-
Notifications
You must be signed in to change notification settings - Fork 2.9k
/
bpf_lxc.c
1490 lines (1275 loc) · 41.4 KB
/
bpf_lxc.c
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
// SPDX-License-Identifier: GPL-2.0
/* Copyright (C) 2016-2020 Authors of Cilium */
#include <bpf/ctx/skb.h>
#include <bpf/api.h>
#include <ep_config.h>
#include <node_config.h>
#include <bpf/verifier.h>
#include <linux/icmpv6.h>
#define EVENT_SOURCE LXC_ID
#include "lib/tailcall.h"
#include "lib/common.h"
#include "lib/config.h"
#include "lib/maps.h"
#include "lib/arp.h"
#include "lib/edt.h"
#include "lib/ipv6.h"
#include "lib/ipv4.h"
#include "lib/icmp6.h"
#include "lib/eth.h"
#include "lib/dbg.h"
#include "lib/l3.h"
#include "lib/lxc.h"
#include "lib/nat46.h"
#include "lib/identity.h"
#include "lib/policy.h"
#include "lib/lb.h"
#include "lib/drop.h"
#include "lib/dbg.h"
#include "lib/trace.h"
#include "lib/csum.h"
#include "lib/encap.h"
#include "lib/eps.h"
#include "lib/nat.h"
#include "lib/fib.h"
#include "lib/nodeport.h"
#include "lib/policy_log.h"
#if defined(ENABLE_ARP_PASSTHROUGH) && defined(ENABLE_ARP_RESPONDER)
#error "Either ENABLE_ARP_PASSTHROUGH or ENABLE_ARP_RESPONDER can be defined"
#endif
#if defined(ENABLE_IPV4) || defined(ENABLE_IPV6)
static __always_inline bool redirect_to_proxy(int verdict, __u8 dir)
{
return is_defined(ENABLE_HOST_REDIRECT) && verdict > 0 &&
(dir == CT_NEW || dir == CT_ESTABLISHED || dir == CT_REOPENED);
}
#endif
#ifdef ENABLE_IPV6
static __always_inline int ipv6_l3_from_lxc(struct __ctx_buff *ctx,
struct ipv6_ct_tuple *tuple,
int l3_off, struct ipv6hdr *ip6,
__u32 *dstID)
{
#ifdef ENABLE_ROUTING
union macaddr router_mac = NODE_MAC;
#endif
int ret, verdict, l4_off, hdrlen;
struct csum_offset csum_off = {};
struct ct_state ct_state_new = {};
struct ct_state ct_state = {};
void *data, *data_end;
union v6addr *daddr, orig_dip;
__u32 tunnel_endpoint = 0;
__u8 encrypt_key = 0;
__u32 monitor = 0;
__u8 reason;
bool hairpin_flow = false; /* endpoint wants to access itself via service IP */
__u8 policy_match_type = POLICY_MATCH_NONE;
__u8 audited = 0;
if (unlikely(!is_valid_lxc_src_ip(ip6)))
return DROP_INVALID_SIP;
ipv6_addr_copy(&tuple->daddr, (union v6addr *) &ip6->daddr);
ipv6_addr_copy(&tuple->saddr, (union v6addr *) &ip6->saddr);
hdrlen = ipv6_hdrlen(ctx, l3_off, &tuple->nexthdr);
if (hdrlen < 0)
return hdrlen;
l4_off = l3_off + hdrlen;
#ifndef ENABLE_HOST_SERVICES_FULL
{
struct lb6_service *svc;
struct lb6_key key = {};
ret = lb6_extract_key(ctx, tuple, l4_off, &key, &csum_off,
CT_EGRESS);
if (IS_ERR(ret)) {
if (ret == DROP_UNKNOWN_L4)
goto skip_service_lookup;
else
return ret;
}
/*
* Check if the destination address is among the address that should
* be load balanced. This operation is performed before we go through
* the connection tracker to allow storing the reverse nat index in
* the CT entry for destination endpoints where we can't encode the
* state in the address.
*/
svc = lb6_lookup_service(&key, is_defined(ENABLE_NODEPORT));
if (svc) {
ret = lb6_local(get_ct_map6(tuple), ctx, l3_off, l4_off,
&csum_off, &key, tuple, svc, &ct_state_new);
if (IS_ERR(ret))
return ret;
hairpin_flow |= ct_state_new.loopback;
}
}
skip_service_lookup:
#endif /* !ENABLE_HOST_SERVICES_FULL */
/* The verifier wants to see this assignment here in case the above goto
* skip_service_lookup is hit. However, in the case the packet
* is _not_ TCP or UDP we should not be using proxy logic anyways. For
* correctness it must be below the service handler in case the service
* logic re-writes the tuple daddr. In "theory" however the assignment
* should be OK to move above goto label.
*/
ipv6_addr_copy(&orig_dip, (union v6addr *) &tuple->daddr);
/* WARNING: ip6 offset check invalidated, revalidate before use */
/* Pass all outgoing packets through conntrack. This will create an
* entry to allow reverse packets and return set cb[CB_POLICY] to
* POLICY_SKIP if the packet is a reply packet to an existing incoming
* connection.
*/
ret = ct_lookup6(get_ct_map6(tuple), tuple, ctx, l4_off, CT_EGRESS,
&ct_state, &monitor);
if (ret < 0)
return ret;
reason = ret;
/* Check it this is return traffic to an ingress proxy. */
if ((ret == CT_REPLY || ret == CT_RELATED) && ct_state.proxy_redirect) {
/* Stack will do a socket match and deliver locally. */
return ctx_redirect_to_proxy6(ctx, tuple, 0, false);
}
if (!revalidate_data(ctx, &data, &data_end, &ip6))
return DROP_INVALID;
/* Determine the destination category for policy fallback. */
if (1) {
struct remote_endpoint_info *info;
info = lookup_ip6_remote_endpoint(&orig_dip);
if (info != NULL && info->sec_label) {
*dstID = info->sec_label;
tunnel_endpoint = info->tunnel_endpoint;
encrypt_key = get_min_encrypt_key(info->key);
} else {
*dstID = WORLD_ID;
}
cilium_dbg(ctx, info ? DBG_IP_ID_MAP_SUCCEED6 : DBG_IP_ID_MAP_FAILED6,
orig_dip.p4, *dstID);
}
/* If the packet is in the establishing direction and it's destined
* within the cluster, it must match policy or be dropped. If it's
* bound for the host/outside, perform the CIDR policy check.
*/
verdict = policy_can_egress6(ctx, tuple, SECLABEL, *dstID,
&policy_match_type, &audited);
if (ret != CT_REPLY && ret != CT_RELATED && verdict < 0) {
send_policy_verdict_notify(ctx, *dstID, tuple->dport,
tuple->nexthdr, POLICY_EGRESS, 1,
verdict, policy_match_type, audited);
return verdict;
}
switch (ret) {
case CT_NEW:
send_policy_verdict_notify(ctx, *dstID, tuple->dport,
tuple->nexthdr, POLICY_EGRESS, 1,
verdict, policy_match_type, audited);
ct_recreate6:
/* New connection implies that rev_nat_index remains untouched
* to the index provided by the loadbalancer (if it applied).
* Create a CT entry which allows to track replies and to
* reverse NAT.
*/
ct_state_new.src_sec_id = SECLABEL;
ret = ct_create6(get_ct_map6(tuple), &CT_MAP_ANY6, tuple, ctx,
CT_EGRESS, &ct_state_new, verdict > 0);
if (IS_ERR(ret))
return ret;
monitor = TRACE_PAYLOAD_LEN;
break;
case CT_REOPENED:
send_policy_verdict_notify(ctx, *dstID, tuple->dport,
tuple->nexthdr, POLICY_EGRESS, 1,
verdict, policy_match_type, audited);
case CT_ESTABLISHED:
/* Did we end up at a stale non-service entry? Recreate if so. */
if (unlikely(ct_state.rev_nat_index != ct_state_new.rev_nat_index))
goto ct_recreate6;
break;
case CT_RELATED:
case CT_REPLY:
policy_mark_skip(ctx);
#ifdef ENABLE_NODEPORT
/* See comment in handle_ipv4_from_lxc(). */
if (ct_state.node_port) {
ctx->tc_index |= TC_INDEX_F_SKIP_RECIRCULATION;
ep_tail_call(ctx, CILIUM_CALL_IPV6_NODEPORT_REVNAT);
return DROP_MISSED_TAIL_CALL;
}
# ifdef ENABLE_DSR
if (ct_state.dsr) {
ret = xlate_dsr_v6(ctx, tuple, l4_off);
if (ret != 0)
return ret;
}
# endif /* ENABLE_DSR */
#endif /* ENABLE_NODEPORT */
if (ct_state.rev_nat_index) {
ret = lb6_rev_nat(ctx, l4_off, &csum_off,
ct_state.rev_nat_index, tuple, 0);
if (IS_ERR(ret))
return ret;
/* A reverse translate packet is always allowed except
* for delivery on the local node in which case this
* marking is cleared again.
*/
policy_mark_skip(ctx);
}
break;
default:
return DROP_UNKNOWN_CT;
}
hairpin_flow |= ct_state.loopback;
if (redirect_to_proxy(verdict, reason)) {
/* Trace the packet before it is forwarded to proxy */
send_trace_notify(ctx, TRACE_TO_PROXY, SECLABEL, 0,
0, 0, reason, monitor);
return ctx_redirect_to_proxy6(ctx, tuple, verdict, false);
}
if (!revalidate_data(ctx, &data, &data_end, &ip6))
return DROP_INVALID;
daddr = (union v6addr *)&ip6->daddr;
/* See handle_ipv4_from_lxc() re hairpin_flow */
if (is_defined(ENABLE_ROUTING) || hairpin_flow) {
struct endpoint_info *ep;
/* Lookup IPv6 address, this will return a match if:
* - The destination IP address belongs to a local endpoint managed by
* cilium
* - The destination IP address is an IP address associated with the
* host itself.
*/
ep = lookup_ip6_endpoint(ip6);
if (ep) {
#ifdef ENABLE_ROUTING
if (ep->flags & ENDPOINT_F_HOST) {
#ifdef HOST_IFINDEX
goto to_host;
#else
return DROP_HOST_UNREACHABLE;
#endif
}
#endif /* ENABLE_ROUTING */
policy_clear_mark(ctx);
return ipv6_local_delivery(ctx, l3_off, SECLABEL, ep,
METRIC_EGRESS, false);
}
}
/* The packet goes to a peer not managed by this agent instance */
#ifdef ENCAP_IFINDEX
{
struct endpoint_key key = {};
/* Lookup the destination prefix in the list of known
* destination prefixes. If there is a match, the packet will
* be encapsulated to that node and then routed by the agent on
* the remote node.
*
* IPv6 lookup key: daddr/96
*/
key.ip6.p1 = daddr->p1;
key.ip6.p2 = daddr->p2;
key.ip6.p3 = daddr->p3;
key.family = ENDPOINT_KEY_IPV6;
/* Three cases exist here either (a) the encap and redirect could
* not find the tunnel so fallthrough to nat46 and stack, (b)
* the packet needs IPSec encap so push ctx to stack for encap, or
* (c) packet was redirected to tunnel device so return.
*/
ret = encap_and_redirect_lxc(ctx, tunnel_endpoint, encrypt_key,
&key, SECLABEL, monitor);
if (ret == IPSEC_ENDPOINT)
goto encrypt_to_stack;
else if (ret != DROP_NO_TUNNEL_ENDPOINT)
return ret;
}
#endif
#ifdef ENABLE_NAT46
if (unlikely(ipv6_addr_is_mapped(daddr))) {
ep_tail_call(ctx, CILIUM_CALL_NAT64);
return DROP_MISSED_TAIL_CALL;
}
#endif
if (is_defined(ENABLE_REDIRECT_FAST))
return redirect_direct_v6(ctx, l3_off, ip6);
goto pass_to_stack;
#ifdef ENABLE_ROUTING
to_host:
if (is_defined(HOST_REDIRECT_TO_INGRESS) ||
(is_defined(ENABLE_HOST_FIREWALL) && *dstID == HOST_ID)) {
if (is_defined(HOST_REDIRECT_TO_INGRESS)) {
union macaddr host_mac = HOST_IFINDEX_MAC;
ret = ipv6_l3(ctx, l3_off, (__u8 *)&router_mac.addr,
(__u8 *)&host_mac.addr, METRIC_EGRESS);
if (ret != CTX_ACT_OK)
return ret;
}
send_trace_notify(ctx, TRACE_TO_HOST, SECLABEL, HOST_ID, 0,
HOST_IFINDEX, reason, monitor);
return redirect(HOST_IFINDEX, BPF_F_INGRESS);
}
#endif
pass_to_stack:
#ifdef ENABLE_ROUTING
ret = ipv6_l3(ctx, l3_off, NULL, (__u8 *) &router_mac.addr, METRIC_EGRESS);
if (unlikely(ret != CTX_ACT_OK))
return ret;
#endif
if (ipv6_store_flowlabel(ctx, l3_off, SECLABEL_NB) < 0)
return DROP_WRITE_ERROR;
#ifndef ENCAP_IFINDEX
#ifdef ENABLE_IPSEC
if (encrypt_key && tunnel_endpoint) {
set_encrypt_key_mark(ctx, encrypt_key);
#ifdef IP_POOLS
set_encrypt_dip(ctx, tunnel_endpoint);
#endif
} else
#endif
#endif
{
#ifdef ENABLE_IDENTITY_MARK
/* Always encode the source identity when passing to the stack.
* If the stack hairpins the packet back to a local endpoint the
* source identity can still be derived even if SNAT is
* performed by a component such as portmap.
*/
ctx->mark |= MARK_MAGIC_IDENTITY;
set_identity_mark(ctx, SECLABEL);
#endif
}
#ifdef ENCAP_IFINDEX
encrypt_to_stack:
#endif
send_trace_notify(ctx, TRACE_TO_STACK, SECLABEL, *dstID, 0, 0,
reason, monitor);
cilium_dbg_capture(ctx, DBG_CAPTURE_DELIVERY, 0);
return CTX_ACT_OK;
}
static __always_inline int handle_ipv6(struct __ctx_buff *ctx, __u32 *dstID)
{
struct ipv6_ct_tuple tuple = {};
void *data, *data_end;
struct ipv6hdr *ip6;
int ret;
if (!revalidate_data(ctx, &data, &data_end, &ip6))
return DROP_INVALID;
/* Handle special ICMPv6 messages. This includes echo requests to the
* logical router address, neighbour advertisements to the router.
* All remaining packets are subjected to forwarding into the container.
*/
if (unlikely(ip6->nexthdr == IPPROTO_ICMPV6)) {
if (data + sizeof(*ip6) + ETH_HLEN + sizeof(struct icmp6hdr) > data_end)
return DROP_INVALID;
ret = icmp6_handle(ctx, ETH_HLEN, ip6, METRIC_EGRESS);
if (IS_ERR(ret))
return ret;
}
/* Perform L3 action on the frame */
tuple.nexthdr = ip6->nexthdr;
return ipv6_l3_from_lxc(ctx, &tuple, ETH_HLEN, ip6, dstID);
}
declare_tailcall_if(__or(__and(is_defined(ENABLE_IPV4), is_defined(ENABLE_IPV6)),
is_defined(DEBUG)), CILIUM_CALL_IPV6_FROM_LXC)
int tail_handle_ipv6(struct __ctx_buff *ctx)
{
__u32 dstID = 0;
int ret = handle_ipv6(ctx, &dstID);
if (IS_ERR(ret)) {
return send_drop_notify(ctx, SECLABEL, dstID, 0, ret,
CTX_ACT_DROP, METRIC_EGRESS);
}
return ret;
}
#endif /* ENABLE_IPV6 */
#ifdef ENABLE_IPV4
static __always_inline int handle_ipv4_from_lxc(struct __ctx_buff *ctx,
__u32 *dstID)
{
struct ipv4_ct_tuple tuple = {};
#ifdef ENABLE_ROUTING
union macaddr router_mac = NODE_MAC;
#endif
void *data, *data_end;
struct iphdr *ip4;
int ret, verdict, l3_off = ETH_HLEN, l4_off;
struct csum_offset csum_off = {};
struct ct_state ct_state_new = {};
struct ct_state ct_state = {};
__be32 orig_dip;
__u32 tunnel_endpoint = 0;
__u8 encrypt_key = 0;
__u32 monitor = 0;
__u8 reason;
bool hairpin_flow = false; /* endpoint wants to access itself via service IP */
__u8 policy_match_type = POLICY_MATCH_NONE;
__u8 audited = 0;
bool has_l4_header = false;
if (!revalidate_data(ctx, &data, &data_end, &ip4))
return DROP_INVALID;
has_l4_header = ipv4_has_l4_header(ip4);
tuple.nexthdr = ip4->protocol;
if (unlikely(!is_valid_lxc_src_ipv4(ip4)))
return DROP_INVALID_SIP;
tuple.daddr = ip4->daddr;
tuple.saddr = ip4->saddr;
l4_off = l3_off + ipv4_hdrlen(ip4);
#ifndef ENABLE_HOST_SERVICES_FULL
{
struct lb4_service *svc;
struct lb4_key key = {};
ret = lb4_extract_key(ctx, ip4, l4_off, &key, &csum_off,
CT_EGRESS);
if (IS_ERR(ret)) {
if (ret == DROP_UNKNOWN_L4)
goto skip_service_lookup;
else
return ret;
}
svc = lb4_lookup_service(&key, is_defined(ENABLE_NODEPORT));
if (svc) {
ret = lb4_local(get_ct_map4(&tuple), ctx, l3_off, l4_off,
&csum_off, &key, &tuple, svc, &ct_state_new,
ip4->saddr, has_l4_header);
if (IS_ERR(ret))
return ret;
hairpin_flow |= ct_state_new.loopback;
}
}
skip_service_lookup:
#endif /* !ENABLE_HOST_SERVICES_FULL */
/* The verifier wants to see this assignment here in case the above goto
* skip_service_lookup is hit. However, in the case the packet
* is _not_ TCP or UDP we should not be using proxy logic anyways. For
* correctness it must be below the service handler in case the service
* logic re-writes the tuple daddr. In "theory" however the assignment
* should be OK to move above goto label.
*/
orig_dip = tuple.daddr;
/* WARNING: ip4 offset check invalidated, revalidate before use */
/* Pass all outgoing packets through conntrack. This will create an
* entry to allow reverse packets and return set cb[CB_POLICY] to
* POLICY_SKIP if the packet is a reply packet to an existing incoming
* connection.
*/
ret = ct_lookup4(get_ct_map4(&tuple), &tuple, ctx, l4_off, CT_EGRESS,
&ct_state, &monitor);
if (ret < 0)
return ret;
reason = ret;
/* Check it this is return traffic to an ingress proxy. */
if ((ret == CT_REPLY || ret == CT_RELATED) && ct_state.proxy_redirect) {
/* Stack will do a socket match and deliver locally. */
return ctx_redirect_to_proxy4(ctx, &tuple, 0, false);
}
/* Determine the destination category for policy fallback. */
if (1) {
struct remote_endpoint_info *info;
info = lookup_ip4_remote_endpoint(orig_dip);
if (info != NULL && info->sec_label) {
*dstID = info->sec_label;
tunnel_endpoint = info->tunnel_endpoint;
encrypt_key = get_min_encrypt_key(info->key);
} else {
*dstID = WORLD_ID;
}
cilium_dbg(ctx, info ? DBG_IP_ID_MAP_SUCCEED4 : DBG_IP_ID_MAP_FAILED4,
orig_dip, *dstID);
}
/* If the packet is in the establishing direction and it's destined
* within the cluster, it must match policy or be dropped. If it's
* bound for the host/outside, perform the CIDR policy check.
*/
verdict = policy_can_egress4(ctx, &tuple, SECLABEL, *dstID,
&policy_match_type, &audited);
if (ret != CT_REPLY && ret != CT_RELATED && verdict < 0) {
send_policy_verdict_notify(ctx, *dstID, tuple.dport,
tuple.nexthdr, POLICY_EGRESS, 0,
verdict, policy_match_type, audited);
return verdict;
}
switch (ret) {
case CT_NEW:
send_policy_verdict_notify(ctx, *dstID, tuple.dport,
tuple.nexthdr, POLICY_EGRESS, 0,
verdict, policy_match_type, audited);
ct_recreate4:
/* New connection implies that rev_nat_index remains untouched
* to the index provided by the loadbalancer (if it applied).
* Create a CT entry which allows to track replies and to
* reverse NAT.
*/
ct_state_new.src_sec_id = SECLABEL;
/* We could avoid creating related entries for legacy ClusterIP
* handling here, but turns out that verifier cannot handle it.
*/
ret = ct_create4(get_ct_map4(&tuple), &CT_MAP_ANY4, &tuple, ctx,
CT_EGRESS, &ct_state_new, verdict > 0);
if (IS_ERR(ret))
return ret;
break;
case CT_REOPENED:
send_policy_verdict_notify(ctx, *dstID, tuple.dport,
tuple.nexthdr, POLICY_EGRESS, 0,
verdict, policy_match_type, audited);
case CT_ESTABLISHED:
/* Did we end up at a stale non-service entry? Recreate if so. */
if (unlikely(ct_state.rev_nat_index != ct_state_new.rev_nat_index))
goto ct_recreate4;
break;
case CT_RELATED:
case CT_REPLY:
policy_mark_skip(ctx);
#ifdef ENABLE_NODEPORT
/* This handles reply traffic for the case where the nodeport EP
* is local to the node. We'll redirect to bpf_host egress to
* perform the reverse DNAT.
*/
if (ct_state.node_port) {
ctx->tc_index |= TC_INDEX_F_SKIP_RECIRCULATION;
ep_tail_call(ctx, CILIUM_CALL_IPV4_NODEPORT_REVNAT);
return DROP_MISSED_TAIL_CALL;
}
# ifdef ENABLE_DSR
if (ct_state.dsr) {
ret = xlate_dsr_v4(ctx, &tuple, l4_off, has_l4_header);
if (ret != 0)
return ret;
}
# endif /* ENABLE_DSR */
#endif /* ENABLE_NODEPORT */
if (ct_state.rev_nat_index) {
ret = lb4_rev_nat(ctx, l3_off, l4_off, &csum_off,
&ct_state, &tuple, 0, has_l4_header);
if (IS_ERR(ret))
return ret;
}
break;
default:
return DROP_UNKNOWN_CT;
}
hairpin_flow |= ct_state.loopback;
if (redirect_to_proxy(verdict, reason)) {
/* Trace the packet before it is forwarded to proxy */
send_trace_notify(ctx, TRACE_TO_PROXY, SECLABEL, 0,
0, 0, reason, monitor);
return ctx_redirect_to_proxy4(ctx, &tuple, verdict, false);
}
/* After L4 write in port mapping: revalidate for direct packet access */
if (!revalidate_data(ctx, &data, &data_end, &ip4))
return DROP_INVALID;
orig_dip = ip4->daddr;
/* Allow a hairpin packet to be redirected even if ENABLE_ROUTING is
* disabled. Otherwise, the packet will be dropped by the kernel if
* it is going to be routed via an interface it came from after it has
* been passed to the stack.
*/
if (is_defined(ENABLE_ROUTING) || hairpin_flow) {
struct endpoint_info *ep;
/* Lookup IPv4 address, this will return a match if:
* - The destination IP address belongs to a local endpoint
* managed by cilium
* - The destination IP address is an IP address associated with the
* host itself
* - The destination IP address belongs to endpoint itself.
*/
ep = lookup_ip4_endpoint(ip4);
if (ep) {
#ifdef ENABLE_ROUTING
if (ep->flags & ENDPOINT_F_HOST) {
#ifdef HOST_IFINDEX
goto to_host;
#else
return DROP_HOST_UNREACHABLE;
#endif
}
#endif /* ENABLE_ROUTING */
policy_clear_mark(ctx);
return ipv4_local_delivery(ctx, l3_off, SECLABEL, ip4,
ep, METRIC_EGRESS, false);
}
}
#ifdef ENCAP_IFINDEX
{
struct endpoint_key key = {};
key.ip4 = orig_dip & IPV4_MASK;
key.family = ENDPOINT_KEY_IPV4;
ret = encap_and_redirect_lxc(ctx, tunnel_endpoint, encrypt_key,
&key, SECLABEL, monitor);
if (ret == DROP_NO_TUNNEL_ENDPOINT)
goto pass_to_stack;
/* If not redirected noteably due to IPSEC then pass up to stack
* for further processing.
*/
else if (ret == IPSEC_ENDPOINT)
goto encrypt_to_stack;
/* This is either redirect by encap code or an error has
* occurred either way return and stack will consume ctx.
*/
else
return ret;
}
#endif
if (is_defined(ENABLE_REDIRECT_FAST))
return redirect_direct_v4(ctx, l3_off, ip4);
goto pass_to_stack;
#ifdef ENABLE_ROUTING
to_host:
if (is_defined(HOST_REDIRECT_TO_INGRESS) ||
(is_defined(ENABLE_HOST_FIREWALL) && *dstID == HOST_ID)) {
if (is_defined(HOST_REDIRECT_TO_INGRESS)) {
union macaddr host_mac = HOST_IFINDEX_MAC;
ret = ipv4_l3(ctx, l3_off, (__u8 *)&router_mac.addr,
(__u8 *)&host_mac.addr, ip4);
if (ret != CTX_ACT_OK)
return ret;
}
send_trace_notify(ctx, TRACE_TO_HOST, SECLABEL, HOST_ID, 0,
HOST_IFINDEX, reason, monitor);
return redirect(HOST_IFINDEX, BPF_F_INGRESS);
}
#endif
pass_to_stack:
#ifdef ENABLE_ROUTING
ret = ipv4_l3(ctx, l3_off, NULL, (__u8 *) &router_mac.addr, ip4);
if (unlikely(ret != CTX_ACT_OK))
return ret;
#endif
#ifndef ENCAP_IFINDEX
#ifdef ENABLE_IPSEC
if (encrypt_key && tunnel_endpoint) {
set_encrypt_key_mark(ctx, encrypt_key);
#ifdef IP_POOLS
set_encrypt_dip(ctx, tunnel_endpoint);
#endif
} else
#endif
#endif
{
#ifdef ENABLE_IDENTITY_MARK
/* Always encode the source identity when passing to the stack.
* If the stack hairpins the packet back to a local endpoint the
* source identity can still be derived even if SNAT is
* performed by a component such as portmap.
*/
ctx->mark |= MARK_MAGIC_IDENTITY;
set_identity_mark(ctx, SECLABEL);
#endif
}
#ifdef ENCAP_IFINDEX
encrypt_to_stack:
#endif
send_trace_notify(ctx, TRACE_TO_STACK, SECLABEL, *dstID, 0, 0,
reason, monitor);
cilium_dbg_capture(ctx, DBG_CAPTURE_DELIVERY, 0);
return CTX_ACT_OK;
}
declare_tailcall_if(__or(__and(is_defined(ENABLE_IPV4), is_defined(ENABLE_IPV6)),
is_defined(DEBUG)), CILIUM_CALL_IPV4_FROM_LXC)
int tail_handle_ipv4(struct __ctx_buff *ctx)
{
__u32 dstID = 0;
int ret = handle_ipv4_from_lxc(ctx, &dstID);
if (IS_ERR(ret))
return send_drop_notify(ctx, SECLABEL, dstID, 0, ret,
CTX_ACT_DROP, METRIC_EGRESS);
return ret;
}
#ifdef ENABLE_ARP_RESPONDER
/*
* ARP responder for ARP requests from container
* Respond to IPV4_GATEWAY with NODE_MAC
*/
__section_tail(CILIUM_MAP_CALLS, CILIUM_CALL_ARP)
int tail_handle_arp(struct __ctx_buff *ctx)
{
union macaddr mac = NODE_MAC;
union macaddr smac;
__be32 sip;
__be32 tip;
/* Pass any unknown ARP requests to the Linux stack */
if (!arp_validate(ctx, &mac, &smac, &sip, &tip))
return CTX_ACT_OK;
/*
* The endpoint is expected to make ARP requests for its gateway IP.
* Most of the time, the gateway IP configured on the endpoint is
* IPV4_GATEWAY but it may not be the case if after cilium agent reload
* a different gateway is chosen. In such a case, existing endpoints
* will have an old gateway configured. Since we don't know the IP of
* previous gateways, we answer requests for all IPs with the exception
* of the LXC IP (to avoid specific problems, like IP duplicate address
* detection checks that might run within the container).
*/
if (tip == LXC_IPV4)
return CTX_ACT_OK;
return arp_respond(ctx, &mac, tip, &smac, sip, 0);
}
#endif /* ENABLE_ARP_RESPONDER */
#endif /* ENABLE_IPV4 */
/* Attachment/entry point is ingress for veth, egress for ipvlan. */
__section("from-container")
int handle_xgress(struct __ctx_buff *ctx)
{
__u16 proto;
int ret;
bpf_clear_meta(ctx);
edt_set_aggregate(ctx, LXC_ID);
send_trace_notify(ctx, TRACE_FROM_LXC, SECLABEL, 0, 0, 0, 0,
TRACE_PAYLOAD_LEN);
if (!validate_ethertype(ctx, &proto)) {
ret = DROP_UNSUPPORTED_L2;
goto out;
}
switch (proto) {
#ifdef ENABLE_IPV6
case bpf_htons(ETH_P_IPV6):
invoke_tailcall_if(__or(__and(is_defined(ENABLE_IPV4), is_defined(ENABLE_IPV6)),
is_defined(DEBUG)),
CILIUM_CALL_IPV6_FROM_LXC, tail_handle_ipv6);
break;
#endif /* ENABLE_IPV6 */
#ifdef ENABLE_IPV4
case bpf_htons(ETH_P_IP):
invoke_tailcall_if(__or(__and(is_defined(ENABLE_IPV4), is_defined(ENABLE_IPV6)),
is_defined(DEBUG)),
CILIUM_CALL_IPV4_FROM_LXC, tail_handle_ipv4);
break;
#ifdef ENABLE_ARP_PASSTHROUGH
case bpf_htons(ETH_P_ARP):
ret = CTX_ACT_OK;
break;
#elif defined(ENABLE_ARP_RESPONDER)
case bpf_htons(ETH_P_ARP):
ep_tail_call(ctx, CILIUM_CALL_ARP);
ret = DROP_MISSED_TAIL_CALL;
break;
#endif /* ENABLE_ARP_RESPONDER */
#endif /* ENABLE_IPV4 */
default:
ret = DROP_UNKNOWN_L3;
}
out:
if (IS_ERR(ret))
return send_drop_notify(ctx, SECLABEL, 0, 0, ret, CTX_ACT_DROP,
METRIC_EGRESS);
return ret;
}
#ifdef ENABLE_IPV6
static __always_inline int
ipv6_policy(struct __ctx_buff *ctx, int ifindex, __u32 src_label, __u8 *reason,
struct ipv6_ct_tuple *tuple_out, __u16 *proxy_port, bool from_host)
{
struct ipv6_ct_tuple tuple = {};
void *data, *data_end;
struct ipv6hdr *ip6;
struct csum_offset csum_off = {};
int ret, l4_off, verdict, hdrlen;
struct ct_state ct_state = {};
struct ct_state ct_state_new = {};
bool skip_ingress_proxy = false;
union v6addr orig_sip;
__u32 monitor = 0;
__u8 policy_match_type = POLICY_MATCH_NONE;
__u8 audited = 0;
if (!revalidate_data(ctx, &data, &data_end, &ip6))
return DROP_INVALID;
policy_clear_mark(ctx);
tuple.nexthdr = ip6->nexthdr;
ipv6_addr_copy(&tuple.daddr, (union v6addr *) &ip6->daddr);
ipv6_addr_copy(&tuple.saddr, (union v6addr *) &ip6->saddr);
ipv6_addr_copy(&orig_sip, (union v6addr *) &ip6->saddr);
/* If packet is coming from the ingress proxy we have to skip
* redirection to the ingress proxy as we would loop forever.
*/
skip_ingress_proxy = tc_index_skip_ingress_proxy(ctx);
hdrlen = ipv6_hdrlen(ctx, ETH_HLEN, &tuple.nexthdr);
if (hdrlen < 0)
return hdrlen;
l4_off = ETH_HLEN + hdrlen;
csum_l4_offset_and_flags(tuple.nexthdr, &csum_off);
ret = ct_lookup6(get_ct_map6(&tuple), &tuple, ctx, l4_off, CT_INGRESS,
&ct_state, &monitor);
if (ret < 0)
return ret;
*reason = ret;
/* Check it this is return traffic to an egress proxy.
* Do not redirect again if the packet is coming from the egress proxy.
*/
if ((ret == CT_REPLY || ret == CT_RELATED) && ct_state.proxy_redirect &&
!tc_index_skip_egress_proxy(ctx)) {
/* This is a reply, the proxy port does not need to be embedded
* into ctx->mark and *proxy_port can be left unset.
*/
send_trace_notify6(ctx, TRACE_TO_PROXY, src_label, SECLABEL, &orig_sip,
0, ifindex, 0, monitor);
if (tuple_out)
memcpy(tuple_out, &tuple, sizeof(tuple));
return POLICY_ACT_PROXY_REDIRECT;
}
if (unlikely(ct_state.rev_nat_index)) {
int ret2;
ret2 = lb6_rev_nat(ctx, l4_off, &csum_off,
ct_state.rev_nat_index, &tuple, 0);
if (IS_ERR(ret2))
return ret2;
}
verdict = policy_can_access_ingress(ctx, src_label, SECLABEL,
tuple.dport, tuple.nexthdr, false,
&policy_match_type, &audited);
/* Reply packets and related packets are allowed, all others must be
* permitted by policy.
*/
if (ret != CT_REPLY && ret != CT_RELATED && verdict < 0) {
send_policy_verdict_notify(ctx, src_label, tuple.dport,
tuple.nexthdr, POLICY_INGRESS, 1,
verdict, policy_match_type, audited);
return verdict;
}
if (skip_ingress_proxy)
verdict = 0;
if (ret == CT_NEW || ret == CT_REOPENED) {
send_policy_verdict_notify(ctx, src_label, tuple.dport,
tuple.nexthdr, POLICY_INGRESS, 1,
verdict, policy_match_type, audited);
}
if (ret == CT_NEW) {
#ifdef ENABLE_DSR
{
bool dsr = false;
ret = handle_dsr_v6(ctx, &dsr);
if (ret != 0)
return ret;
ct_state_new.dsr = dsr;
}
#endif /* ENABLE_DSR */
ct_state_new.src_sec_id = src_label;
ct_state_new.node_port = ct_state.node_port;
ct_state_new.ifindex = ct_state.ifindex;
ret = ct_create6(get_ct_map6(&tuple), &CT_MAP_ANY6, &tuple, ctx, CT_INGRESS,
&ct_state_new, verdict > 0);
if (IS_ERR(ret))
return ret;
/* NOTE: tuple has been invalidated after this */
}
if (!revalidate_data(ctx, &data, &data_end, &ip6))
return DROP_INVALID;
if (redirect_to_proxy(verdict, *reason)) {
*proxy_port = verdict;
send_trace_notify6(ctx, TRACE_TO_PROXY, src_label, SECLABEL, &orig_sip,
0, ifindex, *reason, monitor);
if (tuple_out)
memcpy(tuple_out, &tuple, sizeof(tuple));
return POLICY_ACT_PROXY_REDIRECT;
}
/* Not redirected to host / proxy. */
send_trace_notify6(ctx, TRACE_TO_LXC, src_label, SECLABEL, &orig_sip,
LXC_ID, ifindex, *reason, monitor);