summaryrefslogtreecommitdiffstats
path: root/src/msg/async/AsyncMessenger.cc
blob: 3971e91ffd015c908fadab3ebb3c7e2adb623bd4 (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
// -*- mode:C++; tab-width:8; c-basic-offset:2; indent-tabs-mode:t -*-
// vim: ts=8 sw=2 smarttab
/*
 * Ceph - scalable distributed file system
 *
 * Copyright (C) 2014 UnitedStack <haomai@unitedstack.com>
 *
 * Author: Haomai Wang <haomaiwang@gmail.com>
 *
 * This is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License version 2.1, as published by the Free Software
 * Foundation.  See file COPYING.
 *
 */

#include "acconfig.h"

#include <iostream>
#include <fstream>

#include "AsyncMessenger.h"

#include "common/config.h"
#include "common/Timer.h"
#include "common/errno.h"

#include "messages/MOSDOp.h"
#include "messages/MOSDOpReply.h"
#include "common/EventTrace.h"

#define dout_subsys ceph_subsys_ms
#undef dout_prefix
#define dout_prefix _prefix(_dout, this)
static std::ostream& _prefix(std::ostream *_dout, AsyncMessenger *m) {
  return *_dout << "-- " << m->get_myaddrs() << " ";
}

static std::ostream& _prefix(std::ostream *_dout, Processor *p) {
  return *_dout << " Processor -- ";
}


/*******************
 * Processor
 */

class Processor::C_processor_accept : public EventCallback {
  Processor *pro;

 public:
  explicit C_processor_accept(Processor *p): pro(p) {}
  void do_request(uint64_t id) override {
    pro->accept();
  }
};

Processor::Processor(AsyncMessenger *r, Worker *w, CephContext *c)
  : msgr(r), net(c), worker(w),
    listen_handler(new C_processor_accept(this)) {}

int Processor::bind(const entity_addrvec_t &bind_addrs,
		    const std::set<int>& avoid_ports,
		    entity_addrvec_t* bound_addrs)
{
  const auto& conf = msgr->cct->_conf;
  // bind to socket(s)
  ldout(msgr->cct, 10) << __func__ << " " << bind_addrs << dendl;

  SocketOptions opts;
  opts.nodelay = msgr->cct->_conf->ms_tcp_nodelay;
  opts.rcbuf_size = msgr->cct->_conf->ms_tcp_rcvbuf;

  listen_sockets.resize(bind_addrs.v.size());
  *bound_addrs = bind_addrs;

  for (unsigned k = 0; k < bind_addrs.v.size(); ++k) {
    auto& listen_addr = bound_addrs->v[k];

    /* bind to port */
    int r = -1;

    for (int i = 0; i < conf->ms_bind_retry_count; i++) {
      if (i > 0) {
	lderr(msgr->cct) << __func__ << " was unable to bind. Trying again in "
			 << conf->ms_bind_retry_delay << " seconds " << dendl;
	sleep(conf->ms_bind_retry_delay);
      }

      if (listen_addr.get_port()) {
	worker->center.submit_to(
	  worker->center.get_id(),
	  [this, k, &listen_addr, &opts, &r]() {
	    r = worker->listen(listen_addr, k, opts, &listen_sockets[k]);
	  }, false);
	if (r < 0) {
	  lderr(msgr->cct) << __func__ << " unable to bind to " << listen_addr
			   << ": " << cpp_strerror(r) << dendl;
	  continue;
	}
      } else {
	// try a range of ports
	for (int port = msgr->cct->_conf->ms_bind_port_min;
	     port <= msgr->cct->_conf->ms_bind_port_max;
	     port++) {
	  if (avoid_ports.count(port))
	    continue;

	  listen_addr.set_port(port);
	  worker->center.submit_to(
	    worker->center.get_id(),
	    [this, k, &listen_addr, &opts, &r]() {
	      r = worker->listen(listen_addr, k, opts, &listen_sockets[k]);
	    }, false);
	  if (r == 0)
	    break;
	}
	if (r < 0) {
	  lderr(msgr->cct) << __func__ << " unable to bind to " << listen_addr
			   << " on any port in range "
			   << msgr->cct->_conf->ms_bind_port_min
			   << "-" << msgr->cct->_conf->ms_bind_port_max << ": "
			   << cpp_strerror(r) << dendl;
	  listen_addr.set_port(0); // Clear port before retry, otherwise we shall fail again.
	  continue;
	}
	ldout(msgr->cct, 10) << __func__ << " bound on random port "
			     << listen_addr << dendl;
      }
      if (r == 0) {
	break;
      }
    }

    // It seems that binding completely failed, return with that exit status
    if (r < 0) {
      lderr(msgr->cct) << __func__ << " was unable to bind after "
		       << conf->ms_bind_retry_count
		       << " attempts: " << cpp_strerror(r) << dendl;
      for (unsigned j = 0; j < k; ++j) {
	// clean up previous bind
	listen_sockets[j].abort_accept();
      }
      return r;
    }
  }

  ldout(msgr->cct, 10) << __func__ << " bound to " << *bound_addrs << dendl;
  return 0;
}

void Processor::start()
{
  ldout(msgr->cct, 1) << __func__ << dendl;

  // start thread
  worker->center.submit_to(worker->center.get_id(), [this]() {
      for (auto& listen_socket : listen_sockets) {
	if (listen_socket) {
          if (listen_socket.fd() == -1) {
            ldout(msgr->cct, 1) << __func__ 
                << " Error: processor restart after listen_socket.fd closed. " 
                << this << dendl;
            return;
          }
	  worker->center.create_file_event(listen_socket.fd(), EVENT_READABLE,
					   listen_handler); }
      }
    }, false);
}

void Processor::accept()
{
  SocketOptions opts;
  opts.nodelay = msgr->cct->_conf->ms_tcp_nodelay;
  opts.rcbuf_size = msgr->cct->_conf->ms_tcp_rcvbuf;
  opts.priority = msgr->get_socket_priority();

  for (auto& listen_socket : listen_sockets) {
    ldout(msgr->cct, 10) << __func__ << " listen_fd=" << listen_socket.fd()
			 << dendl;
    unsigned accept_error_num = 0;

    while (true) {
      entity_addr_t addr;
      ConnectedSocket cli_socket;
      Worker *w = worker;
      if (!msgr->get_stack()->support_local_listen_table())
	w = msgr->get_stack()->get_worker();
      else
	++w->references;
      int r = listen_socket.accept(&cli_socket, opts, &addr, w);
      if (r == 0) {
	ldout(msgr->cct, 10) << __func__ << " accepted incoming on sd "
			     << cli_socket.fd() << dendl;

	msgr->add_accept(
	  w, std::move(cli_socket),
	  msgr->get_myaddrs().v[listen_socket.get_addr_slot()],
	  addr);
	accept_error_num = 0;
	continue;
      } else {
	--w->references;
	if (r == -EINTR) {
	  continue;
	} else if (r == -EAGAIN) {
	  break;
	} else if (r == -EMFILE || r == -ENFILE) {
	  lderr(msgr->cct) << __func__ << " open file descriptions limit reached sd = " << listen_socket.fd()
			   << " errno " << r << " " << cpp_strerror(r) << dendl;
	  if (++accept_error_num > msgr->cct->_conf->ms_max_accept_failures) {
	    lderr(msgr->cct) << "Proccessor accept has encountered enough error numbers, just do ceph_abort()." << dendl;
	    ceph_abort();
	  }
	  continue;
	} else if (r == -ECONNABORTED) {
	  ldout(msgr->cct, 0) << __func__ << " it was closed because of rst arrived sd = " << listen_socket.fd()
			      << " errno " << r << " " << cpp_strerror(r) << dendl;
	  continue;
	} else {
	  lderr(msgr->cct) << __func__ << " no incoming connection?"
			   << " errno " << r << " " << cpp_strerror(r) << dendl;
	  if (++accept_error_num > msgr->cct->_conf->ms_max_accept_failures) {
	    lderr(msgr->cct) << "Proccessor accept has encountered enough error numbers, just do ceph_abort()." << dendl;
	    ceph_abort();
	  }
	  continue;
	}
      }
    }
  }
}

void Processor::stop()
{
  ldout(msgr->cct,10) << __func__ << dendl;

  worker->center.submit_to(worker->center.get_id(), [this]() {
      for (auto& listen_socket : listen_sockets) {
	if (listen_socket) {
	  worker->center.delete_file_event(listen_socket.fd(), EVENT_READABLE);
	  listen_socket.abort_accept();
	}
      }
    }, false);
}


struct StackSingleton {
  CephContext *cct;
  std::shared_ptr<NetworkStack> stack;

  explicit StackSingleton(CephContext *c): cct(c) {}
  void ready(std::string &type) {
    if (!stack)
      stack = NetworkStack::create(cct, type);
  }
  ~StackSingleton() {
    stack->stop();
  }
};


class C_handle_reap : public EventCallback {
  AsyncMessenger *msgr;

  public:
  explicit C_handle_reap(AsyncMessenger *m): msgr(m) {}
  void do_request(uint64_t id) override {
    // judge whether is a time event
    msgr->reap_dead();
  }
};

/*******************
 * AsyncMessenger
 */

AsyncMessenger::AsyncMessenger(CephContext *cct, entity_name_t name,
                               const std::string &type, std::string mname, uint64_t _nonce)
  : SimplePolicyMessenger(cct, name),
    dispatch_queue(cct, this, mname),
    nonce(_nonce)
{
  std::string transport_type = "posix";
  if (type.find("rdma") != std::string::npos)
    transport_type = "rdma";
  else if (type.find("dpdk") != std::string::npos)
    transport_type = "dpdk";

  auto single = &cct->lookup_or_create_singleton_object<StackSingleton>(
    "AsyncMessenger::NetworkStack::" + transport_type, true, cct);
  single->ready(transport_type);
  stack = single->stack.get();
  stack->start();
  local_worker = stack->get_worker();
  local_connection = ceph::make_ref<AsyncConnection>(cct, this, &dispatch_queue,
					 local_worker, true, true);
  init_local_connection();
  reap_handler = new C_handle_reap(this);
  unsigned processor_num = 1;
  if (stack->support_local_listen_table())
    processor_num = stack->get_num_worker();
  for (unsigned i = 0; i < processor_num; ++i)
    processors.push_back(new Processor(this, stack->get_worker(i), cct));
}

/**
 * Destroy the AsyncMessenger. Pretty simple since all the work is done
 * elsewhere.
 */
AsyncMessenger::~AsyncMessenger()
{
  delete reap_handler;
  ceph_assert(!did_bind); // either we didn't bind or we shut down the Processor
  for (auto &&p : processors)
    delete p;
}

void AsyncMessenger::ready()
{
  ldout(cct,10) << __func__ << " " << get_myaddrs() << dendl;

  stack->ready();
  if (pending_bind) {
    int err = bindv(pending_bind_addrs, saved_public_addrs);
    if (err) {
      lderr(cct) << __func__ << " postponed bind failed" << dendl;
      ceph_abort();
    }
  }

  std::lock_guard l{lock};
  for (auto &&p : processors)
    p->start();
  dispatch_queue.start();
}

int AsyncMessenger::shutdown()
{
  ldout(cct,10) << __func__ << " " << get_myaddrs() << dendl;

  // done!  clean up.
  for (auto &&p : processors)
    p->stop();
  mark_down_all();
  // break ref cycles on the loopback connection
  local_connection->clear_priv();
  local_connection->mark_down();
  did_bind = false;
  lock.lock();
  stop_cond.notify_all();
  stopped = true;
  lock.unlock();
  stack->drain();
  return 0;
}

int AsyncMessenger::bind(const entity_addr_t &bind_addr,
                         std::optional<entity_addrvec_t> public_addrs)
{
  ldout(cct, 10) << __func__ << " " << bind_addr
                 << " public " << public_addrs << dendl;
  // old bind() can take entity_addr_t(). new bindv() can take a
  // 0.0.0.0-like address but needs type and family to be set.
  auto a = bind_addr;
  if (a == entity_addr_t()) {
    a.set_type(entity_addr_t::TYPE_LEGACY);
    if (cct->_conf->ms_bind_ipv6) {
      a.set_family(AF_INET6);
    } else {
      a.set_family(AF_INET);
    }
  }
  return bindv(entity_addrvec_t(a), public_addrs);
}

int AsyncMessenger::bindv(const entity_addrvec_t &bind_addrs,
                          std::optional<entity_addrvec_t> public_addrs)
{
  lock.lock();

  if (!pending_bind && started) {
    ldout(cct,10) << __func__ << " already started" << dendl;
    lock.unlock();
    return -1;
  }

  ldout(cct, 10) << __func__ << " " << bind_addrs
                 << " public " << public_addrs << dendl;
  if (public_addrs && bind_addrs != public_addrs) {
    // for the sake of rebind() and the is-not-ready case let's
    // store public_addrs. there is no point in that if public
    // addrs are indifferent from bind_addrs.
    saved_public_addrs = std::move(public_addrs);
  }

  if (!stack->is_ready()) {
    ldout(cct, 10) << __func__ << " Network Stack is not ready for bind yet - postponed" << dendl;
    pending_bind_addrs = bind_addrs;
    pending_bind = true;
    lock.unlock();
    return 0;
  }

  lock.unlock();

  // bind to a socket
  std::set<int> avoid_ports;
  entity_addrvec_t bound_addrs;
  unsigned i = 0;
  for (auto &&p : processors) {
    int r = p->bind(bind_addrs, avoid_ports, &bound_addrs);
    if (r) {
      // Note: this is related to local tcp listen table problem.
      // Posix(default kernel implementation) backend shares listen table
      // in the kernel, so all threads can use the same listen table naturally
      // and only one thread need to bind. But other backends(like dpdk) uses local
      // listen table, we need to bind/listen tcp port for each worker. So if the
      // first worker failed to bind, it could be think the normal error then handle
      // it, like port is used case. But if the first worker successfully to bind
      // but the second worker failed, it's not expected and we need to assert
      // here
      ceph_assert(i == 0);
      return r;
    }
    ++i;
  }
  _finish_bind(bind_addrs, bound_addrs);
  return 0;
}

int AsyncMessenger::rebind(const std::set<int>& avoid_ports)
{
  ldout(cct,1) << __func__ << " rebind avoid " << avoid_ports << dendl;
  ceph_assert(did_bind);

  for (auto &&p : processors)
    p->stop();
  mark_down_all();

  // adjust the nonce; we want our entity_addr_t to be truly unique.
  nonce += 1000000;
  ldout(cct, 10) << __func__ << " new nonce " << nonce
		 << " and addr " << get_myaddrs() << dendl;

  entity_addrvec_t bound_addrs;
  entity_addrvec_t bind_addrs = get_myaddrs();
  std::set<int> new_avoid(avoid_ports);
  for (auto& a : bind_addrs.v) {
    new_avoid.insert(a.get_port());
    a.set_port(0);
  }
  ldout(cct, 10) << __func__ << " will try " << bind_addrs
		 << " and avoid ports " << new_avoid << dendl;
  unsigned i = 0;
  for (auto &&p : processors) {
    int r = p->bind(bind_addrs, avoid_ports, &bound_addrs);
    if (r) {
      ceph_assert(i == 0);
      return r;
    }
    ++i;
  }
  _finish_bind(bind_addrs, bound_addrs);
  for (auto &&p : processors) {
    p->start();
  }
  return 0;
}

int AsyncMessenger::client_bind(const entity_addr_t &bind_addr)
{
  if (!cct->_conf->ms_bind_before_connect)
    return 0;
  std::lock_guard l{lock};
  if (did_bind) {
    return 0;
  }
  if (started) {
    ldout(cct, 10) << __func__ << " already started" << dendl;
    return -1;
  }
  ldout(cct, 10) << __func__ << " " << bind_addr << dendl;

  set_myaddrs(entity_addrvec_t(bind_addr));
  return 0;
}

void AsyncMessenger::_finish_bind(const entity_addrvec_t& bind_addrs,
				  const entity_addrvec_t& listen_addrs)
{
  set_myaddrs(bind_addrs);
  for (auto& a : bind_addrs.v) {
    if (!a.is_blank_ip()) {
      learned_addr(a);
    }
  }

  if (get_myaddrs().front().get_port() == 0) {
    set_myaddrs(listen_addrs);
  }

  entity_addrvec_t newaddrs;
  if (saved_public_addrs) {
    newaddrs = *saved_public_addrs;
    for (auto& public_addr : newaddrs.v) {
      public_addr.set_nonce(nonce);
      if (public_addr.is_ip() && public_addr.get_port() == 0) {
	// port is not explicitly set. This is fine as it can be figured
	// out by msgr. For instance, the low-level `Processor::bind`
	// scans for free ports in a range controlled by ms_bind_port_min
	// and ms_bind_port_max.
        for (const auto& a : my_addrs->v) {
          if (public_addr.get_type() == a.get_type() && a.is_ip()) {
             public_addr.set_port(a.get_port());
          }
        }
      }
    }
  } else {
    newaddrs = *my_addrs;
    for (auto& a : newaddrs.v) {
      a.set_nonce(nonce);
    }
  }
  set_myaddrs(newaddrs);

  init_local_connection();

  ldout(cct,1) << __func__ << " bind my_addrs is " << get_myaddrs() << dendl;
  did_bind = true;
}

int AsyncMessenger::client_reset()
{
  mark_down_all();

  std::scoped_lock l{lock};
  // adjust the nonce; we want our entity_addr_t to be truly unique.
  nonce += 1000000;
  ldout(cct, 10) << __func__ << " new nonce " << nonce << dendl;

  entity_addrvec_t newaddrs = *my_addrs;
  for (auto& a : newaddrs.v) {
    a.set_nonce(nonce);
  }
  set_myaddrs(newaddrs);
  _init_local_connection();
  return 0;
}

int AsyncMessenger::start()
{
  std::scoped_lock l{lock};
  ldout(cct,1) << __func__ << " start" << dendl;

  // register at least one entity, first!
  ceph_assert(my_name.type() >= 0);

  ceph_assert(!started);
  started = true;
  stopped = false;

  if (!did_bind) {
    entity_addrvec_t newaddrs = *my_addrs;
    for (auto& a : newaddrs.v) {
      a.nonce = nonce;
    }
    set_myaddrs(newaddrs);
    _init_local_connection();
  }

  return 0;
}

void AsyncMessenger::wait()
{
  {
    std::unique_lock locker{lock};
    if (!started) {
      return;
    }
    if (!stopped)
      stop_cond.wait(locker);
  }
  dispatch_queue.shutdown();
  if (dispatch_queue.is_started()) {
    ldout(cct, 10) << __func__ << ": waiting for dispatch queue" << dendl;
    dispatch_queue.wait();
    dispatch_queue.discard_local();
    ldout(cct, 10) << __func__ << ": dispatch queue is stopped" << dendl;
  }

  // close all connections
  shutdown_connections(false);
  stack->drain();

  ldout(cct, 10) << __func__ << ": done." << dendl;
  ldout(cct, 1) << __func__ << " complete." << dendl;
  started = false;
}

void AsyncMessenger::add_accept(Worker *w, ConnectedSocket cli_socket,
				const entity_addr_t &listen_addr,
				const entity_addr_t &peer_addr)
{
  std::lock_guard l{lock};
  auto conn = ceph::make_ref<AsyncConnection>(cct, this, &dispatch_queue, w,
						listen_addr.is_msgr2(), false);
  conn->accept(std::move(cli_socket), listen_addr, peer_addr);
  accepting_conns.insert(conn);
}

AsyncConnectionRef AsyncMessenger::create_connect(
  const entity_addrvec_t& addrs, int type, bool anon)
{
  ceph_assert(ceph_mutex_is_locked(lock));

  ldout(cct, 10) << __func__ << " " << addrs
      << ", creating connection and registering" << dendl;

  // here is where we decide which of the addrs to connect to.  always prefer
  // the first one, if we support it.
  entity_addr_t target;
  for (auto& a : addrs.v) {
    if (!a.is_msgr2() && !a.is_legacy()) {
      continue;
    }
    // FIXME: for ipv4 vs ipv6, check whether local host can handle ipv6 before
    // trying it?  for now, just pick whichever is listed first.
    target = a;
    break;
  }

  // create connection
  Worker *w = stack->get_worker();
  auto conn = ceph::make_ref<AsyncConnection>(cct, this, &dispatch_queue, w,
						target.is_msgr2(), false);
  conn->anon = anon;
  conn->connect(addrs, type, target);
  if (anon) {
    anon_conns.insert(conn);
  } else {
    ceph_assert(!conns.count(addrs));
    ldout(cct, 10) << __func__ << " " << conn << " " << addrs << " "
		   << *conn->peer_addrs << dendl;
    conns[addrs] = conn;
  }
  w->get_perf_counter()->inc(l_msgr_active_connections);

  return conn;
}


ConnectionRef AsyncMessenger::get_loopback_connection()
{
  return local_connection;
}

bool AsyncMessenger::should_use_msgr2()
{
  // if we are bound to v1 only, and we are connecting to a v2 peer,
  // we cannot use the peer's v2 address. otherwise the connection
  // is assymetrical, because they would have to use v1 to connect
  // to us, and we would use v2, and connection race detection etc
  // would totally break down (among other things).  or, the other
  // end will be confused that we advertise ourselve with a v1
  // address only (that we bound to) but connected with protocol v2.
  return !did_bind || get_myaddrs().has_msgr2();
}

entity_addrvec_t AsyncMessenger::_filter_addrs(const entity_addrvec_t& addrs)
{
  if (!should_use_msgr2()) {
    ldout(cct, 10) << __func__ << " " << addrs << " limiting to v1 ()" << dendl;
    entity_addrvec_t r;
    for (auto& i : addrs.v) {
      if (i.is_msgr2()) {
	continue;
      }
      r.v.push_back(i);
    }
    return r;
  } else {
    return addrs;
  }
}

int AsyncMessenger::send_to(Message *m, int type, const entity_addrvec_t& addrs)
{
  FUNCTRACE(cct);
  ceph_assert(m);

#if defined(WITH_EVENTTRACE)
  if (m->get_type() == CEPH_MSG_OSD_OP)
    OID_EVENT_TRACE(((MOSDOp *)m)->get_oid().name.c_str(), "SEND_MSG_OSD_OP");
  else if (m->get_type() == CEPH_MSG_OSD_OPREPLY)
    OID_EVENT_TRACE(((MOSDOpReply *)m)->get_oid().name.c_str(), "SEND_MSG_OSD_OP_REPLY");
#endif

  ldout(cct, 1) << __func__ << "--> " << ceph_entity_type_name(type) << " "
      << addrs << " -- " << *m << " -- ?+"
      << m->get_data().length() << " " << m << dendl;

  if (addrs.empty()) {
    ldout(cct,0) << __func__ <<  " message " << *m
        << " with empty dest " << addrs << dendl;
    m->put();
    return -EINVAL;
  }

  if (cct->_conf->ms_dump_on_send) {
    m->encode(-1, MSG_CRC_ALL);
    ldout(cct, 0) << __func__ << " submit_message " << *m << "\n";
    m->get_payload().hexdump(*_dout);
    if (m->get_data().length() > 0) {
      *_dout << " data:\n";
      m->get_data().hexdump(*_dout);
    }
    *_dout << dendl;
    m->clear_payload();
  }

  connect_to(type, addrs, false)->send_message(m);
  return 0;
}

ConnectionRef AsyncMessenger::connect_to(int type,
					 const entity_addrvec_t& addrs,
					 bool anon, bool not_local_dest)
{
  if (!not_local_dest) {
    if (*my_addrs == addrs ||
	(addrs.v.size() == 1 &&
	 my_addrs->contains(addrs.front()))) {
      // local
      return local_connection;
    }
  }

  auto av = _filter_addrs(addrs);
  std::lock_guard l{lock};
  if (anon) {
    return create_connect(av, type, anon);
  }

  AsyncConnectionRef conn = _lookup_conn(av);
  if (conn) {
    ldout(cct, 10) << __func__ << " " << av << " existing " << conn << dendl;
  } else {
    conn = create_connect(av, type, false);
    ldout(cct, 10) << __func__ << " " << av << " new " << conn << dendl;
  }

  return conn;
}

/**
 * If my_addr doesn't have an IP set, this function
 * will fill it in from the passed addr. Otherwise it does nothing and returns.
 */
bool AsyncMessenger::set_addr_unknowns(const entity_addrvec_t &addrs)
{
  ldout(cct,1) << __func__ << " " << addrs << dendl;
  bool ret = false;
  std::lock_guard l{lock};

  entity_addrvec_t newaddrs = *my_addrs;
  for (auto& a : newaddrs.v) {
    if (a.is_blank_ip()) {
      int type = a.get_type();
      int port = a.get_port();
      uint32_t nonce = a.get_nonce();
      for (auto& b : addrs.v) {
	if (a.get_family() == b.get_family()) {
	  ldout(cct,1) << __func__ << " assuming my addr " << a
		       << " matches provided addr " << b << dendl;
	  a = b;
	  a.set_nonce(nonce);
	  a.set_type(type);
	  a.set_port(port);
	  ret = true;
	  break;
	}
      }
    }
  }
  set_myaddrs(newaddrs);
  if (ret) {
    _init_local_connection();
  }
  ldout(cct,1) << __func__ << " now " << *my_addrs << dendl;
  return ret;
}

void AsyncMessenger::shutdown_connections(bool queue_reset)
{
  ldout(cct,1) << __func__ << " " << dendl;
  std::lock_guard l{lock};
  for (const auto& c : accepting_conns) {
    ldout(cct, 5) << __func__ << " accepting_conn " << c << dendl;
    c->stop(queue_reset);
  }
  accepting_conns.clear();

  for (const auto& [e, c] : conns) {
    ldout(cct, 5) << __func__ << " mark down " << e << " " << c << dendl;
    c->stop(queue_reset);
  }
  conns.clear();

  for (const auto& c : anon_conns) {
    ldout(cct, 5) << __func__ << " mark down " << c << dendl;
    c->stop(queue_reset);
  }
  anon_conns.clear();

  {
    std::lock_guard l{deleted_lock};
    for (const auto& c : deleted_conns) {
      ldout(cct, 5) << __func__ << " delete " << c << dendl;
      c->get_perf_counter()->dec(l_msgr_active_connections);
    }
    deleted_conns.clear();
  }
}

void AsyncMessenger::mark_down_addrs(const entity_addrvec_t& addrs)
{
  std::lock_guard l{lock};
  const AsyncConnectionRef& conn = _lookup_conn(addrs);
  if (conn) {
    ldout(cct, 1) << __func__ << " " << addrs << " -- " << conn << dendl;
    conn->stop(true);
  } else {
    ldout(cct, 1) << __func__ << " " << addrs << " -- connection dne" << dendl;
  }
}

int AsyncMessenger::get_proto_version(int peer_type, bool connect) const
{
  int my_type = my_name.type();

  // set reply protocol version
  if (peer_type == my_type) {
    // internal
    return cluster_protocol;
  } else {
    // public
    switch (connect ? peer_type : my_type) {
      case CEPH_ENTITY_TYPE_OSD: return CEPH_OSDC_PROTOCOL;
      case CEPH_ENTITY_TYPE_MDS: return CEPH_MDSC_PROTOCOL;
      case CEPH_ENTITY_TYPE_MON: return CEPH_MONC_PROTOCOL;
    }
  }
  return 0;
}

int AsyncMessenger::accept_conn(const AsyncConnectionRef& conn)
{
  std::lock_guard l{lock};
  if (conn->policy.server &&
      conn->policy.lossy &&
      !conn->policy.register_lossy_clients) {
    anon_conns.insert(conn);
    conn->get_perf_counter()->inc(l_msgr_active_connections);
    return 0;
  }
  auto it = conns.find(*conn->peer_addrs);
  if (it != conns.end()) {
    auto& existing = it->second;

    // lazy delete, see "deleted_conns"
    // If conn already in, we will return 0
    std::lock_guard l{deleted_lock};
    if (deleted_conns.erase(existing)) {
      it->second->get_perf_counter()->dec(l_msgr_active_connections);
      conns.erase(it);
    } else if (conn != existing) {
      return -1;
    }
  }
  ldout(cct, 10) << __func__ << " " << conn << " " << *conn->peer_addrs << dendl;
  conns[*conn->peer_addrs] = conn;
  conn->get_perf_counter()->inc(l_msgr_active_connections);
  accepting_conns.erase(conn);
  return 0;
}


bool AsyncMessenger::learned_addr(const entity_addr_t &peer_addr_for_me)
{
  // be careful here: multiple threads may block here, and readers of
  // my_addr do NOT hold any lock.

  // this always goes from true -> false under the protection of the
  // mutex.  if it is already false, we need not retake the mutex at
  // all.
  if (!need_addr)
    return false;
  std::lock_guard l(lock);
  if (need_addr) {
    if (my_addrs->empty()) {
      auto a = peer_addr_for_me;
      a.set_type(entity_addr_t::TYPE_ANY);
      a.set_nonce(nonce);
      if (!did_bind) {
	a.set_port(0);
      }
      set_myaddrs(entity_addrvec_t(a));
      ldout(cct,10) << __func__ << " had no addrs" << dendl;
    } else {
      // fix all addrs of the same family, regardless of type (msgr2 vs legacy)
      entity_addrvec_t newaddrs = *my_addrs;
      for (auto& a : newaddrs.v) {
	if (a.is_blank_ip() &&
	    a.get_family() == peer_addr_for_me.get_family()) {
	  entity_addr_t t = peer_addr_for_me;
	  if (!did_bind) {
	    t.set_type(entity_addr_t::TYPE_ANY);
	    t.set_port(0);
	  } else {	  
	    t.set_type(a.get_type());
	    t.set_port(a.get_port());
	  }
	  t.set_nonce(a.get_nonce());
	  ldout(cct,10) << __func__ << " " << a << " -> " << t << dendl;
	  a = t;
	}
      }
      set_myaddrs(newaddrs);
    }
    ldout(cct, 1) << __func__ << " learned my addr " << *my_addrs
		  << " (peer_addr_for_me " << peer_addr_for_me << ")" << dendl;
    _init_local_connection();
    need_addr = false;
    return true;
  }
  return false;
}

void AsyncMessenger::reap_dead()
{
  ldout(cct, 1) << __func__ << " start" << dendl;

  std::lock_guard l1{lock};

  {
    std::lock_guard l2{deleted_lock};
    for (auto& c : deleted_conns) {
      ldout(cct, 5) << __func__ << " delete " << c << dendl;
      auto conns_it = conns.find(*c->peer_addrs);
      if (conns_it != conns.end() && conns_it->second == c)
        conns.erase(conns_it);
      accepting_conns.erase(c);
      anon_conns.erase(c);
      c->get_perf_counter()->dec(l_msgr_active_connections);
    }
    deleted_conns.clear();
  }
}