[zbin] Fix check for existence of most recent output byte
[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_LINKLOCAL ( *dest ) ||
294 IN6_IS_ADDR_MULTICAST ( *dest ) ) {
295
296 /* If destination is non-global, and the scope ID
297 * matches this network device, then use this route.
298 */
299 if ( miniroute->netdev->index == scope_id )
300 return miniroute;
301
302 } else {
303
304 /* If destination is an on-link global
305 * address, then use this route.
306 */
307 if ( ipv6_is_on_link ( miniroute, *dest ) )
308 return miniroute;
309
310 /* If destination is an off-link global
311 * address, and we have a default gateway,
312 * then use this route.
313 */
314 if ( miniroute->flags & IPV6_HAS_ROUTER ) {
315 *dest = &miniroute->router;
316 return miniroute;
317 }
318 }
319 }
320
321 return NULL;
322 }
323
324 /**
325 * Determine transmitting network device
326 *
327 * @v st_dest Destination network-layer address
328 * @ret netdev Transmitting network device, or NULL
329 */
330 static struct net_device * ipv6_netdev ( struct sockaddr_tcpip *st_dest ) {
331 struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
332 struct in6_addr *dest = &sin6_dest->sin6_addr;
333 struct ipv6_miniroute *miniroute;
334
335 /* Find routing table entry */
336 miniroute = ipv6_route ( sin6_dest->sin6_scope_id, &dest );
337 if ( ! miniroute )
338 return NULL;
339
340 return miniroute->netdev;
341 }
342
343 /**
344 * Check that received options can be safely ignored
345 *
346 * @v iphdr IPv6 header
347 * @v options Options extension header
348 * @v len Maximum length of header
349 * @ret rc Return status code
350 */
351 static int ipv6_check_options ( struct ipv6_header *iphdr,
352 struct ipv6_options_header *options,
353 size_t len ) {
354 struct ipv6_option *option = options->options;
355 struct ipv6_option *end = ( ( ( void * ) options ) + len );
356
357 while ( option < end ) {
358 if ( ! IPV6_CAN_IGNORE_OPT ( option->type ) ) {
359 DBGC ( ipv6col ( &iphdr->src ), "IPv6 unrecognised "
360 "option type %#02x:\n", option->type );
361 DBGC_HDA ( ipv6col ( &iphdr->src ), 0,
362 options, len );
363 return -ENOTSUP_OPT;
364 }
365 if ( option->type == IPV6_OPT_PAD1 ) {
366 option = ( ( ( void * ) option ) + 1 );
367 } else {
368 option = ( ( ( void * ) option->value ) + option->len );
369 }
370 }
371 return 0;
372 }
373
374 /**
375 * Check if fragment matches fragment reassembly buffer
376 *
377 * @v fragment Fragment reassembly buffer
378 * @v iobuf I/O buffer
379 * @v hdrlen Length of non-fragmentable potion of I/O buffer
380 * @ret is_fragment Fragment matches this reassembly buffer
381 */
382 static int ipv6_is_fragment ( struct fragment *fragment,
383 struct io_buffer *iobuf, size_t hdrlen ) {
384 struct ipv6_header *frag_iphdr = fragment->iobuf->data;
385 struct ipv6_fragment_header *frag_fhdr =
386 ( fragment->iobuf->data + fragment->hdrlen -
387 sizeof ( *frag_fhdr ) );
388 struct ipv6_header *iphdr = iobuf->data;
389 struct ipv6_fragment_header *fhdr =
390 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
391
392 return ( ( memcmp ( &iphdr->src, &frag_iphdr->src,
393 sizeof ( iphdr->src ) ) == 0 ) &&
394 ( fhdr->ident == frag_fhdr->ident ) );
395 }
396
397 /**
398 * Get fragment offset
399 *
400 * @v iobuf I/O buffer
401 * @v hdrlen Length of non-fragmentable potion of I/O buffer
402 * @ret offset Offset
403 */
404 static size_t ipv6_fragment_offset ( struct io_buffer *iobuf, size_t hdrlen ) {
405 struct ipv6_fragment_header *fhdr =
406 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
407
408 return ( ntohs ( fhdr->offset_more ) & IPV6_MASK_OFFSET );
409 }
410
411 /**
412 * Check if more fragments exist
413 *
414 * @v iobuf I/O buffer
415 * @v hdrlen Length of non-fragmentable potion of I/O buffer
416 * @ret more_frags More fragments exist
417 */
418 static int ipv6_more_fragments ( struct io_buffer *iobuf, size_t hdrlen ) {
419 struct ipv6_fragment_header *fhdr =
420 ( iobuf->data + hdrlen - sizeof ( *fhdr ) );
421
422 return ( fhdr->offset_more & htons ( IPV6_MASK_MOREFRAGS ) );
423 }
424
425 /** Fragment reassembler */
426 static struct fragment_reassembler ipv6_reassembler = {
427 .list = LIST_HEAD_INIT ( ipv6_reassembler.list ),
428 .is_fragment = ipv6_is_fragment,
429 .fragment_offset = ipv6_fragment_offset,
430 .more_fragments = ipv6_more_fragments,
431 .stats = &ipv6_stats,
432 };
433
434 /**
435 * Calculate IPv6 pseudo-header checksum
436 *
437 * @v iphdr IPv6 header
438 * @v len Payload length
439 * @v next_header Next header type
440 * @v csum Existing checksum
441 * @ret csum Updated checksum
442 */
443 static uint16_t ipv6_pshdr_chksum ( struct ipv6_header *iphdr, size_t len,
444 int next_header, uint16_t csum ) {
445 struct ipv6_pseudo_header pshdr;
446
447 /* Build pseudo-header */
448 memcpy ( &pshdr.src, &iphdr->src, sizeof ( pshdr.src ) );
449 memcpy ( &pshdr.dest, &iphdr->dest, sizeof ( pshdr.dest ) );
450 pshdr.len = htonl ( len );
451 memset ( pshdr.zero, 0, sizeof ( pshdr.zero ) );
452 pshdr.next_header = next_header;
453
454 /* Update the checksum value */
455 return tcpip_continue_chksum ( csum, &pshdr, sizeof ( pshdr ) );
456 }
457
458 /**
459 * Transmit IPv6 packet
460 *
461 * @v iobuf I/O buffer
462 * @v tcpip Transport-layer protocol
463 * @v st_src Source network-layer address
464 * @v st_dest Destination network-layer address
465 * @v netdev Network device to use if no route found, or NULL
466 * @v trans_csum Transport-layer checksum to complete, or NULL
467 * @ret rc Status
468 *
469 * This function expects a transport-layer segment and prepends the
470 * IPv6 header
471 */
472 static int ipv6_tx ( struct io_buffer *iobuf,
473 struct tcpip_protocol *tcpip_protocol,
474 struct sockaddr_tcpip *st_src,
475 struct sockaddr_tcpip *st_dest,
476 struct net_device *netdev,
477 uint16_t *trans_csum ) {
478 struct sockaddr_in6 *sin6_src = ( ( struct sockaddr_in6 * ) st_src );
479 struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
480 struct ipv6_miniroute *miniroute;
481 struct ipv6_header *iphdr;
482 struct in6_addr *src = NULL;
483 struct in6_addr *next_hop;
484 uint8_t ll_dest_buf[MAX_LL_ADDR_LEN];
485 const void *ll_dest;
486 size_t len;
487 int rc;
488
489 /* Update statistics */
490 ipv6_stats.out_requests++;
491
492 /* Fill up the IPv6 header, except source address */
493 len = iob_len ( iobuf );
494 iphdr = iob_push ( iobuf, sizeof ( *iphdr ) );
495 memset ( iphdr, 0, sizeof ( *iphdr ) );
496 iphdr->ver_tc_label = htonl ( IPV6_VER );
497 iphdr->len = htons ( len );
498 iphdr->next_header = tcpip_protocol->tcpip_proto;
499 iphdr->hop_limit = IPV6_HOP_LIMIT;
500 memcpy ( &iphdr->dest, &sin6_dest->sin6_addr, sizeof ( iphdr->dest ) );
501
502 /* Use routing table to identify next hop and transmitting netdev */
503 next_hop = &iphdr->dest;
504 if ( ( miniroute = ipv6_route ( sin6_dest->sin6_scope_id,
505 &next_hop ) ) != NULL ) {
506 src = &miniroute->address;
507 netdev = miniroute->netdev;
508 }
509 if ( ! netdev ) {
510 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 has no route to %s\n",
511 inet6_ntoa ( &iphdr->dest ) );
512 ipv6_stats.out_no_routes++;
513 rc = -ENETUNREACH;
514 goto err;
515 }
516 if ( sin6_src && ! IN6_IS_ADDR_UNSPECIFIED ( &sin6_src->sin6_addr ) )
517 src = &sin6_src->sin6_addr;
518 if ( src )
519 memcpy ( &iphdr->src, src, sizeof ( iphdr->src ) );
520
521 /* Fix up checksums */
522 if ( trans_csum ) {
523 *trans_csum = ipv6_pshdr_chksum ( iphdr, len,
524 tcpip_protocol->tcpip_proto,
525 *trans_csum );
526 }
527
528 /* Print IPv6 header for debugging */
529 DBGC2 ( ipv6col ( &iphdr->dest ), "IPv6 TX %s->",
530 inet6_ntoa ( &iphdr->src ) );
531 DBGC2 ( ipv6col ( &iphdr->dest ), "%s len %zd next %d\n",
532 inet6_ntoa ( &iphdr->dest ), len, iphdr->next_header );
533
534 /* Calculate link-layer destination address, if possible */
535 if ( IN6_IS_ADDR_MULTICAST ( next_hop ) ) {
536 /* Multicast address */
537 ipv6_stats.out_mcast_pkts++;
538 if ( ( rc = netdev->ll_protocol->mc_hash ( AF_INET6, next_hop,
539 ll_dest_buf ) ) !=0){
540 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not hash "
541 "multicast %s: %s\n", inet6_ntoa ( next_hop ),
542 strerror ( rc ) );
543 goto err;
544 }
545 ll_dest = ll_dest_buf;
546 } else {
547 /* Unicast address */
548 ll_dest = NULL;
549 }
550
551 /* Update statistics */
552 ipv6_stats.out_transmits++;
553 ipv6_stats.out_octets += iob_len ( iobuf );
554
555 /* Hand off to link layer (via NDP if applicable) */
556 if ( ll_dest ) {
557 if ( ( rc = net_tx ( iobuf, netdev, &ipv6_protocol, ll_dest,
558 netdev->ll_addr ) ) != 0 ) {
559 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
560 "transmit packet via %s: %s\n",
561 netdev->name, strerror ( rc ) );
562 return rc;
563 }
564 } else {
565 if ( ( rc = ndp_tx ( iobuf, netdev, next_hop, &iphdr->src,
566 netdev->ll_addr ) ) != 0 ) {
567 DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
568 "transmit packet via %s: %s\n",
569 netdev->name, strerror ( rc ) );
570 return rc;
571 }
572 }
573
574 return 0;
575
576 err:
577 free_iob ( iobuf );
578 return rc;
579 }
580
581 /**
582 * Process incoming IPv6 packets
583 *
584 * @v iobuf I/O buffer
585 * @v netdev Network device
586 * @v ll_dest Link-layer destination address
587 * @v ll_source Link-layer destination source
588 * @v flags Packet flags
589 * @ret rc Return status code
590 *
591 * This function expects an IPv6 network datagram. It processes the
592 * headers and sends it to the transport layer.
593 */
594 static int ipv6_rx ( struct io_buffer *iobuf, struct net_device *netdev,
595 const void *ll_dest __unused,
596 const void *ll_source __unused,
597 unsigned int flags __unused ) {
598 struct ipv6_header *iphdr = iobuf->data;
599 union ipv6_extension_header *ext;
600 union {
601 struct sockaddr_in6 sin6;
602 struct sockaddr_tcpip st;
603 } src, dest;
604 uint16_t pshdr_csum;
605 size_t len;
606 size_t hdrlen;
607 size_t extlen;
608 int this_header;
609 int next_header;
610 int rc;
611
612 /* Update statistics */
613 ipv6_stats.in_receives++;
614 ipv6_stats.in_octets += iob_len ( iobuf );
615 if ( flags & LL_BROADCAST ) {
616 ipv6_stats.in_bcast_pkts++;
617 } else if ( flags & LL_MULTICAST ) {
618 ipv6_stats.in_mcast_pkts++;
619 }
620
621 /* Sanity check the IPv6 header */
622 if ( iob_len ( iobuf ) < sizeof ( *iphdr ) ) {
623 DBGC ( ipv6col ( &iphdr->src ), "IPv6 packet too short at %zd "
624 "bytes (min %zd bytes)\n", iob_len ( iobuf ),
625 sizeof ( *iphdr ) );
626 rc = -EINVAL_LEN;
627 goto err_header;
628 }
629 if ( ( iphdr->ver_tc_label & htonl ( IPV6_MASK_VER ) ) !=
630 htonl ( IPV6_VER ) ) {
631 DBGC ( ipv6col ( &iphdr->src ), "IPv6 version %#08x not "
632 "supported\n", ntohl ( iphdr->ver_tc_label ) );
633 rc = -ENOTSUP_VER;
634 goto err_header;
635 }
636
637 /* Truncate packet to specified length */
638 len = ntohs ( iphdr->len );
639 if ( len > iob_len ( iobuf ) ) {
640 DBGC ( ipv6col ( &iphdr->src ), "IPv6 length too long at %zd "
641 "bytes (packet is %zd bytes)\n", len, iob_len ( iobuf ));
642 ipv6_stats.in_truncated_pkts++;
643 rc = -EINVAL_LEN;
644 goto err_other;
645 }
646 iob_unput ( iobuf, ( iob_len ( iobuf ) - len - sizeof ( *iphdr ) ) );
647 hdrlen = sizeof ( *iphdr );
648
649 /* Print IPv6 header for debugging */
650 DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
651 inet6_ntoa ( &iphdr->dest ) );
652 DBGC2 ( ipv6col ( &iphdr->src ), "%s len %zd next %d\n",
653 inet6_ntoa ( &iphdr->src ), len, iphdr->next_header );
654
655 /* Discard unicast packets not destined for us */
656 if ( ( ! ( flags & LL_MULTICAST ) ) &&
657 ( ! ipv6_has_addr ( netdev, &iphdr->dest ) ) ) {
658 DBGC ( ipv6col ( &iphdr->src ), "IPv6 discarding non-local "
659 "unicast packet for %s\n", inet6_ntoa ( &iphdr->dest ) );
660 ipv6_stats.in_addr_errors++;
661 rc = -EPIPE;
662 goto err_other;
663 }
664
665 /* Process any extension headers */
666 next_header = iphdr->next_header;
667 while ( 1 ) {
668
669 /* Extract extension header */
670 this_header = next_header;
671 ext = ( iobuf->data + hdrlen );
672 extlen = sizeof ( ext->pad );
673 if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
674 DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
675 "extension header type %d at %zd bytes (min "
676 "%zd bytes)\n", this_header,
677 ( iob_len ( iobuf ) - hdrlen ), extlen );
678 rc = -EINVAL_LEN;
679 goto err_header;
680 }
681
682 /* Determine size of extension header (if applicable) */
683 if ( ( this_header == IPV6_HOPBYHOP ) ||
684 ( this_header == IPV6_DESTINATION ) ||
685 ( this_header == IPV6_ROUTING ) ) {
686 /* Length field is present */
687 extlen += ext->common.len;
688 } else if ( this_header == IPV6_FRAGMENT ) {
689 /* Length field is reserved and ignored (RFC2460) */
690 } else {
691 /* Not an extension header; assume rest is payload */
692 break;
693 }
694 if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
695 DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
696 "extension header type %d at %zd bytes (min "
697 "%zd bytes)\n", this_header,
698 ( iob_len ( iobuf ) - hdrlen ), extlen );
699 rc = -EINVAL_LEN;
700 goto err_header;
701 }
702 hdrlen += extlen;
703 next_header = ext->common.next_header;
704 DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
705 inet6_ntoa ( &iphdr->dest ) );
706 DBGC2 ( ipv6col ( &iphdr->src ), "%s ext type %d len %zd next "
707 "%d\n", inet6_ntoa ( &iphdr->src ), this_header,
708 extlen, next_header );
709
710 /* Process this extension header */
711 if ( ( this_header == IPV6_HOPBYHOP ) ||
712 ( this_header == IPV6_DESTINATION ) ) {
713
714 /* Check that all options can be ignored */
715 if ( ( rc = ipv6_check_options ( iphdr, &ext->options,
716 extlen ) ) != 0 )
717 goto err_header;
718
719 } else if ( this_header == IPV6_FRAGMENT ) {
720
721 /* Reassemble fragments */
722 iobuf = fragment_reassemble ( &ipv6_reassembler, iobuf,
723 &hdrlen );
724 if ( ! iobuf )
725 return 0;
726 iphdr = iobuf->data;
727 }
728 }
729
730 /* Construct socket address, calculate pseudo-header checksum,
731 * and hand off to transport layer
732 */
733 memset ( &src, 0, sizeof ( src ) );
734 src.sin6.sin6_family = AF_INET6;
735 memcpy ( &src.sin6.sin6_addr, &iphdr->src,
736 sizeof ( src.sin6.sin6_addr ) );
737 src.sin6.sin6_scope_id = netdev->index;
738 memset ( &dest, 0, sizeof ( dest ) );
739 dest.sin6.sin6_family = AF_INET6;
740 memcpy ( &dest.sin6.sin6_addr, &iphdr->dest,
741 sizeof ( dest.sin6.sin6_addr ) );
742 dest.sin6.sin6_scope_id = netdev->index;
743 iob_pull ( iobuf, hdrlen );
744 pshdr_csum = ipv6_pshdr_chksum ( iphdr, iob_len ( iobuf ),
745 next_header, TCPIP_EMPTY_CSUM );
746 if ( ( rc = tcpip_rx ( iobuf, netdev, next_header, &src.st, &dest.st,
747 pshdr_csum, &ipv6_stats ) ) != 0 ) {
748 DBGC ( ipv6col ( &src.sin6.sin6_addr ), "IPv6 received packet "
749 "rejected by stack: %s\n", strerror ( rc ) );
750 return rc;
751 }
752
753 return 0;
754
755 err_header:
756 ipv6_stats.in_hdr_errors++;
757 err_other:
758 free_iob ( iobuf );
759 return rc;
760 }
761
762 /**
763 * Parse IPv6 address
764 *
765 * @v string IPv6 address string
766 * @ret in IPv6 address to fill in
767 * @ret rc Return status code
768 */
769 int inet6_aton ( const char *string, struct in6_addr *in ) {
770 uint16_t *word = in->s6_addr16;
771 uint16_t *end = ( word + ( sizeof ( in->s6_addr16 ) /
772 sizeof ( in->s6_addr16[0] ) ) );
773 uint16_t *pad = NULL;
774 const char *nptr = string;
775 char *endptr;
776 unsigned long value;
777 size_t pad_len;
778 size_t move_len;
779
780 /* Parse string */
781 while ( 1 ) {
782
783 /* Parse current word */
784 value = strtoul ( nptr, &endptr, 16 );
785 if ( value > 0xffff ) {
786 DBG ( "IPv6 invalid word value %#lx in \"%s\"\n",
787 value, string );
788 return -EINVAL;
789 }
790 *(word++) = htons ( value );
791
792 /* Parse separator */
793 if ( ! *endptr )
794 break;
795 if ( *endptr != ':' ) {
796 DBG ( "IPv6 invalid separator '%c' in \"%s\"\n",
797 *endptr, string );
798 return -EINVAL;
799 }
800 if ( ( endptr == nptr ) && ( nptr != string ) ) {
801 if ( pad ) {
802 DBG ( "IPv6 invalid multiple \"::\" in "
803 "\"%s\"\n", string );
804 return -EINVAL;
805 }
806 pad = word;
807 }
808 nptr = ( endptr + 1 );
809
810 /* Check for overrun */
811 if ( word == end ) {
812 DBG ( "IPv6 too many words in \"%s\"\n", string );
813 return -EINVAL;
814 }
815 }
816
817 /* Insert padding if specified */
818 if ( pad ) {
819 move_len = ( ( ( void * ) word ) - ( ( void * ) pad ) );
820 pad_len = ( ( ( void * ) end ) - ( ( void * ) word ) );
821 memmove ( ( ( ( void * ) pad ) + pad_len ), pad, move_len );
822 memset ( pad, 0, pad_len );
823 } else if ( word != end ) {
824 DBG ( "IPv6 underlength address \"%s\"\n", string );
825 return -EINVAL;
826 }
827
828 return 0;
829 }
830
831 /**
832 * Convert IPv6 address to standard notation
833 *
834 * @v in IPv6 address
835 * @ret string IPv6 address string in canonical format
836 *
837 * RFC5952 defines the canonical format for IPv6 textual representation.
838 */
839 char * inet6_ntoa ( const struct in6_addr *in ) {
840 static char buf[41]; /* ":xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */
841 char *out = buf;
842 char *longest_start = NULL;
843 char *start = NULL;
844 int longest_len = 1;
845 int len = 0;
846 char *dest;
847 unsigned int i;
848 uint16_t value;
849
850 /* Format address, keeping track of longest run of zeros */
851 for ( i = 0 ; i < ( sizeof ( in->s6_addr16 ) /
852 sizeof ( in->s6_addr16[0] ) ) ; i++ ) {
853 value = ntohs ( in->s6_addr16[i] );
854 if ( value == 0 ) {
855 if ( len++ == 0 )
856 start = out;
857 if ( len > longest_len ) {
858 longest_start = start;
859 longest_len = len;
860 }
861 } else {
862 len = 0;
863 }
864 out += sprintf ( out, ":%x", value );
865 }
866
867 /* Abbreviate longest run of zeros, if applicable */
868 if ( longest_start ) {
869 dest = strcpy ( ( longest_start + 1 ),
870 ( longest_start + ( 2 * longest_len ) ) );
871 if ( dest[0] == '\0' )
872 dest[1] = '\0';
873 dest[0] = ':';
874 }
875 return ( ( longest_start == buf ) ? buf : ( buf + 1 ) );
876 }
877
878 /**
879 * Transcribe IPv6 address
880 *
881 * @v net_addr IPv6 address
882 * @ret string IPv6 address in standard notation
883 *
884 */
885 static const char * ipv6_ntoa ( const void *net_addr ) {
886 return inet6_ntoa ( net_addr );
887 }
888
889 /**
890 * Transcribe IPv6 socket address
891 *
892 * @v sa Socket address
893 * @ret string Socket address in standard notation
894 */
895 static const char * ipv6_sock_ntoa ( struct sockaddr *sa ) {
896 static char buf[ 39 /* "xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */ +
897 1 /* "%" */ + NETDEV_NAME_LEN + 1 /* NUL */ ];
898 struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
899 struct in6_addr *in = &sin6->sin6_addr;
900 struct net_device *netdev;
901 const char *netdev_name;
902
903 /* Identify network device, if applicable */
904 if ( IN6_IS_ADDR_LINKLOCAL ( in ) || IN6_IS_ADDR_MULTICAST ( in ) ) {
905 netdev = find_netdev_by_index ( sin6->sin6_scope_id );
906 netdev_name = ( netdev ? netdev->name : "UNKNOWN" );
907 } else {
908 netdev_name = NULL;
909 }
910
911 /* Format socket address */
912 snprintf ( buf, sizeof ( buf ), "%s%s%s", inet6_ntoa ( in ),
913 ( netdev_name ? "%" : "" ),
914 ( netdev_name ? netdev_name : "" ) );
915 return buf;
916 }
917
918 /**
919 * Parse IPv6 socket address
920 *
921 * @v string Socket address string
922 * @v sa Socket address to fill in
923 * @ret rc Return status code
924 */
925 static int ipv6_sock_aton ( const char *string, struct sockaddr *sa ) {
926 struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
927 struct in6_addr in;
928 struct net_device *netdev;
929 size_t len;
930 char *tmp;
931 char *in_string;
932 char *netdev_string;
933 int rc;
934
935 /* Create modifiable copy of string */
936 tmp = strdup ( string );
937 if ( ! tmp ) {
938 rc = -ENOMEM;
939 goto err_alloc;
940 }
941 in_string = tmp;
942
943 /* Strip surrounding "[...]", if present */
944 len = strlen ( in_string );
945 if ( ( in_string[0] == '[' ) && ( in_string[ len - 1 ] == ']' ) ) {
946 in_string[ len - 1 ] = '\0';
947 in_string++;
948 }
949
950 /* Split at network device name, if present */
951 netdev_string = strchr ( in_string, '%' );
952 if ( netdev_string )
953 *(netdev_string++) = '\0';
954
955 /* Parse IPv6 address portion */
956 if ( ( rc = inet6_aton ( in_string, &in ) ) != 0 )
957 goto err_inet6_aton;
958
959 /* Parse network device name, if present */
960 if ( netdev_string ) {
961 netdev = find_netdev ( netdev_string );
962 if ( ! netdev ) {
963 rc = -ENODEV;
964 goto err_find_netdev;
965 }
966 sin6->sin6_scope_id = netdev->index;
967 }
968
969 /* Copy IPv6 address portion to socket address */
970 memcpy ( &sin6->sin6_addr, &in, sizeof ( sin6->sin6_addr ) );
971
972 err_find_netdev:
973 err_inet6_aton:
974 free ( tmp );
975 err_alloc:
976 return rc;
977 }
978
979 /** IPv6 protocol */
980 struct net_protocol ipv6_protocol __net_protocol = {
981 .name = "IPv6",
982 .net_proto = htons ( ETH_P_IPV6 ),
983 .net_addr_len = sizeof ( struct in6_addr ),
984 .rx = ipv6_rx,
985 .ntoa = ipv6_ntoa,
986 };
987
988 /** IPv6 TCPIP net protocol */
989 struct tcpip_net_protocol ipv6_tcpip_protocol __tcpip_net_protocol = {
990 .name = "IPv6",
991 .sa_family = AF_INET6,
992 .header_len = sizeof ( struct ipv6_header ),
993 .tx = ipv6_tx,
994 .netdev = ipv6_netdev,
995 };
996
997 /** IPv6 socket address converter */
998 struct sockaddr_converter ipv6_sockaddr_converter __sockaddr_converter = {
999 .family = AF_INET6,
1000 .ntoa = ipv6_sock_ntoa,
1001 .aton = ipv6_sock_aton,
1002 };
1003
1004 /**
1005 * Parse IPv6 address setting value
1006 *
1007 * @v type Setting type
1008 * @v value Formatted setting value
1009 * @v buf Buffer to contain raw value
1010 * @v len Length of buffer
1011 * @ret len Length of raw value, or negative error
1012 */
1013 int parse_ipv6_setting ( const struct setting_type *type __unused,
1014 const char *value, void *buf, size_t len ) {
1015 struct in6_addr ipv6;
1016 int rc;
1017
1018 /* Parse IPv6 address */
1019 if ( ( rc = inet6_aton ( value, &ipv6 ) ) != 0 )
1020 return rc;
1021
1022 /* Copy to buffer */
1023 if ( len > sizeof ( ipv6 ) )
1024 len = sizeof ( ipv6 );
1025 memcpy ( buf, &ipv6, len );
1026
1027 return ( sizeof ( ipv6 ) );
1028 }
1029
1030 /**
1031 * Format IPv6 address setting value
1032 *
1033 * @v type Setting type
1034 * @v raw Raw setting value
1035 * @v raw_len Length of raw setting value
1036 * @v buf Buffer to contain formatted value
1037 * @v len Length of buffer
1038 * @ret len Length of formatted value, or negative error
1039 */
1040 int format_ipv6_setting ( const struct setting_type *type __unused,
1041 const void *raw, size_t raw_len, char *buf,
1042 size_t len ) {
1043 const struct in6_addr *ipv6 = raw;
1044
1045 if ( raw_len < sizeof ( *ipv6 ) )
1046 return -EINVAL;
1047 return snprintf ( buf, len, "%s", inet6_ntoa ( ipv6 ) );
1048 }
1049
1050 /**
1051 * Create IPv6 network device
1052 *
1053 * @v netdev Network device
1054 * @ret rc Return status code
1055 */
1056 static int ipv6_probe ( struct net_device *netdev ) {
1057 struct ipv6_miniroute *miniroute;
1058 struct in6_addr address;
1059 int prefix_len;
1060 int rc;
1061
1062 /* Construct link-local address from EUI-64 as per RFC 2464 */
1063 memset ( &address, 0, sizeof ( address ) );
1064 prefix_len = ipv6_link_local ( &address, netdev );
1065 if ( prefix_len < 0 ) {
1066 rc = prefix_len;
1067 DBGC ( netdev, "IPv6 %s could not construct link-local "
1068 "address: %s\n", netdev->name, strerror ( rc ) );
1069 return rc;
1070 }
1071
1072 /* Create link-local address for this network device */
1073 miniroute = ipv6_add_miniroute ( netdev, &address, prefix_len,
1074 IPV6_HAS_ADDRESS );
1075 if ( ! miniroute )
1076 return -ENOMEM;
1077
1078 return 0;
1079 }
1080
1081 /**
1082 * Destroy IPv6 network device
1083 *
1084 * @v netdev Network device
1085 */
1086 static void ipv6_remove ( struct net_device *netdev ) {
1087 struct ipv6_miniroute *miniroute;
1088 struct ipv6_miniroute *tmp;
1089
1090 /* Delete all miniroutes for this network device */
1091 list_for_each_entry_safe ( miniroute, tmp, &ipv6_miniroutes, list ) {
1092 if ( miniroute->netdev == netdev ) {
1093 netdev_put ( miniroute->netdev );
1094 list_del ( &miniroute->list );
1095 free ( miniroute );
1096 }
1097 }
1098 }
1099
1100 /** IPv6 network device driver */
1101 struct net_driver ipv6_driver __net_driver = {
1102 .name = "IPv6",
1103 .probe = ipv6_probe,
1104 .remove = ipv6_remove,
1105 };
1106
1107 /* Drag in ICMPv6 */
1108 REQUIRE_OBJECT ( icmpv6 );
1109
1110 /* Drag in NDP */
1111 REQUIRE_OBJECT ( ndp );