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 <strings.h>
27 #include <errno.h>
28 #include <assert.h>
29 #include <byteswap.h>
30 #include <ipxe/iobuf.h>
31 #include <ipxe/tcpip.h>
32 #include <ipxe/if_ether.h>
33 #include <ipxe/crc32.h>
34 #include <ipxe/fragment.h>
35 #include <ipxe/ipstat.h>
36 #include <ipxe/ndp.h>
37 #include <ipxe/ipv6.h>
38 
39 /** @file
40  *
41  * IPv6 protocol
42  *
43  */
44 
45 /* Disambiguate the various error causes */
46 #define EINVAL_LEN __einfo_error ( EINFO_EINVAL_LEN )
47 #define EINFO_EINVAL_LEN \
48 	__einfo_uniqify ( EINFO_EINVAL, 0x01, "Invalid length" )
49 #define ENOTSUP_VER __einfo_error ( EINFO_ENOTSUP_VER )
50 #define EINFO_ENOTSUP_VER \
51 	__einfo_uniqify ( EINFO_ENOTSUP, 0x01, "Unsupported version" )
52 #define ENOTSUP_HDR __einfo_error ( EINFO_ENOTSUP_HDR )
53 #define EINFO_ENOTSUP_HDR \
54 	__einfo_uniqify ( EINFO_ENOTSUP, 0x02, "Unsupported header type" )
55 #define ENOTSUP_OPT __einfo_error ( EINFO_ENOTSUP_OPT )
56 #define EINFO_ENOTSUP_OPT \
57 	__einfo_uniqify ( EINFO_ENOTSUP, 0x03, "Unsupported option" )
58 
59 /** List of IPv6 miniroutes */
60 struct list_head ipv6_miniroutes = LIST_HEAD_INIT ( ipv6_miniroutes );
61 
62 /** IPv6 statistics */
63 static struct ip_statistics ipv6_stats;
64 
65 /** IPv6 statistics family */
66 struct ip_statistics_family
67 ipv6_statistics_family __ip_statistics_family ( IP_STATISTICS_IPV6 ) = {
68 	.version = 6,
69 	.stats = &ipv6_stats,
70 };
71 
72 /**
73  * Determine debugging colour for IPv6 debug messages
74  *
75  * @v in		IPv6 address
76  * @ret col		Debugging colour (for DBGC())
77  */
ipv6col(struct in6_addr * in)78 static uint32_t ipv6col ( struct in6_addr *in ) {
79 	return crc32_le ( 0, in, sizeof ( *in ) );
80 }
81 
82 /**
83  * Determine IPv6 address scope
84  *
85  * @v addr		IPv6 address
86  * @ret scope		Address scope
87  */
ipv6_scope(const struct in6_addr * addr)88 static unsigned int ipv6_scope ( const struct in6_addr *addr ) {
89 
90 	/* Multicast addresses directly include a scope field */
91 	if ( IN6_IS_ADDR_MULTICAST ( addr ) )
92 		return ipv6_multicast_scope ( addr );
93 
94 	/* Link-local addresses have link-local scope */
95 	if ( IN6_IS_ADDR_LINKLOCAL ( addr ) )
96 		return IPV6_SCOPE_LINK_LOCAL;
97 
98 	/* Site-local addresses have site-local scope */
99 	if ( IN6_IS_ADDR_SITELOCAL ( addr ) )
100 		return IPV6_SCOPE_SITE_LOCAL;
101 
102 	/* Unique local addresses do not directly map to a defined
103 	 * scope.  They effectively have a scope which is wider than
104 	 * link-local but narrower than global.  Since the only
105 	 * multicast packets that we transmit are link-local, we can
106 	 * simply choose an arbitrary scope between link-local and
107 	 * global.
108 	 */
109 	if ( IN6_IS_ADDR_ULA ( addr ) )
110 		return IPV6_SCOPE_ORGANISATION_LOCAL;
111 
112 	/* All other addresses are assumed to be global */
113 	return IPV6_SCOPE_GLOBAL;
114 }
115 
116 /**
117  * Dump IPv6 routing table entry
118  *
119  * @v miniroute		Routing table entry
120  */
121 static inline __attribute__ (( always_inline )) void
ipv6_dump_miniroute(struct ipv6_miniroute * miniroute)122 ipv6_dump_miniroute ( struct ipv6_miniroute *miniroute ) {
123 	struct net_device *netdev = miniroute->netdev;
124 
125 	DBGC ( netdev, "IPv6 %s has %s %s/%d", netdev->name,
126 	       ( ( miniroute->flags & IPV6_HAS_ADDRESS ) ?
127 		 "address" : "prefix" ),
128 	       inet6_ntoa ( &miniroute->address ), miniroute->prefix_len );
129 	if ( miniroute->flags & IPV6_HAS_ROUTER )
130 		DBGC ( netdev, " router %s", inet6_ntoa ( &miniroute->router ));
131 	DBGC ( netdev, "\n" );
132 }
133 
134 /**
135  * Check if network device has a specific IPv6 address
136  *
137  * @v netdev		Network device
138  * @v addr		IPv6 address
139  * @ret has_addr	Network device has this IPv6 address
140  */
ipv6_has_addr(struct net_device * netdev,struct in6_addr * addr)141 int ipv6_has_addr ( struct net_device *netdev, struct in6_addr *addr ) {
142 	struct ipv6_miniroute *miniroute;
143 
144 	list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
145 		if ( ( miniroute->netdev == netdev ) &&
146 		     ( miniroute->flags & IPV6_HAS_ADDRESS ) &&
147 		     ( memcmp ( &miniroute->address, addr,
148 				sizeof ( miniroute->address ) ) == 0 ) ) {
149 			/* Found matching address */
150 			return 1;
151 		}
152 	}
153 	return 0;
154 }
155 
156 /**
157  * Count matching bits of an IPv6 routing table entry prefix
158  *
159  * @v miniroute		Routing table entry
160  * @v address		IPv6 address
161  * @ret match_len	Number of matching prefix bits
162  */
ipv6_match_len(struct ipv6_miniroute * miniroute,struct in6_addr * address)163 static unsigned int ipv6_match_len ( struct ipv6_miniroute *miniroute,
164 				     struct in6_addr *address ) {
165 	unsigned int match_len = 0;
166 	unsigned int i;
167 	uint32_t diff;
168 
169 	for ( i = 0 ; i < ( sizeof ( address->s6_addr32 ) /
170 			    sizeof ( address->s6_addr32[0] ) ) ; i++ ) {
171 
172 		diff = ntohl ( ~( ( ~( address->s6_addr32[i] ^
173 				       miniroute->address.s6_addr32[i] ) )
174 				  & miniroute->prefix_mask.s6_addr32[i] ) );
175 		match_len += 32;
176 		if ( diff ) {
177 			match_len -= flsl ( diff );
178 			break;
179 		}
180 	}
181 
182 	return match_len;
183 }
184 
185 /**
186  * Find IPv6 routing table entry for a given address
187  *
188  * @v netdev		Network device
189  * @v address		IPv6 address
190  * @ret miniroute	Routing table entry, or NULL if not found
191  */
ipv6_miniroute(struct net_device * netdev,struct in6_addr * address)192 static struct ipv6_miniroute * ipv6_miniroute ( struct net_device *netdev,
193 						struct in6_addr *address ) {
194 	struct ipv6_miniroute *miniroute;
195 	unsigned int match_len;
196 
197 	list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
198 		if ( miniroute->netdev != netdev )
199 			continue;
200 		match_len = ipv6_match_len ( miniroute, address );
201 		if ( match_len < miniroute->prefix_len )
202 			continue;
203 		return miniroute;
204 	}
205 	return NULL;
206 }
207 
208 /**
209  * Add IPv6 routing table entry
210  *
211  * @v netdev		Network device
212  * @v address		IPv6 address (or prefix)
213  * @v prefix_len	Prefix length
214  * @v router		Router address (if any)
215  * @ret rc		Return status code
216  */
ipv6_add_miniroute(struct net_device * netdev,struct in6_addr * address,unsigned int prefix_len,struct in6_addr * router)217 int ipv6_add_miniroute ( struct net_device *netdev, struct in6_addr *address,
218 			 unsigned int prefix_len, struct in6_addr *router ) {
219 	struct ipv6_miniroute *miniroute;
220 	uint8_t *prefix_mask;
221 	unsigned int remaining;
222 	unsigned int i;
223 
224 	/* Find or create routing table entry */
225 	miniroute = ipv6_miniroute ( netdev, address );
226 	if ( miniroute ) {
227 
228 		/* Remove from existing position in routing table */
229 		list_del ( &miniroute->list );
230 
231 	} else {
232 
233 		/* Create new routing table entry */
234 		miniroute = zalloc ( sizeof ( *miniroute ) );
235 		if ( ! miniroute )
236 			return -ENOMEM;
237 		miniroute->netdev = netdev_get ( netdev );
238 		memcpy ( &miniroute->address, address,
239 			 sizeof ( miniroute->address ) );
240 
241 		/* Default to prefix length of 64 if none specified */
242 		if ( ! prefix_len )
243 			prefix_len = IPV6_DEFAULT_PREFIX_LEN;
244 		miniroute->prefix_len = prefix_len;
245 		assert ( prefix_len <= IPV6_MAX_PREFIX_LEN );
246 
247 		/* Construct prefix mask */
248 		remaining = prefix_len;
249 		for ( prefix_mask = miniroute->prefix_mask.s6_addr ;
250 		      remaining >= 8 ; prefix_mask++, remaining -= 8 ) {
251 			*prefix_mask = 0xff;
252 		}
253 		if ( remaining )
254 			*prefix_mask <<= ( 8 - remaining );
255 	}
256 
257 	/* Add to start of routing table */
258 	list_add ( &miniroute->list, &ipv6_miniroutes );
259 
260 	/* Set or update address, if applicable */
261 	for ( i = 0 ; i < ( sizeof ( address->s6_addr32 ) /
262 			    sizeof ( address->s6_addr32[0] ) ) ; i++ ) {
263 		if ( ( address->s6_addr32[i] &
264 		       ~miniroute->prefix_mask.s6_addr32[i] ) != 0 ) {
265 			memcpy ( &miniroute->address, address,
266 				 sizeof ( miniroute->address ) );
267 			miniroute->flags |= IPV6_HAS_ADDRESS;
268 		}
269 	}
270 	if ( miniroute->prefix_len == IPV6_MAX_PREFIX_LEN )
271 		miniroute->flags |= IPV6_HAS_ADDRESS;
272 
273 	/* Update scope */
274 	miniroute->scope = ipv6_scope ( &miniroute->address );
275 
276 	/* Set or update router, if applicable */
277 	if ( router ) {
278 		memcpy ( &miniroute->router, router,
279 			 sizeof ( miniroute->router ) );
280 		miniroute->flags |= IPV6_HAS_ROUTER;
281 	}
282 
283 	ipv6_dump_miniroute ( miniroute );
284 	return 0;
285 }
286 
287 /**
288  * Delete IPv6 minirouting table entry
289  *
290  * @v miniroute		Routing table entry
291  */
ipv6_del_miniroute(struct ipv6_miniroute * miniroute)292 void ipv6_del_miniroute ( struct ipv6_miniroute *miniroute ) {
293 
294 	netdev_put ( miniroute->netdev );
295 	list_del ( &miniroute->list );
296 	free ( miniroute );
297 }
298 
299 /**
300  * Perform IPv6 routing
301  *
302  * @v scope_id		Destination address scope ID (for link-local addresses)
303  * @v dest		Final destination address
304  * @ret dest		Next hop destination address
305  * @ret miniroute	Routing table entry to use, or NULL if no route
306  */
ipv6_route(unsigned int scope_id,struct in6_addr ** dest)307 struct ipv6_miniroute * ipv6_route ( unsigned int scope_id,
308 				     struct in6_addr **dest ) {
309 	struct ipv6_miniroute *miniroute;
310 	struct ipv6_miniroute *chosen = NULL;
311 	unsigned int best = 0;
312 	unsigned int match_len;
313 	unsigned int score;
314 	unsigned int scope;
315 
316 	/* Calculate destination address scope */
317 	scope = ipv6_scope ( *dest );
318 
319 	/* Find first usable route in routing table */
320 	list_for_each_entry ( miniroute, &ipv6_miniroutes, list ) {
321 
322 		/* Skip closed network devices */
323 		if ( ! netdev_is_open ( miniroute->netdev ) )
324 			continue;
325 
326 		/* Skip entries with no usable source address */
327 		if ( ! ( miniroute->flags & IPV6_HAS_ADDRESS ) )
328 			continue;
329 
330 		/* Skip entries with a non-matching scope ID, if
331 		 * destination specifies a scope ID.
332 		 */
333 		if ( scope_id && ( miniroute->netdev->index != scope_id ) )
334 			continue;
335 
336 		/* Skip entries that are out of scope */
337 		if ( miniroute->scope < scope )
338 			continue;
339 
340 		/* Calculate match length */
341 		match_len = ipv6_match_len ( miniroute, *dest );
342 
343 		/* If destination is on-link, then use this route */
344 		if ( match_len >= miniroute->prefix_len )
345 			return miniroute;
346 
347 		/* If destination is unicast, then skip off-link
348 		 * entries with no router.
349 		 */
350 		if ( ! ( IN6_IS_ADDR_MULTICAST ( *dest ) ||
351 			 ( miniroute->flags & IPV6_HAS_ROUTER ) ) )
352 			continue;
353 
354 		/* Choose best route, defined as being the route with
355 		 * the smallest viable scope.  If two routes both have
356 		 * the same scope, then prefer the route with the
357 		 * longest match length.
358 		 */
359 		score = ( ( ( IPV6_SCOPE_MAX + 1 - miniroute->scope ) << 8 )
360 			  + match_len );
361 		if ( score > best ) {
362 			chosen = miniroute;
363 			best = score;
364 		}
365 	}
366 
367 	/* Return chosen route, if any */
368 	if ( chosen ) {
369 		if ( ! IN6_IS_ADDR_MULTICAST ( *dest ) )
370 			*dest = &chosen->router;
371 		return chosen;
372 	}
373 
374 	return NULL;
375 }
376 
377 /**
378  * Determine transmitting network device
379  *
380  * @v st_dest		Destination network-layer address
381  * @ret netdev		Transmitting network device, or NULL
382  */
ipv6_netdev(struct sockaddr_tcpip * st_dest)383 static struct net_device * ipv6_netdev ( struct sockaddr_tcpip *st_dest ) {
384 	struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
385 	struct in6_addr *dest = &sin6_dest->sin6_addr;
386 	struct ipv6_miniroute *miniroute;
387 
388 	/* Find routing table entry */
389 	miniroute = ipv6_route ( sin6_dest->sin6_scope_id, &dest );
390 	if ( ! miniroute )
391 		return NULL;
392 
393 	return miniroute->netdev;
394 }
395 
396 /**
397  * Check that received options can be safely ignored
398  *
399  * @v iphdr		IPv6 header
400  * @v options		Options extension header
401  * @v len		Maximum length of header
402  * @ret rc		Return status code
403  */
ipv6_check_options(struct ipv6_header * iphdr,struct ipv6_options_header * options,size_t len)404 static int ipv6_check_options ( struct ipv6_header *iphdr,
405 				struct ipv6_options_header *options,
406 				size_t len ) {
407 	struct ipv6_option *option = options->options;
408 	struct ipv6_option *end = ( ( ( void * ) options ) + len );
409 
410 	while ( option < end ) {
411 		if ( ! IPV6_CAN_IGNORE_OPT ( option->type ) ) {
412 			DBGC ( ipv6col ( &iphdr->src ), "IPv6 unrecognised "
413 			       "option type %#02x:\n", option->type );
414 			DBGC_HDA ( ipv6col ( &iphdr->src ), 0,
415 				   options, len );
416 			return -ENOTSUP_OPT;
417 		}
418 		if ( option->type == IPV6_OPT_PAD1 ) {
419 			option = ( ( ( void * ) option ) + 1 );
420 		} else {
421 			option = ( ( ( void * ) option->value ) + option->len );
422 		}
423 	}
424 	return 0;
425 }
426 
427 /**
428  * Check if fragment matches fragment reassembly buffer
429  *
430  * @v fragment		Fragment reassembly buffer
431  * @v iobuf		I/O buffer
432  * @v hdrlen		Length of non-fragmentable potion of I/O buffer
433  * @ret is_fragment	Fragment matches this reassembly buffer
434  */
ipv6_is_fragment(struct fragment * fragment,struct io_buffer * iobuf,size_t hdrlen)435 static int ipv6_is_fragment ( struct fragment *fragment,
436 			      struct io_buffer *iobuf, size_t hdrlen ) {
437 	struct ipv6_header *frag_iphdr = fragment->iobuf->data;
438 	struct ipv6_fragment_header *frag_fhdr =
439 		( fragment->iobuf->data + fragment->hdrlen -
440 		  sizeof ( *frag_fhdr ) );
441 	struct ipv6_header *iphdr = iobuf->data;
442 	struct ipv6_fragment_header *fhdr =
443 		( iobuf->data + hdrlen - sizeof ( *fhdr ) );
444 
445 	return ( ( memcmp ( &iphdr->src, &frag_iphdr->src,
446 			    sizeof ( iphdr->src ) ) == 0 ) &&
447 		 ( fhdr->ident == frag_fhdr->ident ) );
448 }
449 
450 /**
451  * Get fragment offset
452  *
453  * @v iobuf		I/O buffer
454  * @v hdrlen		Length of non-fragmentable potion of I/O buffer
455  * @ret offset		Offset
456  */
ipv6_fragment_offset(struct io_buffer * iobuf,size_t hdrlen)457 static size_t ipv6_fragment_offset ( struct io_buffer *iobuf, size_t hdrlen ) {
458 	struct ipv6_fragment_header *fhdr =
459 		( iobuf->data + hdrlen - sizeof ( *fhdr ) );
460 
461 	return ( ntohs ( fhdr->offset_more ) & IPV6_MASK_OFFSET );
462 }
463 
464 /**
465  * Check if more fragments exist
466  *
467  * @v iobuf		I/O buffer
468  * @v hdrlen		Length of non-fragmentable potion of I/O buffer
469  * @ret more_frags	More fragments exist
470  */
ipv6_more_fragments(struct io_buffer * iobuf,size_t hdrlen)471 static int ipv6_more_fragments ( struct io_buffer *iobuf, size_t hdrlen ) {
472 	struct ipv6_fragment_header *fhdr =
473 		( iobuf->data + hdrlen - sizeof ( *fhdr ) );
474 
475 	return ( fhdr->offset_more & htons ( IPV6_MASK_MOREFRAGS ) );
476 }
477 
478 /** Fragment reassembler */
479 static struct fragment_reassembler ipv6_reassembler = {
480 	.list = LIST_HEAD_INIT ( ipv6_reassembler.list ),
481 	.is_fragment = ipv6_is_fragment,
482 	.fragment_offset = ipv6_fragment_offset,
483 	.more_fragments = ipv6_more_fragments,
484 	.stats = &ipv6_stats,
485 };
486 
487 /**
488  * Calculate IPv6 pseudo-header checksum
489  *
490  * @v iphdr		IPv6 header
491  * @v len		Payload length
492  * @v next_header	Next header type
493  * @v csum		Existing checksum
494  * @ret csum		Updated checksum
495  */
ipv6_pshdr_chksum(struct ipv6_header * iphdr,size_t len,int next_header,uint16_t csum)496 static uint16_t ipv6_pshdr_chksum ( struct ipv6_header *iphdr, size_t len,
497 				    int next_header, uint16_t csum ) {
498 	struct ipv6_pseudo_header pshdr;
499 
500 	/* Build pseudo-header */
501 	memcpy ( &pshdr.src, &iphdr->src, sizeof ( pshdr.src ) );
502 	memcpy ( &pshdr.dest, &iphdr->dest, sizeof ( pshdr.dest ) );
503 	pshdr.len = htonl ( len );
504 	memset ( pshdr.zero, 0, sizeof ( pshdr.zero ) );
505 	pshdr.next_header = next_header;
506 
507 	/* Update the checksum value */
508 	return tcpip_continue_chksum ( csum, &pshdr, sizeof ( pshdr ) );
509 }
510 
511 /**
512  * Transmit IPv6 packet
513  *
514  * @v iobuf		I/O buffer
515  * @v tcpip		Transport-layer protocol
516  * @v st_src		Source network-layer address
517  * @v st_dest		Destination network-layer address
518  * @v netdev		Network device to use if no route found, or NULL
519  * @v trans_csum	Transport-layer checksum to complete, or NULL
520  * @ret rc		Status
521  *
522  * This function expects a transport-layer segment and prepends the
523  * IPv6 header
524  */
ipv6_tx(struct io_buffer * iobuf,struct tcpip_protocol * tcpip_protocol,struct sockaddr_tcpip * st_src,struct sockaddr_tcpip * st_dest,struct net_device * netdev,uint16_t * trans_csum)525 static int ipv6_tx ( struct io_buffer *iobuf,
526 		     struct tcpip_protocol *tcpip_protocol,
527 		     struct sockaddr_tcpip *st_src,
528 		     struct sockaddr_tcpip *st_dest,
529 		     struct net_device *netdev,
530 		     uint16_t *trans_csum ) {
531 	struct sockaddr_in6 *sin6_src = ( ( struct sockaddr_in6 * ) st_src );
532 	struct sockaddr_in6 *sin6_dest = ( ( struct sockaddr_in6 * ) st_dest );
533 	struct ipv6_miniroute *miniroute;
534 	struct ipv6_header *iphdr;
535 	struct in6_addr *src = NULL;
536 	struct in6_addr *next_hop;
537 	uint8_t ll_dest_buf[MAX_LL_ADDR_LEN];
538 	const void *ll_dest;
539 	size_t len;
540 	int rc;
541 
542 	/* Update statistics */
543 	ipv6_stats.out_requests++;
544 
545 	/* Fill up the IPv6 header, except source address */
546 	len = iob_len ( iobuf );
547 	iphdr = iob_push ( iobuf, sizeof ( *iphdr ) );
548 	memset ( iphdr, 0, sizeof ( *iphdr ) );
549 	iphdr->ver_tc_label = htonl ( IPV6_VER );
550 	iphdr->len = htons ( len );
551 	iphdr->next_header = tcpip_protocol->tcpip_proto;
552 	iphdr->hop_limit = IPV6_HOP_LIMIT;
553 	memcpy ( &iphdr->dest, &sin6_dest->sin6_addr, sizeof ( iphdr->dest ) );
554 
555 	/* Use routing table to identify next hop and transmitting netdev */
556 	next_hop = &iphdr->dest;
557 	if ( ( miniroute = ipv6_route ( sin6_dest->sin6_scope_id,
558 					&next_hop ) ) != NULL ) {
559 		src = &miniroute->address;
560 		netdev = miniroute->netdev;
561 	}
562 	if ( ! netdev ) {
563 		DBGC ( ipv6col ( &iphdr->dest ), "IPv6 has no route to %s\n",
564 		       inet6_ntoa ( &iphdr->dest ) );
565 		ipv6_stats.out_no_routes++;
566 		rc = -ENETUNREACH;
567 		goto err;
568 	}
569 	if ( sin6_src && ! IN6_IS_ADDR_UNSPECIFIED ( &sin6_src->sin6_addr ) )
570 		src = &sin6_src->sin6_addr;
571 	if ( src )
572 		memcpy ( &iphdr->src, src, sizeof ( iphdr->src ) );
573 
574 	/* Fix up checksums */
575 	if ( trans_csum ) {
576 		*trans_csum = ipv6_pshdr_chksum ( iphdr, len,
577 						  tcpip_protocol->tcpip_proto,
578 						  *trans_csum );
579 		if ( ! *trans_csum )
580 			*trans_csum = tcpip_protocol->zero_csum;
581 	}
582 
583 	/* Print IPv6 header for debugging */
584 	DBGC2 ( ipv6col ( &iphdr->dest ), "IPv6 TX %s->",
585 		inet6_ntoa ( &iphdr->src ) );
586 	DBGC2 ( ipv6col ( &iphdr->dest ), "%s len %zd next %d\n",
587 		inet6_ntoa ( &iphdr->dest ), len, iphdr->next_header );
588 
589 	/* Calculate link-layer destination address, if possible */
590 	if ( IN6_IS_ADDR_MULTICAST ( next_hop ) ) {
591 		/* Multicast address */
592 		ipv6_stats.out_mcast_pkts++;
593 		if ( ( rc = netdev->ll_protocol->mc_hash ( AF_INET6, next_hop,
594 							   ll_dest_buf ) ) !=0){
595 			DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not hash "
596 			       "multicast %s: %s\n", inet6_ntoa ( next_hop ),
597 			       strerror ( rc ) );
598 			goto err;
599 		}
600 		ll_dest = ll_dest_buf;
601 	} else {
602 		/* Unicast address */
603 		ll_dest = NULL;
604 	}
605 
606 	/* Update statistics */
607 	ipv6_stats.out_transmits++;
608 	ipv6_stats.out_octets += iob_len ( iobuf );
609 
610 	/* Hand off to link layer (via NDP if applicable) */
611 	if ( ll_dest ) {
612 		if ( ( rc = net_tx ( iobuf, netdev, &ipv6_protocol, ll_dest,
613 				     netdev->ll_addr ) ) != 0 ) {
614 			DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
615 			       "transmit packet via %s: %s\n",
616 			       netdev->name, strerror ( rc ) );
617 			return rc;
618 		}
619 	} else {
620 		if ( ( rc = ndp_tx ( iobuf, netdev, next_hop, &iphdr->src,
621 				     netdev->ll_addr ) ) != 0 ) {
622 			DBGC ( ipv6col ( &iphdr->dest ), "IPv6 could not "
623 			       "transmit packet via %s: %s\n",
624 			       netdev->name, strerror ( rc ) );
625 			return rc;
626 		}
627 	}
628 
629 	return 0;
630 
631  err:
632 	free_iob ( iobuf );
633 	return rc;
634 }
635 
636 /**
637  * Process incoming IPv6 packets
638  *
639  * @v iobuf		I/O buffer
640  * @v netdev		Network device
641  * @v ll_dest		Link-layer destination address
642  * @v ll_source		Link-layer destination source
643  * @v flags		Packet flags
644  * @ret rc		Return status code
645  *
646  * This function expects an IPv6 network datagram. It processes the
647  * headers and sends it to the transport layer.
648  */
ipv6_rx(struct io_buffer * iobuf,struct net_device * netdev,const void * ll_dest __unused,const void * ll_source __unused,unsigned int flags __unused)649 static int ipv6_rx ( struct io_buffer *iobuf, struct net_device *netdev,
650 		     const void *ll_dest __unused,
651 		     const void *ll_source __unused,
652 		     unsigned int flags __unused ) {
653 	struct ipv6_header *iphdr = iobuf->data;
654 	union ipv6_extension_header *ext;
655 	union {
656 		struct sockaddr_in6 sin6;
657 		struct sockaddr_tcpip st;
658 	} src, dest;
659 	uint16_t pshdr_csum;
660 	size_t len;
661 	size_t hdrlen;
662 	size_t extlen;
663 	int this_header;
664 	int next_header;
665 	int rc;
666 
667 	/* Update statistics */
668 	ipv6_stats.in_receives++;
669 	ipv6_stats.in_octets += iob_len ( iobuf );
670 	if ( flags & LL_BROADCAST ) {
671 		ipv6_stats.in_bcast_pkts++;
672 	} else if ( flags & LL_MULTICAST ) {
673 		ipv6_stats.in_mcast_pkts++;
674 	}
675 
676 	/* Sanity check the IPv6 header */
677 	if ( iob_len ( iobuf ) < sizeof ( *iphdr ) ) {
678 		DBGC ( ipv6col ( &iphdr->src ), "IPv6 packet too short at %zd "
679 		       "bytes (min %zd bytes)\n", iob_len ( iobuf ),
680 		       sizeof ( *iphdr ) );
681 		rc = -EINVAL_LEN;
682 		goto err_header;
683 	}
684 	if ( ( iphdr->ver_tc_label & htonl ( IPV6_MASK_VER ) ) !=
685 	     htonl ( IPV6_VER ) ) {
686 		DBGC ( ipv6col ( &iphdr->src ), "IPv6 version %#08x not "
687 		       "supported\n", ntohl ( iphdr->ver_tc_label ) );
688 		rc = -ENOTSUP_VER;
689 		goto err_header;
690 	}
691 
692 	/* Truncate packet to specified length */
693 	len = ntohs ( iphdr->len );
694 	if ( len > iob_len ( iobuf ) ) {
695 		DBGC ( ipv6col ( &iphdr->src ), "IPv6 length too long at %zd "
696 		       "bytes (packet is %zd bytes)\n", len, iob_len ( iobuf ));
697 		ipv6_stats.in_truncated_pkts++;
698 		rc = -EINVAL_LEN;
699 		goto err_other;
700 	}
701 	iob_unput ( iobuf, ( iob_len ( iobuf ) - len - sizeof ( *iphdr ) ) );
702 	hdrlen = sizeof ( *iphdr );
703 
704 	/* Print IPv6 header for debugging */
705 	DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
706 		inet6_ntoa ( &iphdr->dest ) );
707 	DBGC2 ( ipv6col ( &iphdr->src ), "%s len %zd next %d\n",
708 		inet6_ntoa ( &iphdr->src ), len, iphdr->next_header );
709 
710 	/* Discard unicast packets not destined for us */
711 	if ( ( ! ( flags & LL_MULTICAST ) ) &&
712 	     ( ! ipv6_has_addr ( netdev, &iphdr->dest ) ) ) {
713 		DBGC ( ipv6col ( &iphdr->src ), "IPv6 discarding non-local "
714 		       "unicast packet for %s\n", inet6_ntoa ( &iphdr->dest ) );
715 		ipv6_stats.in_addr_errors++;
716 		rc = -EPIPE;
717 		goto err_other;
718 	}
719 
720 	/* Process any extension headers */
721 	next_header = iphdr->next_header;
722 	while ( 1 ) {
723 
724 		/* Extract extension header */
725 		this_header = next_header;
726 		ext = ( iobuf->data + hdrlen );
727 		extlen = sizeof ( ext->pad );
728 		if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
729 			DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
730 			       "extension header type %d at %zd bytes (min "
731 			       "%zd bytes)\n", this_header,
732 			       ( iob_len ( iobuf ) - hdrlen ), extlen );
733 			rc = -EINVAL_LEN;
734 			goto err_header;
735 		}
736 
737 		/* Determine size of extension header (if applicable) */
738 		if ( ( this_header == IPV6_HOPBYHOP ) ||
739 		     ( this_header == IPV6_DESTINATION ) ||
740 		     ( this_header == IPV6_ROUTING ) ) {
741 			/* Length field is present */
742 			extlen += ext->common.len;
743 		} else if ( this_header == IPV6_FRAGMENT ) {
744 			/* Length field is reserved and ignored (RFC2460) */
745 		} else {
746 			/* Not an extension header; assume rest is payload */
747 			break;
748 		}
749 		if ( iob_len ( iobuf ) < ( hdrlen + extlen ) ) {
750 			DBGC ( ipv6col ( &iphdr->src ), "IPv6 too short for "
751 			       "extension header type %d at %zd bytes (min "
752 			       "%zd bytes)\n", this_header,
753 			       ( iob_len ( iobuf ) - hdrlen ), extlen );
754 			rc = -EINVAL_LEN;
755 			goto err_header;
756 		}
757 		hdrlen += extlen;
758 		next_header = ext->common.next_header;
759 		DBGC2 ( ipv6col ( &iphdr->src ), "IPv6 RX %s<-",
760 			inet6_ntoa ( &iphdr->dest ) );
761 		DBGC2 ( ipv6col ( &iphdr->src ), "%s ext type %d len %zd next "
762 			"%d\n", inet6_ntoa ( &iphdr->src ), this_header,
763 			extlen, next_header );
764 
765 		/* Process this extension header */
766 		if ( ( this_header == IPV6_HOPBYHOP ) ||
767 		     ( this_header == IPV6_DESTINATION ) ) {
768 
769 			/* Check that all options can be ignored */
770 			if ( ( rc = ipv6_check_options ( iphdr, &ext->options,
771 							 extlen ) ) != 0 )
772 				goto err_header;
773 
774 		} else if ( this_header == IPV6_FRAGMENT ) {
775 
776 			/* Reassemble fragments */
777 			iobuf = fragment_reassemble ( &ipv6_reassembler, iobuf,
778 						      &hdrlen );
779 			if ( ! iobuf )
780 				return 0;
781 			iphdr = iobuf->data;
782 		}
783 	}
784 
785 	/* Construct socket address, calculate pseudo-header checksum,
786 	 * and hand off to transport layer
787 	 */
788 	memset ( &src, 0, sizeof ( src ) );
789 	src.sin6.sin6_family = AF_INET6;
790 	memcpy ( &src.sin6.sin6_addr, &iphdr->src,
791 		 sizeof ( src.sin6.sin6_addr ) );
792 	src.sin6.sin6_scope_id = netdev->index;
793 	memset ( &dest, 0, sizeof ( dest ) );
794 	dest.sin6.sin6_family = AF_INET6;
795 	memcpy ( &dest.sin6.sin6_addr, &iphdr->dest,
796 		 sizeof ( dest.sin6.sin6_addr ) );
797 	dest.sin6.sin6_scope_id = netdev->index;
798 	iob_pull ( iobuf, hdrlen );
799 	pshdr_csum = ipv6_pshdr_chksum ( iphdr, iob_len ( iobuf ),
800 					 next_header, TCPIP_EMPTY_CSUM );
801 	if ( ( rc = tcpip_rx ( iobuf, netdev, next_header, &src.st, &dest.st,
802 			       pshdr_csum, &ipv6_stats ) ) != 0 ) {
803 		DBGC ( ipv6col ( &src.sin6.sin6_addr ), "IPv6 received packet "
804 				"rejected by stack: %s\n", strerror ( rc ) );
805 		return rc;
806 	}
807 
808 	return 0;
809 
810  err_header:
811 	ipv6_stats.in_hdr_errors++;
812  err_other:
813 	free_iob ( iobuf );
814 	return rc;
815 }
816 
817 /**
818  * Parse IPv6 address
819  *
820  * @v string		IPv6 address string
821  * @ret in		IPv6 address to fill in
822  * @ret rc		Return status code
823  */
inet6_aton(const char * string,struct in6_addr * in)824 int inet6_aton ( const char *string, struct in6_addr *in ) {
825 	uint16_t *word = in->s6_addr16;
826 	uint16_t *end = ( word + ( sizeof ( in->s6_addr16 ) /
827 				   sizeof ( in->s6_addr16[0] ) ) );
828 	uint16_t *pad = NULL;
829 	const char *nptr = string;
830 	char *endptr;
831 	unsigned long value;
832 	size_t pad_len;
833 	size_t move_len;
834 
835 	/* Parse string */
836 	while ( 1 ) {
837 
838 		/* Parse current word */
839 		value = strtoul ( nptr, &endptr, 16 );
840 		if ( value > 0xffff ) {
841 			DBG ( "IPv6 invalid word value %#lx in \"%s\"\n",
842 			      value, string );
843 			return -EINVAL;
844 		}
845 		*(word++) = htons ( value );
846 
847 		/* Parse separator */
848 		if ( ! *endptr )
849 			break;
850 		if ( *endptr != ':' ) {
851 			DBG ( "IPv6 invalid separator '%c' in \"%s\"\n",
852 			      *endptr, string );
853 			return -EINVAL;
854 		}
855 		if ( ( endptr == nptr ) && ( nptr != string ) ) {
856 			if ( pad ) {
857 				DBG ( "IPv6 invalid multiple \"::\" in "
858 				      "\"%s\"\n", string );
859 				return -EINVAL;
860 			}
861 			pad = word;
862 		}
863 		nptr = ( endptr + 1 );
864 
865 		/* Check for overrun */
866 		if ( word == end ) {
867 			DBG ( "IPv6 too many words in \"%s\"\n", string );
868 			return -EINVAL;
869 		}
870 	}
871 
872 	/* Insert padding if specified */
873 	if ( pad ) {
874 		move_len = ( ( ( void * ) word ) - ( ( void * ) pad ) );
875 		pad_len = ( ( ( void * ) end ) - ( ( void * ) word ) );
876 		memmove ( ( ( ( void * ) pad ) + pad_len ), pad, move_len );
877 		memset ( pad, 0, pad_len );
878 	} else if ( word != end ) {
879 		DBG ( "IPv6 underlength address \"%s\"\n", string );
880 		return -EINVAL;
881 	}
882 
883 	return 0;
884 }
885 
886 /**
887  * Convert IPv6 address to standard notation
888  *
889  * @v in		IPv6 address
890  * @ret string		IPv6 address string in canonical format
891  *
892  * RFC5952 defines the canonical format for IPv6 textual representation.
893  */
inet6_ntoa(const struct in6_addr * in)894 char * inet6_ntoa ( const struct in6_addr *in ) {
895 	static char buf[41]; /* ":xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */
896 	char *out = buf;
897 	char *longest_start = NULL;
898 	char *start = NULL;
899 	int longest_len = 1;
900 	int len = 0;
901 	char *dest;
902 	unsigned int i;
903 	uint16_t value;
904 
905 	/* Format address, keeping track of longest run of zeros */
906 	for ( i = 0 ; i < ( sizeof ( in->s6_addr16 ) /
907 			    sizeof ( in->s6_addr16[0] ) ) ; i++ ) {
908 		value = ntohs ( in->s6_addr16[i] );
909 		if ( value == 0 ) {
910 			if ( len++ == 0 )
911 				start = out;
912 			if ( len > longest_len ) {
913 				longest_start = start;
914 				longest_len = len;
915 			}
916 		} else {
917 			len = 0;
918 		}
919 		out += sprintf ( out, ":%x", value );
920 	}
921 
922 	/* Abbreviate longest run of zeros, if applicable */
923 	if ( longest_start ) {
924 		dest = strcpy ( ( longest_start + 1 ),
925 				( longest_start + ( 2 * longest_len ) ) );
926 		if ( dest[0] == '\0' )
927 			dest[1] = '\0';
928 		dest[0] = ':';
929 	}
930 	return ( ( longest_start == buf ) ? buf : ( buf + 1 ) );
931 }
932 
933 /**
934  * Transcribe IPv6 address
935  *
936  * @v net_addr		IPv6 address
937  * @ret string		IPv6 address in standard notation
938  *
939  */
ipv6_ntoa(const void * net_addr)940 static const char * ipv6_ntoa ( const void *net_addr ) {
941 	return inet6_ntoa ( net_addr );
942 }
943 
944 /**
945  * Transcribe IPv6 socket address
946  *
947  * @v sa		Socket address
948  * @ret string		Socket address in standard notation
949  */
ipv6_sock_ntoa(struct sockaddr * sa)950 static const char * ipv6_sock_ntoa ( struct sockaddr *sa ) {
951 	static char buf[ 39 /* "xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx:xxxx" */ +
952 			 1 /* "%" */ + NETDEV_NAME_LEN + 1 /* NUL */ ];
953 	struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
954 	struct in6_addr *in = &sin6->sin6_addr;
955 	struct net_device *netdev;
956 	const char *netdev_name;
957 
958 	/* Identify network device, if applicable */
959 	if ( IN6_IS_ADDR_LINKLOCAL ( in ) || IN6_IS_ADDR_MULTICAST ( in ) ) {
960 		netdev = find_netdev_by_index ( sin6->sin6_scope_id );
961 		netdev_name = ( netdev ? netdev->name : "UNKNOWN" );
962 	} else {
963 		netdev_name = NULL;
964 	}
965 
966 	/* Format socket address */
967 	snprintf ( buf, sizeof ( buf ), "%s%s%s", inet6_ntoa ( in ),
968 		   ( netdev_name ? "%" : "" ),
969 		   ( netdev_name ? netdev_name : "" ) );
970 	return buf;
971 }
972 
973 /**
974  * Parse IPv6 socket address
975  *
976  * @v string		Socket address string
977  * @v sa		Socket address to fill in
978  * @ret rc		Return status code
979  */
ipv6_sock_aton(const char * string,struct sockaddr * sa)980 static int ipv6_sock_aton ( const char *string, struct sockaddr *sa ) {
981 	struct sockaddr_in6 *sin6 = ( ( struct sockaddr_in6 * ) sa );
982 	struct in6_addr in;
983 	struct net_device *netdev;
984 	size_t len;
985 	char *tmp;
986 	char *in_string;
987 	char *netdev_string;
988 	int rc;
989 
990 	/* Create modifiable copy of string */
991 	tmp = strdup ( string );
992 	if ( ! tmp ) {
993 		rc = -ENOMEM;
994 		goto err_alloc;
995 	}
996 	in_string = tmp;
997 
998 	/* Strip surrounding "[...]", if present */
999 	len = strlen ( in_string );
1000 	if ( ( in_string[0] == '[' ) && ( in_string[ len - 1 ] == ']' ) ) {
1001 		in_string[ len - 1 ] = '\0';
1002 		in_string++;
1003 	}
1004 
1005 	/* Split at network device name, if present */
1006 	netdev_string = strchr ( in_string, '%' );
1007 	if ( netdev_string )
1008 		*(netdev_string++) = '\0';
1009 
1010 	/* Parse IPv6 address portion */
1011 	if ( ( rc = inet6_aton ( in_string, &in ) ) != 0 )
1012 		goto err_inet6_aton;
1013 
1014 	/* Parse scope ID, if applicable */
1015 	if ( netdev_string ) {
1016 
1017 		/* Parse explicit network device name, if present */
1018 		netdev = find_netdev ( netdev_string );
1019 		if ( ! netdev ) {
1020 			rc = -ENODEV;
1021 			goto err_find_netdev;
1022 		}
1023 		sin6->sin6_scope_id = netdev->index;
1024 
1025 	} else if ( IN6_IS_ADDR_LINKLOCAL ( &in ) ||
1026 		    IN6_IS_ADDR_MULTICAST ( &in ) ) {
1027 
1028 		/* If no network device is explicitly specified for a
1029 		 * link-local or multicast address, default to using
1030 		 * "netX" (if existent).
1031 		 */
1032 		netdev = last_opened_netdev();
1033 		if ( netdev )
1034 			sin6->sin6_scope_id = netdev->index;
1035 	}
1036 
1037 	/* Copy IPv6 address portion to socket address */
1038 	memcpy ( &sin6->sin6_addr, &in, sizeof ( sin6->sin6_addr ) );
1039 
1040  err_find_netdev:
1041  err_inet6_aton:
1042 	free ( tmp );
1043  err_alloc:
1044 	return rc;
1045 }
1046 
1047 /** IPv6 protocol */
1048 struct net_protocol ipv6_protocol __net_protocol = {
1049 	.name = "IPv6",
1050 	.net_proto = htons ( ETH_P_IPV6 ),
1051 	.net_addr_len = sizeof ( struct in6_addr ),
1052 	.rx = ipv6_rx,
1053 	.ntoa = ipv6_ntoa,
1054 };
1055 
1056 /** IPv6 TCPIP net protocol */
1057 struct tcpip_net_protocol ipv6_tcpip_protocol __tcpip_net_protocol = {
1058 	.name = "IPv6",
1059 	.sa_family = AF_INET6,
1060 	.header_len = sizeof ( struct ipv6_header ),
1061 	.net_protocol = &ipv6_protocol,
1062 	.tx = ipv6_tx,
1063 	.netdev = ipv6_netdev,
1064 };
1065 
1066 /** IPv6 socket address converter */
1067 struct sockaddr_converter ipv6_sockaddr_converter __sockaddr_converter = {
1068 	.family = AF_INET6,
1069 	.ntoa = ipv6_sock_ntoa,
1070 	.aton = ipv6_sock_aton,
1071 };
1072 
1073 /**
1074  * Parse IPv6 address setting value
1075  *
1076  * @v type		Setting type
1077  * @v value		Formatted setting value
1078  * @v buf		Buffer to contain raw value
1079  * @v len		Length of buffer
1080  * @ret len		Length of raw value, or negative error
1081  */
parse_ipv6_setting(const struct setting_type * type __unused,const char * value,void * buf,size_t len)1082 int parse_ipv6_setting ( const struct setting_type *type __unused,
1083 			 const char *value, void *buf, size_t len ) {
1084 	struct in6_addr ipv6;
1085 	int rc;
1086 
1087 	/* Parse IPv6 address */
1088 	if ( ( rc = inet6_aton ( value, &ipv6 ) ) != 0 )
1089 		return rc;
1090 
1091 	/* Copy to buffer */
1092 	if ( len > sizeof ( ipv6 ) )
1093 		len = sizeof ( ipv6 );
1094 	memcpy ( buf, &ipv6, len );
1095 
1096 	return ( sizeof ( ipv6 ) );
1097 }
1098 
1099 /**
1100  * Format IPv6 address setting value
1101  *
1102  * @v type		Setting type
1103  * @v raw		Raw setting value
1104  * @v raw_len		Length of raw setting value
1105  * @v buf		Buffer to contain formatted value
1106  * @v len		Length of buffer
1107  * @ret len		Length of formatted value, or negative error
1108  */
format_ipv6_setting(const struct setting_type * type __unused,const void * raw,size_t raw_len,char * buf,size_t len)1109 int format_ipv6_setting ( const struct setting_type *type __unused,
1110 			  const void *raw, size_t raw_len, char *buf,
1111 			  size_t len ) {
1112 	const struct in6_addr *ipv6 = raw;
1113 
1114 	if ( raw_len < sizeof ( *ipv6 ) )
1115 		return -EINVAL;
1116 	return snprintf ( buf, len, "%s", inet6_ntoa ( ipv6 ) );
1117 }
1118 
1119 /** IPv6 settings scope */
1120 const struct settings_scope ipv6_settings_scope;
1121 
1122 /** IPv6 address setting */
1123 const struct setting ip6_setting __setting ( SETTING_IP6, ip6 ) = {
1124 	.name = "ip6",
1125 	.description = "IPv6 address",
1126 	.type = &setting_type_ipv6,
1127 	.scope = &ipv6_settings_scope,
1128 };
1129 
1130 /** IPv6 prefix length setting */
1131 const struct setting len6_setting __setting ( SETTING_IP6, len6 ) = {
1132 	.name = "len6",
1133 	.description = "IPv6 prefix length",
1134 	.type = &setting_type_int8,
1135 	.scope = &ipv6_settings_scope,
1136 };
1137 
1138 /** Default gateway setting */
1139 const struct setting gateway6_setting __setting ( SETTING_IP6, gateway6 ) = {
1140 	.name = "gateway6",
1141 	.description = "IPv6 gateway",
1142 	.type = &setting_type_ipv6,
1143 	.scope = &ipv6_settings_scope,
1144 };
1145 
1146 /**
1147  * Check applicability of IPv6 link-local address setting
1148  *
1149  * @v settings		Settings block
1150  * @v setting		Setting to fetch
1151  * @ret applies		Setting applies within this settings block
1152  */
ipv6_applies(struct settings * settings __unused,const struct setting * setting)1153 static int ipv6_applies ( struct settings *settings __unused,
1154 			  const struct setting *setting ) {
1155 
1156 	return ( setting->scope == &ipv6_settings_scope );
1157 }
1158 
1159 /**
1160  * Fetch IPv6 link-local address setting
1161  *
1162  * @v settings		Settings block
1163  * @v setting		Setting to fetch
1164  * @v data		Buffer to fill with setting data
1165  * @v len		Length of buffer
1166  * @ret len		Length of setting data, or negative error
1167  */
ipv6_fetch(struct settings * settings,struct setting * setting,void * data,size_t len)1168 static int ipv6_fetch ( struct settings *settings, struct setting *setting,
1169 			void *data, size_t len ) {
1170 	struct net_device *netdev =
1171 		container_of ( settings->parent, struct net_device,
1172 			       settings.settings );
1173 	struct in6_addr ip6;
1174 	uint8_t *len6;
1175 	int prefix_len;
1176 	int rc;
1177 
1178 	/* Construct link-local address from EUI-64 as per RFC 2464 */
1179 	memset ( &ip6, 0, sizeof ( ip6 ) );
1180 	prefix_len = ipv6_link_local ( &ip6, netdev );
1181 	if ( prefix_len < 0 ) {
1182 		rc = prefix_len;
1183 		return rc;
1184 	}
1185 
1186 	/* Handle setting */
1187 	if ( setting_cmp ( setting, &ip6_setting ) == 0 ) {
1188 
1189 		/* Return link-local ip6 */
1190 		if ( len > sizeof ( ip6 ) )
1191 			len = sizeof ( ip6 );
1192 		memcpy ( data, &ip6, len );
1193 		return sizeof ( ip6 );
1194 
1195 	} else if ( setting_cmp ( setting, &len6_setting ) == 0 ) {
1196 
1197 		/* Return prefix length */
1198 		if ( len ) {
1199 			len6 = data;
1200 			*len6 = prefix_len;
1201 		}
1202 		return sizeof ( *len6 );
1203 
1204 	}
1205 
1206 	return -ENOENT;
1207 }
1208 
1209 /** IPv6 link-local address settings operations */
1210 static struct settings_operations ipv6_settings_operations = {
1211 	.applies = ipv6_applies,
1212 	.fetch = ipv6_fetch,
1213 };
1214 
1215 /** IPv6 link-local address settings */
1216 struct ipv6_settings {
1217 	/** Reference counter */
1218 	struct refcnt refcnt;
1219 	/** Settings interface */
1220 	struct settings settings;
1221 };
1222 
1223 /**
1224  * Register IPv6 link-local address settings
1225  *
1226  * @v netdev		Network device
1227  * @ret rc		Return status code
1228  */
ipv6_register_settings(struct net_device * netdev)1229 static int ipv6_register_settings ( struct net_device *netdev ) {
1230 	struct settings *parent = netdev_settings ( netdev );
1231 	struct ipv6_settings *ipv6set;
1232 	int rc;
1233 
1234 	/* Allocate and initialise structure */
1235 	ipv6set = zalloc ( sizeof ( *ipv6set ) );
1236 	if ( ! ipv6set ) {
1237 		rc = -ENOMEM;
1238 		goto err_alloc;
1239 	}
1240 	ref_init ( &ipv6set->refcnt, NULL );
1241 	settings_init ( &ipv6set->settings, &ipv6_settings_operations,
1242 			&ipv6set->refcnt, &ipv6_settings_scope );
1243 	ipv6set->settings.order = IPV6_ORDER_LINK_LOCAL;
1244 
1245 	/* Register settings */
1246 	if ( ( rc = register_settings ( &ipv6set->settings, parent,
1247 					IPV6_SETTINGS_NAME ) ) != 0 )
1248 		goto err_register;
1249 
1250  err_register:
1251 	ref_put ( &ipv6set->refcnt );
1252  err_alloc:
1253 	return rc;
1254 }
1255 
1256 /** IPv6 network device driver */
1257 struct net_driver ipv6_driver __net_driver = {
1258 	.name = "IPv6",
1259 	.probe = ipv6_register_settings,
1260 };
1261 
1262 /**
1263  * Create IPv6 routing table based on configured settings
1264  *
1265  * @v netdev		Network device
1266  * @v settings		Settings block
1267  * @ret rc		Return status code
1268  */
ipv6_create_routes(struct net_device * netdev,struct settings * settings)1269 static int ipv6_create_routes ( struct net_device *netdev,
1270 				struct settings *settings ) {
1271 	struct settings *child;
1272 	struct settings *origin;
1273 	struct in6_addr ip6_buf;
1274 	struct in6_addr gateway6_buf;
1275 	struct in6_addr *ip6 = &ip6_buf;
1276 	struct in6_addr *gateway6 = &gateway6_buf;
1277 	uint8_t len6;
1278 	size_t len;
1279 	int rc;
1280 
1281 	/* First, create routing table for any child settings.  We do
1282 	 * this depth-first and in reverse order so that the end
1283 	 * result reflects the relative priorities of the settings
1284 	 * blocks.
1285 	 */
1286 	list_for_each_entry_reverse ( child, &settings->children, siblings )
1287 		ipv6_create_routes ( netdev, child );
1288 
1289 	/* Fetch IPv6 address, if any */
1290 	len = fetch_setting ( settings, &ip6_setting, &origin, NULL,
1291 			      ip6, sizeof ( *ip6 ) );
1292 	if ( ( len != sizeof ( *ip6 ) ) || ( origin != settings ) )
1293 		return 0;
1294 
1295 	/* Fetch prefix length, if defined */
1296 	len = fetch_setting ( settings, &len6_setting, &origin, NULL,
1297 			      &len6, sizeof ( len6 ) );
1298 	if ( ( len != sizeof ( len6 ) ) || ( origin != settings ) )
1299 		len6 = 0;
1300 	if ( len6 > IPV6_MAX_PREFIX_LEN )
1301 		len6 = IPV6_MAX_PREFIX_LEN;
1302 
1303 	/* Fetch gateway, if defined */
1304 	len = fetch_setting ( settings, &gateway6_setting, &origin, NULL,
1305 			      gateway6, sizeof ( *gateway6 ) );
1306 	if ( ( len != sizeof ( *gateway6 ) ) || ( origin != settings ) )
1307 		gateway6 = NULL;
1308 
1309 	/* Create or update route */
1310 	if ( ( rc = ipv6_add_miniroute ( netdev, ip6, len6, gateway6 ) ) != 0){
1311 		DBGC ( netdev, "IPv6 %s could not add route: %s\n",
1312 		       netdev->name, strerror ( rc ) );
1313 		return rc;
1314 	}
1315 
1316 	return 0;
1317 }
1318 
1319 /**
1320  * Create IPv6 routing table based on configured settings
1321  *
1322  * @ret rc		Return status code
1323  */
ipv6_create_all_routes(void)1324 static int ipv6_create_all_routes ( void ) {
1325 	struct ipv6_miniroute *miniroute;
1326 	struct ipv6_miniroute *tmp;
1327 	struct net_device *netdev;
1328 	struct settings *settings;
1329 	int rc;
1330 
1331 	/* Delete all existing routes */
1332 	list_for_each_entry_safe ( miniroute, tmp, &ipv6_miniroutes, list )
1333 		ipv6_del_miniroute ( miniroute );
1334 
1335 	/* Create routes for each configured network device */
1336 	for_each_netdev ( netdev ) {
1337 		settings = netdev_settings ( netdev );
1338 		if ( ( rc = ipv6_create_routes ( netdev, settings ) ) != 0 )
1339 			return rc;
1340 	}
1341 
1342 	return 0;
1343 }
1344 
1345 /** IPv6 settings applicator */
1346 struct settings_applicator ipv6_settings_applicator __settings_applicator = {
1347 	.apply = ipv6_create_all_routes,
1348 };
1349 
1350 /* Drag in objects via ipv6_protocol */
1351 REQUIRING_SYMBOL ( ipv6_protocol );
1352 
1353 /* Drag in ICMPv6 */
1354 REQUIRE_OBJECT ( icmpv6 );
1355 
1356 /* Drag in NDP */
1357 REQUIRE_OBJECT ( ndp );
1358