @v value Formatted setting value
1021 * @v buf Buffer to contain raw value
1022 * @v len Length of buffer
1023 * @ret len Length of raw value, or negative error
1024 */
1025 int parse_ipv6_setting ( const struct setting_type *type __unused,
1026 const char *value, void *buf, size_t len ) {
1027 struct in6_addr ipv6;
1028 int rc;
1029
1030 /* Parse IPv6 address */
1031 if ( ( rc = inet6_aton ( value, &ipv6 ) ) != 0 )
1032 return rc;
1033
1034 /* Copy to buffer */
1035 if ( len > sizeof ( ipv6 ) )
1036 len = sizeof ( ipv6 );
1037 memcpy ( buf, &ipv6, len );
1038
1039 return ( sizeof ( ipv6 ) );
1040 }
1041
1042 /**
1043 * Format IPv6 address setting value
1044 *
1045 * @v type Setting type
1046 * @v raw Raw setting value
1047 * @v raw_len Length of raw setting value
1048 * @v buf Buffer to contain formatted value
1049 * @v len Length of buffer
1050 * @ret len Length of formatted value, or negative error
1051 */
1052 int format_ipv6_setting ( const struct setting_type *type __unused,
1053 const void *raw, size_t raw_len, char *buf,
1054 size_t len ) {
1055 const struct in6_addr *ipv6 = raw;
1056
1057 if ( raw_len < sizeof ( *ipv6 ) )
1058 return -EINVAL;
1059 return snprintf ( buf, len, "%s", inet6_ntoa ( ipv6 ) );
1060 }
1061
1062 /**
1063 * Create IPv6 network device
1064 *
1065 * @v netdev Network device
1066 * @ret rc Return status code
1067 */
1068 static int ipv6_probe ( struct net_device *netdev ) {
1069 struct ipv6_miniroute *miniroute;
1070 struct in6_addr address;
1071 int prefix_len;
1072 int rc;
1073
1074 /* Construct link-local address from EUI-64 as per RFC 2464 */
1075 memset ( &address, 0, sizeof ( address ) );
1076 prefix_len = ipv6_link_local ( &address, netdev );
1077 if ( prefix_len < 0 ) {
1078 rc = prefix_len;
1079 DBGC ( netdev, "IPv6 %s could not construct link-local "
1080 "address: %s\n", netdev->name, strerror ( rc ) );
1081 return rc;
1082 }
1083
1084 /* Create link-local address for this network device */
1085 miniroute = ipv6_add_miniroute ( netdev, &address, prefix_len,
1086 IPV6_HAS_ADDRESS );
1087 if ( ! miniroute )
1088 return -ENOMEM;
1089
1090 return 0;
1091 }
1092
1093 /**
1094 * Destroy IPv6 network device
1095 *
1096 * @v netdev Network device
1097 */
1098 static void ipv6_remove ( struct net_device *netdev ) {
1099 struct ipv6_miniroute *miniroute;
1100 struct ipv6_miniroute *tmp;
1101
1102 /* Delete all miniroutes for this network device */
1103 list_for_each_entry_safe ( miniroute, tmp, &ipv6_miniroutes, list ) {
1104 if ( miniroute->netdev == netdev ) {
1105 netdev_put ( miniroute->netdev );
1106 list_del ( &miniroute->list );
1107 free ( miniroute );
1108 }
1109 }
1110 }
1111
1112 /** IPv6 network device driver */
1113 struct net_driver ipv6_driver __net_driver = {
1114 .name = "IPv6",
1115 .probe = ipv6_probe,
1116 .remove = ipv6_remove,
1117 };
1118
1119 /* Drag in objects via ipv6_protocol */
1120 REQUIRING_SYMBOL ( ipv6_protocol );
1121
1122 /* Drag in ICMPv6 */
1123 REQUIRE_OBJECT ( icmpv6 );
1124
1125 /* Drag in NDP */
1126 REQUIRE_OBJECT ( ndp );