Repository navigation
Expand file tree
/
Copy pathping.c
More file actions
1836 lines (1567 loc) · 62.3 KB
/
Copy pathping.c
File metadata and controls
1836 lines (1567 loc) · 62.3 KB
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
/*
ex: set tabstop=4 shiftwidth=4 autoindent:
+-------------------------------------------------------------------------+
| Copyright (C) 2004-2026 The Cacti Group |
| |
| This program is free software; you can redistribute it and/or |
| modify it under the terms of the GNU Lesser General Public |
| License as published by the Free Software Foundation; either |
| version 2.1 of the License, or (at your option) any later version. |
| |
| This program is distributed in the hope that it will be useful, |
| but WITHOUT ANY WARRANTY; without even the implied warranty of |
| MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| GNU Lesser General Public License for more details. |
| |
| You should have received a copy of the GNU Lesser General Public |
| License along with this library; if not, write to the Free Software |
| Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA |
| 02110-1301, USA |
| |
+-------------------------------------------------------------------------+
| spine: a backend data gatherer for cacti |
+-------------------------------------------------------------------------+
| This poller would not have been possible without: |
| - Larry Adams (current development and enhancements) |
| - Rivo Nurges (rrd support, mysql poller cache, misc functions) |
| - RTG (core poller code, pthreads, snmp, autoconf examples) |
| - Brady Alleman/Doug Warner (threading ideas, implementation details) |
+-------------------------------------------------------------------------+
| - Cacti - http://www.cacti.net/ |
+-------------------------------------------------------------------------+
*/
#include "common.h"
#include "spine.h"
#include <fcntl.h>
#ifdef SPINE_HAVE_ICMPV6
static int ping_icmp_ipv6(const host_t *host, ping_t *ping);
#endif
/*! \fn int ping_host(host_t *host, ping_t *ping)
* \brief ping a host to determine if it is reachable for polling
* \param host a pointer to the current host structure
* \param ping a pointer to the current hosts ping structure
*
* This function pings a host using the method specified within the system
* configuration and then returns the host status to the calling function.
*
* \return HOST_UP if the host is reachable, HOST_DOWN otherwise.
*/
static int ping_network(host_t *host, ping_t *ping) {
if (host->availability.ping_method == PING_ICMP && !set.availability.icmp_avail) {
SPINE_LOG(("Device[%i] DEBUG Falling back to UDP Ping Due to SetUID Issues", host->id));
host->availability.ping_method = PING_UDP;
}
if (strstr(host->hostname, "localhost")) {
STRNCOPY(ping->ping_status, "0.000");
STRNCOPY(ping->ping_response, "PING: Device does not require ping.");
return HOST_UP;
}
int address_type = get_address_type(host);
#ifdef SPINE_HAVE_ICMPV6
if (address_type == SPINE_IPV6 && host->availability.ping_method == PING_ICMP) {
return ping_icmp_ipv6(host, ping);
}
#endif
if (address_type != SPINE_IPV4) {
if (host->availability.method == AVAIL_PING) {
STRNCOPY(ping->ping_status, "0.000");
STRNCOPY(ping->ping_response, "PING: Device is Unknown or is IPV6. Please use the SNMP ping options only.");
}
return HOST_DOWN;
}
switch (host->availability.ping_method) {
case PING_ICMP: return ping_icmp(host, ping);
case PING_UDP: return ping_udp(host, ping);
case PING_TCP:
case PING_TCP_CLOSED: return ping_tcp(host, ping);
default: return HOST_DOWN;
}
}
static int ping_snmp_availability(host_t *host, ping_t *ping, int ping_result) {
if (host->availability.method == AVAIL_SNMP_AND_PING && ping_result != HOST_UP) return HOST_DOWN;
if (host->availability.method == AVAIL_SNMP_OR_PING && ping_result == HOST_UP) return HOST_UP;
/* Preserve the configured no-SNMP contract for v1/v2 without a community. */
if (host->snmp.profile.community[0] == '\0' && host->snmp.profile.version < 3) return HOST_UP;
double begin = get_time_as_double();
int result = ping_snmp(host, ping);
double elapsed = (get_time_as_double() - begin) * 1000.0;
SPINE_LOG_DEVICE(host->id, POLLER_VERBOSITY_MEDIUM,
("Device[%i] INFO: SNMP Device %s, Time:%.4f ms", host->id, result == HOST_UP ? "Alive" : "Down", elapsed));
return result;
}
int ping_host(host_t *host, ping_t *ping) {
int network_result = HOST_DOWN;
if (host->availability.method == AVAIL_SNMP_AND_PING ||
host->availability.method == AVAIL_PING ||
host->availability.method == AVAIL_SNMP_OR_PING) {
network_result = ping_network(host, ping);
}
switch (host->availability.method) {
case AVAIL_SNMP_AND_PING:
case AVAIL_SNMP_OR_PING:
case AVAIL_SNMP:
case AVAIL_SNMP_GET_NEXT:
case AVAIL_SNMP_GET_SYSDESC:
return ping_snmp_availability(host, ping, network_result);
case AVAIL_PING: return network_result;
case AVAIL_NONE: return HOST_UP;
default: return HOST_DOWN;
}
}
/*! \fn int ping_snmp(host_t *host, ping_t *ping)
* \brief ping a host using snmp sysUptime
* \param host a pointer to the current host structure
* \param ping a pointer to the current hosts ping structure
*
* This function pings a host using snmp. It polls sysUptime by default.
* It will modify the ping structure to include the specifics of the ping results.
*
* \return HOST_UP if the host is reachable, HOST_DOWN otherwise.
*
*/
int ping_snmp(host_t *host, ping_t *ping) {
SPINE_LOG_DEVICE(host->id, POLLER_VERBOSITY_DEBUG, ("Device[%i] DEBUG: Entering SNMP Ping", host->id));
if (!host->snmp.session) {
STRNCOPY(ping->snmp_status, "0.00");
STRNCOPY(ping->snmp_response, "Invalid SNMP Session");
return HOST_DOWN;
}
if (host->snmp.profile.community[0] == '\0' && host->snmp.profile.version != 3) {
STRNCOPY(ping->snmp_status, "0.00");
STRNCOPY(ping->snmp_response, "Device does not require SNMP");
return HOST_UP;
}
char oid[32];
switch (host->availability.method) {
case AVAIL_SNMP_GET_NEXT: STRNCOPY(oid, ".1.3"); break;
case AVAIL_SNMP_GET_SYSDESC: STRNCOPY(oid, ".1.3.6.1.2.1.1.1.0"); break;
default: STRNCOPY(oid, ".1.3.6.1.2.1.1.3.0"); break;
}
double begin = get_time_as_double();
char *result = host->availability.method == AVAIL_SNMP_GET_NEXT ? snmp_getnext(host, oid) : snmp_get(host, oid);
double elapsed = (get_time_as_double() - begin) * 1000.0;
SPINE_FREE(result);
if (host->snmp.status == SNMPERR_SUCCESS || host->snmp.status == SNMPERR_UNKNOWN_OBJID) {
STRNCOPY(ping->snmp_response, "Device responded to SNMP");
spine_snprintf(ping->snmp_status, sizeof(ping->snmp_status), "%.5f", elapsed);
return HOST_UP;
}
SPINE_LOG_DEVICE(host->id, POLLER_VERBOSITY_HIGH,
("Device[%i] SNMP Ping %s", host->id, host->snmp.status == STAT_TIMEOUT ? "Timeout" : "Unknown Error"));
STRNCOPY(ping->snmp_response, "Device did not respond to SNMP");
return HOST_DOWN;
}
static int ping_down(ping_t *ping, const char *message);
#ifdef __CYGWIN__
/*! \fn static void icmp_discard_reply(int icmp_socket, char *buffer)
* \brief consumes a datagram that ping_icmp() has only peeked at
*
* Cygwin cannot use MSG_WAITALL on a raw socket, so ping_icmp() reads with
* MSG_PEEK. A peeked datagram stays queued, so a reply we decline has to be
* drained here. Otherwise the next pass reads the same bytes again and the
* loop spins until the device timeout expires.
*/
static void icmp_discard_reply(int icmp_socket, char *buffer) {
while (recvfrom(icmp_socket, buffer, BUFSIZE, 0, NULL, NULL) < 0 && errno == EINTR) {
/* interrupted before the datagram was removed, try again */
}
}
#define ICMP_DISCARD_PEEKED(sock, buf) icmp_discard_reply((sock), (buf))
#else
#define ICMP_DISCARD_PEEKED(sock, buf) ((void)0)
#endif
/*! \fn spine_icmp_reply_t spine_icmp_classify_reply(const unsigned char *reply, ssize_t len, uint16_t want_id, uint16_t want_seq, const struct icmp **out_pkt)
* \brief bounds-check a raw ICMP reply and decide whether it answers our echo
*
* The raw socket is shared across every poller thread and is fed by whatever
* the network sends, so this walks the IP header length field before touching
* the ICMP header. Split out of ping_icmp() so the fuzz target exercises the
* same code the poller runs.
*
* \return the reply classification; *out_pkt is set only for SPINE_ICMP_REPLY_OK
*/
spine_icmp_reply_t spine_icmp_classify_reply(const unsigned char *reply, ssize_t len,
uint16_t want_id, uint16_t want_seq, const struct icmp **out_pkt) {
size_t ihl;
uint16_t received_id;
uint16_t received_seq;
if (out_pkt != NULL) {
*out_pkt = NULL;
}
if (reply == NULL || len < (ssize_t) sizeof(struct ip)) {
return SPINE_ICMP_REPLY_TOO_SHORT;
}
/* Read header fields as bytes: the receive buffer has no alignment
* guarantee, and a raw socket can deliver a non-IPv4 or non-ICMP datagram. */
ihl = (size_t) (reply[0] & 0x0F) << 2;
if ((reply[0] >> 4) != 4 || reply[9] != IPPROTO_ICMP ||
ihl < sizeof(struct ip) || len < (ssize_t) (ihl + ICMP_HDR_SIZE)) {
return SPINE_ICMP_REPLY_BAD_HEADER;
}
/* An echo reply always carries code 0. */
if (reply[ihl] != ICMP_ECHOREPLY || reply[ihl + 1] != 0) {
return SPINE_ICMP_REPLY_NOT_ECHO;
}
memcpy(&received_id, reply + ihl + 4, sizeof(received_id));
memcpy(&received_seq, reply + ihl + 6, sizeof(received_seq));
if (received_id != want_id || received_seq != want_seq) {
return SPINE_ICMP_REPLY_NOT_OURS;
}
if (out_pkt != NULL) {
*out_pkt = (const struct icmp *) (reply + ihl);
}
return SPINE_ICMP_REPLY_OK;
}
/*! \fn spine_icmp_reply_t spine_icmp_classify_dgram_reply(const unsigned char *reply, ssize_t len, uint16_t want_seq, const struct icmp **out_pkt)
* \brief classifies an echo reply read from an unprivileged datagram socket
*
* A SOCK_DGRAM ICMP socket delivers the reply with the IPv4 header already
* stripped, and the kernel assigns the echo id rather than honouring the one
* we wrote, so the sequence is all that is left to match on. The caller has
* already checked the source address.
*/
spine_icmp_reply_t spine_icmp_classify_dgram_reply(const unsigned char *reply, ssize_t len,
uint16_t want_seq, const struct icmp **out_pkt) {
const struct icmp *pkt;
uint16_t received_seq;
if (out_pkt != NULL) {
*out_pkt = NULL;
}
if (reply == NULL || len < (ssize_t) ICMP_HDR_SIZE) {
return SPINE_ICMP_REPLY_TOO_SHORT;
}
pkt = (const struct icmp *) reply;
/* Fields are read as bytes; the receive buffer has no alignment guarantee. */
if (reply[0] != ICMP_ECHOREPLY || reply[1] != 0) {
return SPINE_ICMP_REPLY_NOT_ECHO;
}
memcpy(&received_seq, reply + 6, sizeof(received_seq));
if (received_seq != want_seq) {
return SPINE_ICMP_REPLY_NOT_OURS;
}
if (out_pkt != NULL) {
*out_pkt = pkt;
}
return SPINE_ICMP_REPLY_OK;
}
#ifdef SPINE_HAVE_ICMPV6
static const char icmp6_payload[] = "cacti-monitoring-system";
/*! \fn static int icmp6_reply_matches(const unsigned char *reply, ssize_t len, uint16_t id, uint16_t seq, int check_id)
* \brief decides whether an ICMPv6 message answers our echo request
*
* ICMPv6 sockets deliver the message without the IPv6 header. id and seq
* are in network order, as written into the request. A datagram socket
* rewrites the echo id, so its caller passes check_id FALSE.
*/
static int icmp6_reply_matches(const unsigned char *reply, ssize_t len, uint16_t id, uint16_t seq, int check_id) {
struct icmp6_hdr hdr;
size_t payload_len = sizeof(icmp6_payload) - 1;
if (len < 0 || (size_t) len < sizeof(hdr) + payload_len) {
return FALSE;
}
memcpy(&hdr, reply, sizeof(hdr));
if (hdr.icmp6_type != ICMP6_ECHO_REPLY || hdr.icmp6_seq != seq) {
return FALSE;
}
if (check_id && hdr.icmp6_id != id) {
return FALSE;
}
return memcmp(reply + sizeof(hdr), icmp6_payload, payload_len) == 0;
}
#endif
/* Raw sockets opened by drop_privileges() before a setuid root spine gave up
* root for good. Without CAP_NET_RAW nothing can open another, so every
* poller thread shares them. A raw socket queues each ICMP message once, so a
* thread that reads a reply meant for another thread must hand it over rather
* than drop it: one thread at a time reads, matches each reply against the
* registered requests, and wakes only the thread it answers. LOCK_ICMP guards
* all of it. */
typedef struct {
int fd;
int reading;
} icmp_shared_t;
typedef struct icmp_waiter {
struct icmp_waiter *next;
icmp_shared_t *shared;
pthread_cond_t wake;
int family;
uint16_t id;
uint16_t seq;
struct in_addr peer;
#ifdef SPINE_HAVE_ICMPV6
struct in6_addr peer6;
#endif
int answered;
int sleeping;
} icmp_waiter_t;
/* A reader empties the queue in batches read without LOCK_ICMP held, so a
* flood delays neither registrations nor the waiters being woken. The bound
* on one pass keeps the reader's own deadline check running. Matching needs
* only the IP header, the ICMP header and the payload marker, so each slot
* keeps the first ICMP_SHARED_SNAPLEN bytes. The queue itself is sized for
* every poller thread rather than one. */
#define ICMP_SHARED_DRAIN_MAX 1024
#define ICMP_SHARED_BATCH 64
#define ICMP_SHARED_SNAPLEN 128
#define ICMP_SHARED_RCVBUF (4 * 1024 * 1024)
#if defined(__linux__)
#ifndef SOL_RAW
#define SOL_RAW 255
#endif
#ifndef ICMP_FILTER
#define ICMP_FILTER 1
#endif
#endif
static icmp_shared_t icmp_shared = {-1, FALSE};
#ifdef SPINE_HAVE_ICMPV6
static icmp_shared_t icmp6_shared = {-1, FALSE};
#endif
static icmp_waiter_t *icmp_waiters = NULL;
/*! \fn static int icmp_open_socket(int family, int type, int protocol)
* \brief opens an ICMP socket that PHP and script children cannot inherit
*
* Poller threads posix_spawn() scripts while others ping, so the flag is set
* atomically where SOCK_CLOEXEC exists; elsewhere fcntl() narrows the window.
*/
static int icmp_open_socket(int family, int type, int protocol) {
int fd;
#ifdef SOCK_CLOEXEC
fd = socket(family, type | SOCK_CLOEXEC, protocol);
if (fd != -1 || errno != EINVAL) {
return fd;
}
#endif
fd = socket(family, type, protocol);
if (fd != -1 && fcntl(fd, F_SETFD, FD_CLOEXEC) == -1) {
close(fd);
return -1;
}
return fd;
}
static int icmp_open_shared(int family, int protocol) {
int fd = icmp_open_socket(family, SOCK_RAW, protocol);
if (fd == -1) {
return -1;
}
/* A descriptor select() cannot watch is no use to the readers below. */
if (fd >= FD_SETSIZE) {
close(fd);
return -1;
}
#if defined(__linux__)
/* The kernel copies every inbound ICMP message to a raw socket. Echo
* requests and errors aimed at this host would otherwise compete with
* the replies for the single shared queue. Best effort, as is
* ICMP6_FILTER on the IPv6 socket. */
if (family == AF_INET) {
uint32_t blocked = ~(1U << ICMP_ECHOREPLY);
(void) setsockopt(fd, SOL_RAW, ICMP_FILTER, &blocked, sizeof(blocked));
}
#endif
/* Best effort as well. SO_RCVBUFFORCE passes net.core.rmem_max, which
* is why this happens here, while spine is still root. */
{
int rcvbuf = ICMP_SHARED_RCVBUF;
#ifdef SO_RCVBUFFORCE
if (setsockopt(fd, SOL_SOCKET, SO_RCVBUFFORCE, &rcvbuf, sizeof(rcvbuf)) != 0)
#endif
(void) setsockopt(fd, SOL_SOCKET, SO_RCVBUF, &rcvbuf, sizeof(rcvbuf));
}
return fd;
}
/*! \fn int ping_icmp_open_shared(void)
* \brief opens the shared raw ICMP sockets; the caller must still be root
*
* \return TRUE if the IPv4 socket is open. IPv6 is best effort.
*/
int ping_icmp_open_shared(void) {
if (icmp_shared.fd == -1) {
icmp_shared.fd = icmp_open_shared(AF_INET, IPPROTO_ICMP);
}
#ifdef SPINE_HAVE_ICMPV6
if (icmp6_shared.fd == -1) {
icmp6_shared.fd = icmp_open_shared(AF_INET6, IPPROTO_ICMPV6);
}
#endif
return icmp_shared.fd != -1;
}
int ping_icmp_shared_available(void) {
return icmp_shared.fd != -1;
}
/* Called with LOCK_ICMP held whenever nobody is reading a socket: wake one
* sleeping waiter so it takes over. A waiter that is not asleep finds the
* socket free on its own before it next sleeps. */
static void icmp_shared_handoff(const icmp_shared_t *shared) {
icmp_waiter_t *waiter;
if (shared->reading) {
return;
}
for (waiter = icmp_waiters; waiter != NULL; waiter = waiter->next) {
if (waiter->shared == shared && waiter->sleeping && !waiter->answered) {
pthread_cond_signal(&waiter->wake);
return;
}
}
}
static void icmp_shared_register(icmp_waiter_t *waiter) {
pthread_cond_init(&waiter->wake, NULL);
thread_mutex_lock(LOCK_ICMP);
waiter->answered = FALSE;
waiter->sleeping = FALSE;
waiter->next = icmp_waiters;
icmp_waiters = waiter;
thread_mutex_unlock(LOCK_ICMP);
}
static void icmp_shared_unregister(icmp_waiter_t *waiter) {
icmp_waiter_t **link;
thread_mutex_lock(LOCK_ICMP);
for (link = &icmp_waiters; *link != NULL; link = &(*link)->next) {
if (*link == waiter) {
*link = waiter->next;
break;
}
}
icmp_shared_handoff(waiter->shared);
thread_mutex_unlock(LOCK_ICMP);
pthread_cond_destroy(&waiter->wake);
}
/* Called with LOCK_ICMP held. A reply nobody waits for is dropped, exactly as
* a per-thread socket would drop a reply that is not its own. */
static void icmp_shared_dispatch(int family, const unsigned char *reply, ssize_t len,
const struct sockaddr_storage *from) {
icmp_waiter_t *waiter;
for (waiter = icmp_waiters; waiter != NULL; waiter = waiter->next) {
if (waiter->family != family || waiter->answered) {
continue;
}
if (family == AF_INET) {
const struct sockaddr_in *source = (const struct sockaddr_in *) from;
if (source->sin_addr.s_addr != waiter->peer.s_addr ||
spine_icmp_classify_reply(reply, len, waiter->id, waiter->seq, NULL) != SPINE_ICMP_REPLY_OK) {
continue;
}
}
#ifdef SPINE_HAVE_ICMPV6
else {
const struct sockaddr_in6 *source = (const struct sockaddr_in6 *) from;
if (memcmp(&source->sin6_addr, &waiter->peer6, sizeof(waiter->peer6)) != 0 ||
!icmp6_reply_matches(reply, len, waiter->id, waiter->seq, TRUE)) {
continue;
}
}
#endif
waiter->answered = TRUE;
pthread_cond_signal(&waiter->wake);
return;
}
}
/*! \fn static int icmp_shared_await(icmp_waiter_t *waiter, double deadline)
* \brief waits until the reply to a registered request arrives, or deadline
*
* Whichever waiting thread finds the socket idle reads it, so a reply is
* never stuck behind a thread whose own deadline is further away: a reader
* returns from select() as soon as anything arrives, drains the queue, and
* wakes the threads whose replies it found.
*
* \return TRUE if the reply arrived in time.
*/
static int icmp_shared_await(icmp_waiter_t *waiter, double deadline) {
icmp_shared_t *shared = waiter->shared;
unsigned char replies[ICMP_SHARED_BATCH][ICMP_SHARED_SNAPLEN];
struct sockaddr_storage sources[ICMP_SHARED_BATCH];
ssize_t lengths[ICMP_SHARED_BATCH];
socklen_t fromlen;
ssize_t received;
struct timeval wait;
struct timespec until;
fd_set fds;
double now;
double remaining;
int readable;
int drained;
int count;
int slot;
int answered;
until.tv_sec = (time_t) deadline;
until.tv_nsec = (long) ((deadline - (double) until.tv_sec) * 1000000000.0);
thread_mutex_lock(LOCK_ICMP);
while (!waiter->answered) {
now = get_time_as_double();
if (now >= deadline) {
break;
}
if (shared->reading) {
waiter->sleeping = TRUE;
pthread_cond_timedwait(&waiter->wake, get_lock(LOCK_ICMP), &until);
waiter->sleeping = FALSE;
continue;
}
shared->reading = TRUE;
thread_mutex_unlock(LOCK_ICMP);
remaining = deadline - now;
wait.tv_sec = (time_t) remaining;
wait.tv_usec = (suseconds_t) ((remaining - (double) wait.tv_sec) * 1000000.0);
FD_ZERO(&fds);
FD_SET(shared->fd, &fds);
readable = select(shared->fd + 1, &fds, NULL, NULL, &wait) > 0;
/* Empty the queue: under a flood, one reply per pass falls behind
* and the kernel drops the replies that matter. */
for (drained = 0; readable && drained < ICMP_SHARED_DRAIN_MAX; drained += count) {
for (count = 0; count < ICMP_SHARED_BATCH; ) {
fromlen = sizeof(sources[count]);
received = recvfrom(shared->fd, replies[count], ICMP_SHARED_SNAPLEN, MSG_DONTWAIT,
(struct sockaddr *) &sources[count], &fromlen);
if (received < 0) {
if (errno == EINTR) {
continue;
}
readable = FALSE;
break;
}
lengths[count++] = received;
}
thread_mutex_lock(LOCK_ICMP);
for (slot = 0; slot < count; slot++) {
if (lengths[slot] > 0) {
icmp_shared_dispatch(waiter->family, replies[slot], lengths[slot], &sources[slot]);
}
}
thread_mutex_unlock(LOCK_ICMP);
}
thread_mutex_lock(LOCK_ICMP);
shared->reading = FALSE;
}
answered = waiter->answered;
/* Leaving with the socket idle: let a sleeping waiter take over. */
icmp_shared_handoff(shared);
thread_mutex_unlock(LOCK_ICMP);
return answered;
}
/*! \fn int ping_icmp(host_t *host, ping_t *ping)
* \brief ping a host using an ICMP packet
* \param host a pointer to the current host structure
* \param ping a pointer to the current hosts ping structure
*
* This function pings a host using ICMP. The ICMP packet contains a marker
* to the "Cacti" application so that firewall's can be configured to allow.
* It will modify the ping structure to include the specifics of the ping results.
*
* \return HOST_UP if the host is reachable, HOST_DOWN otherwise.
*
*/
int ping_icmp(const host_t *host, ping_t *ping) {
int icmp_socket = -1;
int icmp_dgram;
int rc = HOST_DOWN;
int socket_errno = 0;
int icmp_use_shared = FALSE;
icmp_waiter_t waiter;
int waiting = FALSE;
double begin_time, end_time, total_time;
double host_timeout;
double one_thousand = 1000.00;
struct timeval timeout;
struct sockaddr_in recvname;
struct sockaddr_in fromname;
char socket_reply[BUFSIZE];
int retry_count;
const char *cacti_msg = "cacti-monitoring-system\0";
int packet_len;
socklen_t fromlen;
ssize_t return_code;
fd_set socket_fds;
static unsigned int seq = 0;
struct icmp *icmp;
unsigned char *packet = NULL;
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] DEBUG: Entering ICMP Ping", host->id));
} else {
SPINE_LOG_DEBUG(("Device[%i] DEBUG: Entering ICMP Ping", host->id));
}
/* Linux hands out SOCK_DGRAM ICMP sockets to the groups listed in
* net.ipv4.ping_group_range, so try that before asking for root. When
* the sysctl excludes us the call fails and the raw path below runs
* unchanged. A datagram reply arrives with the IPv4 header already
* stripped, which the receive path has to account for. */
icmp_dgram = FALSE;
icmp_socket = icmp_open_socket(AF_INET, SOCK_DGRAM, IPPROTO_ICMP);
if (icmp_socket != -1) {
icmp_dgram = TRUE;
} else if (icmp_shared.fd != -1) {
icmp_socket = icmp_shared.fd;
icmp_use_shared = TRUE;
}
/* Otherwise open a raw socket of our own, which needs CAP_NET_RAW or a
* real root user. Spine never regains root to do it: a setuid install
* without capabilities uses the shared socket above instead. */
retry_count = 0;
while (icmp_socket == -1) {
icmp_socket = icmp_open_socket(AF_INET, SOCK_RAW, IPPROTO_ICMP);
socket_errno = errno;
if (icmp_socket != -1) {
break;
}
SPINE_LOG_MEDIUM(("WARNING: Device[%i] raw ICMP socket creation failed: %s",
host->id, strerror(socket_errno)));
retry_count++;
if (retry_count > 4) {
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Ping unable to create ICMP Socket");
snprintf(ping->ping_status, 50, "down");
rc = HOST_DOWN;
goto cleanup;
}
usleep(500000);
}
/* convert the host timeout to a double precision number in seconds */
host_timeout = host->availability.timeout;
/* allocate the packet in memory */
packet_len = ICMP_HDR_SIZE + strlen(cacti_msg);
if (!(packet = malloc(packet_len))) {
die("ERROR: Fatal malloc error: ping.c ping_icmp!");
}
memset(packet, 0, packet_len);
/* set the memory of the ping address */
memset(&fromname, 0, sizeof(struct sockaddr_in));
memset(&recvname, 0, sizeof(struct sockaddr_in));
icmp = (struct icmp*) packet;
icmp->icmp_type = ICMP_ECHO;
icmp->icmp_code = 0;
icmp->icmp_id = getpid() & 0xFFFF;
/* lock set/get the sequence and unlock */
thread_mutex_lock(LOCK_GHBN);
icmp->icmp_seq = seq++;
thread_mutex_unlock(LOCK_GHBN);
icmp->icmp_cksum = 0;
memcpy(packet+ICMP_HDR_SIZE, cacti_msg, strlen(cacti_msg));
icmp->icmp_cksum = get_checksum(packet, packet_len);
/* hostname must be nonblank */
if ((strlen(host->hostname) != 0) && (icmp_socket != -1)) {
/* initialize variables */
snprintf(ping->ping_status, 50, "down");
snprintf(ping->ping_response, SMALL_BUFSIZE, "default");
/* get address of hostname */
if (init_sockaddr(&fromname, host->hostname, 7)) {
retry_count = 0;
total_time = 0;
begin_time = get_time_as_double();
if (icmp_use_shared) {
memset(&waiter, 0, sizeof(waiter));
waiter.shared = &icmp_shared;
waiter.family = AF_INET;
waiter.id = (uint16_t) (getpid() & 0xFFFF);
waiter.seq = icmp->icmp_seq;
waiter.peer = fromname.sin_addr;
icmp_shared_register(&waiter);
waiting = TRUE;
}
while (1) {
if (retry_count > host->availability.retries) {
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Ping timed out");
snprintf(ping->ping_status, 50, "down");
rc = HOST_DOWN;
goto cleanup;
}
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] DEBUG: Attempting to ping %s, seq %d (Retry %d of %d)", host->id, host->hostname, icmp->icmp_seq, retry_count, host->availability.retries));
} else {
SPINE_LOG_DEBUG(("Device[%i] DEBUG: Attempting to ping %s, seq %d (Retry %d of %d)", host->id, host->hostname, icmp->icmp_seq, retry_count, host->availability.retries));
}
if (waiting) {
/* Socket options would change every other thread's socket too. */
double attempt_begin = get_time_as_double();
if (sendto(icmp_socket, packet, packet_len, 0, (struct sockaddr *) &fromname, sizeof(fromname)) >= 0 &&
icmp_shared_await(&waiter, attempt_begin + host_timeout / one_thousand)) {
total_time = (get_time_as_double() - attempt_begin) * one_thousand;
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] INFO: ICMP Device Alive, Try Count:%i, Time:%.4f ms", host->id, retry_count+1, (total_time)));
} else {
SPINE_LOG_MEDIUM(("Device[%i] INFO: ICMP Device Alive, Try Count:%i, Time:%.4f ms", host->id, retry_count+1, (total_time)));
}
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Device is Alive");
snprintf(ping->ping_status, 50, "%.5f", total_time);
rc = HOST_UP;
goto cleanup;
}
retry_count++;
continue;
}
/* decrement the timeout value by the total time */
timeout.tv_sec = rint((host_timeout - total_time) / 1000);
timeout.tv_usec = ((int) (host_timeout - total_time) % 1000) * 1000;
/* set the socket send and receive timeout */
setsockopt(icmp_socket, SOL_SOCKET, SO_RCVTIMEO, (char*)&timeout, sizeof(timeout));
setsockopt(icmp_socket, SOL_SOCKET, SO_SNDTIMEO, (char*)&timeout, sizeof(timeout));
/* send packet to destination */
return_code = sendto(icmp_socket, packet, packet_len, 0, (struct sockaddr *) &fromname, sizeof(fromname));
fromlen = sizeof(fromname);
/* wait for a response on the socket */
/* reinitialize fd_set -- select(2) clears bits in place on return */
keep_listening:
FD_ZERO(&socket_fds);
if (icmp_socket >= FD_SETSIZE) {
SPINE_LOG(("ERROR: Device[%i] ICMP socket %d exceeds FD_SETSIZE %d", host->id, icmp_socket, FD_SETSIZE));
snprintf(ping->ping_status, 50, "down");
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: fd exceeds FD_SETSIZE");
rc = HOST_DOWN;
goto cleanup;
}
FD_SET(icmp_socket,&socket_fds);
return_code = select(icmp_socket + 1, &socket_fds, NULL, NULL, &timeout);
/* record end time */
end_time = get_time_as_double();
/* calculate total time */
total_time = (end_time - begin_time) * one_thousand;
if (total_time < host_timeout) {
#if !(defined(__CYGWIN__))
return_code = recvfrom(icmp_socket, socket_reply, BUFSIZE, MSG_WAITALL, (struct sockaddr *) &recvname, &fromlen);
#else
return_code = recvfrom(icmp_socket, socket_reply, BUFSIZE, MSG_PEEK, (struct sockaddr *) &recvname, &fromlen);
#endif
if (return_code < 0) {
if (errno == EINTR) {
/* call was interrupted by some system event */
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] DEBUG: Received EINTR", host->id));
} else {
SPINE_LOG_DEBUG(("Device[%i] DEBUG: Received EINTR", host->id));
}
goto keep_listening;
}
} else {
const struct icmp *reply_pkt = NULL;
const struct icmp *seen = NULL;
spine_icmp_reply_t verdict;
if (icmp_dgram) {
verdict = spine_icmp_classify_dgram_reply((const unsigned char *) socket_reply,
return_code, icmp->icmp_seq, &reply_pkt);
} else {
verdict = spine_icmp_classify_reply((const unsigned char *) socket_reply,
return_code, (uint16_t)(getpid() & 0xFFFF), icmp->icmp_seq, &reply_pkt);
}
if (verdict == SPINE_ICMP_REPLY_NOT_OURS) {
/* a middlebox that rewrites ICMP query ids is indistinguishable
* from a dead device once the reply is dropped, so record what
* arrived. Both classifiers validate the length before they can
* return this, so the header is safe to read here */
if (icmp_dgram) {
seen = (const struct icmp *) socket_reply;
} else {
seen = (const struct icmp *) (socket_reply + (((struct ip *) socket_reply)->ip_hl << 2));
}
SPINE_LOG_DEBUG(("DEBUG: Device[%i] Discarded ICMP reply, expected id:%d seq:%d, received id:%d seq:%d", host->id, (int)(getpid() & 0xFFFF), icmp->icmp_seq, seen->icmp_id, seen->icmp_seq));
}
if (verdict != SPINE_ICMP_REPLY_OK && verdict != SPINE_ICMP_REPLY_NOT_ECHO) {
ICMP_DISCARD_PEEKED(icmp_socket, socket_reply);
goto keep_listening;
}
if (fromname.sin_addr.s_addr == recvname.sin_addr.s_addr) {
if (verdict == SPINE_ICMP_REPLY_OK) {
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] INFO: ICMP Device Alive, Try Count:%i, Time:%.4f ms", host->id, retry_count+1, (total_time)));
} else {
SPINE_LOG_MEDIUM(("Device[%i] INFO: ICMP Device Alive, Try Count:%i, Time:%.4f ms", host->id, retry_count+1, (total_time)));
}
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Device is Alive");
snprintf(ping->ping_status, 50, "%.5f", total_time);
rc = HOST_UP;
goto cleanup;
} else {
/* received a response other than an echo reply */
ICMP_DISCARD_PEEKED(icmp_socket, socket_reply);
if (total_time > host_timeout) {
retry_count++;
total_time = 0;
}
continue;
}
} else {
/* another host responded */
ICMP_DISCARD_PEEKED(icmp_socket, socket_reply);
goto keep_listening;
}
}
} else {
if (is_debug_device(host->id)) {
SPINE_LOG(("Device[%i] DEBUG: Exceeded Device Timeout, Retrying", host->id));
} else {
SPINE_LOG_DEBUG(("Device[%i] DEBUG: Exceeded Device Timeout, Retrying", host->id));
}
}
total_time = 0;
retry_count++;
#ifndef SOLAR_THREAD
usleep(1000);
#endif
}
} else {
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Destination hostname invalid");
snprintf(ping->ping_status, 50, "down");
rc = HOST_DOWN;
goto cleanup;
}
} else {
snprintf(ping->ping_response, SMALL_BUFSIZE, "ICMP: Destination address not specified");
snprintf(ping->ping_status, 50, "down");
rc = HOST_DOWN;
goto cleanup;
}
cleanup:
/* One owner for both resources. Five exits used to spell this out
* separately and they had already drifted: the FD_SETSIZE guard closed the
* socket and returned without freeing the packet, once per affected device
* per cycle. See #593.
*
* The shared socket outlives this call; only a socket opened here is
* closed. */
SPINE_FREE(packet);
if (waiting) {
icmp_shared_unregister(&waiter);
}
if (icmp_socket != -1 && !icmp_use_shared) {
close(icmp_socket);
}
return rc;
}
#ifdef SPINE_HAVE_ICMPV6
/*! \fn static int init_sockaddr6(struct sockaddr_in6 *name, const char *hostname)
* \brief converts a hostname or IPv6 literal to an IPv6 socket address
*
* \return TRUE if successful, FALSE otherwise.
*
*/
static int init_sockaddr6(struct sockaddr_in6 *name, const char *hostname) {
struct addrinfo hints, *hostinfo;
int rv, retry_count;
memset(&hints, 0, sizeof(hints));
/* AI_ADDRCONFIG is deliberately omitted: it hides IPv6 results on hosts
* that only carry a loopback address, which is exactly where a ::1 test
* has to work */
hints.ai_family = AF_INET6;
hints.ai_socktype = SOCK_DGRAM;
retry_count = 0;
hostinfo = NULL;
while (TRUE) {
rv = getaddrinfo(hostname, NULL, &hints, &hostinfo);
if (rv == 0) {
break;
}