1 /*
2 * Copyright (C) 2006 Michael Brown <mbrown@fensystems.co.uk>.
3 * Copyright (C) 2006 Nikhil Chandru Rao
4 *
5 * This program is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU General Public License as
7 * published by the Free Software Foundation; either version 2 of the
8 * License, or (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
18 * 02110-1301, USA.
19 *
20 * You can also choose to distribute this program under the terms of
21 * the Unmodified Binary Distribution Licence (as given in the file
22 * COPYING.UBDL), provided that you have satisfied its requirements.
23 */
24
25 #include <string.h>
26 #include <stdint.h>
27 #include <stdlib.h>
28 #include <stdio.h>
29 #include <errno.h>
30 #include <byteswap.h>
31 #include <ipxe/list.h>
32 #include <ipxe/in.h>
33 #include <ipxe/arp.h>
34 #include <ipxe/if_ether.h>
35 #include <ipxe/iobuf.h>
36 #include <ipxe/netdevice.h>
37 #include <ipxe/ip.h>
38 #include <ipxe/tcpip.h>
39 #include <ipxe/dhcp.h>
40 #include <ipxe/settings.h>
41 #include <ipxe/fragment.h>
42 #include <ipxe/ipstat.h>
43 #include <ipxe/profile.h>
44
45 /** @file
46 *
47 * IPv4 protocol
48 *
49 */
50
51 FILE_LICENCE ( GPL2_OR_LATER_OR_UBDL );
52
53 /* Unique IP datagram identification number (high byte) */
54 static uint8_t next_ident_high = 0;
55
56 /** List of IPv4 miniroutes */
57 struct list_head ipv4_miniroutes = LIST_HEAD_INIT ( ipv4_miniroutes );
58
59 /** IPv4 statistics */
60 static struct ip_statistics ipv4_stats;
61
62 /** IPv4 statistics family */
63 struct ip_statistics_family
64 ipv4_stats_family __ip_statistics_family ( IP_STATISTICS_IPV4 ) = {
65 .version = 4,
66 .stats = &ipv4_stats,
67 };
68
69 /** Transmit profiler */
70 static struct profiler ipv4_tx_profiler __profiler = { .name = "ipv4.tx" };
71
72 /** Receive profiler */
73 static struct profiler ipv4_rx_profiler __profiler = { .name = "ipv4.rx" };
74
75 /**
76 * Add IPv4 minirouting table entry
77 *
78 * @v netdev Network device
79 * @v address IPv4 address
80 * @v netmask Subnet mask
81 * @v gateway Gateway address (if any)
82 * @ret rc Return status code
83 */
add_ipv4_miniroute(struct net_device * netdev,struct in_addr address,struct in_addr netmask,struct in_addr gateway)84 static int add_ipv4_miniroute ( struct net_device *netdev,
85 struct in_addr address, struct in_addr netmask,
86 struct in_addr gateway ) {
87 struct ipv4_miniroute *miniroute;
88
89 DBGC ( netdev, "IPv4 add %s", inet_ntoa ( address ) );
90 DBGC ( netdev, "/%s ", inet_ntoa ( netmask ) );
91 if ( gateway.s_addr )
92 DBGC ( netdev, "gw %s ", inet_ntoa ( gateway ) );
93 DBGC ( netdev, "via %s\n", netdev->name );
94
95 /* Allocate and populate miniroute structure */
96 miniroute = malloc ( sizeof ( *miniroute ) );
97 if ( ! miniroute ) {
98 DBGC ( netdev, "IPv4 could not add miniroute\n" );
99 return -ENOMEM;
100 }
101
102 /* Record routing information */
103 miniroute->netdev = netdev_get ( netdev );
104 miniroute->address = address;
105 miniroute->netmask = netmask;
106 miniroute->gateway = gateway;
107
108 /* Add to end of list if we have a gateway, otherwise
109 * to start of list.
110 */
111 if ( gateway.s_addr ) {
112 list_add_tail ( &miniroute->list, &ipv4_miniroutes );
113 } else {
114 list_add ( &miniroute->list, &ipv4_miniroutes );
115 }
116
117 return 0;
118 }
119
120 /**
121 * Delete IPv4 minirouting table entry
122 *
123 * @v miniroute Routing table entry
124 */
del_ipv4_miniroute(struct ipv4_miniroute * miniroute)125 static void del_ipv4_miniroute ( struct ipv4_miniroute *miniroute ) {
126 struct net_device *netdev = miniroute->netdev;
127
128 DBGC ( netdev, "IPv4 del %s", inet_ntoa ( miniroute->address ) );
129 DBGC ( netdev, "/%s ", inet_ntoa ( miniroute->netmask ) );
130 if ( miniroute->gateway.s_addr )
131 DBGC ( netdev, "gw %s ", inet_ntoa ( miniroute->gateway ) );
132 DBGC ( netdev, "via %s\n", miniroute->netdev->name );
133
134 netdev_put ( miniroute->netdev );
135 list_del ( &miniroute->list );
136 free ( miniroute );
137 }
138
139 /**
140 * Perform IPv4 routing
141 *
142 * @v scope_id Destination address scope ID
143 * @v dest Final destination address
144 * @ret dest Next hop destination address
145 * @ret miniroute Routing table entry to use, or NULL if no route
146 *
147 * If the route requires use of a gateway, the next hop destination
148 * address will be overwritten with the gateway address.
149 */
ipv4_route(unsigned int scope_id,struct in_addr * dest)150 static struct ipv4_miniroute * ipv4_route ( unsigned int scope_id,
151 struct in_addr *dest ) {
152 struct ipv4_miniroute *miniroute;
153
154 /* Find first usable route in routing table */
155 list_for_each_entry ( miniroute, &ipv4_miniroutes, list ) {
156
157 /* Skip closed network devices */
158 if ( ! netdev_is_open ( miniroute->netdev ) )
159 continue;
160
161 if ( IN_IS_MULTICAST ( dest->s_addr ) ) {
162
163 /* If destination is non-global, and the scope ID
164 * matches this network device, then use this route.
165 */
166 if ( miniroute->netdev->index == scope_id )
167 return miniroute;
168
169 } else {
170
171 /* If destination is an on-link global
172 * address, then use this route.
173 */
174 if ( ( ( dest->s_addr ^ miniroute->address.s_addr )
175 & miniroute->netmask.s_addr ) == 0 )
176 return miniroute;
177
178 /* If destination is an off-link global
179 * address, and we have a default gateway,
180 * then use this route.
181 */
182 if ( miniroute->gateway.s_addr ) {
183 *dest = miniroute->gateway;
184 return miniroute;
185 }
186 }
187 }
188
189 return NULL;
190 }
191
192 /**
193 * Determine transmitting network device
194 *
195 * @v st_dest Destination network-layer address
196 * @ret netdev Transmitting network device, or NULL
197 */
ipv4_netdev(struct sockaddr_tcpip * st_dest)198 static struct net_device * ipv4_netdev ( struct sockaddr_tcpip *st_dest ) {
199 struct sockaddr_in *sin_dest = ( ( struct sockaddr_in * ) st_dest );
200 struct in_addr dest = sin_dest->sin_addr;
201 struct ipv4_miniroute *miniroute;
202
203 /* Find routing table entry */
204 miniroute = ipv4_route ( sin_dest->sin_scope_id, &dest );
205 if ( ! miniroute )
206 return NULL;
207
208 return miniroute->netdev;
209 }
210
211 /**
212 * Check if IPv4 fragment matches fragment reassembly buffer
213 *
214 * @v fragment Fragment reassembly buffer
215 * @v iobuf I/O buffer
216 * @v hdrlen Length of non-fragmentable potion of I/O buffer
217 * @ret is_fragment Fragment matches this reassembly buffer
218 */
ipv4_is_fragment(struct fragment * fragment,struct io_buffer * iobuf,size_t hdrlen __unused)219 static int ipv4_is_fragment ( struct fragment *fragment,
220 struct io_buffer *iobuf,
221 size_t hdrlen __unused ) {
222 struct iphdr *frag_iphdr = fragment->iobuf->data;
223 struct iphdr *iphdr = iobuf->data;
224
225 return ( ( iphdr->src.s_addr == frag_iphdr->src.s_addr ) &&
226 ( iphdr->ident == frag_iphdr->ident ) );
227 }
228
229 /**
230 * Get IPv4 fragment offset
231 *
232 * @v iobuf I/O buffer
233 * @v hdrlen Length of non-fragmentable potion of I/O buffer
234 * @ret offset Offset
235 */
ipv4_fragment_offset(struct io_buffer * iobuf,size_t hdrlen __unused)236 static size_t ipv4_fragment_offset ( struct io_buffer *iobuf,
237 size_t hdrlen __unused ) {
238 struct iphdr *iphdr = iobuf->data;
239
240 return ( ( ntohs ( iphdr->frags ) & IP_MASK_OFFSET ) << 3 );
241 }
242
243 /**
244 * Check if more fragments exist
245 *
246 * @v iobuf I/O buffer
247 * @v hdrlen Length of non-fragmentable potion of I/O buffer
248 * @ret more_frags More fragments exist
249 */
ipv4_more_fragments(struct io_buffer * iobuf,size_t hdrlen __unused)250 static int ipv4_more_fragments ( struct io_buffer *iobuf,
251 size_t hdrlen __unused ) {
252 struct iphdr *iphdr = iobuf->data;
253
254 return ( iphdr->frags & htons ( IP_MASK_MOREFRAGS ) );
255 }
256
257 /** IPv4 fragment reassembler */
258 static struct fragment_reassembler ipv4_reassembler = {
259 .list = LIST_HEAD_INIT ( ipv4_reassembler.list ),
260 .is_fragment = ipv4_is_fragment,
261 .fragment_offset = ipv4_fragment_offset,
262 .more_fragments = ipv4_more_fragments,
263 .stats = &ipv4_stats,
264 };
265
266 /**
267 * Add IPv4 pseudo-header checksum to existing checksum
268 *
269 * @v iobuf I/O buffer
270 * @v csum Existing checksum
271 * @ret csum Updated checksum
272 */
ipv4_pshdr_chksum(struct io_buffer * iobuf,uint16_t csum)273 static uint16_t ipv4_pshdr_chksum ( struct io_buffer *iobuf, uint16_t csum ) {
274 struct ipv4_pseudo_header pshdr;
275 struct iphdr *iphdr = iobuf->data;
276 size_t hdrlen = ( ( iphdr->verhdrlen & IP_MASK_HLEN ) * 4 );
277
278 /* Build pseudo-header */
279 pshdr.src = iphdr->src;
280 pshdr.dest = iphdr->dest;
281 pshdr.zero_padding = 0x00;
282 pshdr.protocol = iphdr->protocol;
283 pshdr.len = htons ( iob_len ( iobuf ) - hdrlen );
284
285 /* Update the checksum value */
286 return tcpip_continue_chksum ( csum, &pshdr, sizeof ( pshdr ) );
287 }
288
289 /**
290 * Transmit IP packet
291 *
292 * @v iobuf I/O buffer
293 * @v tcpip Transport-layer protocol
294 * @v st_src Source network-layer address
295 * @v st_dest Destination network-layer address
296 * @v netdev Network device to use if no route found, or NULL
297 * @v trans_csum Transport-layer checksum to complete, or NULL
298 * @ret rc Status
299 *
300 * This function expects a transport-layer segment and prepends the IP header
301 */
ipv4_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)302 static int ipv4_tx ( struct io_buffer *iobuf,
303 struct tcpip_protocol *tcpip_protocol,
304 struct sockaddr_tcpip *st_src,
305 struct sockaddr_tcpip *st_dest,
306 struct net_device *netdev,
307 uint16_t *trans_csum ) {
308 struct iphdr *iphdr = iob_push ( iobuf, sizeof ( *iphdr ) );
309 struct sockaddr_in *sin_src = ( ( struct sockaddr_in * ) st_src );
310 struct sockaddr_in *sin_dest = ( ( struct sockaddr_in * ) st_dest );
311 struct ipv4_miniroute *miniroute;
312 struct in_addr next_hop;
313 struct in_addr netmask = { .s_addr = 0 };
314 uint8_t ll_dest_buf[MAX_LL_ADDR_LEN];
315 const void *ll_dest;
316 int rc;
317
318 /* Start profiling */
319 profile_start ( &ipv4_tx_profiler );
320
321 /* Update statistics */
322 ipv4_stats.out_requests++;
323
324 /* Fill up the IP header, except source address */
325 memset ( iphdr, 0, sizeof ( *iphdr ) );
326 iphdr->verhdrlen = ( IP_VER | ( sizeof ( *iphdr ) / 4 ) );
327 iphdr->service = IP_TOS;
328 iphdr->len = htons ( iob_len ( iobuf ) );
329 iphdr->ttl = IP_TTL;
330 iphdr->protocol = tcpip_protocol->tcpip_proto;
331 iphdr->dest = sin_dest->sin_addr;
332
333 /* Use routing table to identify next hop and transmitting netdev */
334 next_hop = iphdr->dest;
335 if ( sin_src )
336 iphdr->src = sin_src->sin_addr;
337 if ( ( next_hop.s_addr != INADDR_BROADCAST ) &&
338 ( ( miniroute = ipv4_route ( sin_dest->sin_scope_id,
339 &next_hop ) ) != NULL ) ) {
340 iphdr->src = miniroute->address;
341 netmask = miniroute->netmask;
342 netdev = miniroute->netdev;
343 }
344 if ( ! netdev ) {
345 DBGC ( sin_dest->sin_addr, "IPv4 has no route to %s\n",
346 inet_ntoa ( iphdr->dest ) );
347 ipv4_stats.out_no_routes++;
348 rc = -ENETUNREACH;
349 goto err;
350 }
351
352 /* (Ab)use the "ident" field to convey metadata about the
353 * network device statistics into packet traces. Useful for
354 * extracting debug information from non-debug builds.
355 */
356 iphdr->ident = htons ( ( (++next_ident_high) << 8 ) |
357 ( ( netdev->rx_stats.bad & 0xf ) << 4 ) |
358 ( ( netdev->rx_stats.good & 0xf ) << 0 ) );
359
360 /* Fix up checksums */
361 if ( trans_csum ) {
362 *trans_csum = ipv4_pshdr_chksum ( iobuf, *trans_csum );
363 if ( ! *trans_csum )
364 *trans_csum = tcpip_protocol->zero_csum;
365 }
366 iphdr->chksum = tcpip_chksum ( iphdr, sizeof ( *iphdr ) );
367
368 /* Print IP4 header for debugging */
369 DBGC2 ( sin_dest->sin_addr, "IPv4 TX %s->", inet_ntoa ( iphdr->src ) );
370 DBGC2 ( sin_dest->sin_addr, "%s len %d proto %d id %04x csum %04x\n",
371 inet_ntoa ( iphdr->dest ), ntohs ( iphdr->len ),
372 iphdr->protocol, ntohs ( iphdr->ident ),
373 ntohs ( iphdr->chksum ) );
374
375 /* Calculate link-layer destination address, if possible */
376 if ( ( ( next_hop.s_addr ^ INADDR_BROADCAST ) & ~netmask.s_addr ) == 0){
377 /* Broadcast address */
378 ipv4_stats.out_bcast_pkts++;
379 ll_dest = netdev->ll_broadcast;
380 } else if ( IN_IS_MULTICAST ( next_hop.s_addr ) ) {
381 /* Multicast address */
382 ipv4_stats.out_mcast_pkts++;
383 if ( ( rc = netdev->ll_protocol->mc_hash ( AF_INET, &next_hop,
384 ll_dest_buf ) ) !=0){
385 DBGC ( sin_dest->sin_addr, "IPv4 could not hash "
386 "multicast %s: %s\n",
387 inet_ntoa ( next_hop ), strerror ( rc ) );
388 goto err;
389 }
390 ll_dest = ll_dest_buf;
391 } else {
392 /* Unicast address */
393 ll_dest = NULL;
394 }
395
396 /* Update statistics */
397 ipv4_stats.out_transmits++;
398 ipv4_stats.out_octets += iob_len ( iobuf );
399
400 /* Hand off to link layer (via ARP if applicable) */
401 if ( ll_dest ) {
402 if ( ( rc = net_tx ( iobuf, netdev, &ipv4_protocol, ll_dest,
403 netdev->ll_addr ) ) != 0 ) {
404 DBGC ( sin_dest->sin_addr, "IPv4 could not transmit "
405 "packet via %s: %s\n",
406 netdev->name, strerror ( rc ) );
407 return rc;
408 }
409 } else {
410 if ( ( rc = arp_tx ( iobuf, netdev, &ipv4_protocol, &next_hop,
411 &iphdr->src, netdev->ll_addr ) ) != 0 ) {
412 DBGC ( sin_dest->sin_addr, "IPv4 could not transmit "
413 "packet via %s: %s\n",
414 netdev->name, strerror ( rc ) );
415 return rc;
416 }
417 }
418
419 profile_stop ( &ipv4_tx_profiler );
420 return 0;
421
422 err:
423 free_iob ( iobuf );
424 return rc;
425 }
426
427 /**
428 * Check if network device has any IPv4 address
429 *
430 * @v netdev Network device
431 * @ret has_any_addr Network device has any IPv4 address
432 */
ipv4_has_any_addr(struct net_device * netdev)433 int ipv4_has_any_addr ( struct net_device *netdev ) {
434 struct ipv4_miniroute *miniroute;
435
436 list_for_each_entry ( miniroute, &ipv4_miniroutes, list ) {
437 if ( miniroute->netdev == netdev )
438 return 1;
439 }
440 return 0;
441 }
442
443 /**
444 * Check if network device has a specific IPv4 address
445 *
446 * @v netdev Network device
447 * @v addr IPv4 address
448 * @ret has_addr Network device has this IPv4 address
449 */
ipv4_has_addr(struct net_device * netdev,struct in_addr addr)450 static int ipv4_has_addr ( struct net_device *netdev, struct in_addr addr ) {
451 struct ipv4_miniroute *miniroute;
452
453 list_for_each_entry ( miniroute, &ipv4_miniroutes, list ) {
454 if ( ( miniroute->netdev == netdev ) &&
455 ( miniroute->address.s_addr == addr.s_addr ) ) {
456 /* Found matching address */
457 return 1;
458 }
459 }
460 return 0;
461 }
462
463 /**
464 * Process incoming packets
465 *
466 * @v iobuf I/O buffer
467 * @v netdev Network device
468 * @v ll_dest Link-layer destination address
469 * @v ll_source Link-layer destination source
470 * @v flags Packet flags
471 * @ret rc Return status code
472 *
473 * This function expects an IP4 network datagram. It processes the headers
474 * and sends it to the transport layer.
475 */
ipv4_rx(struct io_buffer * iobuf,struct net_device * netdev,const void * ll_dest __unused,const void * ll_source __unused,unsigned int flags)476 static int ipv4_rx ( struct io_buffer *iobuf,
477 struct net_device *netdev,
478 const void *ll_dest __unused,
479 const void *ll_source __unused,
480 unsigned int flags ) {
481 struct iphdr *iphdr = iobuf->data;
482 size_t hdrlen;
483 size_t len;
484 union {
485 struct sockaddr_in sin;
486 struct sockaddr_tcpip st;
487 } src, dest;
488 uint16_t csum;
489 uint16_t pshdr_csum;
490 int rc;
491
492 /* Start profiling */
493 profile_start ( &ipv4_rx_profiler );
494
495 /* Update statistics */
496 ipv4_stats.in_receives++;
497 ipv4_stats.in_octets += iob_len ( iobuf );
498 if ( flags & LL_BROADCAST ) {
499 ipv4_stats.in_bcast_pkts++;
500 } else if ( flags & LL_MULTICAST ) {
501 ipv4_stats.in_mcast_pkts++;
502 }
503
504 /* Sanity check the IPv4 header */
505 if ( iob_len ( iobuf ) < sizeof ( *iphdr ) ) {
506 DBGC ( iphdr->src, "IPv4 packet too short at %zd bytes (min "
507 "%zd bytes)\n", iob_len ( iobuf ), sizeof ( *iphdr ) );
508 goto err_header;
509 }
510 if ( ( iphdr->verhdrlen & IP_MASK_VER ) != IP_VER ) {
511 DBGC ( iphdr->src, "IPv4 version %#02x not supported\n",
512 iphdr->verhdrlen );
513 goto err_header;
514 }
515 hdrlen = ( ( iphdr->verhdrlen & IP_MASK_HLEN ) * 4 );
516 if ( hdrlen < sizeof ( *iphdr ) ) {
517 DBGC ( iphdr->src, "IPv4 header too short at %zd bytes (min "
518 "%zd bytes)\n", hdrlen, sizeof ( *iphdr ) );
519 goto err_header;
520 }
521 if ( hdrlen > iob_len ( iobuf ) ) {
522 DBGC ( iphdr->src, "IPv4 header too long at %zd bytes "
523 "(packet is %zd bytes)\n", hdrlen, iob_len ( iobuf ) );
524 goto err_header;
525 }
526 if ( ( csum = tcpip_chksum ( iphdr, hdrlen ) ) != 0 ) {
527 DBGC ( iphdr->src, "IPv4 checksum incorrect (is %04x "
528 "including checksum field, should be 0000)\n", csum );
529 goto err_header;
530 }
531 len = ntohs ( iphdr->len );
532 if ( len < hdrlen ) {
533 DBGC ( iphdr->src, "IPv4 length too short at %zd bytes "
534 "(header is %zd bytes)\n", len, hdrlen );
535 goto err_header;
536 }
537 if ( len > iob_len ( iobuf ) ) {
538 DBGC ( iphdr->src, "IPv4 length too long at %zd bytes "
539 "(packet is %zd bytes)\n", len, iob_len ( iobuf ) );
540 ipv4_stats.in_truncated_pkts++;
541 goto err_other;
542 }
543
544 /* Truncate packet to correct length */
545 iob_unput ( iobuf, ( iob_len ( iobuf ) - len ) );
546
547 /* Print IPv4 header for debugging */
548 DBGC2 ( iphdr->src, "IPv4 RX %s<-", inet_ntoa ( iphdr->dest ) );
549 DBGC2 ( iphdr->src, "%s len %d proto %d id %04x csum %04x\n",
550 inet_ntoa ( iphdr->src ), ntohs ( iphdr->len ), iphdr->protocol,
551 ntohs ( iphdr->ident ), ntohs ( iphdr->chksum ) );
552
553 /* Discard unicast packets not destined for us */
554 if ( ( ! ( flags & LL_MULTICAST ) ) &&
555 ( iphdr->dest.s_addr != INADDR_BROADCAST ) &&
556 ipv4_has_any_addr ( netdev ) &&
557 ( ! ipv4_has_addr ( netdev, iphdr->dest ) ) ) {
558 DBGC ( iphdr->src, "IPv4 discarding non-local unicast packet "
559 "for %s\n", inet_ntoa ( iphdr->dest ) );
560 ipv4_stats.in_addr_errors++;
561 goto err_other;
562 }
563
564 /* Perform fragment reassembly if applicable */
565 if ( iphdr->frags & htons ( IP_MASK_OFFSET | IP_MASK_MOREFRAGS ) ) {
566 /* Pass the fragment to fragment_reassemble() which returns
567 * either a fully reassembled I/O buffer or NULL.
568 */
569 iobuf = fragment_reassemble ( &ipv4_reassembler, iobuf,
570 &hdrlen );
571 if ( ! iobuf )
572 return 0;
573 iphdr = iobuf->data;
574 }
575
576 /* Construct socket addresses, calculate pseudo-header
577 * checksum, and hand off to transport layer
578 */
579 memset ( &src, 0, sizeof ( src ) );
580 src.sin.sin_family = AF_INET;
581 src.sin.sin_addr = iphdr->src;
582 memset ( &dest, 0, sizeof ( dest ) );
583 dest.sin.sin_family = AF_INET;
584 dest.sin.sin_addr = iphdr->dest;
585 pshdr_csum = ipv4_pshdr_chksum ( iobuf, TCPIP_EMPTY_CSUM );
586 iob_pull ( iobuf, hdrlen );
587 if ( ( rc = tcpip_rx ( iobuf, netdev, iphdr->protocol, &src.st,
588 &dest.st, pshdr_csum, &ipv4_stats ) ) != 0 ) {
589 DBGC ( src.sin.sin_addr, "IPv4 received packet rejected by "
590 "stack: %s\n", strerror ( rc ) );
591 return rc;
592 }
593
594 profile_stop ( &ipv4_rx_profiler );
595 return 0;
596
597 err_header:
598 ipv4_stats.in_hdr_errors++;
599 err_other:
600 free_iob ( iobuf );
601 return -EINVAL;
602 }
603
604 /**
605 * Check existence of IPv4 address for ARP
606 *
607 * @v netdev Network device
608 * @v net_addr Network-layer address
609 * @ret rc Return status code
610 */
ipv4_arp_check(struct net_device * netdev,const void * net_addr)611 static int ipv4_arp_check ( struct net_device *netdev, const void *net_addr ) {
612 const struct in_addr *address = net_addr;
613
614 if ( ipv4_has_addr ( netdev, *address ) )
615 return 0;
616
617 return -ENOENT;
618 }
619
620 /**
621 * Parse IPv4 address
622 *
623 * @v string IPv4 address string
624 * @ret in IPv4 address to fill in
625 * @ret ok IPv4 address is valid
626 *
627 * Note that this function returns nonzero iff the address is valid,
628 * to match the standard BSD API function of the same name. Unlike
629 * most other iPXE functions, a zero therefore indicates failure.
630 */
inet_aton(const char * string,struct in_addr * in)631 int inet_aton ( const char *string, struct in_addr *in ) {
632 const char *separator = "...";
633 uint8_t *byte = ( ( uint8_t * ) in );
634 char *endp;
635 unsigned long value;
636
637 while ( 1 ) {
638 value = strtoul ( string, &endp, 0 );
639 if ( string == endp )
640 return 0;
641 if ( value > 0xff )
642 return 0;
643 *(byte++) = value;
644 if ( *endp != *separator )
645 return 0;
646 if ( ! *(separator++) )
647 return 1;
648 string = ( endp + 1 );
649 }
650 }
651
652 /**
653 * Convert IPv4 address to dotted-quad notation
654 *
655 * @v in IPv4 address
656 * @ret string IPv4 address in dotted-quad notation
657 */
inet_ntoa(struct in_addr in)658 char * inet_ntoa ( struct in_addr in ) {
659 static char buf[16]; /* "xxx.xxx.xxx.xxx" */
660 uint8_t *bytes = ( uint8_t * ) ∈
661
662 sprintf ( buf, "%d.%d.%d.%d", bytes[0], bytes[1], bytes[2], bytes[3] );
663 return buf;
664 }
665
666 /**
667 * Transcribe IPv4 address
668 *
669 * @v net_addr IPv4 address
670 * @ret string IPv4 address in dotted-quad notation
671 *
672 */
ipv4_ntoa(const void * net_addr)673 static const char * ipv4_ntoa ( const void *net_addr ) {
674 return inet_ntoa ( * ( ( struct in_addr * ) net_addr ) );
675 }
676
677 /**
678 * Transcribe IPv4 socket address
679 *
680 * @v sa Socket address
681 * @ret string Socket address in standard notation
682 */
ipv4_sock_ntoa(struct sockaddr * sa)683 static const char * ipv4_sock_ntoa ( struct sockaddr *sa ) {
684 struct sockaddr_in *sin = ( ( struct sockaddr_in * ) sa );
685
686 return inet_ntoa ( sin->sin_addr );
687 }
688
689 /**
690 * Parse IPv4 socket address
691 *
692 * @v string Socket address string
693 * @v sa Socket address to fill in
694 * @ret rc Return status code
695 */
ipv4_sock_aton(const char * string,struct sockaddr * sa)696 static int ipv4_sock_aton ( const char *string, struct sockaddr *sa ) {
697 struct sockaddr_in *sin = ( ( struct sockaddr_in * ) sa );
698 struct in_addr in;
699
700 if ( inet_aton ( string, &in ) ) {
701 sin->sin_addr = in;
702 return 0;
703 }
704 return -EINVAL;
705 }
706
707 /** IPv4 protocol */
708 struct net_protocol ipv4_protocol __net_protocol = {
709 .name = "IP",
710 .net_proto = htons ( ETH_P_IP ),
711 .net_addr_len = sizeof ( struct in_addr ),
712 .rx = ipv4_rx,
713 .ntoa = ipv4_ntoa,
714 };
715
716 /** IPv4 TCPIP net protocol */
717 struct tcpip_net_protocol ipv4_tcpip_protocol __tcpip_net_protocol = {
718 .name = "IPv4",
719 .sa_family = AF_INET,
720 .header_len = sizeof ( struct iphdr ),
721 .net_protocol = &ipv4_protocol,
722 .tx = ipv4_tx,
723 .netdev = ipv4_netdev,
724 };
725
726 /** IPv4 ARP protocol */
727 struct arp_net_protocol ipv4_arp_protocol __arp_net_protocol = {
728 .net_protocol = &ipv4_protocol,
729 .check = ipv4_arp_check,
730 };
731
732 /** IPv4 socket address converter */
733 struct sockaddr_converter ipv4_sockaddr_converter __sockaddr_converter = {
734 .family = AF_INET,
735 .ntoa = ipv4_sock_ntoa,
736 .aton = ipv4_sock_aton,
737 };
738
739 /******************************************************************************
740 *
741 * Settings
742 *
743 ******************************************************************************
744 */
745
746 /**
747 * Parse IPv4 address setting value
748 *
749 * @v type Setting type
750 * @v value Formatted setting value
751 * @v buf Buffer to contain raw value
752 * @v len Length of buffer
753 * @ret len Length of raw value, or negative error
754 */
parse_ipv4_setting(const struct setting_type * type __unused,const char * value,void * buf,size_t len)755 int parse_ipv4_setting ( const struct setting_type *type __unused,
756 const char *value, void *buf, size_t len ) {
757 struct in_addr ipv4;
758
759 /* Parse IPv4 address */
760 if ( inet_aton ( value, &ipv4 ) == 0 )
761 return -EINVAL;
762
763 /* Copy to buffer */
764 if ( len > sizeof ( ipv4 ) )
765 len = sizeof ( ipv4 );
766 memcpy ( buf, &ipv4, len );
767
768 return ( sizeof ( ipv4 ) );
769 }
770
771 /**
772 * Format IPv4 address setting value
773 *
774 * @v type Setting type
775 * @v raw Raw setting value
776 * @v raw_len Length of raw setting value
777 * @v buf Buffer to contain formatted value
778 * @v len Length of buffer
779 * @ret len Length of formatted value, or negative error
780 */
format_ipv4_setting(const struct setting_type * type __unused,const void * raw,size_t raw_len,char * buf,size_t len)781 int format_ipv4_setting ( const struct setting_type *type __unused,
782 const void *raw, size_t raw_len, char *buf,
783 size_t len ) {
784 const struct in_addr *ipv4 = raw;
785
786 if ( raw_len < sizeof ( *ipv4 ) )
787 return -EINVAL;
788 return snprintf ( buf, len, "%s", inet_ntoa ( *ipv4 ) );
789 }
790
791 /** IPv4 address setting */
792 const struct setting ip_setting __setting ( SETTING_IP4, ip ) = {
793 .name = "ip",
794 .description = "IP address",
795 .tag = DHCP_EB_YIADDR,
796 .type = &setting_type_ipv4,
797 };
798
799 /** IPv4 subnet mask setting */
800 const struct setting netmask_setting __setting ( SETTING_IP4, netmask ) = {
801 .name = "netmask",
802 .description = "Subnet mask",
803 .tag = DHCP_SUBNET_MASK,
804 .type = &setting_type_ipv4,
805 };
806
807 /** Default gateway setting */
808 const struct setting gateway_setting __setting ( SETTING_IP4, gateway ) = {
809 .name = "gateway",
810 .description = "Default gateway",
811 .tag = DHCP_ROUTERS,
812 .type = &setting_type_ipv4,
813 };
814
815 /**
816 * Send gratuitous ARP, if applicable
817 *
818 * @v netdev Network device
819 * @v address IPv4 address
820 * @v netmask Subnet mask
821 * @v gateway Gateway address (if any)
822 * @ret rc Return status code
823 */
ipv4_gratuitous_arp(struct net_device * netdev,struct in_addr address,struct in_addr netmask __unused,struct in_addr gateway __unused)824 static int ipv4_gratuitous_arp ( struct net_device *netdev,
825 struct in_addr address,
826 struct in_addr netmask __unused,
827 struct in_addr gateway __unused ) {
828 int rc;
829
830 /* Do nothing if network device already has this IPv4 address */
831 if ( ipv4_has_addr ( netdev, address ) )
832 return 0;
833
834 /* Transmit gratuitous ARP */
835 DBGC ( netdev, "IPv4 sending gratuitous ARP for %s via %s\n",
836 inet_ntoa ( address ), netdev->name );
837 if ( ( rc = arp_tx_request ( netdev, &ipv4_protocol, &address,
838 &address ) ) != 0 ) {
839 DBGC ( netdev, "IPv4 could not transmit gratuitous ARP: %s\n",
840 strerror ( rc ) );
841 /* Treat failures as non-fatal */
842 }
843
844 return 0;
845 }
846
847 /**
848 * Process IPv4 network device settings
849 *
850 * @v apply Application method
851 * @ret rc Return status code
852 */
ipv4_settings(int (* apply)(struct net_device * netdev,struct in_addr address,struct in_addr netmask,struct in_addr gateway))853 static int ipv4_settings ( int ( * apply ) ( struct net_device *netdev,
854 struct in_addr address,
855 struct in_addr netmask,
856 struct in_addr gateway ) ) {
857 struct net_device *netdev;
858 struct settings *settings;
859 struct in_addr address = { 0 };
860 struct in_addr netmask = { 0 };
861 struct in_addr gateway = { 0 };
862 int rc;
863
864 /* Process settings for each network device */
865 for_each_netdev ( netdev ) {
866
867 /* Get network device settings */
868 settings = netdev_settings ( netdev );
869
870 /* Get IPv4 address */
871 address.s_addr = 0;
872 fetch_ipv4_setting ( settings, &ip_setting, &address );
873 if ( ! address.s_addr )
874 continue;
875
876 /* Get subnet mask */
877 fetch_ipv4_setting ( settings, &netmask_setting, &netmask );
878
879 /* Calculate default netmask, if necessary */
880 if ( ! netmask.s_addr ) {
881 if ( IN_IS_CLASSA ( address.s_addr ) ) {
882 netmask.s_addr = INADDR_NET_CLASSA;
883 } else if ( IN_IS_CLASSB ( address.s_addr ) ) {
884 netmask.s_addr = INADDR_NET_CLASSB;
885 } else if ( IN_IS_CLASSC ( address.s_addr ) ) {
886 netmask.s_addr = INADDR_NET_CLASSC;
887 }
888 }
889
890 /* Get default gateway, if present */
891 fetch_ipv4_setting ( settings, &gateway_setting, &gateway );
892
893 /* Apply settings */
894 if ( ( rc = apply ( netdev, address, netmask, gateway ) ) != 0 )
895 return rc;
896 }
897
898 return 0;
899 }
900
901 /**
902 * Create IPv4 routing table based on configured settings
903 *
904 * @ret rc Return status code
905 */
ipv4_create_routes(void)906 static int ipv4_create_routes ( void ) {
907 struct ipv4_miniroute *miniroute;
908 struct ipv4_miniroute *tmp;
909 int rc;
910
911 /* Send gratuitous ARPs for any new IPv4 addresses */
912 ipv4_settings ( ipv4_gratuitous_arp );
913
914 /* Delete all existing routes */
915 list_for_each_entry_safe ( miniroute, tmp, &ipv4_miniroutes, list )
916 del_ipv4_miniroute ( miniroute );
917
918 /* Create a route for each configured network device */
919 if ( ( rc = ipv4_settings ( add_ipv4_miniroute ) ) != 0 )
920 return rc;
921
922 return 0;
923 }
924
925 /** IPv4 settings applicator */
926 struct settings_applicator ipv4_settings_applicator __settings_applicator = {
927 .apply = ipv4_create_routes,
928 };
929
930 /* Drag in objects via ipv4_protocol */
931 REQUIRING_SYMBOL ( ipv4_protocol );
932
933 /* Drag in ICMPv4 */
934 REQUIRE_OBJECT ( icmpv4 );
935