summaryrefslogtreecommitdiff
path: root/plugins/imttcp/imttcp.c
blob: 4bd44dd57ede3b3367a9683630594d418b97059d (plain)
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
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
/* imttcp.c
 * This is an experimental plain tcp input module which follows the
 * multiple thread paradigm.
 *
 * WARNING
 * This module is unfinished. It seems to work, but there also seems to be a problem
 * if it is under large stress (e.g. tcpflood with more than 500 to 1000 concurrent
 * connections). I quickly put together this module after I worked on a larger paper
 * and read [1], which claims that using massively threaded applications is
 * preferrable to an event driven approach. So I put this to test, especially as
 * that would also lead to a much simpler programming paradigm. Unfortuantely, the
 * performance results are devastive: while there is a very slight speedup with 
 * a low connection number (close to the number of cores on the system), there
 * is a dramatic negative speedup if running with many threads. Even at only 50
 * connections, rsyslog is dramatically slower (80 seconds for the same workload
 * which was processed in 60 seconds with traditional imtcp or when running on
 * a single connection). At 1,000 connections, the run was *extremely* slow. So
 * this is definitely a dead-end. To be honest, Behren, condit and Brewer claim
 * that the problem lies in the current implementation of thread libraries.
 * As one cure, they propose user-level threads. However, as far as I could
 * find out, User-Level threads seem not to be much faster under Linux than
 * Kernel-Level threads (which I used in my approach). 
 *
 * Even more convincing is, from the rsyslog PoV, that there are clear reasons
 * why the highly threaded input must be slower:
 * o batch sizes are smaller, leading to much more overhead
 * o many more context switches are needed to switch between the various
 *   i/o handlers
 * o more OS API calls are required because in this model we get more
 *   frequent wakeups on new incoming data, so we have less data available
 *   to read at each instant
 * o more lock contention because many more threads compete on the
 *   main queue mutex
 *
 * All in all, this means that the approach is not the right one, at least
 * not for rsyslog (it may work better if the input can be processed
 * totally independent, but I have note evaluated this). So I will look into
 * an enhanced event-based model with a small set of input workers pulling
 * off data (I assume this is useful for e.g. TLS, as TLS transport is much
 * more computebound than other inputs, and this computation becomes a
 * limiting factor for the overall processing speed under some
 * circumstances - see [2]).
 *
 * For obvious reasons, I will not try to finish imttcp. However, I have
 * decided to leave it included in the source tree, so that
 * a) someone else can build on it, if he sees value in that
 * b) I may use it for some other tests in the future
 *
 * But if you intend to actually use this module unmodified, be prepared
 * for problems.
 *
 * [1] R. Von Behren, J. Condit, and E. Brewer. Why events are a bad idea
 *     (for high-concurrency servers). In Proceedings of the 9th conference on Hot
 *     Topics in Operating Systems-Volume 9, page 4. USENIX Association, 2003.
 *
 * [2] http://kb.monitorware.com/tls-limited-17800-messages-per-second-t10598.html
 *
 * File begun on 2011-01-24 by RGerhards
 *
 * Copyright 2007-2011 Rainer Gerhards and Adiscon GmbH.
 *
 * This file is part of rsyslog.
 *
 * Rsyslog is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * Rsyslog 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 General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Rsyslog.  If not, see <http://www.gnu.org/licenses/>.
 *
 * A copy of the GPL can be found in the file "COPYING" in this distribution.
 */
#include "config.h"
#if !defined(HAVE_EPOLL_CREATE)
#	error imttcp requires OS support for epoll - can not build
	/* imttcp gains speed by using modern Linux capabilities. As such,
	 * it can only be build on platforms supporting the epoll API.
	 */
#endif

#include <stdio.h>
#include <stdlib.h>
#include <assert.h>
#include <string.h>
#include <errno.h>
#include <unistd.h>
#include <stdarg.h>
#include <ctype.h>
#include <netinet/in.h>
#include <netdb.h>
#include <signal.h>
#include <sys/types.h>
#include <sys/socket.h>
#include <sys/epoll.h>
#if HAVE_FCNTL_H
#include <fcntl.h>
#endif
#include "rsyslog.h"
#include "cfsysline.h"
#include "prop.h"
#include "dirty.h"
#include "module-template.h"
#include "unicode-helper.h"
#include "glbl.h"
#include "prop.h"
#include "errmsg.h"
#include "srUtils.h"
#include "datetime.h"
#include "ruleset.h"
#include "msg.h"
#include "net.h" /* for permittedPeers, may be removed when this is removed */

/* the define is from tcpsrv.h, we need to find a new (but easier!!!) abstraction layer some time ... */
#define TCPSRV_NO_ADDTL_DELIMITER -1 /* specifies that no additional delimiter is to be used in TCP framing */


MODULE_TYPE_INPUT
MODULE_TYPE_NOKEEP
MODULE_CNFNAME("imttcp")

/* static data */
DEF_IMOD_STATIC_DATA
DEFobjCurrIf(glbl)
DEFobjCurrIf(net)
DEFobjCurrIf(prop)
DEFobjCurrIf(datetime)
DEFobjCurrIf(errmsg)
DEFobjCurrIf(ruleset)



/* config settings */
struct modConfData_s {
	EMPTY_STRUCT;
};

typedef struct configSettings_s {
	int bEmitMsgOnClose;		/* emit an informational message on close by remote peer */
	int iAddtlFrameDelim;		/* addtl frame delimiter, e.g. for netscreen, default none */
	uchar *pszInputName;		/* value for inputname property, NULL is OK and handled by core engine */
	uchar *lstnIP;			/* which IP we should listen on? */
	ruleset_t *pRuleset;		/* ruleset to bind listener to (use system default if unspecified) */
} configSettings_t;

static configSettings_t cs;

/* data elements describing our running config */
typedef struct ttcpsrv_s ttcpsrv_t;
typedef struct ttcplstn_s ttcplstn_t;
typedef struct ttcpsess_s ttcpsess_t;
typedef struct epolld_s epolld_t;

/* the ttcp server (listener) object
 * Note that the object contains support for forming a linked list
 * of them. It does not make sense to do this seperately.
 */
struct ttcpsrv_s {
	ttcpsrv_t *pNext;		/* linked list maintenance */
	uchar *port;			/* Port to listen to */
	uchar *lstnIP;			/* which IP we should listen on? */
	int bEmitMsgOnClose;
	int iAddtlFrameDelim;
	uchar *pszInputName;
	prop_t *pInputName;		/* InputName in (fast to process) property format */
	ruleset_t *pRuleset;
	ttcplstn_t *pLstn;		/* root of our listeners */
	ttcpsess_t *pSess;		/* root of our sessions */
	pthread_mutex_t mutSess;	/* mutex for session list updates */
};

/* the ttcp session object. Describes a single active session.
 * includes support for doubly-linked list.
 */
struct ttcpsess_s {
	ttcpsrv_t *pSrv;	/* our server */
	ttcpsess_t *prev, *next;
	int sock;
	pthread_t tid;
//--- from tcps_sess.h
	int iMsg;		 /* index of next char to store in msg */
	int bAtStrtOfFram;	/* are we at the very beginning of a new frame? */
	enum {
		eAtStrtFram,
		eInOctetCnt,
		eInMsg
	} inputState;		/* our current state */
	int iOctetsRemain;	/* Number of Octets remaining in message */
	TCPFRAMINGMODE eFraming;
	uchar *pMsg;		/* message (fragment) received */
	prop_t *peerName;	/* host name we received messages from */
	prop_t *peerIP;
//--- END from tcps_sess.h
};


/* the ttcp listener object. Describes a single active listener.
 */
struct ttcplstn_s {
	ttcpsrv_t *pSrv;	/* our server */
	ttcplstn_t *prev, *next;
	int sock;
	pthread_t tid;		/* ID of our listener thread */
};


/* type of object stored in epoll descriptor */
typedef enum {
	epolld_lstn,
	epolld_sess
} epolld_type_t;

/* an epoll descriptor. contains all information necessary to process
 * the result of epoll.
 */
struct epolld_s {
	epolld_type_t typ;
	void *ptr;
	struct epoll_event ev;
};


/* global data */
static ttcpsrv_t *pSrvRoot = NULL;
static int iMaxLine; /* maximum size of a single message */
pthread_attr_t sessThrdAttr;	/* Attribute for session threads; read only after startup */

/* forward definitions */
static rsRetVal resetConfigVariables(uchar __attribute__((unused)) *pp, void __attribute__((unused)) *pVal);
static rsRetVal addLstn(ttcpsrv_t *pSrv, int sock);
static void * sessThrd(void *arg);


/* some simple constructors/destructors */
static void
destructSess(ttcpsess_t *pSess)
{
	free(pSess->pMsg);
	prop.Destruct(&pSess->peerName);
	prop.Destruct(&pSess->peerIP);
	/* TODO: make these inits compile-time switch depending: */
	pSess->pMsg = NULL;
	free(pSess);
}

static void
destructSrv(ttcpsrv_t *pSrv)
{
	prop.Destruct(&pSrv->pInputName);
	free(pSrv->port);
	free(pSrv);
}


/* common initialisation for new threads */
static inline void
initThrd(void)
{
	/* block all signals */
	sigset_t sigSet;
	sigfillset(&sigSet);
	pthread_sigmask(SIG_BLOCK, &sigSet, NULL);

	/* but ignore SIGTTN, which we (ab)use to signal the thread to shutdown -- rgerhards, 2009-07-20 */
	sigemptyset(&sigSet);
	sigaddset(&sigSet, SIGTTIN);
	pthread_sigmask(SIG_UNBLOCK, &sigSet, NULL);

}



/****************************************** TCP SUPPORT FUNCTIONS ***********************************/
/* We may later think about moving this into a helper library again. But the whole point
 * so far was to keep everything related close togehter. -- rgerhards, 2010-08-10
 */


/* Start up a server. That means all of its listeners are created.
 * Does NOT yet accept/process any incoming data (but binds ports). Hint: this
 * code is to be executed before dropping privileges.
 */
static rsRetVal
createSrv(ttcpsrv_t *pSrv)
{
	DEFiRet;
        int error, maxs, on = 1;
	int sock = -1;
	int numSocks;
        struct addrinfo hints, *res = NULL, *r;
	uchar *lstnIP;

	lstnIP = pSrv->lstnIP == NULL ? UCHAR_CONSTANT("") : pSrv->lstnIP;

	DBGPRINTF("imttcp creating listen socket on server '%s', port %s\n", lstnIP, pSrv->port);

        memset(&hints, 0, sizeof(hints));
        hints.ai_flags = AI_PASSIVE;
        hints.ai_family = glbl.GetDefPFFamily();
        hints.ai_socktype = SOCK_STREAM;

        error = getaddrinfo((char*)pSrv->lstnIP, (char*) pSrv->port, &hints, &res);
        if(error) {
		DBGPRINTF("error %d querying server '%s', port '%s'\n", error, pSrv->lstnIP, pSrv->port);
		ABORT_FINALIZE(RS_RET_INVALID_PORT);
	}

        /* Count max number of sockets we may open */
        for(maxs = 0, r = res; r != NULL ; r = r->ai_next, maxs++)
		/* EMPTY */;

        numSocks = 0;   /* num of sockets counter at start of array */
	for(r = res; r != NULL ; r = r->ai_next) {
               sock = socket(r->ai_family, r->ai_socktype, r->ai_protocol);
        	if(sock < 0) {
			if(!(r->ai_family == PF_INET6 && errno == EAFNOSUPPORT))
				DBGPRINTF("error %d creating tcp listen socket", errno);
				/* it is debatable if PF_INET with EAFNOSUPPORT should
				 * also be ignored...
				 */
                        continue;
                }

#ifdef IPV6_V6ONLY
                if(r->ai_family == AF_INET6) {
                	int iOn = 1;
			if(setsockopt(sock, IPPROTO_IPV6, IPV6_V6ONLY,
			      (char *)&iOn, sizeof (iOn)) < 0) {
				close(sock);
				sock = -1;
				continue;
                	}
                }
#endif
       		if(setsockopt(sock, SOL_SOCKET, SO_REUSEADDR, (char *) &on, sizeof(on)) < 0 ) {
			DBGPRINTF("error %d setting tcp socket option\n", errno);
                        close(sock);
			sock = -1;
			continue;
		}

		/* We need to enable BSD compatibility. Otherwise an attacker
		 * could flood our log files by sending us tons of ICMP errors.
		 */
#ifndef BSD	
		if(net.should_use_so_bsdcompat()) {
			if (setsockopt(sock, SOL_SOCKET, SO_BSDCOMPAT,
					(char *) &on, sizeof(on)) < 0) {
				errmsg.LogError(errno, NO_ERRCODE, "TCP setsockopt(BSDCOMPAT)");
                                close(sock);
				sock = -1;
				continue;
			}
		}
#endif

	        if( (bind(sock, r->ai_addr, r->ai_addrlen) < 0)
#ifndef IPV6_V6ONLY
		     && (errno != EADDRINUSE)
#endif
	           ) {
			/* TODO: check if *we* bound the socket - else we *have* an error! */
			char errStr[1024];
			rs_strerror_r(errno, errStr, sizeof(errStr));
                        dbgprintf("error %d while binding tcp socket: %s\n", errno, errStr);
                	close(sock);
			sock = -1;
                        continue;
                }

		if(listen(sock, 511) < 0) {
			DBGPRINTF("tcp listen error %d, suspending\n", errno);
			close(sock);
			sock = -1;
			continue;
		}

		/* if we reach this point, we were able to obtain a valid socket, so we can
		 * create our listener object. -- rgerhards, 2010-08-10
		 */
		CHKiRet(addLstn(pSrv, sock));
		++numSocks;
	}

	if(numSocks != maxs)
		DBGPRINTF("We could initialize %d TCP listen sockets out of %d we received "
		 	  "- this may or may not be an error indication.\n", numSocks, maxs);

        if(numSocks == 0) {
		DBGPRINTF("No TCP listen sockets could successfully be initialized");
		ABORT_FINALIZE(RS_RET_COULD_NOT_BIND);
	}

finalize_it:
	if(res != NULL)
		freeaddrinfo(res);

	if(iRet != RS_RET_OK) {
		if(sock != -1)
			close(sock);
	}

	RETiRet;
}


/* Set pRemHost based on the address provided. This is to be called upon accept()ing
 * a connection request. It must be provided by the socket we received the
 * message on as well as a NI_MAXHOST size large character buffer for the FQDN.
 * Please see http://www.hmug.org/man/3/getnameinfo.php (under Caveats)
 * for some explanation of the code found below. If we detect a malicious
 * hostname, we return RS_RET_MALICIOUS_HNAME and let the caller decide
 * on how to deal with that.
 * rgerhards, 2008-03-31
 */
static rsRetVal
getPeerNames(prop_t **peerName, prop_t **peerIP, struct sockaddr *pAddr)
{
	int error;
	uchar szIP[NI_MAXHOST] = "";
	uchar szHname[NI_MAXHOST] = "";
	struct addrinfo hints, *res;
	
	DEFiRet;

        error = getnameinfo(pAddr, SALEN(pAddr), (char*)szIP, sizeof(szIP), NULL, 0, NI_NUMERICHOST);

        if(error) {
                DBGPRINTF("Malformed from address %s\n", gai_strerror(error));
		strcpy((char*)szHname, "???");
		strcpy((char*)szIP, "???");
		ABORT_FINALIZE(RS_RET_INVALID_HNAME);
	}

	if(!glbl.GetDisableDNS()) {
		error = getnameinfo(pAddr, SALEN(pAddr), (char*)szHname, NI_MAXHOST, NULL, 0, NI_NAMEREQD);
		if(error == 0) {
			memset (&hints, 0, sizeof (struct addrinfo));
			hints.ai_flags = AI_NUMERICHOST;
			hints.ai_socktype = SOCK_STREAM;
			/* we now do a lookup once again. This one should fail,
			 * because we should not have obtained a non-numeric address. If
			 * we got a numeric one, someone messed with DNS!
			 */
			if(getaddrinfo((char*)szHname, NULL, &hints, &res) == 0) {
				freeaddrinfo (res);
				/* OK, we know we have evil, so let's indicate this to our caller */
				snprintf((char*)szHname, NI_MAXHOST, "[MALICIOUS:IP=%s]", szIP);
				DBGPRINTF("Malicious PTR record, IP = \"%s\" HOST = \"%s\"", szIP, szHname);
				iRet = RS_RET_MALICIOUS_HNAME;
			}
		} else {
			strcpy((char*)szHname, (char*)szIP);
		}
	} else {
		strcpy((char*)szHname, (char*)szIP);
	}

	/* We now have the names, so now let's allocate memory and store them permanently. */
	CHKiRet(prop.Construct(peerName));
	CHKiRet(prop.SetString(*peerName, szHname, ustrlen(szHname)));
	CHKiRet(prop.ConstructFinalize(*peerName));
	CHKiRet(prop.Construct(peerIP));
	CHKiRet(prop.SetString(*peerIP, szIP, ustrlen(szIP)));
	CHKiRet(prop.ConstructFinalize(*peerIP));

finalize_it:
	RETiRet;
}



/* accept an incoming connection request
 * rgerhards, 2008-04-22
 */
static rsRetVal
AcceptConnReq(int sock, int *newSock, prop_t **peerName, prop_t **peerIP)
{
	struct sockaddr_storage addr;
	socklen_t addrlen = sizeof(addr);
	int iNewSock = -1;

	DEFiRet;

	iNewSock = accept(sock, (struct sockaddr*) &addr, &addrlen);
	if(iNewSock < 0) {
		if(errno == EAGAIN || errno == EWOULDBLOCK)
			ABORT_FINALIZE(RS_RET_NO_MORE_DATA);
		ABORT_FINALIZE(RS_RET_ACCEPT_ERR);
	}

	CHKiRet(getPeerNames(peerName, peerIP, (struct sockaddr*) &addr));

	*newSock = iNewSock;

finalize_it:
	if(iRet != RS_RET_OK) {
		/* the close may be redundant, but that doesn't hurt... */
		if(iNewSock != -1)
			close(iNewSock);
	}

	RETiRet;
}


/* This is a helper for submitting the message to the rsyslog core.
 * It does some common processing, including resetting the various
 * state variables to a "processed" state.
 * Note that this function is also called if we had a buffer overflow
 * due to a too-long message. So far, there is no indication this 
 * happened and it may be worth thinking about different handling
 * of this case (what obviously would require a change to this
 * function or some related code).
 * rgerhards, 2009-04-23
 * EXTRACT from tcps_sess.c
 */
static rsRetVal
doSubmitMsg(ttcpsess_t *pThis, struct syslogTime *stTime, time_t ttGenTime, multi_submit_t *pMultiSub)
{
	msg_t *pMsg;
	DEFiRet;

	if(pThis->iMsg == 0) {
		DBGPRINTF("discarding zero-sized message\n");
		FINALIZE;
	}

	/* we now create our own message object and submit it to the queue */
	CHKiRet(msgConstructWithTime(&pMsg, stTime, ttGenTime));
	MsgSetRawMsg(pMsg, (char*)pThis->pMsg, pThis->iMsg);
	MsgSetInputName(pMsg, pThis->pSrv->pInputName);
	MsgSetFlowControlType(pMsg, eFLOWCTL_LIGHT_DELAY);
	pMsg->msgFlags  = NEEDS_PARSING | PARSE_HOSTNAME;
	MsgSetRcvFrom(pMsg, pThis->peerName);
	CHKiRet(MsgSetRcvFromIP(pMsg, pThis->peerIP));
	MsgSetRuleset(pMsg, pThis->pSrv->pRuleset);

	if(pMultiSub == NULL) {
		CHKiRet(submitMsg(pMsg));
	} else {
		pMultiSub->ppMsgs[pMultiSub->nElem++] = pMsg;
		if(pMultiSub->nElem == pMultiSub->maxElem)
			CHKiRet(multiSubmitMsg(pMultiSub));
	}


finalize_it:
	/* reset status variables */
	pThis->bAtStrtOfFram = 1;
	pThis->iMsg = 0;

	RETiRet;
}


/* process the data received. As TCP is stream based, we need to process the
 * data inside a state machine. The actual data received is passed in byte-by-byte
 * from DataRcvd, and this function here compiles messages from them and submits
 * the end result to the queue. Introducing this function fixes a long-term bug ;)
 * rgerhards, 2008-03-14
 * EXTRACT from tcps_sess.c
 */
static inline rsRetVal
processDataRcvd(ttcpsess_t *pThis, char c, struct syslogTime *stTime, time_t ttGenTime, multi_submit_t *pMultiSub)
{
	DEFiRet;

	if(pThis->inputState == eAtStrtFram) {
		if(isdigit((int) c)) {
			pThis->inputState = eInOctetCnt;
			pThis->iOctetsRemain = 0;
			pThis->eFraming = TCP_FRAMING_OCTET_COUNTING;
		} else {
			pThis->inputState = eInMsg;
			pThis->eFraming = TCP_FRAMING_OCTET_STUFFING;
		}
	}

	if(pThis->inputState == eInOctetCnt) {
		if(isdigit(c)) {
			pThis->iOctetsRemain = pThis->iOctetsRemain * 10 + c - '0';
		} else { /* done with the octet count, so this must be the SP terminator */
			DBGPRINTF("TCP Message with octet-counter, size %d.\n", pThis->iOctetsRemain);
			if(c != ' ') {
				errmsg.LogError(0, NO_ERRCODE, "Framing Error in received TCP message: "
					    "delimiter is not SP but has ASCII value %d.", c);
			}
			if(pThis->iOctetsRemain < 1) {
				/* TODO: handle the case where the octet count is 0! */
				DBGPRINTF("Framing Error: invalid octet count\n");
				errmsg.LogError(0, NO_ERRCODE, "Framing Error in received TCP message: "
					    "invalid octet count %d.", pThis->iOctetsRemain);
			} else if(pThis->iOctetsRemain > iMaxLine) {
				/* while we can not do anything against it, we can at least log an indication
				 * that something went wrong) -- rgerhards, 2008-03-14
				 */
				DBGPRINTF("truncating message with %d octets - max msg size is %d\n",
					  pThis->iOctetsRemain, iMaxLine);
				errmsg.LogError(0, NO_ERRCODE, "received oversize message: size is %d bytes, "
					        "max msg size is %d, truncating...", pThis->iOctetsRemain, iMaxLine);
			}
			pThis->inputState = eInMsg;
		}
	} else {
		assert(pThis->inputState == eInMsg);
		if(pThis->iMsg >= iMaxLine) {
			/* emergency, we now need to flush, no matter if we are at end of message or not... */
			DBGPRINTF("error: message received is larger than max msg size, we split it\n");
			doSubmitMsg(pThis, stTime, ttGenTime, pMultiSub);
			/* we might think if it is better to ignore the rest of the
			 * message than to treat it as a new one. Maybe this is a good
			 * candidate for a configuration parameter...
			 * rgerhards, 2006-12-04
			 */
		}

		if((   (c == '\n')
		   || ((pThis->pSrv->iAddtlFrameDelim != TCPSRV_NO_ADDTL_DELIMITER) && (c == pThis->pSrv->iAddtlFrameDelim))
		   ) && pThis->eFraming == TCP_FRAMING_OCTET_STUFFING) { /* record delimiter? */
			doSubmitMsg(pThis, stTime, ttGenTime, pMultiSub);
			pThis->inputState = eAtStrtFram;
		} else {
			/* IMPORTANT: here we copy the actual frame content to the message - for BOTH framing modes!
			 * If we have a message that is larger than the max msg size, we truncate it. This is the best
			 * we can do in light of what the engine supports. -- rgerhards, 2008-03-14
			 */
			if(pThis->iMsg < iMaxLine) {
				*(pThis->pMsg + pThis->iMsg++) = c;
			}
		}

		if(pThis->eFraming == TCP_FRAMING_OCTET_COUNTING) {
			/* do we need to find end-of-frame via octet counting? */
			pThis->iOctetsRemain--;
			if(pThis->iOctetsRemain < 1) {
				/* we have end of frame! */
				doSubmitMsg(pThis, stTime, ttGenTime, pMultiSub);
				pThis->inputState = eAtStrtFram;
			}
		}
	}

	RETiRet;
}


/* Processes the data received via a TCP session. If there
 * is no other way to handle it, data is discarded.
 * Input parameter data is the data received, iLen is its
 * len as returned from recv(). iLen must be 1 or more (that
 * is errors must be handled by caller!). iTCPSess must be
 * the index of the TCP session that received the data.
 * rgerhards 2005-07-04
 * And another change while generalizing. We now return either
 * RS_RET_OK, which means the session should be kept open
 * or anything else, which means it must be closed.
 * rgerhards, 2008-03-01
 * As a performance optimization, we pick up the timestamp here. Acutally,
 * this *is* the *correct* reception step for all the data we received, because
 * we have just received a bunch of data! -- rgerhards, 2009-06-16
 * EXTRACT from tcps_sess.c
 */
#define NUM_MULTISUB 1024
static rsRetVal
DataRcvd(ttcpsess_t *pThis, char *pData, size_t iLen)
{
	multi_submit_t multiSub;
	msg_t *pMsgs[NUM_MULTISUB];
	struct syslogTime stTime;
	time_t ttGenTime;
	char *pEnd;
	DEFiRet;

	assert(pData != NULL);
	assert(iLen > 0);

	datetime.getCurrTime(&stTime, &ttGenTime);
	multiSub.ppMsgs = pMsgs;
	multiSub.maxElem = NUM_MULTISUB;
	multiSub.nElem = 0;

	 /* We now copy the message to the session buffer. */
	pEnd = pData + iLen; /* this is one off, which is intensional */

	while(pData < pEnd) {
		CHKiRet(processDataRcvd(pThis, *pData++, &stTime, ttGenTime, &multiSub));
	}

	if(multiSub.nElem > 0) {
		/* submit anything that was not yet submitted */
		CHKiRet(multiSubmitMsg(&multiSub));
	}

finalize_it:
	RETiRet;
}
#undef NUM_MULTISUB


/****************************************** --END-- TCP SUPPORT FUNCTIONS ***********************************/


static inline void
initConfigSettings(void)
{
	cs.bEmitMsgOnClose = 0;
	cs.iAddtlFrameDelim = TCPSRV_NO_ADDTL_DELIMITER;
	cs.pszInputName = NULL;
	cs.pRuleset = NULL;
	cs.lstnIP = NULL;
}


/* add a listener to the server 
 */
static rsRetVal
addLstn(ttcpsrv_t *pSrv, int sock)
{
	DEFiRet;
	ttcplstn_t *pLstn;

	CHKmalloc(pLstn = malloc(sizeof(ttcplstn_t)));
	pLstn->pSrv = pSrv;
	pLstn->sock = sock;

	/* add to start of server's listener list */
	pLstn->prev = NULL;
	pLstn->next = pSrv->pLstn;
	if(pSrv->pLstn != NULL)
		pSrv->pLstn->prev = pLstn;
	pSrv->pLstn = pLstn;

finalize_it:
	RETiRet;
}


/* add a session to the server 
 */
static rsRetVal
addSess(ttcpsrv_t *pSrv, int sock, prop_t *peerName, prop_t *peerIP)
{
	DEFiRet;
	ttcpsess_t *pSess = NULL;

	CHKmalloc(pSess = malloc(sizeof(ttcpsess_t)));
	CHKmalloc(pSess->pMsg = malloc(iMaxLine * sizeof(uchar)));
	pSess->pSrv = pSrv;
	pSess->sock = sock;
	pSess->inputState = eAtStrtFram;
	pSess->iMsg = 0;
	pSess->bAtStrtOfFram = 1;
	pSess->peerName = peerName;
	pSess->peerIP = peerIP;

	/* add to start of server's listener list */
	pSess->prev = NULL;
	pthread_mutex_lock(&pSrv->mutSess);
	pSess->next = pSrv->pSess;
	if(pSrv->pSess != NULL)
		pSrv->pSess->prev = pSess;
	pSrv->pSess = pSess;
	pthread_mutex_unlock(&pSrv->mutSess);

	/* finally run session handler */
	pthread_create(&pSess->tid, &sessThrdAttr, sessThrd, (void*) pSess);

finalize_it:
	RETiRet;
}


/* close/remove a session
 * NOTE: we must first remove the fd from the epoll set and then close it -- else we
 * get an error "bad file descriptor" from epoll.
 */
static inline rsRetVal
closeSess(ttcpsess_t *pSess)
{
	int sock;
	DEFiRet;
	
	sock = pSess->sock;
	close(sock);

	/* finally unlink session from structures */
	pthread_mutex_lock(&pSess->pSrv->mutSess);
	if(pSess->next != NULL)
		pSess->next->prev = pSess->prev;
	if(pSess->prev == NULL) {
		/* need to update root! */
		pSess->pSrv->pSess = pSess->next;
	} else {
		pSess->prev->next = pSess->next;
	}
	pthread_mutex_unlock(&pSess->pSrv->mutSess);

	/* unlinked, now remove structure */
	destructSess(pSess);

	RETiRet;
}


/* accept a new ruleset to bind. Checks if it exists and complains, if not */
static rsRetVal setRuleset(void __attribute__((unused)) *pVal, uchar *pszName)
{
	ruleset_t *pRuleset;
	rsRetVal localRet;
	DEFiRet;

	localRet = ruleset.GetRuleset(ourConf, &pRuleset, pszName);
	if(localRet == RS_RET_NOT_FOUND) {
		errmsg.LogError(0, NO_ERRCODE, "error: ruleset '%s' not found - ignored", pszName);
	}
	CHKiRet(localRet);
	cs.pRuleset = pRuleset;
	DBGPRINTF("imttcp current bind ruleset %p: '%s'\n", pRuleset, pszName);

finalize_it:
	free(pszName); /* no longer needed */
	RETiRet;
}


static rsRetVal addTCPListener(void __attribute__((unused)) *pVal, uchar *pNewVal)
{
	DEFiRet;
	ttcpsrv_t *pSrv;

	CHKmalloc(pSrv = malloc(sizeof(ttcpsrv_t)));
	pthread_mutex_init(&pSrv->mutSess, NULL);
	pSrv->pSess = NULL;
	pSrv->pLstn = NULL;
	pSrv->bEmitMsgOnClose = cs.bEmitMsgOnClose;
	pSrv->port = pNewVal;
	pSrv->iAddtlFrameDelim = cs.iAddtlFrameDelim;
	cs.pszInputName = NULL;	/* moved over to pSrv, we do not own */
	pSrv->lstnIP = cs.lstnIP;
	cs.lstnIP = NULL;	/* moved over to pSrv, we do not own */
	pSrv->pRuleset = cs.pRuleset;
	pSrv->pszInputName = (cs.pszInputName == NULL) ?  UCHAR_CONSTANT("imttcp") : cs.pszInputName;
	CHKiRet(prop.Construct(&pSrv->pInputName));
	CHKiRet(prop.SetString(pSrv->pInputName, pSrv->pszInputName, ustrlen(pSrv->pszInputName)));
	CHKiRet(prop.ConstructFinalize(pSrv->pInputName));

	/* add to linked list */
	pSrv->pNext = pSrvRoot;
	pSrvRoot = pSrv;

	/* all config vars are auto-reset -- this also is very useful with the
	 * new config format effort (v6).
	 */
	resetConfigVariables(NULL, NULL);

finalize_it:
	if(iRet != RS_RET_OK) {
		errmsg.LogError(0, NO_ERRCODE, "error %d trying to add listener", iRet);
	}
	RETiRet;
}


/* create up all listeners 
 * This is a one-time stop once the module is set to start.
 */
static inline rsRetVal
createServers()
{
	DEFiRet;
	ttcpsrv_t *pSrv;

	pSrv = pSrvRoot;
	while(pSrv != NULL) {
		DBGPRINTF("Starting up ttcp server for port %s, name '%s'\n", pSrv->port, pSrv->pszInputName);
		createSrv(pSrv);
		pSrv = pSrv->pNext;
	}

	RETiRet;
}


/* This function implements the thread to be used for listeners.
 * The function terminates if shutdown is required.
 */
static void *
lstnThrd(void *arg)
{
	ttcplstn_t *pLstn = (ttcplstn_t *) arg;
	rsRetVal iRet = RS_RET_OK;
	int newSock;
	prop_t *peerName;
	prop_t *peerIP;
	rsRetVal localRet;

	initThrd();
	
	while(glbl.GetGlobalInputTermState() == 0) {
		localRet = AcceptConnReq(pLstn->sock, &newSock, &peerName, &peerIP);
		if(glbl.GetGlobalInputTermState() == 1)
			break; /* terminate input! */
		CHKiRet(localRet);
		DBGPRINTF("imttcp: new connection %d on listen socket %d\n", newSock, pLstn->sock);
		CHKiRet(addSess(pLstn->pSrv, newSock, peerName, peerIP));
	}

finalize_it:
	close(pLstn->sock);
	DBGPRINTF("imttcp shutdown listen socket %d\n", pLstn->sock);
	/* Note: we do NOT unlink the deleted session. While this sounds not 100% clean,
	 * it is fine with the current implementation as we will never reuse these elements.
	 * However, it make sense (and not cost notable performance) to do it "right"...
	 */
	return NULL;
}


/* This function implements the thread to be used for a session
 * The function terminates if shutdown is required.
 */
static void *
sessThrd(void *arg)
{
	ttcpsess_t *pSess = (ttcpsess_t*) arg;
	rsRetVal iRet = RS_RET_OK;
	int lenRcv;
	int lenBuf;
	char rcvBuf[64*1024];
	
	initThrd();
	
	while(glbl.GetGlobalInputTermState() == 0) {
		lenBuf = sizeof(rcvBuf);
		lenRcv = recv(pSess->sock, rcvBuf, lenBuf, 0);

		if(glbl.GetGlobalInputTermState() == 1)
			ABORT_FINALIZE(RS_RET_FORCE_TERM);

		if(lenRcv > 0) {
			/* have data, process it */
			DBGPRINTF("imttcp: data(%d) on socket %d: %s\n", lenRcv, pSess->sock, rcvBuf);
			CHKiRet(DataRcvd(pSess, rcvBuf, lenRcv));
		} else if (lenRcv == 0) {
			/* session was closed, do clean-up */
			if(pSess->pSrv->bEmitMsgOnClose) {
				uchar *peerName;
				int lenPeer;
				prop.GetString(pSess->peerName, &peerName, &lenPeer);
				errmsg.LogError(0, RS_RET_PEER_CLOSED_CONN, "imttcp session %d closed by remote peer %s.",
						pSess->sock, peerName);
			}
			break;
		} else {
			if(errno == EAGAIN)
				break;
			DBGPRINTF("imttcp: error on session socket %d - closing.\n", pSess->sock);
			break;
		}
	}

finalize_it:
	DBGPRINTF("imttcp: session thread terminates, socket %d\n", pSess->sock);
	closeSess(pSess); /* try clean-up by dropping session */
	return NULL;
}

/* startup all listeners 
 */
static inline rsRetVal
startupListeners()
{
	DEFiRet;
	ttcpsrv_t *pSrv;
	ttcplstn_t *pLstn;

	pSrv = pSrvRoot;
	while(pSrv != NULL) {
		for(pLstn = pSrv->pLstn ; pLstn != NULL ; pLstn = pLstn->next) {
			pthread_create(&pLstn->tid, NULL, lstnThrd, (void*) pLstn);
		}
		pSrv = pSrv->pNext;
	}

	RETiRet;
}


/* This function is called to gather input.
 */
BEGINrunInput
	struct timeval tvSelectTimeout;
CODESTARTrunInput
	DBGPRINTF("imttcp: now beginning to process input data\n");
	CHKiRet(startupListeners());

	// TODO: this loop is a quick hack, do it right!
	while(glbl.GetGlobalInputTermState() == 0) {
		tvSelectTimeout.tv_sec =  86400 /*1 day*/;
		tvSelectTimeout.tv_usec = 0;
		select(1, NULL, NULL, NULL, &tvSelectTimeout);
	}
finalize_it: ;
ENDrunInput


/* initialize and return if will run or not */
BEGINwillRun
CODESTARTwillRun
	/* first apply some config settings */
	iMaxLine = glbl.GetMaxLine(); /* get maximum size we currently support */

	if(pSrvRoot == NULL) {
		errmsg.LogError(0, RS_RET_NO_LSTN_DEFINED, "error: no ttcp server defined, module can not run.");
		ABORT_FINALIZE(RS_RET_NO_RUN);
	}

	/* start up servers, but do not yet read input data */
	CHKiRet(createServers());
	DBGPRINTF("imttcp started up, but not yet receiving data\n");
finalize_it:
ENDwillRun


/* completely shut down a server. All we need to do is unblock the
 * various session and listerner threads as they then check the termination
 * praedicate themselves.
 */
static inline void
shutdownSrv(ttcpsrv_t *pSrv)
{
	ttcplstn_t *pLstn;
	ttcplstn_t *pLstnDel;
	ttcpsess_t *pSess;
	pthread_t tid;

	pLstn = pSrv->pLstn;
	while(pLstn != NULL) {
		tid = pLstn->tid; /* pSess will be destructed! */
		pthread_kill(tid, SIGTTIN);
		DBGPRINTF("imttcp: termination request for listen thread %x\n", (unsigned) tid);
		pthread_join(tid, NULL);
		DBGPRINTF("imttcp: listen thread %x terminated \n", (unsigned) tid);
		pLstnDel = pLstn;
		pLstn = pLstn->next;
		free(pLstnDel);
	}

	pSess = pSrv->pSess;
	while(pSess != NULL) {
		tid = pSess->tid; /* pSess will be destructed! */
		pSess = pSess->next;
		pthread_kill(tid, SIGTTIN);
		DBGPRINTF("imttcp: termination request for session thread %x\n", (unsigned) tid);
		//pthread_join(tid, NULL);
		DBGPRINTF("imttcp: session thread %x terminated \n", (unsigned) tid);
	}
}


BEGINafterRun
	ttcpsrv_t *pSrv, *srvDel;
CODESTARTafterRun
	/* do cleanup here */
	/* we need to close everything that is still open */
	pSrv = pSrvRoot;
	while(pSrv != NULL) {
		srvDel = pSrv;
		pSrv = pSrv->pNext;
		shutdownSrv(srvDel);
		destructSrv(srvDel);
	}
ENDafterRun


BEGINmodExit
CODESTARTmodExit
	pthread_attr_destroy(&sessThrdAttr);

	/* release objects we used */
	objRelease(glbl, CORE_COMPONENT);
	objRelease(prop, CORE_COMPONENT);
	objRelease(net, LM_NET_FILENAME);
	objRelease(datetime, CORE_COMPONENT);
	objRelease(errmsg, CORE_COMPONENT);
	objRelease(ruleset, CORE_COMPONENT);
ENDmodExit


static rsRetVal
resetConfigVariables(uchar __attribute__((unused)) *pp, void __attribute__((unused)) *pVal)
{
	cs.bEmitMsgOnClose = 0;
	cs.iAddtlFrameDelim = TCPSRV_NO_ADDTL_DELIMITER;
	free(cs.pszInputName);
	cs.pszInputName = NULL;
	free(cs.lstnIP);
	cs.lstnIP = NULL;
	return RS_RET_OK;
}



BEGINqueryEtryPt
CODESTARTqueryEtryPt
CODEqueryEtryPt_STD_IMOD_QUERIES
ENDqueryEtryPt


BEGINmodInit()
CODESTARTmodInit
	*ipIFVersProvided = CURR_MOD_IF_VERSION; /* we only support the current interface specification */
CODEmodInit_QueryRegCFSLineHdlr
	initConfigSettings();
	/* request objects we use */
	CHKiRet(objUse(glbl, CORE_COMPONENT));
	CHKiRet(objUse(prop, CORE_COMPONENT));
	CHKiRet(objUse(net, LM_NET_FILENAME));
	CHKiRet(objUse(errmsg, CORE_COMPONENT));
	CHKiRet(objUse(datetime, CORE_COMPONENT));
	CHKiRet(objUse(ruleset, CORE_COMPONENT));

	/* initialize "read-only" thread attributes */
	pthread_attr_init(&sessThrdAttr);
	pthread_attr_setdetachstate(&sessThrdAttr, PTHREAD_CREATE_DETACHED);
	pthread_attr_setstacksize(&sessThrdAttr, 4096*1024);

	/* register config file handlers */
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpserverrun"), 0, eCmdHdlrGetWord,
				   addTCPListener, NULL, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpservernotifyonconnectionclose"), 0,
				   eCmdHdlrBinary, NULL, &cs.bEmitMsgOnClose, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpserveraddtlframedelimiter"), 0, eCmdHdlrInt,
				   NULL, &cs.iAddtlFrameDelim, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpserverinputname"), 0,
				   eCmdHdlrGetWord, NULL, &cs.pszInputName, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpserverlistenip"), 0,
				   eCmdHdlrGetWord, NULL, &cs.lstnIP, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("inputttcpserverbindruleset"), 0,
				   eCmdHdlrGetWord, setRuleset, NULL, STD_LOADABLE_MODULE_ID));
	CHKiRet(omsdRegCFSLineHdlr(UCHAR_CONSTANT("resetconfigvariables"), 1, eCmdHdlrCustomHandler,
		resetConfigVariables, NULL, STD_LOADABLE_MODULE_ID));
ENDmodInit


/* vim:set ai:
 */