]> git.proxmox.com Git - mirror_frr.git/blob - bgpd/bgp_pbr.c
projects / mirror_frr.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge pull request #12397 from opensourcerouting/vscode-intro
[mirror_frr.git] / bgpd / bgp_pbr.c
1 /*
2 * BGP pbr
3 * Copyright (C) 6WIND
4 *
5 * FRR is free software; you can redistribute it and/or modify it
6 * under the terms of the GNU General Public License as published by the
7 * Free Software Foundation; either version 2, or (at your option) any
8 * later version.
9 *
10 * FRR 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 along
16 * with this program; see the file COPYING; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
18 */
19
20 #include "zebra.h"
21 #include "prefix.h"
22 #include "zclient.h"
23 #include "jhash.h"
24 #include "pbr.h"
25
26 #include "lib/printfrr.h"
27
28 #include "bgpd/bgpd.h"
29 #include "bgpd/bgp_pbr.h"
30 #include "bgpd/bgp_debug.h"
31 #include "bgpd/bgp_flowspec_util.h"
32 #include "bgpd/bgp_ecommunity.h"
33 #include "bgpd/bgp_route.h"
34 #include "bgpd/bgp_attr.h"
35 #include "bgpd/bgp_zebra.h"
36 #include "bgpd/bgp_mplsvpn.h"
37 #include "bgpd/bgp_flowspec_private.h"
38 #include "bgpd/bgp_errors.h"
39
40 DEFINE_MTYPE_STATIC(BGPD, PBR_MATCH_ENTRY, "PBR match entry");
41 DEFINE_MTYPE_STATIC(BGPD, PBR_MATCH, "PBR match");
42 DEFINE_MTYPE_STATIC(BGPD, PBR_ACTION, "PBR action");
43 DEFINE_MTYPE_STATIC(BGPD, PBR_RULE, "PBR rule");
44 DEFINE_MTYPE_STATIC(BGPD, PBR, "BGP PBR Context");
45 DEFINE_MTYPE_STATIC(BGPD, PBR_VALMASK, "BGP PBR Val Mask Value");
46
47 /* chain strings too long to fit in one line */
48 #define FSPEC_ACTION_EXCEED_LIMIT "flowspec actions exceeds limit"
49 #define IPV6_FRAGMENT_INVALID "fragment not valid for IPv6 for this implementation"
50
51 RB_GENERATE(bgp_pbr_interface_head, bgp_pbr_interface,
52 id_entry, bgp_pbr_interface_compare);
53 struct bgp_pbr_interface_head ifaces_by_name_ipv4 =
54 RB_INITIALIZER(&ifaces_by_name_ipv4);
55
56 static int bgp_pbr_match_counter_unique;
57 static int bgp_pbr_match_entry_counter_unique;
58 static int bgp_pbr_action_counter_unique;
59 static int bgp_pbr_match_iptable_counter_unique;
60
61 struct bgp_pbr_match_iptable_unique {
62 uint32_t unique;
63 struct bgp_pbr_match *bpm_found;
64 };
65
66 struct bgp_pbr_match_entry_unique {
67 uint32_t unique;
68 struct bgp_pbr_match_entry *bpme_found;
69 };
70
71 struct bgp_pbr_action_unique {
72 uint32_t unique;
73 struct bgp_pbr_action *bpa_found;
74 };
75
76 struct bgp_pbr_rule_unique {
77 uint32_t unique;
78 struct bgp_pbr_rule *bpr_found;
79 };
80
81 static int bgp_pbr_rule_walkcb(struct hash_bucket *bucket, void *arg)
82 {
83 struct bgp_pbr_rule *bpr = (struct bgp_pbr_rule *)bucket->data;
84 struct bgp_pbr_rule_unique *bpru = (struct bgp_pbr_rule_unique *)
85 arg;
86 uint32_t unique = bpru->unique;
87
88 if (bpr->unique == unique) {
89 bpru->bpr_found = bpr;
90 return HASHWALK_ABORT;
91 }
92 return HASHWALK_CONTINUE;
93 }
94
95 static int bgp_pbr_action_walkcb(struct hash_bucket *bucket, void *arg)
96 {
97 struct bgp_pbr_action *bpa = (struct bgp_pbr_action *)bucket->data;
98 struct bgp_pbr_action_unique *bpau = (struct bgp_pbr_action_unique *)
99 arg;
100 uint32_t unique = bpau->unique;
101
102 if (bpa->unique == unique) {
103 bpau->bpa_found = bpa;
104 return HASHWALK_ABORT;
105 }
106 return HASHWALK_CONTINUE;
107 }
108
109 static int bgp_pbr_match_entry_walkcb(struct hash_bucket *bucket, void *arg)
110 {
111 struct bgp_pbr_match_entry *bpme =
112 (struct bgp_pbr_match_entry *)bucket->data;
113 struct bgp_pbr_match_entry_unique *bpmeu =
114 (struct bgp_pbr_match_entry_unique *)arg;
115 uint32_t unique = bpmeu->unique;
116
117 if (bpme->unique == unique) {
118 bpmeu->bpme_found = bpme;
119 return HASHWALK_ABORT;
120 }
121 return HASHWALK_CONTINUE;
122 }
123
124 struct bgp_pbr_match_ipsetname {
125 char *ipsetname;
126 struct bgp_pbr_match *bpm_found;
127 };
128
129 static int bgp_pbr_match_pername_walkcb(struct hash_bucket *bucket, void *arg)
130 {
131 struct bgp_pbr_match *bpm = (struct bgp_pbr_match *)bucket->data;
132 struct bgp_pbr_match_ipsetname *bpmi =
133 (struct bgp_pbr_match_ipsetname *)arg;
134 char *ipset_name = bpmi->ipsetname;
135
136 if (!strncmp(ipset_name, bpm->ipset_name,
137 ZEBRA_IPSET_NAME_SIZE)) {
138 bpmi->bpm_found = bpm;
139 return HASHWALK_ABORT;
140 }
141 return HASHWALK_CONTINUE;
142 }
143
144 static int bgp_pbr_match_iptable_walkcb(struct hash_bucket *bucket, void *arg)
145 {
146 struct bgp_pbr_match *bpm = (struct bgp_pbr_match *)bucket->data;
147 struct bgp_pbr_match_iptable_unique *bpmiu =
148 (struct bgp_pbr_match_iptable_unique *)arg;
149 uint32_t unique = bpmiu->unique;
150
151 if (bpm->unique2 == unique) {
152 bpmiu->bpm_found = bpm;
153 return HASHWALK_ABORT;
154 }
155 return HASHWALK_CONTINUE;
156 }
157
158 struct bgp_pbr_match_unique {
159 uint32_t unique;
160 struct bgp_pbr_match *bpm_found;
161 };
162
163 static int bgp_pbr_match_walkcb(struct hash_bucket *bucket, void *arg)
164 {
165 struct bgp_pbr_match *bpm = (struct bgp_pbr_match *)bucket->data;
166 struct bgp_pbr_match_unique *bpmu = (struct bgp_pbr_match_unique *)
167 arg;
168 uint32_t unique = bpmu->unique;
169
170 if (bpm->unique == unique) {
171 bpmu->bpm_found = bpm;
172 return HASHWALK_ABORT;
173 }
174 return HASHWALK_CONTINUE;
175 }
176
177 static int snprintf_bgp_pbr_match_val(char *str, int len,
178 struct bgp_pbr_match_val *mval,
179 const char *prepend)
180 {
181 char *ptr = str;
182 int delta;
183
184 if (prepend) {
185 delta = snprintf(ptr, len, "%s", prepend);
186 ptr += delta;
187 len -= delta;
188 } else {
189 if (mval->unary_operator & OPERATOR_UNARY_OR) {
190 delta = snprintf(ptr, len, ", or ");
191 ptr += delta;
192 len -= delta;
193 }
194 if (mval->unary_operator & OPERATOR_UNARY_AND) {
195 delta = snprintf(ptr, len, ", and ");
196 ptr += delta;
197 len -= delta;
198 }
199 }
200 if (mval->compare_operator & OPERATOR_COMPARE_LESS_THAN) {
201 delta = snprintf(ptr, len, "<");
202 ptr += delta;
203 len -= delta;
204 }
205 if (mval->compare_operator & OPERATOR_COMPARE_GREATER_THAN) {
206 delta = snprintf(ptr, len, ">");
207 ptr += delta;
208 len -= delta;
209 }
210 if (mval->compare_operator & OPERATOR_COMPARE_EQUAL_TO) {
211 delta = snprintf(ptr, len, "=");
212 ptr += delta;
213 len -= delta;
214 }
215 if (mval->compare_operator & OPERATOR_COMPARE_EXACT_MATCH) {
216 delta = snprintf(ptr, len, "match");
217 ptr += delta;
218 len -= delta;
219 }
220 ptr += snprintf(ptr, len, " %u", mval->value);
221 return (int)(ptr - str);
222 }
223
224 #define INCREMENT_DISPLAY(_ptr, _cnt, _len) do { \
225 int sn_delta; \
226 \
227 if (_cnt) { \
228 sn_delta = snprintf((_ptr), (_len), "; ");\
229 (_len) -= sn_delta; \
230 (_ptr) += sn_delta; \
231 } \
232 (_cnt)++; \
233 } while (0)
234
235 /* this structure can be used for port range,
236 * but also for other values range like packet length range
237 */
238 struct bgp_pbr_range_port {
239 uint16_t min_port;
240 uint16_t max_port;
241 };
242
243 /* this structure can be used to filter with a mask
244 * for instance it supports not instructions like for
245 * tcpflags
246 */
247 struct bgp_pbr_val_mask {
248 uint16_t val;
249 uint16_t mask;
250 };
251
252 /* this structure is used to pass instructs
253 * so that BGP can create pbr instructions to ZEBRA
254 */
255 struct bgp_pbr_filter {
256 uint8_t type;
257 vrf_id_t vrf_id;
258 uint8_t family;
259 struct prefix *src;
260 struct prefix *dst;
261 uint8_t bitmask_iprule;
262 uint8_t protocol;
263 struct bgp_pbr_range_port *pkt_len;
264 struct bgp_pbr_range_port *src_port;
265 struct bgp_pbr_range_port *dst_port;
266 struct bgp_pbr_val_mask *tcp_flags;
267 struct bgp_pbr_val_mask *dscp;
268 struct bgp_pbr_val_mask *flow_label;
269 struct bgp_pbr_val_mask *pkt_len_val;
270 struct bgp_pbr_val_mask *fragment;
271 };
272
273 /* this structure is used to contain OR instructions
274 * so that BGP can create multiple pbr instructions
275 * to ZEBRA
276 */
277 struct bgp_pbr_or_filter {
278 struct list *tcpflags;
279 struct list *dscp;
280 struct list *flowlabel;
281 struct list *pkt_len;
282 struct list *fragment;
283 struct list *icmp_type;
284 struct list *icmp_code;
285 };
286
287 static void bgp_pbr_policyroute_add_to_zebra_unit(struct bgp *bgp,
288 struct bgp_path_info *path,
289 struct bgp_pbr_filter *bpf,
290 struct nexthop *nh,
291 float *rate);
292
293 static void bgp_pbr_dump_entry(struct bgp_pbr_filter *bpf, bool add);
294
295 static bool bgp_pbr_extract_enumerate_unary_opposite(
296 uint8_t unary_operator,
297 struct bgp_pbr_val_mask *and_valmask,
298 struct list *or_valmask, uint32_t value,
299 uint8_t type_entry)
300 {
301 if (unary_operator == OPERATOR_UNARY_AND && and_valmask) {
302 if (type_entry == FLOWSPEC_TCP_FLAGS) {
303 and_valmask->mask |=
304 TCP_HEADER_ALL_FLAGS &
305 ~(value);
306 } else if (type_entry == FLOWSPEC_DSCP ||
307 type_entry == FLOWSPEC_FLOW_LABEL ||
308 type_entry == FLOWSPEC_PKT_LEN ||
309 type_entry == FLOWSPEC_FRAGMENT) {
310 and_valmask->val = value;
311 and_valmask->mask = 1; /* inverse */
312 }
313 } else if (unary_operator == OPERATOR_UNARY_OR && or_valmask) {
314 and_valmask = XCALLOC(MTYPE_PBR_VALMASK,
315 sizeof(struct bgp_pbr_val_mask));
316 if (type_entry == FLOWSPEC_TCP_FLAGS) {
317 and_valmask->val = TCP_HEADER_ALL_FLAGS;
318 and_valmask->mask |=
319 TCP_HEADER_ALL_FLAGS &
320 ~(value);
321 } else if (type_entry == FLOWSPEC_DSCP ||
322 type_entry == FLOWSPEC_FLOW_LABEL ||
323 type_entry == FLOWSPEC_FRAGMENT ||
324 type_entry == FLOWSPEC_PKT_LEN) {
325 and_valmask->val = value;
326 and_valmask->mask = 1; /* inverse */
327 }
328 listnode_add(or_valmask, and_valmask);
329 } else if (type_entry == FLOWSPEC_ICMP_CODE ||
330 type_entry == FLOWSPEC_ICMP_TYPE)
331 return false;
332 return true;
333 }
334
335 /* TCP : FIN and SYN -> val = ALL; mask = 3
336 * TCP : not (FIN and SYN) -> val = ALL; mask = ALL & ~(FIN|RST)
337 * other variables type: dscp, pkt len, fragment, flow label
338 * - value is copied in bgp_pbr_val_mask->val value
339 * - if negate form is identifierd, bgp_pbr_val_mask->mask set to 1
340 */
341 static bool bgp_pbr_extract_enumerate_unary(struct bgp_pbr_match_val list[],
342 int num, uint8_t unary_operator,
343 void *valmask, uint8_t type_entry)
344 {
345 int i = 0;
346 struct bgp_pbr_val_mask *and_valmask = NULL;
347 struct list *or_valmask = NULL;
348 bool ret;
349
350 if (valmask) {
351 if (unary_operator == OPERATOR_UNARY_AND) {
352 and_valmask = (struct bgp_pbr_val_mask *)valmask;
353 memset(and_valmask, 0, sizeof(struct bgp_pbr_val_mask));
354 } else if (unary_operator == OPERATOR_UNARY_OR) {
355 or_valmask = (struct list *)valmask;
356 }
357 }
358 for (i = 0; i < num; i++) {
359 if (i != 0 && list[i].unary_operator !=
360 unary_operator)
361 return false;
362 if (!(list[i].compare_operator &
363 OPERATOR_COMPARE_EQUAL_TO) &&
364 !(list[i].compare_operator &
365 OPERATOR_COMPARE_EXACT_MATCH)) {
366 if ((list[i].compare_operator &
367 OPERATOR_COMPARE_LESS_THAN) &&
368 (list[i].compare_operator &
369 OPERATOR_COMPARE_GREATER_THAN)) {
370 ret = bgp_pbr_extract_enumerate_unary_opposite(
371 unary_operator, and_valmask,
372 or_valmask, list[i].value,
373 type_entry);
374 if (!ret)
375 return ret;
376 continue;
377 }
378 return false;
379 }
380 if (unary_operator == OPERATOR_UNARY_AND && and_valmask) {
381 if (type_entry == FLOWSPEC_TCP_FLAGS)
382 and_valmask->mask |=
383 TCP_HEADER_ALL_FLAGS & list[i].value;
384 } else if (unary_operator == OPERATOR_UNARY_OR && or_valmask) {
385 and_valmask = XCALLOC(MTYPE_PBR_VALMASK,
386 sizeof(struct bgp_pbr_val_mask));
387 if (type_entry == FLOWSPEC_TCP_FLAGS) {
388 and_valmask->val = TCP_HEADER_ALL_FLAGS;
389 and_valmask->mask |=
390 TCP_HEADER_ALL_FLAGS & list[i].value;
391 } else if (type_entry == FLOWSPEC_DSCP ||
392 type_entry == FLOWSPEC_FLOW_LABEL ||
393 type_entry == FLOWSPEC_ICMP_TYPE ||
394 type_entry == FLOWSPEC_ICMP_CODE ||
395 type_entry == FLOWSPEC_FRAGMENT ||
396 type_entry == FLOWSPEC_PKT_LEN)
397 and_valmask->val = list[i].value;
398 listnode_add(or_valmask, and_valmask);
399 }
400 }
401 if (unary_operator == OPERATOR_UNARY_AND && and_valmask
402 && type_entry == FLOWSPEC_TCP_FLAGS)
403 and_valmask->val = TCP_HEADER_ALL_FLAGS;
404 return true;
405 }
406
407 /* if unary operator can either be UNARY_OR/AND/OR-AND.
408 * in the latter case, combinationf of both is not handled
409 */
410 static bool bgp_pbr_extract_enumerate(struct bgp_pbr_match_val list[],
411 int num, uint8_t unary_operator,
412 void *valmask, uint8_t type_entry)
413 {
414 bool ret;
415 uint8_t unary_operator_val;
416 bool double_check = false;
417
418 if ((unary_operator & OPERATOR_UNARY_OR) &&
419 (unary_operator & OPERATOR_UNARY_AND)) {
420 unary_operator_val = OPERATOR_UNARY_AND;
421 double_check = true;
422 } else
423 unary_operator_val = unary_operator;
424 ret = bgp_pbr_extract_enumerate_unary(list, num, unary_operator_val,
425 valmask, type_entry);
426 if (!ret && double_check)
427 ret = bgp_pbr_extract_enumerate_unary(list, num,
428 OPERATOR_UNARY_OR,
429 valmask,
430 type_entry);
431 return ret;
432 }
433
434 /* returns the unary operator that is in the list
435 * return 0 if both operators are used
436 */
437 static uint8_t bgp_pbr_match_val_get_operator(struct bgp_pbr_match_val list[],
438 int num)
439
440 {
441 int i;
442 uint8_t unary_operator = OPERATOR_UNARY_AND;
443
444 for (i = 0; i < num; i++) {
445 if (i == 0)
446 continue;
447 if (list[i].unary_operator & OPERATOR_UNARY_OR)
448 unary_operator = OPERATOR_UNARY_OR;
449 if ((list[i].unary_operator & OPERATOR_UNARY_AND
450 && unary_operator == OPERATOR_UNARY_OR) ||
451 (list[i].unary_operator & OPERATOR_UNARY_OR
452 && unary_operator == OPERATOR_UNARY_AND))
453 return 0;
454 }
455 return unary_operator;
456 }
457
458
459 /* return true if extraction ok
460 */
461 static bool bgp_pbr_extract(struct bgp_pbr_match_val list[],
462 int num,
463 struct bgp_pbr_range_port *range)
464 {
465 int i = 0;
466 bool exact_match = false;
467
468 if (range)
469 memset(range, 0, sizeof(struct bgp_pbr_range_port));
470
471 if (num > 2)
472 return false;
473 for (i = 0; i < num; i++) {
474 if (i != 0 && (list[i].compare_operator ==
475 OPERATOR_COMPARE_EQUAL_TO))
476 return false;
477 if (i == 0 && (list[i].compare_operator ==
478 OPERATOR_COMPARE_EQUAL_TO)) {
479 if (range)
480 range->min_port = list[i].value;
481 exact_match = true;
482 }
483 if (exact_match && i > 0)
484 return false;
485 if (list[i].compare_operator ==
486 (OPERATOR_COMPARE_GREATER_THAN +
487 OPERATOR_COMPARE_EQUAL_TO)) {
488 if (range)
489 range->min_port = list[i].value;
490 } else if (list[i].compare_operator ==
491 (OPERATOR_COMPARE_LESS_THAN +
492 OPERATOR_COMPARE_EQUAL_TO)) {
493 if (range)
494 range->max_port = list[i].value;
495 } else if (list[i].compare_operator ==
496 OPERATOR_COMPARE_LESS_THAN) {
497 if (range)
498 range->max_port = list[i].value - 1;
499 } else if (list[i].compare_operator ==
500 OPERATOR_COMPARE_GREATER_THAN) {
501 if (range)
502 range->min_port = list[i].value + 1;
503 }
504 }
505 return true;
506 }
507
508 static int bgp_pbr_validate_policy_route(struct bgp_pbr_entry_main *api)
509 {
510 bool enumerate_icmp = false;
511
512 if (api->type == BGP_PBR_UNDEFINED) {
513 if (BGP_DEBUG(pbr, PBR))
514 zlog_debug("BGP: pbr entry undefined. cancel.");
515 return 0;
516 }
517 /* because bgp pbr entry may contain unsupported
518 * combinations, a message will be displayed here if
519 * not supported.
520 * for now, only match/set supported is
521 * - combination src/dst => redirect nexthop [ + rate]
522 * - combination src/dst => redirect VRF [ + rate]
523 * - combination src/dst => drop
524 * - combination srcport + @IP
525 */
526 if (api->match_protocol_num > 1) {
527 if (BGP_DEBUG(pbr, PBR))
528 zlog_debug("BGP: match protocol operations:multiple protocols ( %d). ignoring.",
529 api->match_protocol_num);
530 return 0;
531 }
532 if (api->src_prefix_offset > 0 ||
533 api->dst_prefix_offset > 0) {
534 if (BGP_DEBUG(pbr, PBR))
535 zlog_debug("BGP: match prefix offset:"
536 "implementation does not support it.");
537 return 0;
538 }
539 if (api->match_protocol_num == 1 &&
540 api->protocol[0].value != PROTOCOL_UDP &&
541 api->protocol[0].value != PROTOCOL_ICMP &&
542 api->protocol[0].value != PROTOCOL_ICMPV6 &&
543 api->protocol[0].value != PROTOCOL_TCP) {
544 if (BGP_DEBUG(pbr, PBR))
545 zlog_debug("BGP: match protocol operations:protocol (%d) not supported. ignoring",
546 api->match_protocol_num);
547 return 0;
548 }
549 if (!bgp_pbr_extract(api->src_port, api->match_src_port_num, NULL)) {
550 if (BGP_DEBUG(pbr, PBR))
551 zlog_debug("BGP: match src port operations:too complex. ignoring.");
552 return 0;
553 }
554 if (!bgp_pbr_extract(api->dst_port, api->match_dst_port_num, NULL)) {
555 if (BGP_DEBUG(pbr, PBR))
556 zlog_debug("BGP: match dst port operations:too complex. ignoring.");
557 return 0;
558 }
559 if (!bgp_pbr_extract_enumerate(api->tcpflags,
560 api->match_tcpflags_num,
561 OPERATOR_UNARY_AND |
562 OPERATOR_UNARY_OR, NULL,
563 FLOWSPEC_TCP_FLAGS)) {
564 if (BGP_DEBUG(pbr, PBR))
565 zlog_debug("BGP: match tcp flags:too complex. ignoring.");
566 return 0;
567 }
568 if (!bgp_pbr_extract(api->icmp_type, api->match_icmp_type_num, NULL)) {
569 if (!bgp_pbr_extract_enumerate(api->icmp_type,
570 api->match_icmp_type_num,
571 OPERATOR_UNARY_OR, NULL,
572 FLOWSPEC_ICMP_TYPE)) {
573 if (BGP_DEBUG(pbr, PBR))
574 zlog_debug("BGP: match icmp type operations:too complex. ignoring.");
575 return 0;
576 }
577 enumerate_icmp = true;
578 }
579 if (!bgp_pbr_extract(api->icmp_code, api->match_icmp_code_num, NULL)) {
580 if (!bgp_pbr_extract_enumerate(api->icmp_code,
581 api->match_icmp_code_num,
582 OPERATOR_UNARY_OR, NULL,
583 FLOWSPEC_ICMP_CODE)) {
584 if (BGP_DEBUG(pbr, PBR))
585 zlog_debug("BGP: match icmp code operations:too complex. ignoring.");
586 return 0;
587 } else if (api->match_icmp_type_num > 1 &&
588 !enumerate_icmp) {
589 if (BGP_DEBUG(pbr, PBR))
590 zlog_debug("BGP: match icmp code is enumerate, and icmp type is not. too complex. ignoring.");
591 return 0;
592 }
593 }
594 if (!bgp_pbr_extract(api->port, api->match_port_num, NULL)) {
595 if (BGP_DEBUG(pbr, PBR))
596 zlog_debug("BGP: match port operations:too complex. ignoring.");
597 return 0;
598 }
599 if (api->match_packet_length_num) {
600 bool ret;
601
602 ret = bgp_pbr_extract(api->packet_length,
603 api->match_packet_length_num, NULL);
604 if (!ret)
605 ret = bgp_pbr_extract_enumerate(api->packet_length,
606 api->match_packet_length_num,
607 OPERATOR_UNARY_OR
608 | OPERATOR_UNARY_AND,
609 NULL, FLOWSPEC_PKT_LEN);
610 if (!ret) {
611 if (BGP_DEBUG(pbr, PBR))
612 zlog_debug("BGP: match packet length operations:too complex. ignoring.");
613 return 0;
614 }
615 }
616 if (api->match_dscp_num) {
617 if (!bgp_pbr_extract_enumerate(api->dscp, api->match_dscp_num,
618 OPERATOR_UNARY_OR | OPERATOR_UNARY_AND,
619 NULL, FLOWSPEC_DSCP)) {
620 if (BGP_DEBUG(pbr, PBR))
621 zlog_debug("BGP: match DSCP operations:too complex. ignoring.");
622 return 0;
623 }
624 }
625 if (api->match_flowlabel_num) {
626 if (api->afi == AFI_IP) {
627 if (BGP_DEBUG(pbr, PBR))
628 zlog_debug("BGP: match Flow Label operations:"
629 "Not for IPv4.");
630 return 0;
631 }
632 if (!bgp_pbr_extract_enumerate(api->flow_label,
633 api->match_flowlabel_num,
634 OPERATOR_UNARY_OR | OPERATOR_UNARY_AND,
635 NULL, FLOWSPEC_FLOW_LABEL)) {
636 if (BGP_DEBUG(pbr, PBR))
637 zlog_debug("BGP: match FlowLabel operations:"
638 "too complex. ignoring.");
639 return 0;
640 }
641 if (BGP_DEBUG(pbr, PBR))
642 zlog_debug("BGP: match FlowLabel operations "
643 "not supported. ignoring.");
644 return 0;
645 }
646 if (api->match_fragment_num) {
647 char fail_str[64];
648 bool success;
649
650 success = bgp_pbr_extract_enumerate(api->fragment,
651 api->match_fragment_num,
652 OPERATOR_UNARY_OR
653 | OPERATOR_UNARY_AND,
654 NULL, FLOWSPEC_FRAGMENT);
655 if (success) {
656 int i;
657
658 for (i = 0; i < api->match_fragment_num; i++) {
659 if (api->fragment[i].value != 1 &&
660 api->fragment[i].value != 2 &&
661 api->fragment[i].value != 4 &&
662 api->fragment[i].value != 8) {
663 success = false;
664 snprintf(
665 fail_str, sizeof(fail_str),
666 "Value not valid (%d) for this implementation",
667 api->fragment[i].value);
668 }
669 if (api->afi == AFI_IP6 &&
670 api->fragment[i].value == 1) {
671 success = false;
672 snprintf(fail_str, sizeof(fail_str),
673 "IPv6 dont fragment match invalid (%d)",
674 api->fragment[i].value);
675 }
676 }
677 if (api->afi == AFI_IP6) {
678 success = false;
679 snprintf(fail_str, sizeof(fail_str),
680 "%s", IPV6_FRAGMENT_INVALID);
681 }
682 } else
683 snprintf(fail_str, sizeof(fail_str),
684 "too complex. ignoring");
685 if (!success) {
686 if (BGP_DEBUG(pbr, PBR))
687 zlog_debug("BGP: match fragment operation (%d) %s",
688 api->match_fragment_num,
689 fail_str);
690 return 0;
691 }
692 }
693
694 /* no combinations with both src_port and dst_port
695 * or port with src_port and dst_port
696 */
697 if (api->match_src_port_num + api->match_dst_port_num +
698 api->match_port_num > 3) {
699 if (BGP_DEBUG(pbr, PBR))
700 zlog_debug("BGP: match multiple port operations: too complex. ignoring.");
701 return 0;
702 }
703 if ((api->match_src_port_num || api->match_dst_port_num
704 || api->match_port_num) && (api->match_icmp_type_num
705 || api->match_icmp_code_num)) {
706 if (BGP_DEBUG(pbr, PBR))
707 zlog_debug("BGP: match multiple port/imcp operations: too complex. ignoring.");
708 return 0;
709 }
710 /* iprule only supports redirect IP */
711 if (api->type == BGP_PBR_IPRULE) {
712 int i;
713
714 for (i = 0; i < api->action_num; i++) {
715 if (api->actions[i].action == ACTION_TRAFFICRATE &&
716 api->actions[i].u.r.rate == 0) {
717 if (BGP_DEBUG(pbr, PBR)) {
718 bgp_pbr_print_policy_route(api);
719 zlog_debug("BGP: iprule match actions drop not supported");
720 }
721 return 0;
722 }
723 if (api->actions[i].action == ACTION_MARKING) {
724 if (BGP_DEBUG(pbr, PBR)) {
725 bgp_pbr_print_policy_route(api);
726 zlog_warn("PBR: iprule set DSCP/Flow Label %u not supported",
727 api->actions[i].u.marking_dscp);
728 }
729 }
730 if (api->actions[i].action == ACTION_REDIRECT) {
731 if (BGP_DEBUG(pbr, PBR)) {
732 bgp_pbr_print_policy_route(api);
733 zlog_warn("PBR: iprule redirect VRF %u not supported",
734 api->actions[i].u.redirect_vrf);
735 }
736 }
737 }
738
739 } else if (!(api->match_bitmask & PREFIX_SRC_PRESENT) &&
740 !(api->match_bitmask & PREFIX_DST_PRESENT)) {
741 if (BGP_DEBUG(pbr, PBR)) {
742 bgp_pbr_print_policy_route(api);
743 zlog_debug("BGP: match actions without src or dst address can not operate. ignoring.");
744 }
745 return 0;
746 }
747 return 1;
748 }
749
750 /* return -1 if build or validation failed */
751
752 int bgp_pbr_build_and_validate_entry(const struct prefix *p,
753 struct bgp_path_info *path,
754 struct bgp_pbr_entry_main *api)
755 {
756 int ret;
757 uint32_t i, action_count = 0;
758 struct ecommunity *ecom;
759 struct ecommunity_val *ecom_eval;
760 struct bgp_pbr_entry_action *api_action;
761 struct prefix *src = NULL, *dst = NULL;
762 int valid_prefix = 0;
763 struct bgp_pbr_entry_action *api_action_redirect_ip = NULL;
764 bool discard_action_found = false;
765 afi_t afi = family2afi(p->u.prefix_flowspec.family);
766
767 /* extract match from flowspec entries */
768 ret = bgp_flowspec_match_rules_fill((uint8_t *)p->u.prefix_flowspec.ptr,
769 p->u.prefix_flowspec.prefixlen, api, afi);
770 if (ret < 0)
771 return -1;
772 /* extract actiosn from flowspec ecom list */
773 if (path && bgp_attr_get_ecommunity(path->attr)) {
774 ecom = bgp_attr_get_ecommunity(path->attr);
775 for (i = 0; i < ecom->size; i++) {
776 ecom_eval = (struct ecommunity_val *)
777 (ecom->val + (i * ECOMMUNITY_SIZE));
778 action_count++;
779 if (action_count > ACTIONS_MAX_NUM) {
780 if (BGP_DEBUG(pbr, PBR_ERROR))
781 flog_err(
782 EC_BGP_FLOWSPEC_PACKET,
783 "%s: %s (max %u)",
784 __func__,
785 FSPEC_ACTION_EXCEED_LIMIT,
786 action_count);
787 break;
788 }
789 api_action = &api->actions[action_count - 1];
790
791 if ((ecom_eval->val[1] ==
792 (char)ECOMMUNITY_REDIRECT_VRF) &&
793 (ecom_eval->val[0] ==
794 (char)ECOMMUNITY_ENCODE_TRANS_EXP ||
795 ecom_eval->val[0] ==
796 (char)ECOMMUNITY_EXTENDED_COMMUNITY_PART_2 ||
797 ecom_eval->val[0] ==
798 (char)ECOMMUNITY_EXTENDED_COMMUNITY_PART_3)) {
799 struct ecommunity *eckey = ecommunity_new();
800 struct ecommunity_val ecom_copy;
801
802 memcpy(&ecom_copy, ecom_eval,
803 sizeof(struct ecommunity_val));
804 ecom_copy.val[0] &=
805 ~ECOMMUNITY_ENCODE_TRANS_EXP;
806 ecom_copy.val[1] = ECOMMUNITY_ROUTE_TARGET;
807 ecommunity_add_val(eckey, &ecom_copy,
808 false, false);
809
810 api_action->action = ACTION_REDIRECT;
811 api_action->u.redirect_vrf =
812 get_first_vrf_for_redirect_with_rt(
813 eckey);
814 ecommunity_free(&eckey);
815 } else if ((ecom_eval->val[0] ==
816 (char)ECOMMUNITY_ENCODE_REDIRECT_IP_NH) &&
817 (ecom_eval->val[1] ==
818 (char)ECOMMUNITY_REDIRECT_IP_NH)) {
819 /* in case the 2 ecom present,
820 * do not overwrite
821 * draft-ietf-idr-flowspec-redirect
822 */
823 if (api_action_redirect_ip &&
824 p->u.prefix_flowspec.family == AF_INET) {
825 if (api_action_redirect_ip->u
826 .zr.redirect_ip_v4.s_addr
827 != INADDR_ANY)
828 continue;
829 if (path->attr->nexthop.s_addr
830 == INADDR_ANY)
831 continue;
832 api_action_redirect_ip->u.zr
833 .redirect_ip_v4.s_addr =
834 path->attr->nexthop.s_addr;
835 api_action_redirect_ip->u.zr.duplicate
836 = ecom_eval->val[7];
837 continue;
838 } else if (api_action_redirect_ip &&
839 p->u.prefix_flowspec.family == AF_INET6) {
840 if (memcmp(&api_action_redirect_ip->u
841 .zr.redirect_ip_v6,
842 &in6addr_any,
843 sizeof(struct in6_addr)))
844 continue;
845 if (path->attr->mp_nexthop_len == 0 ||
846 path->attr->mp_nexthop_len ==
847 BGP_ATTR_NHLEN_IPV4 ||
848 path->attr->mp_nexthop_len ==
849 BGP_ATTR_NHLEN_VPNV4)
850 continue;
851 memcpy(&api_action_redirect_ip->u
852 .zr.redirect_ip_v6,
853 &path->attr->mp_nexthop_global,
854 sizeof(struct in6_addr));
855 api_action_redirect_ip->u.zr.duplicate
856 = ecom_eval->val[7];
857 continue;
858 } else if (p->u.prefix_flowspec.family ==
859 AF_INET) {
860 api_action->action = ACTION_REDIRECT_IP;
861 api_action->u.zr.redirect_ip_v4.s_addr =
862 path->attr->nexthop.s_addr;
863 api_action->u.zr.duplicate =
864 ecom_eval->val[7];
865 api_action_redirect_ip = api_action;
866 } else if (p->u.prefix_flowspec.family ==
867 AF_INET6) {
868 api_action->action = ACTION_REDIRECT_IP;
869 memcpy(&api_action->u
870 .zr.redirect_ip_v6,
871 &path->attr->mp_nexthop_global,
872 sizeof(struct in6_addr));
873 api_action->u.zr.duplicate
874 = ecom_eval->val[7];
875 api_action_redirect_ip = api_action;
876 }
877 } else if ((ecom_eval->val[0] ==
878 (char)ECOMMUNITY_ENCODE_IP) &&
879 (ecom_eval->val[1] ==
880 (char)ECOMMUNITY_FLOWSPEC_REDIRECT_IPV4)) {
881 /* in case the 2 ecom present,
882 * overwrite simpson draft
883 * update redirect ip fields
884 */
885 if (api_action_redirect_ip) {
886 memcpy(&(api_action_redirect_ip->u
887 .zr.redirect_ip_v4.s_addr),
888 (ecom_eval->val+2), 4);
889 api_action_redirect_ip->u
890 .zr.duplicate =
891 ecom_eval->val[7];
892 continue;
893 } else {
894 api_action->action = ACTION_REDIRECT_IP;
895 memcpy(&(api_action->u
896 .zr.redirect_ip_v4.s_addr),
897 (ecom_eval->val+2), 4);
898 api_action->u.zr.duplicate =
899 ecom_eval->val[7];
900 api_action_redirect_ip = api_action;
901 }
902 } else {
903 if (ecom_eval->val[0] !=
904 (char)ECOMMUNITY_ENCODE_TRANS_EXP)
905 continue;
906 ret = ecommunity_fill_pbr_action(ecom_eval,
907 api_action,
908 afi);
909 if (ret != 0)
910 continue;
911 if ((api_action->action == ACTION_TRAFFICRATE)
912 && api->actions[i].u.r.rate == 0)
913 discard_action_found = true;
914 }
915 api->action_num++;
916 }
917 }
918 if (path && path->attr && bgp_attr_get_ipv6_ecommunity(path->attr)) {
919 struct ecommunity_val_ipv6 *ipv6_ecom_eval;
920
921 ecom = bgp_attr_get_ipv6_ecommunity(path->attr);
922 for (i = 0; i < ecom->size; i++) {
923 ipv6_ecom_eval = (struct ecommunity_val_ipv6 *)
924 (ecom->val + (i * ecom->unit_size));
925 action_count++;
926 if (action_count > ACTIONS_MAX_NUM) {
927 if (BGP_DEBUG(pbr, PBR_ERROR))
928 flog_err(
929 EC_BGP_FLOWSPEC_PACKET,
930 "%s: flowspec actions exceeds limit (max %u)",
931 __func__, action_count);
932 break;
933 }
934 api_action = &api->actions[action_count - 1];
935 if ((ipv6_ecom_eval->val[1] ==
936 (char)ECOMMUNITY_FLOWSPEC_REDIRECT_IPV6) &&
937 (ipv6_ecom_eval->val[0] ==
938 (char)ECOMMUNITY_ENCODE_TRANS_EXP)) {
939 struct ecommunity *eckey = ecommunity_new();
940 struct ecommunity_val_ipv6 ecom_copy;
941
942 eckey->unit_size = IPV6_ECOMMUNITY_SIZE;
943 memcpy(&ecom_copy, ipv6_ecom_eval,
944 sizeof(struct ecommunity_val_ipv6));
945 ecom_copy.val[1] = ECOMMUNITY_ROUTE_TARGET;
946 ecommunity_add_val_ipv6(eckey, &ecom_copy,
947 false, false);
948 api_action->action = ACTION_REDIRECT;
949 api_action->u.redirect_vrf =
950 get_first_vrf_for_redirect_with_rt(
951 eckey);
952 ecommunity_free(&eckey);
953 api->action_num++;
954 }
955 }
956 }
957 /* if ECOMMUNITY_TRAFFIC_RATE = 0 as action
958 * then reduce the API action list to that action
959 */
960 if (api->action_num > 1 && discard_action_found) {
961 api->action_num = 1;
962 memset(&api->actions[0], 0,
963 sizeof(struct bgp_pbr_entry_action));
964 api->actions[0].action = ACTION_TRAFFICRATE;
965 }
966
967 /* validate if incoming matc/action is compatible
968 * with our policy routing engine
969 */
970 if (!bgp_pbr_validate_policy_route(api))
971 return -1;
972
973 /* check inconsistency in the match rule */
974 if (api->match_bitmask & PREFIX_SRC_PRESENT) {
975 src = &api->src_prefix;
976 afi = family2afi(src->family);
977 valid_prefix = 1;
978 }
979 if (api->match_bitmask & PREFIX_DST_PRESENT) {
980 dst = &api->dst_prefix;
981 if (valid_prefix && afi != family2afi(dst->family)) {
982 if (BGP_DEBUG(pbr, PBR)) {
983 bgp_pbr_print_policy_route(api);
984 zlog_debug("%s: inconsistency: no match for afi src and dst (%u/%u)",
985 __func__, afi, family2afi(dst->family));
986 }
987 return -1;
988 }
989 }
990 return 0;
991 }
992
993 static void bgp_pbr_match_entry_free(void *arg)
994 {
995 struct bgp_pbr_match_entry *bpme;
996
997 bpme = (struct bgp_pbr_match_entry *)arg;
998
999 if (bpme->installed) {
1000 bgp_send_pbr_ipset_entry_match(bpme, false);
1001 bpme->installed = false;
1002 bpme->backpointer = NULL;
1003 }
1004 XFREE(MTYPE_PBR_MATCH_ENTRY, bpme);
1005 }
1006
1007 static void bgp_pbr_match_free(void *arg)
1008 {
1009 struct bgp_pbr_match *bpm;
1010
1011 bpm = (struct bgp_pbr_match *)arg;
1012
1013 hash_clean(bpm->entry_hash, bgp_pbr_match_entry_free);
1014
1015 if (hashcount(bpm->entry_hash) == 0) {
1016 /* delete iptable entry first */
1017 /* then delete ipset match */
1018 if (bpm->installed) {
1019 if (bpm->installed_in_iptable) {
1020 bgp_send_pbr_iptable(bpm->action,
1021 bpm, false);
1022 bpm->installed_in_iptable = false;
1023 bpm->action->refcnt--;
1024 }
1025 bgp_send_pbr_ipset_match(bpm, false);
1026 bpm->installed = false;
1027 bpm->action = NULL;
1028 }
1029 }
1030 hash_free(bpm->entry_hash);
1031
1032 XFREE(MTYPE_PBR_MATCH, bpm);
1033 }
1034
1035 static void *bgp_pbr_match_alloc_intern(void *arg)
1036 {
1037 struct bgp_pbr_match *bpm, *new;
1038
1039 bpm = (struct bgp_pbr_match *)arg;
1040
1041 new = XCALLOC(MTYPE_PBR_MATCH, sizeof(*new));
1042 memcpy(new, bpm, sizeof(*bpm));
1043
1044 return new;
1045 }
1046
1047 static void bgp_pbr_rule_free(void *arg)
1048 {
1049 struct bgp_pbr_rule *bpr;
1050
1051 bpr = (struct bgp_pbr_rule *)arg;
1052
1053 /* delete iprule */
1054 if (bpr->installed) {
1055 bgp_send_pbr_rule_action(bpr->action, bpr, false);
1056 bpr->installed = false;
1057 bpr->action->refcnt--;
1058 bpr->action = NULL;
1059 }
1060 XFREE(MTYPE_PBR_RULE, bpr);
1061 }
1062
1063 static void *bgp_pbr_rule_alloc_intern(void *arg)
1064 {
1065 struct bgp_pbr_rule *bpr, *new;
1066
1067 bpr = (struct bgp_pbr_rule *)arg;
1068
1069 new = XCALLOC(MTYPE_PBR_RULE, sizeof(*new));
1070 memcpy(new, bpr, sizeof(*bpr));
1071
1072 return new;
1073 }
1074
1075 static void bgp_pbr_bpa_remove(struct bgp_pbr_action *bpa)
1076 {
1077 if ((bpa->refcnt == 0) && bpa->installed && bpa->table_id != 0) {
1078 bgp_send_pbr_rule_action(bpa, NULL, false);
1079 bgp_zebra_announce_default(bpa->bgp, &bpa->nh, bpa->afi,
1080 bpa->table_id, false);
1081 bpa->installed = false;
1082 }
1083 }
1084
1085 static void bgp_pbr_bpa_add(struct bgp_pbr_action *bpa)
1086 {
1087 if (!bpa->installed && !bpa->install_in_progress) {
1088 bgp_send_pbr_rule_action(bpa, NULL, true);
1089 bgp_zebra_announce_default(bpa->bgp, &bpa->nh, bpa->afi,
1090 bpa->table_id, true);
1091 }
1092 }
1093
1094 static void bgp_pbr_action_free(void *arg)
1095 {
1096 struct bgp_pbr_action *bpa = arg;
1097
1098 bgp_pbr_bpa_remove(bpa);
1099
1100 XFREE(MTYPE_PBR_ACTION, bpa);
1101 }
1102
1103 static void *bgp_pbr_action_alloc_intern(void *arg)
1104 {
1105 struct bgp_pbr_action *bpa, *new;
1106
1107 bpa = (struct bgp_pbr_action *)arg;
1108
1109 new = XCALLOC(MTYPE_PBR_ACTION, sizeof(*new));
1110
1111 memcpy(new, bpa, sizeof(*bpa));
1112
1113 return new;
1114 }
1115
1116 static void *bgp_pbr_match_entry_alloc_intern(void *arg)
1117 {
1118 struct bgp_pbr_match_entry *bpme, *new;
1119
1120 bpme = (struct bgp_pbr_match_entry *)arg;
1121
1122 new = XCALLOC(MTYPE_PBR_MATCH_ENTRY, sizeof(*new));
1123
1124 memcpy(new, bpme, sizeof(*bpme));
1125
1126 return new;
1127 }
1128
1129 uint32_t bgp_pbr_match_hash_key(const void *arg)
1130 {
1131 const struct bgp_pbr_match *pbm = arg;
1132 uint32_t key;
1133
1134 key = jhash_1word(pbm->vrf_id, 0x4312abde);
1135 key = jhash_1word(pbm->flags, key);
1136 key = jhash_1word(pbm->family, key);
1137 key = jhash(&pbm->pkt_len_min, 2, key);
1138 key = jhash(&pbm->pkt_len_max, 2, key);
1139 key = jhash(&pbm->tcp_flags, 2, key);
1140 key = jhash(&pbm->tcp_mask_flags, 2, key);
1141 key = jhash(&pbm->dscp_value, 1, key);
1142 key = jhash(&pbm->flow_label, 2, key);
1143 key = jhash(&pbm->fragment, 1, key);
1144 key = jhash(&pbm->protocol, 1, key);
1145 return jhash_1word(pbm->type, key);
1146 }
1147
1148 bool bgp_pbr_match_hash_equal(const void *arg1, const void *arg2)
1149 {
1150 const struct bgp_pbr_match *r1, *r2;
1151
1152 r1 = (const struct bgp_pbr_match *)arg1;
1153 r2 = (const struct bgp_pbr_match *)arg2;
1154
1155 if (r1->vrf_id != r2->vrf_id)
1156 return false;
1157
1158 if (r1->family != r2->family)
1159 return false;
1160
1161 if (r1->type != r2->type)
1162 return false;
1163
1164 if (r1->flags != r2->flags)
1165 return false;
1166
1167 if (r1->action != r2->action)
1168 return false;
1169
1170 if (r1->pkt_len_min != r2->pkt_len_min)
1171 return false;
1172
1173 if (r1->pkt_len_max != r2->pkt_len_max)
1174 return false;
1175
1176 if (r1->tcp_flags != r2->tcp_flags)
1177 return false;
1178
1179 if (r1->tcp_mask_flags != r2->tcp_mask_flags)
1180 return false;
1181
1182 if (r1->dscp_value != r2->dscp_value)
1183 return false;
1184
1185 if (r1->flow_label != r2->flow_label)
1186 return false;
1187
1188 if (r1->fragment != r2->fragment)
1189 return false;
1190
1191 if (r1->protocol != r2->protocol)
1192 return false;
1193 return true;
1194 }
1195
1196 uint32_t bgp_pbr_rule_hash_key(const void *arg)
1197 {
1198 const struct bgp_pbr_rule *pbr = arg;
1199 uint32_t key;
1200
1201 key = prefix_hash_key(&pbr->src);
1202 key = jhash_1word(pbr->vrf_id, key);
1203 key = jhash_1word(pbr->flags, key);
1204 return jhash_1word(prefix_hash_key(&pbr->dst), key);
1205 }
1206
1207 bool bgp_pbr_rule_hash_equal(const void *arg1, const void *arg2)
1208 {
1209 const struct bgp_pbr_rule *r1, *r2;
1210
1211 r1 = (const struct bgp_pbr_rule *)arg1;
1212 r2 = (const struct bgp_pbr_rule *)arg2;
1213
1214 if (r1->vrf_id != r2->vrf_id)
1215 return false;
1216
1217 if (r1->flags != r2->flags)
1218 return false;
1219
1220 if (r1->action != r2->action)
1221 return false;
1222
1223 if ((r1->flags & MATCH_IP_SRC_SET) &&
1224 !prefix_same(&r1->src, &r2->src))
1225 return false;
1226
1227 if ((r1->flags & MATCH_IP_DST_SET) &&
1228 !prefix_same(&r1->dst, &r2->dst))
1229 return false;
1230
1231 return true;
1232 }
1233
1234 uint32_t bgp_pbr_match_entry_hash_key(const void *arg)
1235 {
1236 const struct bgp_pbr_match_entry *pbme;
1237 uint32_t key;
1238
1239 pbme = arg;
1240 key = prefix_hash_key(&pbme->src);
1241 key = jhash_1word(prefix_hash_key(&pbme->dst), key);
1242 key = jhash(&pbme->dst_port_min, 2, key);
1243 key = jhash(&pbme->src_port_min, 2, key);
1244 key = jhash(&pbme->dst_port_max, 2, key);
1245 key = jhash(&pbme->src_port_max, 2, key);
1246 key = jhash(&pbme->proto, 1, key);
1247
1248 return key;
1249 }
1250
1251 bool bgp_pbr_match_entry_hash_equal(const void *arg1, const void *arg2)
1252 {
1253 const struct bgp_pbr_match_entry *r1, *r2;
1254
1255 r1 = (const struct bgp_pbr_match_entry *)arg1;
1256 r2 = (const struct bgp_pbr_match_entry *)arg2;
1257
1258 /*
1259 * on updates, comparing backpointer is not necessary
1260 * unique value is self calculated
1261 * rate is ignored for now
1262 */
1263
1264 if (!prefix_same(&r1->src, &r2->src))
1265 return false;
1266
1267 if (!prefix_same(&r1->dst, &r2->dst))
1268 return false;
1269
1270 if (r1->src_port_min != r2->src_port_min)
1271 return false;
1272
1273 if (r1->dst_port_min != r2->dst_port_min)
1274 return false;
1275
1276 if (r1->src_port_max != r2->src_port_max)
1277 return false;
1278
1279 if (r1->dst_port_max != r2->dst_port_max)
1280 return false;
1281
1282 if (r1->proto != r2->proto)
1283 return false;
1284
1285 return true;
1286 }
1287
1288 uint32_t bgp_pbr_action_hash_key(const void *arg)
1289 {
1290 const struct bgp_pbr_action *pbra;
1291 uint32_t key;
1292
1293 pbra = arg;
1294 key = jhash_1word(pbra->table_id, 0x4312abde);
1295 key = jhash_1word(pbra->fwmark, key);
1296 key = jhash_1word(pbra->afi, key);
1297 return key;
1298 }
1299
1300 bool bgp_pbr_action_hash_equal(const void *arg1, const void *arg2)
1301 {
1302 const struct bgp_pbr_action *r1, *r2;
1303
1304 r1 = (const struct bgp_pbr_action *)arg1;
1305 r2 = (const struct bgp_pbr_action *)arg2;
1306
1307 /* unique value is self calculated
1308 * table and fwmark is self calculated
1309 * rate is ignored
1310 */
1311 if (r1->vrf_id != r2->vrf_id)
1312 return false;
1313
1314 if (r1->afi != r2->afi)
1315 return false;
1316
1317 return nexthop_same(&r1->nh, &r2->nh);
1318 }
1319
1320 struct bgp_pbr_rule *bgp_pbr_rule_lookup(vrf_id_t vrf_id,
1321 uint32_t unique)
1322 {
1323 struct bgp *bgp = bgp_lookup_by_vrf_id(vrf_id);
1324 struct bgp_pbr_rule_unique bpru;
1325
1326 if (!bgp || unique == 0)
1327 return NULL;
1328 bpru.unique = unique;
1329 bpru.bpr_found = NULL;
1330 hash_walk(bgp->pbr_rule_hash, bgp_pbr_rule_walkcb, &bpru);
1331 return bpru.bpr_found;
1332 }
1333
1334 struct bgp_pbr_action *bgp_pbr_action_rule_lookup(vrf_id_t vrf_id,
1335 uint32_t unique)
1336 {
1337 struct bgp *bgp = bgp_lookup_by_vrf_id(vrf_id);
1338 struct bgp_pbr_action_unique bpau;
1339
1340 if (!bgp || unique == 0)
1341 return NULL;
1342 bpau.unique = unique;
1343 bpau.bpa_found = NULL;
1344 hash_walk(bgp->pbr_action_hash, bgp_pbr_action_walkcb, &bpau);
1345 return bpau.bpa_found;
1346 }
1347
1348 struct bgp_pbr_match *bgp_pbr_match_ipset_lookup(vrf_id_t vrf_id,
1349 uint32_t unique)
1350 {
1351 struct bgp *bgp = bgp_lookup_by_vrf_id(vrf_id);
1352 struct bgp_pbr_match_unique bpmu;
1353
1354 if (!bgp || unique == 0)
1355 return NULL;
1356 bpmu.unique = unique;
1357 bpmu.bpm_found = NULL;
1358 hash_walk(bgp->pbr_match_hash, bgp_pbr_match_walkcb, &bpmu);
1359 return bpmu.bpm_found;
1360 }
1361
1362 struct bgp_pbr_match_entry *bgp_pbr_match_ipset_entry_lookup(vrf_id_t vrf_id,
1363 char *ipset_name,
1364 uint32_t unique)
1365 {
1366 struct bgp *bgp = bgp_lookup_by_vrf_id(vrf_id);
1367 struct bgp_pbr_match_entry_unique bpmeu;
1368 struct bgp_pbr_match_ipsetname bpmi;
1369
1370 if (!bgp || unique == 0)
1371 return NULL;
1372 bpmi.ipsetname = XCALLOC(MTYPE_TMP, ZEBRA_IPSET_NAME_SIZE);
1373 snprintf(bpmi.ipsetname, ZEBRA_IPSET_NAME_SIZE, "%s", ipset_name);
1374 bpmi.bpm_found = NULL;
1375 hash_walk(bgp->pbr_match_hash, bgp_pbr_match_pername_walkcb, &bpmi);
1376 XFREE(MTYPE_TMP, bpmi.ipsetname);
1377 if (!bpmi.bpm_found)
1378 return NULL;
1379 bpmeu.bpme_found = NULL;
1380 bpmeu.unique = unique;
1381 hash_walk(bpmi.bpm_found->entry_hash,
1382 bgp_pbr_match_entry_walkcb, &bpmeu);
1383 return bpmeu.bpme_found;
1384 }
1385
1386 struct bgp_pbr_match *bgp_pbr_match_iptable_lookup(vrf_id_t vrf_id,
1387 uint32_t unique)
1388 {
1389 struct bgp *bgp = bgp_lookup_by_vrf_id(vrf_id);
1390 struct bgp_pbr_match_iptable_unique bpmiu;
1391
1392 if (!bgp || unique == 0)
1393 return NULL;
1394 bpmiu.unique = unique;
1395 bpmiu.bpm_found = NULL;
1396 hash_walk(bgp->pbr_match_hash, bgp_pbr_match_iptable_walkcb, &bpmiu);
1397 return bpmiu.bpm_found;
1398 }
1399
1400 void bgp_pbr_cleanup(struct bgp *bgp)
1401 {
1402 if (bgp->pbr_match_hash) {
1403 hash_clean(bgp->pbr_match_hash, bgp_pbr_match_free);
1404 hash_free(bgp->pbr_match_hash);
1405 bgp->pbr_match_hash = NULL;
1406 }
1407 if (bgp->pbr_rule_hash) {
1408 hash_clean(bgp->pbr_rule_hash, bgp_pbr_rule_free);
1409 hash_free(bgp->pbr_rule_hash);
1410 bgp->pbr_rule_hash = NULL;
1411 }
1412 if (bgp->pbr_action_hash) {
1413 hash_clean(bgp->pbr_action_hash, bgp_pbr_action_free);
1414 hash_free(bgp->pbr_action_hash);
1415 bgp->pbr_action_hash = NULL;
1416 }
1417 if (bgp->bgp_pbr_cfg == NULL)
1418 return;
1419 bgp_pbr_reset(bgp, AFI_IP);
1420 bgp_pbr_reset(bgp, AFI_IP6);
1421 XFREE(MTYPE_PBR, bgp->bgp_pbr_cfg);
1422 }
1423
1424 void bgp_pbr_init(struct bgp *bgp)
1425 {
1426 bgp->pbr_match_hash =
1427 hash_create_size(8, bgp_pbr_match_hash_key,
1428 bgp_pbr_match_hash_equal,
1429 "Match Hash");
1430 bgp->pbr_action_hash =
1431 hash_create_size(8, bgp_pbr_action_hash_key,
1432 bgp_pbr_action_hash_equal,
1433 "Match Hash Entry");
1434
1435 bgp->pbr_rule_hash =
1436 hash_create_size(8, bgp_pbr_rule_hash_key,
1437 bgp_pbr_rule_hash_equal,
1438 "Match Rule");
1439
1440 bgp->bgp_pbr_cfg = XCALLOC(MTYPE_PBR, sizeof(struct bgp_pbr_config));
1441 bgp->bgp_pbr_cfg->pbr_interface_any_ipv4 = true;
1442 }
1443
1444 void bgp_pbr_print_policy_route(struct bgp_pbr_entry_main *api)
1445 {
1446 int i = 0;
1447 char return_string[512];
1448 char *ptr = return_string;
1449 int nb_items = 0;
1450 int delta, len = sizeof(return_string);
1451
1452 delta = snprintf(ptr, sizeof(return_string), "MATCH : ");
1453 len -= delta;
1454 ptr += delta;
1455 if (api->match_bitmask & PREFIX_SRC_PRESENT) {
1456 struct prefix *p = &(api->src_prefix);
1457
1458 if (api->src_prefix_offset)
1459 delta = snprintfrr(ptr, len, "@src %pFX/off%u", p,
1460 api->src_prefix_offset);
1461 else
1462 delta = snprintfrr(ptr, len, "@src %pFX", p);
1463 len -= delta;
1464 ptr += delta;
1465 INCREMENT_DISPLAY(ptr, nb_items, len);
1466 }
1467 if (api->match_bitmask & PREFIX_DST_PRESENT) {
1468 struct prefix *p = &(api->dst_prefix);
1469
1470 INCREMENT_DISPLAY(ptr, nb_items, len);
1471 if (api->dst_prefix_offset)
1472 delta = snprintfrr(ptr, len, "@dst %pFX/off%u", p,
1473 api->dst_prefix_offset);
1474 else
1475 delta = snprintfrr(ptr, len, "@dst %pFX", p);
1476 len -= delta;
1477 ptr += delta;
1478 }
1479
1480 if (api->match_protocol_num)
1481 INCREMENT_DISPLAY(ptr, nb_items, len);
1482 for (i = 0; i < api->match_protocol_num; i++) {
1483 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->protocol[i],
1484 i > 0 ? NULL : "@proto ");
1485 len -= delta;
1486 ptr += delta;
1487 }
1488
1489 if (api->match_src_port_num)
1490 INCREMENT_DISPLAY(ptr, nb_items, len);
1491 for (i = 0; i < api->match_src_port_num; i++) {
1492 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->src_port[i],
1493 i > 0 ? NULL : "@srcport ");
1494 len -= delta;
1495 ptr += delta;
1496 }
1497
1498 if (api->match_dst_port_num)
1499 INCREMENT_DISPLAY(ptr, nb_items, len);
1500 for (i = 0; i < api->match_dst_port_num; i++) {
1501 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->dst_port[i],
1502 i > 0 ? NULL : "@dstport ");
1503 len -= delta;
1504 ptr += delta;
1505 }
1506
1507 if (api->match_port_num)
1508 INCREMENT_DISPLAY(ptr, nb_items, len);
1509 for (i = 0; i < api->match_port_num; i++) {
1510 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->port[i],
1511 i > 0 ? NULL : "@port ");
1512 len -= delta;
1513 ptr += delta;
1514 }
1515
1516 if (api->match_icmp_type_num)
1517 INCREMENT_DISPLAY(ptr, nb_items, len);
1518 for (i = 0; i < api->match_icmp_type_num; i++) {
1519 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->icmp_type[i],
1520 i > 0 ? NULL : "@icmptype ");
1521 len -= delta;
1522 ptr += delta;
1523 }
1524
1525 if (api->match_icmp_code_num)
1526 INCREMENT_DISPLAY(ptr, nb_items, len);
1527 for (i = 0; i < api->match_icmp_code_num; i++) {
1528 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->icmp_code[i],
1529 i > 0 ? NULL : "@icmpcode ");
1530 len -= delta;
1531 ptr += delta;
1532 }
1533
1534 if (api->match_packet_length_num)
1535 INCREMENT_DISPLAY(ptr, nb_items, len);
1536 for (i = 0; i < api->match_packet_length_num; i++) {
1537 delta = snprintf_bgp_pbr_match_val(ptr, len,
1538 &api->packet_length[i],
1539 i > 0 ? NULL : "@plen ");
1540 len -= delta;
1541 ptr += delta;
1542 }
1543
1544 if (api->match_dscp_num)
1545 INCREMENT_DISPLAY(ptr, nb_items, len);
1546 for (i = 0; i < api->match_dscp_num; i++) {
1547 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->dscp[i],
1548 i > 0 ? NULL : "@dscp ");
1549 len -= delta;
1550 ptr += delta;
1551 }
1552
1553 if (api->match_flowlabel_num)
1554 INCREMENT_DISPLAY(ptr, nb_items, len);
1555 for (i = 0; i < api->match_flowlabel_num; i++) {
1556 delta = snprintf_bgp_pbr_match_val(ptr, len,
1557 &api->flow_label[i],
1558 i > 0 ? NULL : "@flowlabel ");
1559 len -= delta;
1560 ptr += delta;
1561 }
1562
1563 if (api->match_tcpflags_num)
1564 INCREMENT_DISPLAY(ptr, nb_items, len);
1565 for (i = 0; i < api->match_tcpflags_num; i++) {
1566 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->tcpflags[i],
1567 i > 0 ? NULL : "@tcpflags ");
1568 len -= delta;
1569 ptr += delta;
1570 }
1571
1572 if (api->match_fragment_num)
1573 INCREMENT_DISPLAY(ptr, nb_items, len);
1574 for (i = 0; i < api->match_fragment_num; i++) {
1575 delta = snprintf_bgp_pbr_match_val(ptr, len, &api->fragment[i],
1576 i > 0 ? NULL : "@fragment ");
1577 len -= delta;
1578 ptr += delta;
1579 }
1580
1581 len = sizeof(return_string);
1582 if (!nb_items) {
1583 ptr = return_string;
1584 } else {
1585 len -= (ptr - return_string);
1586 delta = snprintf(ptr, len, "; ");
1587 len -= delta;
1588 ptr += delta;
1589 }
1590 if (api->action_num) {
1591 delta = snprintf(ptr, len, "SET : ");
1592 len -= delta;
1593 ptr += delta;
1594 }
1595 nb_items = 0;
1596 for (i = 0; i < api->action_num; i++) {
1597 switch (api->actions[i].action) {
1598 case ACTION_TRAFFICRATE:
1599 INCREMENT_DISPLAY(ptr, nb_items, len);
1600 delta = snprintf(ptr, len, "@set rate %f",
1601 api->actions[i].u.r.rate);
1602 len -= delta;
1603 ptr += delta;
1604 break;
1605 case ACTION_TRAFFIC_ACTION:
1606 INCREMENT_DISPLAY(ptr, nb_items, len);
1607 delta = snprintf(ptr, len, "@action ");
1608 len -= delta;
1609 ptr += delta;
1610 if (api->actions[i].u.za.filter
1611 & TRAFFIC_ACTION_TERMINATE) {
1612 delta = snprintf(ptr, len,
1613 " terminate (apply filter(s))");
1614 len -= delta;
1615 ptr += delta;
1616 }
1617 if (api->actions[i].u.za.filter
1618 & TRAFFIC_ACTION_DISTRIBUTE) {
1619 delta = snprintf(ptr, len, " distribute");
1620 len -= delta;
1621 ptr += delta;
1622 }
1623 if (api->actions[i].u.za.filter
1624 & TRAFFIC_ACTION_SAMPLE) {
1625 delta = snprintf(ptr, len, " sample");
1626 len -= delta;
1627 ptr += delta;
1628 }
1629 break;
1630 case ACTION_REDIRECT_IP: {
1631 char local_buff[INET6_ADDRSTRLEN];
1632 void *ptr_ip;
1633
1634 INCREMENT_DISPLAY(ptr, nb_items, len);
1635 if (api->afi == AF_INET)
1636 ptr_ip = &api->actions[i].u.zr.redirect_ip_v4;
1637 else
1638 ptr_ip = &api->actions[i].u.zr.redirect_ip_v6;
1639 if (inet_ntop(afi2family(api->afi), ptr_ip, local_buff,
1640 sizeof(local_buff)) != NULL) {
1641 delta = snprintf(ptr, len,
1642 "@redirect ip nh %s", local_buff);
1643 len -= delta;
1644 ptr += delta;
1645 }
1646 break;
1647 }
1648 case ACTION_REDIRECT: {
1649 struct vrf *vrf;
1650
1651 vrf = vrf_lookup_by_id(api->actions[i].u.redirect_vrf);
1652 INCREMENT_DISPLAY(ptr, nb_items, len);
1653 delta = snprintf(ptr, len, "@redirect vrf %s(%u)",
1654 VRF_LOGNAME(vrf),
1655 api->actions[i].u.redirect_vrf);
1656 len -= delta;
1657 ptr += delta;
1658 break;
1659 }
1660 case ACTION_MARKING:
1661 INCREMENT_DISPLAY(ptr, nb_items, len);
1662 delta = snprintf(ptr, len, "@set dscp/flowlabel %u",
1663 api->actions[i].u.marking_dscp);
1664 len -= delta;
1665 ptr += delta;
1666 break;
1667 default:
1668 break;
1669 }
1670 }
1671 zlog_info("%s", return_string);
1672 }
1673
1674 static void bgp_pbr_flush_iprule(struct bgp *bgp, struct bgp_pbr_action *bpa,
1675 struct bgp_pbr_rule *bpr)
1676 {
1677 /* if bpr is null, do nothing
1678 */
1679 if (bpr == NULL)
1680 return;
1681 if (bpr->installed) {
1682 bgp_send_pbr_rule_action(bpa, bpr, false);
1683 bpr->installed = false;
1684 bpr->action->refcnt--;
1685 bpr->action = NULL;
1686 if (bpr->path) {
1687 struct bgp_path_info *path;
1688 struct bgp_path_info_extra *extra;
1689
1690 /* unlink path to bpme */
1691 path = (struct bgp_path_info *)bpr->path;
1692 extra = bgp_path_info_extra_get(path);
1693 if (extra->bgp_fs_iprule)
1694 listnode_delete(extra->bgp_fs_iprule, bpr);
1695 bpr->path = NULL;
1696 }
1697 }
1698 hash_release(bgp->pbr_rule_hash, bpr);
1699 bgp_pbr_bpa_remove(bpa);
1700 }
1701
1702 static void bgp_pbr_flush_entry(struct bgp *bgp, struct bgp_pbr_action *bpa,
1703 struct bgp_pbr_match *bpm,
1704 struct bgp_pbr_match_entry *bpme)
1705 {
1706 /* if bpme is null, bpm is also null
1707 */
1708 if (bpme == NULL)
1709 return;
1710 /* ipset del entry */
1711 if (bpme->installed) {
1712 bgp_send_pbr_ipset_entry_match(bpme, false);
1713 bpme->installed = false;
1714 bpme->backpointer = NULL;
1715 if (bpme->path) {
1716 struct bgp_path_info *path;
1717 struct bgp_path_info_extra *extra;
1718
1719 /* unlink path to bpme */
1720 path = (struct bgp_path_info *)bpme->path;
1721 extra = bgp_path_info_extra_get(path);
1722 if (extra->bgp_fs_pbr)
1723 listnode_delete(extra->bgp_fs_pbr, bpme);
1724 bpme->path = NULL;
1725 }
1726 }
1727 hash_release(bpm->entry_hash, bpme);
1728 if (hashcount(bpm->entry_hash) == 0) {
1729 /* delete iptable entry first */
1730 /* then delete ipset match */
1731 if (bpm->installed) {
1732 if (bpm->installed_in_iptable) {
1733 bgp_send_pbr_iptable(bpm->action,
1734 bpm, false);
1735 bpm->installed_in_iptable = false;
1736 bpm->action->refcnt--;
1737 }
1738 bgp_send_pbr_ipset_match(bpm, false);
1739 bpm->installed = false;
1740 bpm->action = NULL;
1741 }
1742 hash_release(bgp->pbr_match_hash, bpm);
1743 /* XXX release pbr_match_action if not used
1744 * note that drop does not need to call send_pbr_action
1745 */
1746 }
1747 bgp_pbr_bpa_remove(bpa);
1748 }
1749
1750 struct bgp_pbr_match_entry_remain {
1751 struct bgp_pbr_match_entry *bpme_to_match;
1752 struct bgp_pbr_match_entry *bpme_found;
1753 };
1754
1755 struct bgp_pbr_rule_remain {
1756 struct bgp_pbr_rule *bpr_to_match;
1757 struct bgp_pbr_rule *bpr_found;
1758 };
1759
1760 static int bgp_pbr_get_same_rule(struct hash_bucket *bucket, void *arg)
1761 {
1762 struct bgp_pbr_rule *r1 = (struct bgp_pbr_rule *)bucket->data;
1763 struct bgp_pbr_rule_remain *ctxt =
1764 (struct bgp_pbr_rule_remain *)arg;
1765 struct bgp_pbr_rule *r2;
1766
1767 r2 = ctxt->bpr_to_match;
1768
1769 if (r1->vrf_id != r2->vrf_id)
1770 return HASHWALK_CONTINUE;
1771
1772 if (r1->flags != r2->flags)
1773 return HASHWALK_CONTINUE;
1774
1775 if ((r1->flags & MATCH_IP_SRC_SET) &&
1776 !prefix_same(&r1->src, &r2->src))
1777 return HASHWALK_CONTINUE;
1778
1779 if ((r1->flags & MATCH_IP_DST_SET) &&
1780 !prefix_same(&r1->dst, &r2->dst))
1781 return HASHWALK_CONTINUE;
1782
1783 /* this function is used for two cases:
1784 * - remove an entry upon withdraw request
1785 * (case r2->action is null)
1786 * - replace an old iprule with different action
1787 * (case r2->action is != null)
1788 * the old one is removed after the new one
1789 * this is to avoid disruption in traffic
1790 */
1791 if (r2->action == NULL ||
1792 r1->action != r2->action) {
1793 ctxt->bpr_found = r1;
1794 return HASHWALK_ABORT;
1795 }
1796 return HASHWALK_CONTINUE;
1797 }
1798
1799 static int bgp_pbr_get_remaining_entry(struct hash_bucket *bucket, void *arg)
1800 {
1801 struct bgp_pbr_match *bpm = (struct bgp_pbr_match *)bucket->data;
1802 struct bgp_pbr_match_entry_remain *bpmer =
1803 (struct bgp_pbr_match_entry_remain *)arg;
1804 struct bgp_pbr_match *bpm_temp;
1805 struct bgp_pbr_match_entry *bpme = bpmer->bpme_to_match;
1806
1807 if (!bpme->backpointer ||
1808 bpm == bpme->backpointer ||
1809 bpme->backpointer->action == bpm->action)
1810 return HASHWALK_CONTINUE;
1811 /* ensure bpm other characteristics are equal */
1812 bpm_temp = bpme->backpointer;
1813 if (bpm_temp->vrf_id != bpm->vrf_id ||
1814 bpm_temp->type != bpm->type ||
1815 bpm_temp->flags != bpm->flags ||
1816 bpm_temp->tcp_flags != bpm->tcp_flags ||
1817 bpm_temp->tcp_mask_flags != bpm->tcp_mask_flags ||
1818 bpm_temp->pkt_len_min != bpm->pkt_len_min ||
1819 bpm_temp->pkt_len_max != bpm->pkt_len_max ||
1820 bpm_temp->dscp_value != bpm->dscp_value ||
1821 bpm_temp->flow_label != bpm->flow_label ||
1822 bpm_temp->family != bpm->family ||
1823 bpm_temp->fragment != bpm->fragment)
1824 return HASHWALK_CONTINUE;
1825
1826 /* look for remaining bpme */
1827 bpmer->bpme_found = hash_lookup(bpm->entry_hash, bpme);
1828 if (!bpmer->bpme_found)
1829 return HASHWALK_CONTINUE;
1830 return HASHWALK_ABORT;
1831 }
1832
1833 static void bgp_pbr_policyroute_remove_from_zebra_unit(
1834 struct bgp *bgp, struct bgp_path_info *path, struct bgp_pbr_filter *bpf)
1835 {
1836 struct bgp_pbr_match temp;
1837 struct bgp_pbr_match_entry temp2;
1838 struct bgp_pbr_rule pbr_rule;
1839 struct bgp_pbr_rule *bpr;
1840 struct bgp_pbr_match *bpm;
1841 struct bgp_pbr_match_entry *bpme;
1842 struct bgp_pbr_match_entry_remain bpmer;
1843 struct bgp_pbr_range_port *src_port;
1844 struct bgp_pbr_range_port *dst_port;
1845 struct bgp_pbr_range_port *pkt_len;
1846 struct bgp_pbr_rule_remain bprr;
1847
1848 if (!bpf)
1849 return;
1850 src_port = bpf->src_port;
1851 dst_port = bpf->dst_port;
1852 pkt_len = bpf->pkt_len;
1853
1854 if (BGP_DEBUG(zebra, ZEBRA))
1855 bgp_pbr_dump_entry(bpf, false);
1856
1857 /* as we don't know information from EC
1858 * look for bpm that have the bpm
1859 * with vrf_id characteristics
1860 */
1861 memset(&temp2, 0, sizeof(temp2));
1862 memset(&temp, 0, sizeof(temp));
1863
1864 if (bpf->type == BGP_PBR_IPRULE) {
1865 memset(&pbr_rule, 0, sizeof(pbr_rule));
1866 pbr_rule.vrf_id = bpf->vrf_id;
1867 if (bpf->src) {
1868 prefix_copy(&pbr_rule.src, bpf->src);
1869 pbr_rule.flags |= MATCH_IP_SRC_SET;
1870 }
1871 if (bpf->dst) {
1872 prefix_copy(&pbr_rule.dst, bpf->dst);
1873 pbr_rule.flags |= MATCH_IP_DST_SET;
1874 }
1875 bpr = &pbr_rule;
1876 /* A previous entry may already exist
1877 * flush previous entry if necessary
1878 */
1879 bprr.bpr_to_match = bpr;
1880 bprr.bpr_found = NULL;
1881 hash_walk(bgp->pbr_rule_hash, bgp_pbr_get_same_rule, &bprr);
1882 if (bprr.bpr_found) {
1883 static struct bgp_pbr_rule *local_bpr;
1884 static struct bgp_pbr_action *local_bpa;
1885
1886 local_bpr = bprr.bpr_found;
1887 local_bpa = local_bpr->action;
1888 bgp_pbr_flush_iprule(bgp, local_bpa,
1889 local_bpr);
1890 }
1891 return;
1892 }
1893
1894 temp.family = bpf->family;
1895 if (bpf->src) {
1896 temp.flags |= MATCH_IP_SRC_SET;
1897 prefix_copy(&temp2.src, bpf->src);
1898 } else
1899 temp2.src.family = bpf->family;
1900 if (bpf->dst) {
1901 temp.flags |= MATCH_IP_DST_SET;
1902 prefix_copy(&temp2.dst, bpf->dst);
1903 } else
1904 temp2.dst.family = bpf->family;
1905 if (src_port && (src_port->min_port || bpf->protocol == IPPROTO_ICMP)) {
1906 if (bpf->protocol == IPPROTO_ICMP)
1907 temp.flags |= MATCH_ICMP_SET;
1908 temp.flags |= MATCH_PORT_SRC_SET;
1909 temp2.src_port_min = src_port->min_port;
1910 if (src_port->max_port) {
1911 temp.flags |= MATCH_PORT_SRC_RANGE_SET;
1912 temp2.src_port_max = src_port->max_port;
1913 }
1914 }
1915 if (dst_port && (dst_port->min_port || bpf->protocol == IPPROTO_ICMP)) {
1916 if (bpf->protocol == IPPROTO_ICMP)
1917 temp.flags |= MATCH_ICMP_SET;
1918 temp.flags |= MATCH_PORT_DST_SET;
1919 temp2.dst_port_min = dst_port->min_port;
1920 if (dst_port->max_port) {
1921 temp.flags |= MATCH_PORT_DST_RANGE_SET;
1922 temp2.dst_port_max = dst_port->max_port;
1923 }
1924 }
1925 temp2.proto = bpf->protocol;
1926
1927 if (pkt_len) {
1928 temp.pkt_len_min = pkt_len->min_port;
1929 if (pkt_len->max_port)
1930 temp.pkt_len_max = pkt_len->max_port;
1931 } else if (bpf->pkt_len_val) {
1932 if (bpf->pkt_len_val->mask)
1933 temp.flags |= MATCH_PKT_LEN_INVERSE_SET;
1934 temp.pkt_len_min = bpf->pkt_len_val->val;
1935 }
1936 if (bpf->tcp_flags) {
1937 temp.tcp_flags = bpf->tcp_flags->val;
1938 temp.tcp_mask_flags = bpf->tcp_flags->mask;
1939 }
1940 if (bpf->dscp) {
1941 if (bpf->dscp->mask)
1942 temp.flags |= MATCH_DSCP_INVERSE_SET;
1943 else
1944 temp.flags |= MATCH_DSCP_SET;
1945 temp.dscp_value = bpf->dscp->val;
1946 }
1947 if (bpf->flow_label) {
1948 if (bpf->flow_label->mask)
1949 temp.flags |= MATCH_FLOW_LABEL_INVERSE_SET;
1950 else
1951 temp.flags |= MATCH_FLOW_LABEL_SET;
1952 temp.flow_label = bpf->flow_label->val;
1953 }
1954
1955 if (bpf->fragment) {
1956 if (bpf->fragment->mask)
1957 temp.flags |= MATCH_FRAGMENT_INVERSE_SET;
1958 temp.fragment = bpf->fragment->val;
1959 }
1960
1961 if (bpf->src == NULL || bpf->dst == NULL) {
1962 if (temp.flags & (MATCH_PORT_DST_SET | MATCH_PORT_SRC_SET))
1963 temp.type = IPSET_NET_PORT;
1964 else
1965 temp.type = IPSET_NET;
1966 } else {
1967 if (temp.flags & (MATCH_PORT_DST_SET | MATCH_PORT_SRC_SET))
1968 temp.type = IPSET_NET_PORT_NET;
1969 else
1970 temp.type = IPSET_NET_NET;
1971 }
1972 if (bpf->vrf_id == VRF_UNKNOWN) /* XXX case BGP destroy */
1973 temp.vrf_id = VRF_DEFAULT;
1974 else
1975 temp.vrf_id = bpf->vrf_id;
1976 bpme = &temp2;
1977 bpm = &temp;
1978 bpme->backpointer = bpm;
1979 /* right now, a previous entry may already exist
1980 * flush previous entry if necessary
1981 */
1982 bpmer.bpme_to_match = bpme;
1983 bpmer.bpme_found = NULL;
1984 hash_walk(bgp->pbr_match_hash, bgp_pbr_get_remaining_entry, &bpmer);
1985 if (bpmer.bpme_found) {
1986 static struct bgp_pbr_match *local_bpm;
1987 static struct bgp_pbr_action *local_bpa;
1988
1989 local_bpm = bpmer.bpme_found->backpointer;
1990 local_bpa = local_bpm->action;
1991 bgp_pbr_flush_entry(bgp, local_bpa,
1992 local_bpm, bpmer.bpme_found);
1993 }
1994 }
1995
1996 static uint8_t bgp_pbr_next_type_entry(uint8_t type_entry)
1997 {
1998 if (type_entry == FLOWSPEC_TCP_FLAGS)
1999 return FLOWSPEC_DSCP;
2000 if (type_entry == FLOWSPEC_DSCP)
2001 return FLOWSPEC_FLOW_LABEL;
2002 if (type_entry == FLOWSPEC_FLOW_LABEL)
2003 return FLOWSPEC_PKT_LEN;
2004 if (type_entry == FLOWSPEC_PKT_LEN)
2005 return FLOWSPEC_FRAGMENT;
2006 if (type_entry == FLOWSPEC_FRAGMENT)
2007 return FLOWSPEC_ICMP_TYPE;
2008 return 0;
2009 }
2010
2011 static void bgp_pbr_icmp_action(struct bgp *bgp, struct bgp_path_info *path,
2012 struct bgp_pbr_filter *bpf,
2013 struct bgp_pbr_or_filter *bpof, bool add,
2014 struct nexthop *nh, float *rate)
2015 {
2016 struct bgp_pbr_range_port srcp, dstp;
2017 struct bgp_pbr_val_mask *icmp_type, *icmp_code;
2018 struct listnode *tnode, *cnode;
2019
2020 if (!bpf)
2021 return;
2022 if (bpf->protocol != IPPROTO_ICMP)
2023 return;
2024
2025 memset(&srcp, 0, sizeof(srcp));
2026 memset(&dstp, 0, sizeof(dstp));
2027 bpf->src_port = &srcp;
2028 bpf->dst_port = &dstp;
2029 /* parse icmp type and lookup appropriate icmp code
2030 * if no icmp code found, create as many entryes as
2031 * there are listed icmp codes for that icmp type
2032 */
2033 if (!bpof->icmp_type) {
2034 srcp.min_port = 0;
2035 srcp.max_port = 255;
2036 for (ALL_LIST_ELEMENTS_RO(bpof->icmp_code, cnode, icmp_code)) {
2037 dstp.min_port = icmp_code->val;
2038 if (add)
2039 bgp_pbr_policyroute_add_to_zebra_unit(
2040 bgp, path, bpf, nh, rate);
2041 else
2042 bgp_pbr_policyroute_remove_from_zebra_unit(
2043 bgp, path, bpf);
2044 }
2045 return;
2046 }
2047 for (ALL_LIST_ELEMENTS_RO(bpof->icmp_type, tnode, icmp_type)) {
2048 srcp.min_port = icmp_type->val;
2049 srcp.max_port = 0;
2050 dstp.max_port = 0;
2051 /* only icmp type. create an entry only with icmp type */
2052 if (!bpof->icmp_code) {
2053 /* icmp type is not one of the above
2054 * forge an entry only based on the icmp type
2055 */
2056 dstp.min_port = 0;
2057 dstp.max_port = 255;
2058 if (add)
2059 bgp_pbr_policyroute_add_to_zebra_unit(
2060 bgp, path, bpf, nh, rate);
2061 else
2062 bgp_pbr_policyroute_remove_from_zebra_unit(
2063 bgp, path, bpf);
2064 continue;
2065 }
2066 for (ALL_LIST_ELEMENTS_RO(bpof->icmp_code, cnode, icmp_code)) {
2067 dstp.min_port = icmp_code->val;
2068 if (add)
2069 bgp_pbr_policyroute_add_to_zebra_unit(
2070 bgp, path, bpf, nh, rate);
2071 else
2072 bgp_pbr_policyroute_remove_from_zebra_unit(
2073 bgp, path, bpf);
2074 }
2075 }
2076
2077 bpf->src_port = NULL;
2078 bpf->dst_port = NULL;
2079 }
2080
2081 static void bgp_pbr_policyroute_remove_from_zebra_recursive(
2082 struct bgp *bgp, struct bgp_path_info *path, struct bgp_pbr_filter *bpf,
2083 struct bgp_pbr_or_filter *bpof, uint8_t type_entry)
2084 {
2085 struct listnode *node, *nnode;
2086 struct bgp_pbr_val_mask *valmask;
2087 uint8_t next_type_entry;
2088 struct list *orig_list;
2089 struct bgp_pbr_val_mask **target_val;
2090
2091 if (type_entry == 0) {
2092 bgp_pbr_policyroute_remove_from_zebra_unit(bgp, path, bpf);
2093 return;
2094 }
2095 next_type_entry = bgp_pbr_next_type_entry(type_entry);
2096 if (type_entry == FLOWSPEC_TCP_FLAGS && bpof->tcpflags) {
2097 orig_list = bpof->tcpflags;
2098 target_val = &bpf->tcp_flags;
2099 } else if (type_entry == FLOWSPEC_DSCP && bpof->dscp) {
2100 orig_list = bpof->dscp;
2101 target_val = &bpf->dscp;
2102 } else if (type_entry == FLOWSPEC_FLOW_LABEL && bpof->flowlabel) {
2103 orig_list = bpof->flowlabel;
2104 target_val = &bpf->flow_label;
2105 } else if (type_entry == FLOWSPEC_PKT_LEN && bpof->pkt_len) {
2106 orig_list = bpof->pkt_len;
2107 target_val = &bpf->pkt_len_val;
2108 } else if (type_entry == FLOWSPEC_FRAGMENT && bpof->fragment) {
2109 orig_list = bpof->fragment;
2110 target_val = &bpf->fragment;
2111 } else if (type_entry == FLOWSPEC_ICMP_TYPE &&
2112 (bpof->icmp_type || bpof->icmp_code)) {
2113 /* enumerate list for icmp - must be last one */
2114 bgp_pbr_icmp_action(bgp, path, bpf, bpof, false, NULL, NULL);
2115 return;
2116 } else {
2117 bgp_pbr_policyroute_remove_from_zebra_recursive(
2118 bgp, path, bpf, bpof, next_type_entry);
2119 return;
2120 }
2121 for (ALL_LIST_ELEMENTS(orig_list, node, nnode, valmask)) {
2122 *target_val = valmask;
2123 bgp_pbr_policyroute_remove_from_zebra_recursive(
2124 bgp, path, bpf, bpof, next_type_entry);
2125 }
2126 }
2127
2128 static void bgp_pbr_policyroute_remove_from_zebra(
2129 struct bgp *bgp, struct bgp_path_info *path, struct bgp_pbr_filter *bpf,
2130 struct bgp_pbr_or_filter *bpof)
2131 {
2132 if (!bpof) {
2133 bgp_pbr_policyroute_remove_from_zebra_unit(bgp, path, bpf);
2134 return;
2135 }
2136 if (bpof->tcpflags)
2137 bgp_pbr_policyroute_remove_from_zebra_recursive(
2138 bgp, path, bpf, bpof, FLOWSPEC_TCP_FLAGS);
2139 else if (bpof->dscp)
2140 bgp_pbr_policyroute_remove_from_zebra_recursive(
2141 bgp, path, bpf, bpof, FLOWSPEC_DSCP);
2142 else if (bpof->flowlabel)
2143 bgp_pbr_policyroute_remove_from_zebra_recursive(
2144 bgp, path, bpf, bpof, FLOWSPEC_FLOW_LABEL);
2145 else if (bpof->pkt_len)
2146 bgp_pbr_policyroute_remove_from_zebra_recursive(
2147 bgp, path, bpf, bpof, FLOWSPEC_PKT_LEN);
2148 else if (bpof->fragment)
2149 bgp_pbr_policyroute_remove_from_zebra_recursive(
2150 bgp, path, bpf, bpof, FLOWSPEC_FRAGMENT);
2151 else if (bpof->icmp_type || bpof->icmp_code)
2152 bgp_pbr_policyroute_remove_from_zebra_recursive(
2153 bgp, path, bpf, bpof, FLOWSPEC_ICMP_TYPE);
2154 else
2155 bgp_pbr_policyroute_remove_from_zebra_unit(bgp, path, bpf);
2156 /* flush bpof */
2157 if (bpof->tcpflags)
2158 list_delete_all_node(bpof->tcpflags);
2159 if (bpof->dscp)
2160 list_delete_all_node(bpof->dscp);
2161 if (bpof->flowlabel)
2162 list_delete_all_node(bpof->flowlabel);
2163 if (bpof->pkt_len)
2164 list_delete_all_node(bpof->pkt_len);
2165 if (bpof->fragment)
2166 list_delete_all_node(bpof->fragment);
2167 }
2168
2169 static void bgp_pbr_dump_entry(struct bgp_pbr_filter *bpf, bool add)
2170 {
2171 struct bgp_pbr_range_port *src_port;
2172 struct bgp_pbr_range_port *dst_port;
2173 struct bgp_pbr_range_port *pkt_len;
2174 char bufsrc[64], bufdst[64];
2175 char buffer[64];
2176 int remaining_len = 0;
2177 char protocol_str[16];
2178
2179 if (!bpf)
2180 return;
2181 src_port = bpf->src_port;
2182 dst_port = bpf->dst_port;
2183 pkt_len = bpf->pkt_len;
2184
2185 protocol_str[0] = '\0';
2186 if (bpf->tcp_flags && bpf->tcp_flags->mask)
2187 bpf->protocol = IPPROTO_TCP;
2188 if (bpf->protocol)
2189 snprintf(protocol_str, sizeof(protocol_str),
2190 "proto %d", bpf->protocol);
2191 buffer[0] = '\0';
2192 if (bpf->protocol == IPPROTO_ICMP && src_port && dst_port)
2193 remaining_len += snprintf(buffer, sizeof(buffer),
2194 "type %d, code %d",
2195 src_port->min_port,
2196 dst_port->min_port);
2197 else if (bpf->protocol == IPPROTO_UDP ||
2198 bpf->protocol == IPPROTO_TCP) {
2199
2200 if (src_port && src_port->min_port)
2201 remaining_len += snprintf(buffer,
2202 sizeof(buffer),
2203 "from [%u:%u]",
2204 src_port->min_port,
2205 src_port->max_port ?
2206 src_port->max_port :
2207 src_port->min_port);
2208 if (dst_port && dst_port->min_port)
2209 remaining_len += snprintf(buffer +
2210 remaining_len,
2211 sizeof(buffer)
2212 - remaining_len,
2213 "to [%u:%u]",
2214 dst_port->min_port,
2215 dst_port->max_port ?
2216 dst_port->max_port :
2217 dst_port->min_port);
2218 }
2219 if (pkt_len && (pkt_len->min_port || pkt_len->max_port)) {
2220 remaining_len += snprintf(buffer + remaining_len,
2221 sizeof(buffer)
2222 - remaining_len,
2223 " len [%u:%u]",
2224 pkt_len->min_port,
2225 pkt_len->max_port ?
2226 pkt_len->max_port :
2227 pkt_len->min_port);
2228 } else if (bpf->pkt_len_val) {
2229 remaining_len += snprintf(buffer + remaining_len,
2230 sizeof(buffer)
2231 - remaining_len,
2232 " %s len %u",
2233 bpf->pkt_len_val->mask
2234 ? "!" : "",
2235 bpf->pkt_len_val->val);
2236 }
2237 if (bpf->tcp_flags) {
2238 remaining_len += snprintf(buffer + remaining_len,
2239 sizeof(buffer)
2240 - remaining_len,
2241 "tcpflags %x/%x",
2242 bpf->tcp_flags->val,
2243 bpf->tcp_flags->mask);
2244 }
2245 if (bpf->dscp) {
2246 snprintf(buffer + remaining_len,
2247 sizeof(buffer)
2248 - remaining_len,
2249 "%s dscp %d",
2250 bpf->dscp->mask
2251 ? "!" : "",
2252 bpf->dscp->val);
2253 }
2254 if (bpf->flow_label) {
2255 snprintf(buffer + remaining_len,
2256 sizeof(buffer)
2257 - remaining_len,
2258 "%s flow_label %d",
2259 bpf->flow_label->mask
2260 ? "!" : "",
2261 bpf->flow_label->val);
2262 }
2263 zlog_debug("BGP: %s FS PBR from %s to %s, %s %s",
2264 add ? "adding" : "removing",
2265 bpf->src == NULL ? "<all>" :
2266 prefix2str(bpf->src, bufsrc, sizeof(bufsrc)),
2267 bpf->dst == NULL ? "<all>" :
2268 prefix2str(bpf->dst, bufdst, sizeof(bufdst)),
2269 protocol_str, buffer);
2270
2271 }
2272
2273 static void bgp_pbr_policyroute_add_to_zebra_unit(struct bgp *bgp,
2274 struct bgp_path_info *path,
2275 struct bgp_pbr_filter *bpf,
2276 struct nexthop *nh,
2277 float *rate)
2278 {
2279 struct bgp_pbr_match temp;
2280 struct bgp_pbr_match_entry temp2;
2281 struct bgp_pbr_match *bpm;
2282 struct bgp_pbr_match_entry *bpme = NULL;
2283 struct bgp_pbr_action temp3;
2284 struct bgp_pbr_action *bpa = NULL;
2285 struct bgp_pbr_match_entry_remain bpmer;
2286 struct bgp_pbr_rule_remain bprr;
2287 struct bgp_pbr_range_port *src_port;
2288 struct bgp_pbr_range_port *dst_port;
2289 struct bgp_pbr_range_port *pkt_len;
2290 struct bgp_pbr_rule pbr_rule;
2291 struct bgp_pbr_rule *bpr;
2292 bool bpr_found = false;
2293 bool bpme_found = false;
2294 struct vrf *vrf = NULL;
2295
2296 if (!bpf)
2297 return;
2298 src_port = bpf->src_port;
2299 dst_port = bpf->dst_port;
2300 pkt_len = bpf->pkt_len;
2301
2302 if (BGP_DEBUG(zebra, ZEBRA))
2303 bgp_pbr_dump_entry(bpf, true);
2304
2305 /* look for bpa first */
2306 memset(&temp3, 0, sizeof(temp3));
2307 if (rate)
2308 temp3.rate = *rate;
2309 if (nh)
2310 memcpy(&temp3.nh, nh, sizeof(struct nexthop));
2311 temp3.vrf_id = bpf->vrf_id;
2312 temp3.afi = family2afi(bpf->family);
2313 bpa = hash_get(bgp->pbr_action_hash, &temp3,
2314 bgp_pbr_action_alloc_intern);
2315
2316 if (nh)
2317 vrf = vrf_lookup_by_id(nh->vrf_id);
2318 if (bpa->fwmark == 0) {
2319 /* drop is handled by iptable */
2320 if (nh && nh->type == NEXTHOP_TYPE_BLACKHOLE) {
2321 bpa->table_id = 0;
2322 bpa->installed = true;
2323 } else {
2324 bpa->fwmark = bgp_zebra_tm_get_id();
2325 /* if action is redirect-vrf, then
2326 * use directly table_id of vrf
2327 */
2328 if (nh && vrf && !vrf_is_backend_netns()
2329 && bpf->vrf_id != vrf->vrf_id)
2330 bpa->table_id = vrf->data.l.table_id;
2331 else
2332 bpa->table_id = bpa->fwmark;
2333 bpa->installed = false;
2334 }
2335 bpa->bgp = bgp;
2336 bpa->unique = ++bgp_pbr_action_counter_unique;
2337 /* 0 value is forbidden */
2338 bpa->install_in_progress = false;
2339 }
2340 if (bpf->type == BGP_PBR_IPRULE) {
2341 memset(&pbr_rule, 0, sizeof(pbr_rule));
2342 pbr_rule.vrf_id = bpf->vrf_id;
2343 pbr_rule.priority = 20;
2344 if (bpf->src) {
2345 pbr_rule.flags |= MATCH_IP_SRC_SET;
2346 prefix_copy(&pbr_rule.src, bpf->src);
2347 }
2348 if (bpf->dst) {
2349 pbr_rule.flags |= MATCH_IP_DST_SET;
2350 prefix_copy(&pbr_rule.dst, bpf->dst);
2351 }
2352 pbr_rule.action = bpa;
2353 bpr = hash_get(bgp->pbr_rule_hash, &pbr_rule,
2354 bgp_pbr_rule_alloc_intern);
2355 if (bpr->unique == 0) {
2356 bpr->unique = ++bgp_pbr_action_counter_unique;
2357 bpr->installed = false;
2358 bpr->install_in_progress = false;
2359 /* link bgp info to bpr */
2360 bpr->path = (void *)path;
2361 } else
2362 bpr_found = true;
2363 /* already installed */
2364 if (bpr_found) {
2365 struct bgp_path_info_extra *extra =
2366 bgp_path_info_extra_get(path);
2367
2368 if (extra &&
2369 listnode_lookup_nocheck(extra->bgp_fs_iprule,
2370 bpr)) {
2371 if (BGP_DEBUG(pbr, PBR_ERROR))
2372 zlog_err("%s: entry %p/%p already installed in bgp pbr iprule",
2373 __func__, path, bpr);
2374 return;
2375 }
2376 }
2377
2378 bgp_pbr_bpa_add(bpa);
2379
2380 /* ip rule add */
2381 if (!bpr->installed)
2382 bgp_send_pbr_rule_action(bpa, bpr, true);
2383
2384 /* A previous entry may already exist
2385 * flush previous entry if necessary
2386 */
2387 bprr.bpr_to_match = bpr;
2388 bprr.bpr_found = NULL;
2389 hash_walk(bgp->pbr_rule_hash, bgp_pbr_get_same_rule, &bprr);
2390 if (bprr.bpr_found) {
2391 static struct bgp_pbr_rule *local_bpr;
2392 static struct bgp_pbr_action *local_bpa;
2393
2394 local_bpr = bprr.bpr_found;
2395 local_bpa = local_bpr->action;
2396 bgp_pbr_flush_iprule(bgp, local_bpa,
2397 local_bpr);
2398 }
2399 return;
2400 }
2401 /* then look for bpm */
2402 memset(&temp, 0, sizeof(temp));
2403 temp.vrf_id = bpf->vrf_id;
2404 temp.family = bpf->family;
2405 if (bpf->src)
2406 temp.flags |= MATCH_IP_SRC_SET;
2407 if (bpf->dst)
2408 temp.flags |= MATCH_IP_DST_SET;
2409
2410 if (src_port && (src_port->min_port || bpf->protocol == IPPROTO_ICMP)) {
2411 if (bpf->protocol == IPPROTO_ICMP)
2412 temp.flags |= MATCH_ICMP_SET;
2413 temp.flags |= MATCH_PORT_SRC_SET;
2414 }
2415 if (dst_port && (dst_port->min_port || bpf->protocol == IPPROTO_ICMP)) {
2416 if (bpf->protocol == IPPROTO_ICMP)
2417 temp.flags |= MATCH_ICMP_SET;
2418 temp.flags |= MATCH_PORT_DST_SET;
2419 }
2420 if (src_port && src_port->max_port)
2421 temp.flags |= MATCH_PORT_SRC_RANGE_SET;
2422 if (dst_port && dst_port->max_port)
2423 temp.flags |= MATCH_PORT_DST_RANGE_SET;
2424
2425 if (bpf->src == NULL || bpf->dst == NULL) {
2426 if (temp.flags & (MATCH_PORT_DST_SET | MATCH_PORT_SRC_SET))
2427 temp.type = IPSET_NET_PORT;
2428 else
2429 temp.type = IPSET_NET;
2430 } else {
2431 if (temp.flags & (MATCH_PORT_DST_SET | MATCH_PORT_SRC_SET))
2432 temp.type = IPSET_NET_PORT_NET;
2433 else
2434 temp.type = IPSET_NET_NET;
2435 }
2436 if (pkt_len) {
2437 temp.pkt_len_min = pkt_len->min_port;
2438 if (pkt_len->max_port)
2439 temp.pkt_len_max = pkt_len->max_port;
2440 } else if (bpf->pkt_len_val) {
2441 if (bpf->pkt_len_val->mask)
2442 temp.flags |= MATCH_PKT_LEN_INVERSE_SET;
2443 temp.pkt_len_min = bpf->pkt_len_val->val;
2444 }
2445 if (bpf->tcp_flags) {
2446 temp.tcp_flags = bpf->tcp_flags->val;
2447 temp.tcp_mask_flags = bpf->tcp_flags->mask;
2448 }
2449 if (bpf->dscp) {
2450 if (bpf->dscp->mask)
2451 temp.flags |= MATCH_DSCP_INVERSE_SET;
2452 else
2453 temp.flags |= MATCH_DSCP_SET;
2454 temp.dscp_value = bpf->dscp->val;
2455 }
2456 if (bpf->flow_label) {
2457 if (bpf->flow_label->mask)
2458 temp.flags |= MATCH_FLOW_LABEL_INVERSE_SET;
2459 else
2460 temp.flags |= MATCH_FLOW_LABEL_SET;
2461 temp.flow_label = bpf->flow_label->val;
2462 }
2463 if (bpf->fragment) {
2464 if (bpf->fragment->mask)
2465 temp.flags |= MATCH_FRAGMENT_INVERSE_SET;
2466 temp.fragment = bpf->fragment->val;
2467 }
2468 if (bpf->protocol) {
2469 temp.protocol = bpf->protocol;
2470 temp.flags |= MATCH_PROTOCOL_SET;
2471 }
2472 temp.action = bpa;
2473 bpm = hash_get(bgp->pbr_match_hash, &temp,
2474 bgp_pbr_match_alloc_intern);
2475
2476 /* new, then self allocate ipset_name and unique */
2477 if (bpm->unique == 0) {
2478 bpm->unique = ++bgp_pbr_match_counter_unique;
2479 /* 0 value is forbidden */
2480 snprintf(bpm->ipset_name, sizeof(bpm->ipset_name),
2481 "match%p", bpm);
2482 bpm->entry_hash = hash_create_size(8,
2483 bgp_pbr_match_entry_hash_key,
2484 bgp_pbr_match_entry_hash_equal,
2485 "Match Entry Hash");
2486 bpm->installed = false;
2487
2488 /* unique2 should be updated too */
2489 bpm->unique2 = ++bgp_pbr_match_iptable_counter_unique;
2490 bpm->installed_in_iptable = false;
2491 bpm->install_in_progress = false;
2492 bpm->install_iptable_in_progress = false;
2493 }
2494
2495 memset(&temp2, 0, sizeof(temp2));
2496 if (bpf->src)
2497 prefix_copy(&temp2.src, bpf->src);
2498 else
2499 temp2.src.family = bpf->family;
2500 if (bpf->dst)
2501 prefix_copy(&temp2.dst, bpf->dst);
2502 else
2503 temp2.dst.family = bpf->family;
2504 temp2.src_port_min = src_port ? src_port->min_port : 0;
2505 temp2.dst_port_min = dst_port ? dst_port->min_port : 0;
2506 temp2.src_port_max = src_port ? src_port->max_port : 0;
2507 temp2.dst_port_max = dst_port ? dst_port->max_port : 0;
2508 temp2.proto = bpf->protocol;
2509 bpme = hash_get(bpm->entry_hash, &temp2,
2510 bgp_pbr_match_entry_alloc_intern);
2511 if (bpme->unique == 0) {
2512 bpme->unique = ++bgp_pbr_match_entry_counter_unique;
2513 /* 0 value is forbidden */
2514 bpme->backpointer = bpm;
2515 bpme->installed = false;
2516 bpme->install_in_progress = false;
2517 /* link bgp info to bpme */
2518 bpme->path = (void *)path;
2519 } else
2520 bpme_found = true;
2521
2522 /* already installed */
2523 if (bpme_found) {
2524 struct bgp_path_info_extra *extra =
2525 bgp_path_info_extra_get(path);
2526
2527 if (extra &&
2528 listnode_lookup_nocheck(extra->bgp_fs_pbr, bpme)) {
2529 if (BGP_DEBUG(pbr, PBR_ERROR))
2530 zlog_err(
2531 "%s: entry %p/%p already installed in bgp pbr",
2532 __func__, path, bpme);
2533 return;
2534 }
2535 }
2536 /* BGP FS: append entry to zebra
2537 * - policies are not routing entries and as such
2538 * route replace semantics don't necessarily follow
2539 * through to policy entries
2540 * - because of that, not all policing information will be stored
2541 * into zebra. and non selected policies will be suppressed from zebra
2542 * - as consequence, in order to bring consistency
2543 * a policy will be added, then ifan ecmp policy exists,
2544 * it will be suppressed subsequently
2545 */
2546 /* ip rule add */
2547 bgp_pbr_bpa_add(bpa);
2548
2549 /* ipset create */
2550 if (!bpm->installed)
2551 bgp_send_pbr_ipset_match(bpm, true);
2552 /* ipset add */
2553 if (!bpme->installed)
2554 bgp_send_pbr_ipset_entry_match(bpme, true);
2555
2556 /* iptables */
2557 if (!bpm->installed_in_iptable)
2558 bgp_send_pbr_iptable(bpa, bpm, true);
2559
2560 /* A previous entry may already exist
2561 * flush previous entry if necessary
2562 */
2563 bpmer.bpme_to_match = bpme;
2564 bpmer.bpme_found = NULL;
2565 hash_walk(bgp->pbr_match_hash, bgp_pbr_get_remaining_entry, &bpmer);
2566 if (bpmer.bpme_found) {
2567 static struct bgp_pbr_match *local_bpm;
2568 static struct bgp_pbr_action *local_bpa;
2569
2570 local_bpm = bpmer.bpme_found->backpointer;
2571 local_bpa = local_bpm->action;
2572 bgp_pbr_flush_entry(bgp, local_bpa,
2573 local_bpm, bpmer.bpme_found);
2574 }
2575
2576
2577 }
2578
2579 static void bgp_pbr_policyroute_add_to_zebra_recursive(
2580 struct bgp *bgp, struct bgp_path_info *path, struct bgp_pbr_filter *bpf,
2581 struct bgp_pbr_or_filter *bpof, struct nexthop *nh, float *rate,
2582 uint8_t type_entry)
2583 {
2584 struct listnode *node, *nnode;
2585 struct bgp_pbr_val_mask *valmask;
2586 uint8_t next_type_entry;
2587 struct list *orig_list;
2588 struct bgp_pbr_val_mask **target_val;
2589
2590 if (type_entry == 0) {
2591 bgp_pbr_policyroute_add_to_zebra_unit(bgp, path, bpf, nh, rate);
2592 return;
2593 }
2594 next_type_entry = bgp_pbr_next_type_entry(type_entry);
2595 if (type_entry == FLOWSPEC_TCP_FLAGS && bpof->tcpflags) {
2596 orig_list = bpof->tcpflags;
2597 target_val = &bpf->tcp_flags;
2598 } else if (type_entry == FLOWSPEC_DSCP && bpof->dscp) {
2599 orig_list = bpof->dscp;
2600 target_val = &bpf->dscp;
2601 } else if (type_entry == FLOWSPEC_PKT_LEN && bpof->pkt_len) {
2602 orig_list = bpof->pkt_len;
2603 target_val = &bpf->pkt_len_val;
2604 } else if (type_entry == FLOWSPEC_FRAGMENT && bpof->fragment) {
2605 orig_list = bpof->fragment;
2606 target_val = &bpf->fragment;
2607 } else if (type_entry == FLOWSPEC_ICMP_TYPE &&
2608 (bpof->icmp_type || bpof->icmp_code)) {
2609 /* enumerate list for icmp - must be last one */
2610 bgp_pbr_icmp_action(bgp, path, bpf, bpof, true, nh, rate);
2611 return;
2612 } else {
2613 bgp_pbr_policyroute_add_to_zebra_recursive(
2614 bgp, path, bpf, bpof, nh, rate, next_type_entry);
2615 return;
2616 }
2617 for (ALL_LIST_ELEMENTS(orig_list, node, nnode, valmask)) {
2618 *target_val = valmask;
2619 bgp_pbr_policyroute_add_to_zebra_recursive(
2620 bgp, path, bpf, bpof, nh, rate, next_type_entry);
2621 }
2622 }
2623
2624 static void bgp_pbr_policyroute_add_to_zebra(struct bgp *bgp,
2625 struct bgp_path_info *path,
2626 struct bgp_pbr_filter *bpf,
2627 struct bgp_pbr_or_filter *bpof,
2628 struct nexthop *nh, float *rate)
2629 {
2630 if (!bpof) {
2631 bgp_pbr_policyroute_add_to_zebra_unit(bgp, path, bpf, nh, rate);
2632 return;
2633 }
2634 if (bpof->tcpflags)
2635 bgp_pbr_policyroute_add_to_zebra_recursive(
2636 bgp, path, bpf, bpof, nh, rate, FLOWSPEC_TCP_FLAGS);
2637 else if (bpof->dscp)
2638 bgp_pbr_policyroute_add_to_zebra_recursive(
2639 bgp, path, bpf, bpof, nh, rate, FLOWSPEC_DSCP);
2640 else if (bpof->pkt_len)
2641 bgp_pbr_policyroute_add_to_zebra_recursive(
2642 bgp, path, bpf, bpof, nh, rate, FLOWSPEC_PKT_LEN);
2643 else if (bpof->fragment)
2644 bgp_pbr_policyroute_add_to_zebra_recursive(
2645 bgp, path, bpf, bpof, nh, rate, FLOWSPEC_FRAGMENT);
2646 else if (bpof->icmp_type || bpof->icmp_code)
2647 bgp_pbr_policyroute_add_to_zebra_recursive(
2648 bgp, path, bpf, bpof, nh, rate, FLOWSPEC_ICMP_TYPE);
2649 else
2650 bgp_pbr_policyroute_add_to_zebra_unit(bgp, path, bpf, nh, rate);
2651 /* flush bpof */
2652 if (bpof->tcpflags)
2653 list_delete_all_node(bpof->tcpflags);
2654 if (bpof->dscp)
2655 list_delete_all_node(bpof->dscp);
2656 if (bpof->pkt_len)
2657 list_delete_all_node(bpof->pkt_len);
2658 if (bpof->fragment)
2659 list_delete_all_node(bpof->fragment);
2660 if (bpof->icmp_type)
2661 list_delete_all_node(bpof->icmp_type);
2662 if (bpof->icmp_code)
2663 list_delete_all_node(bpof->icmp_code);
2664 }
2665
2666 static void bgp_pbr_handle_entry(struct bgp *bgp, struct bgp_path_info *path,
2667 struct bgp_pbr_entry_main *api, bool add)
2668 {
2669 struct nexthop nh;
2670 int i = 0;
2671 int continue_loop = 1;
2672 float rate = 0;
2673 struct prefix *src = NULL, *dst = NULL;
2674 uint8_t proto = 0;
2675 struct bgp_pbr_range_port *srcp = NULL, *dstp = NULL;
2676 struct bgp_pbr_range_port range, range_icmp_code;
2677 struct bgp_pbr_range_port pkt_len;
2678 struct bgp_pbr_filter bpf;
2679 uint8_t kind_enum;
2680 struct bgp_pbr_or_filter bpof;
2681 struct bgp_pbr_val_mask bpvm;
2682
2683 memset(&range, 0, sizeof(range));
2684 memset(&nh, 0, sizeof(nh));
2685 memset(&bpf, 0, sizeof(bpf));
2686 memset(&bpof, 0, sizeof(bpof));
2687 if (api->match_bitmask & PREFIX_SRC_PRESENT ||
2688 (api->type == BGP_PBR_IPRULE &&
2689 api->match_bitmask_iprule & PREFIX_SRC_PRESENT))
2690 src = &api->src_prefix;
2691 if (api->match_bitmask & PREFIX_DST_PRESENT ||
2692 (api->type == BGP_PBR_IPRULE &&
2693 api->match_bitmask_iprule & PREFIX_DST_PRESENT))
2694 dst = &api->dst_prefix;
2695 if (api->type == BGP_PBR_IPRULE)
2696 bpf.type = api->type;
2697 memset(&nh, 0, sizeof(nh));
2698 nh.vrf_id = VRF_UNKNOWN;
2699 if (api->match_protocol_num) {
2700 proto = (uint8_t)api->protocol[0].value;
2701 if (api->afi == AF_INET6 && proto == IPPROTO_ICMPV6)
2702 proto = IPPROTO_ICMP;
2703 }
2704 /* if match_port is selected, then either src or dst port will be parsed
2705 * but not both at the same time
2706 */
2707 if (api->match_port_num >= 1) {
2708 bgp_pbr_extract(api->port,
2709 api->match_port_num,
2710 &range);
2711 srcp = dstp = &range;
2712 } else if (api->match_src_port_num >= 1) {
2713 bgp_pbr_extract(api->src_port,
2714 api->match_src_port_num,
2715 &range);
2716 srcp = &range;
2717 dstp = NULL;
2718 } else if (api->match_dst_port_num >= 1) {
2719 bgp_pbr_extract(api->dst_port,
2720 api->match_dst_port_num,
2721 &range);
2722 dstp = &range;
2723 srcp = NULL;
2724 }
2725 if (api->match_icmp_type_num >= 1) {
2726 proto = IPPROTO_ICMP;
2727 if (bgp_pbr_extract(api->icmp_type,
2728 api->match_icmp_type_num,
2729 &range))
2730 srcp = &range;
2731 else {
2732 bpof.icmp_type = list_new();
2733 bgp_pbr_extract_enumerate(api->icmp_type,
2734 api->match_icmp_type_num,
2735 OPERATOR_UNARY_OR,
2736 bpof.icmp_type,
2737 FLOWSPEC_ICMP_TYPE);
2738 }
2739 }
2740 if (api->match_icmp_code_num >= 1) {
2741 proto = IPPROTO_ICMP;
2742 if (bgp_pbr_extract(api->icmp_code,
2743 api->match_icmp_code_num,
2744 &range_icmp_code))
2745 dstp = &range_icmp_code;
2746 else {
2747 bpof.icmp_code = list_new();
2748 bgp_pbr_extract_enumerate(api->icmp_code,
2749 api->match_icmp_code_num,
2750 OPERATOR_UNARY_OR,
2751 bpof.icmp_code,
2752 FLOWSPEC_ICMP_CODE);
2753 }
2754 }
2755
2756 if (api->match_tcpflags_num) {
2757 kind_enum = bgp_pbr_match_val_get_operator(api->tcpflags,
2758 api->match_tcpflags_num);
2759 if (kind_enum == OPERATOR_UNARY_AND) {
2760 bpf.tcp_flags = &bpvm;
2761 bgp_pbr_extract_enumerate(api->tcpflags,
2762 api->match_tcpflags_num,
2763 OPERATOR_UNARY_AND,
2764 bpf.tcp_flags,
2765 FLOWSPEC_TCP_FLAGS);
2766 } else if (kind_enum == OPERATOR_UNARY_OR) {
2767 bpof.tcpflags = list_new();
2768 bgp_pbr_extract_enumerate(api->tcpflags,
2769 api->match_tcpflags_num,
2770 OPERATOR_UNARY_OR,
2771 bpof.tcpflags,
2772 FLOWSPEC_TCP_FLAGS);
2773 }
2774 }
2775 if (api->match_packet_length_num) {
2776 bool ret;
2777
2778 ret = bgp_pbr_extract(api->packet_length,
2779 api->match_packet_length_num,
2780 &pkt_len);
2781 if (ret)
2782 bpf.pkt_len = &pkt_len;
2783 else {
2784 bpof.pkt_len = list_new();
2785 bgp_pbr_extract_enumerate(api->packet_length,
2786 api->match_packet_length_num,
2787 OPERATOR_UNARY_OR,
2788 bpof.pkt_len,
2789 FLOWSPEC_PKT_LEN);
2790 }
2791 }
2792 if (api->match_dscp_num >= 1) {
2793 bpof.dscp = list_new();
2794 bgp_pbr_extract_enumerate(api->dscp, api->match_dscp_num,
2795 OPERATOR_UNARY_OR,
2796 bpof.dscp, FLOWSPEC_DSCP);
2797 }
2798 if (api->match_fragment_num) {
2799 bpof.fragment = list_new();
2800 bgp_pbr_extract_enumerate(api->fragment,
2801 api->match_fragment_num,
2802 OPERATOR_UNARY_OR,
2803 bpof.fragment,
2804 FLOWSPEC_FRAGMENT);
2805 }
2806 bpf.vrf_id = api->vrf_id;
2807 bpf.src = src;
2808 bpf.dst = dst;
2809 bpf.protocol = proto;
2810 bpf.src_port = srcp;
2811 bpf.dst_port = dstp;
2812 bpf.family = afi2family(api->afi);
2813 if (!add) {
2814 bgp_pbr_policyroute_remove_from_zebra(bgp, path, &bpf, &bpof);
2815 return;
2816 }
2817 /* no action for add = true */
2818 for (i = 0; i < api->action_num; i++) {
2819 switch (api->actions[i].action) {
2820 case ACTION_TRAFFICRATE:
2821 /* drop packet */
2822 if (api->actions[i].u.r.rate == 0) {
2823 nh.vrf_id = api->vrf_id;
2824 nh.type = NEXTHOP_TYPE_BLACKHOLE;
2825 bgp_pbr_policyroute_add_to_zebra(
2826 bgp, path, &bpf, &bpof, &nh, &rate);
2827 } else {
2828 /* update rate. can be reentrant */
2829 rate = api->actions[i].u.r.rate;
2830 if (BGP_DEBUG(pbr, PBR)) {
2831 bgp_pbr_print_policy_route(api);
2832 zlog_warn("PBR: ignoring Set action rate %f",
2833 api->actions[i].u.r.rate);
2834 }
2835 }
2836 break;
2837 case ACTION_TRAFFIC_ACTION:
2838 if (api->actions[i].u.za.filter
2839 & TRAFFIC_ACTION_SAMPLE) {
2840 if (BGP_DEBUG(pbr, PBR)) {
2841 bgp_pbr_print_policy_route(api);
2842 zlog_warn("PBR: Sample action Ignored");
2843 }
2844 }
2845 /* terminate action: run other filters
2846 */
2847 break;
2848 case ACTION_REDIRECT_IP:
2849 nh.vrf_id = api->vrf_id;
2850 if (api->afi == AFI_IP) {
2851 nh.type = NEXTHOP_TYPE_IPV4;
2852 nh.gate.ipv4.s_addr =
2853 api->actions[i].u.zr.
2854 redirect_ip_v4.s_addr;
2855 } else {
2856 nh.type = NEXTHOP_TYPE_IPV6;
2857 memcpy(&nh.gate.ipv6,
2858 &api->actions[i].u.zr.redirect_ip_v6,
2859 sizeof(struct in6_addr));
2860 }
2861 bgp_pbr_policyroute_add_to_zebra(bgp, path, &bpf, &bpof,
2862 &nh, &rate);
2863 /* XXX combination with REDIRECT_VRF
2864 * + REDIRECT_NH_IP not done
2865 */
2866 continue_loop = 0;
2867 break;
2868 case ACTION_REDIRECT:
2869 if (api->afi == AFI_IP)
2870 nh.type = NEXTHOP_TYPE_IPV4;
2871 else
2872 nh.type = NEXTHOP_TYPE_IPV6;
2873 nh.vrf_id = api->actions[i].u.redirect_vrf;
2874 bgp_pbr_policyroute_add_to_zebra(bgp, path, &bpf, &bpof,
2875 &nh, &rate);
2876 continue_loop = 0;
2877 break;
2878 case ACTION_MARKING:
2879 if (BGP_DEBUG(pbr, PBR)) {
2880 bgp_pbr_print_policy_route(api);
2881 zlog_warn("PBR: Set DSCP/FlowLabel %u Ignored",
2882 api->actions[i].u.marking_dscp);
2883 }
2884 break;
2885 default:
2886 break;
2887 }
2888 if (continue_loop == 0)
2889 break;
2890 }
2891 }
2892
2893 void bgp_pbr_update_entry(struct bgp *bgp, const struct prefix *p,
2894 struct bgp_path_info *info, afi_t afi, safi_t safi,
2895 bool nlri_update)
2896 {
2897 struct bgp_pbr_entry_main api;
2898
2899 if (safi != SAFI_FLOWSPEC)
2900 return; /* not supported */
2901 /* Make Zebra API structure. */
2902 memset(&api, 0, sizeof(api));
2903 api.vrf_id = bgp->vrf_id;
2904 api.afi = afi;
2905
2906 if (!bgp_zebra_tm_chunk_obtained()) {
2907 if (BGP_DEBUG(pbr, PBR_ERROR))
2908 flog_err(EC_BGP_TABLE_CHUNK,
2909 "%s: table chunk not obtained yet", __func__);
2910 return;
2911 }
2912
2913 if (bgp_pbr_build_and_validate_entry(p, info, &api) < 0) {
2914 if (BGP_DEBUG(pbr, PBR_ERROR))
2915 flog_err(EC_BGP_FLOWSPEC_INSTALLATION,
2916 "%s: cancel updating entry %p in bgp pbr",
2917 __func__, info);
2918 return;
2919 }
2920 bgp_pbr_handle_entry(bgp, info, &api, nlri_update);
2921 }
2922
2923 int bgp_pbr_interface_compare(const struct bgp_pbr_interface *a,
2924 const struct bgp_pbr_interface *b)
2925 {
2926 return strcmp(a->name, b->name);
2927 }
2928
2929 struct bgp_pbr_interface *bgp_pbr_interface_lookup(const char *name,
2930 struct bgp_pbr_interface_head *head)
2931 {
2932 struct bgp_pbr_interface pbr_if;
2933
2934 strlcpy(pbr_if.name, name, sizeof(pbr_if.name));
2935 return (RB_FIND(bgp_pbr_interface_head,
2936 head, &pbr_if));
2937 }
2938
2939 /* this function resets to the default policy routing
2940 * go back to default status
2941 */
2942 void bgp_pbr_reset(struct bgp *bgp, afi_t afi)
2943 {
2944 struct bgp_pbr_config *bgp_pbr_cfg = bgp->bgp_pbr_cfg;
2945 struct bgp_pbr_interface_head *head;
2946 struct bgp_pbr_interface *pbr_if;
2947
2948 if (!bgp_pbr_cfg)
2949 return;
2950 if (afi == AFI_IP)
2951 head = &(bgp_pbr_cfg->ifaces_by_name_ipv4);
2952 else
2953 head = &(bgp_pbr_cfg->ifaces_by_name_ipv6);
2954 while (!RB_EMPTY(bgp_pbr_interface_head, head)) {
2955 pbr_if = RB_ROOT(bgp_pbr_interface_head, head);
2956 RB_REMOVE(bgp_pbr_interface_head, head, pbr_if);
2957 XFREE(MTYPE_TMP, pbr_if);
2958 }
2959 }
FRRouting mirror