summaryrefslogtreecommitdiffstats
path: root/fs/ocfs2/cluster/tcp.c
blob: 9aa426e4212330994255aaa25c85a9f60ec3874b (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
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
/* -*- mode: c; c-basic-offset: 8; -*-
 *
 * vim: noexpandtab sw=8 ts=8 sts=0:
 *
 * Copyright (C) 2004 Oracle.  All rights reserved.
 *
 * This program 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 2 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
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 *
 * ----
 *
 * Callers for this were originally written against a very simple synchronus
 * API.  This implementation reflects those simple callers.  Some day I'm sure
 * we'll need to move to a more robust posting/callback mechanism.
 *
 * Transmit calls pass in kernel virtual addresses and block copying this into
 * the socket's tx buffers via a usual blocking sendmsg.  They'll block waiting
 * for a failed socket to timeout.  TX callers can also pass in a poniter to an
 * 'int' which gets filled with an errno off the wire in response to the
 * message they send.
 *
 * Handlers for unsolicited messages are registered.  Each socket has a page
 * that incoming data is copied into.  First the header, then the data.
 * Handlers are called from only one thread with a reference to this per-socket
 * page.  This page is destroyed after the handler call, so it can't be
 * referenced beyond the call.  Handlers may block but are discouraged from
 * doing so.
 *
 * Any framing errors (bad magic, large payload lengths) close a connection.
 *
 * Our sock_container holds the state we associate with a socket.  It's current
 * framing state is held there as well as the refcounting we do around when it
 * is safe to tear down the socket.  The socket is only finally torn down from
 * the container when the container loses all of its references -- so as long
 * as you hold a ref on the container you can trust that the socket is valid
 * for use with kernel socket APIs.
 *
 * Connections are initiated between a pair of nodes when the node with the
 * higher node number gets a heartbeat callback which indicates that the lower
 * numbered node has started heartbeating.  The lower numbered node is passive
 * and only accepts the connection if the higher numbered node is heartbeating.
 */

#include <linux/kernel.h>
#include <linux/jiffies.h>
#include <linux/slab.h>
#include <linux/idr.h>
#include <linux/kref.h>
#include <linux/net.h>
#include <net/tcp.h>

#include <asm/uaccess.h>

#include "heartbeat.h"
#include "tcp.h"
#include "nodemanager.h"
#define MLOG_MASK_PREFIX ML_TCP
#include "masklog.h"
#include "quorum.h"

#include "tcp_internal.h"

#define SC_NODEF_FMT "node %s (num %u) at %pI4:%u"
#define SC_NODEF_ARGS(sc) sc->sc_node->nd_name, sc->sc_node->nd_num,	\
			  &sc->sc_node->nd_ipv4_address,		\
			  ntohs(sc->sc_node->nd_ipv4_port)

/*
 * In the following two log macros, the whitespace after the ',' just
 * before ##args is intentional. Otherwise, gcc 2.95 will eat the
 * previous token if args expands to nothing.
 */
#define msglog(hdr, fmt, args...) do {					\
	typeof(hdr) __hdr = (hdr);					\
	mlog(ML_MSG, "[mag %u len %u typ %u stat %d sys_stat %d "	\
	     "key %08x num %u] " fmt,					\
	     be16_to_cpu(__hdr->magic), be16_to_cpu(__hdr->data_len), 	\
	     be16_to_cpu(__hdr->msg_type), be32_to_cpu(__hdr->status),	\
	     be32_to_cpu(__hdr->sys_status), be32_to_cpu(__hdr->key),	\
	     be32_to_cpu(__hdr->msg_num) ,  ##args);			\
} while (0)

#define sclog(sc, fmt, args...) do {					\
	typeof(sc) __sc = (sc);						\
	mlog(ML_SOCKET, "[sc %p refs %d sock %p node %u page %p "	\
	     "pg_off %zu] " fmt, __sc,					\
	     atomic_read(&__sc->sc_kref.refcount), __sc->sc_sock,	\
	    __sc->sc_node->nd_num, __sc->sc_page, __sc->sc_page_off ,	\
	    ##args);							\
} while (0)

static DEFINE_RWLOCK(o2net_handler_lock);
static struct rb_root o2net_handler_tree = RB_ROOT;

static struct o2net_node o2net_nodes[O2NM_MAX_NODES];

/* XXX someday we'll need better accounting */
static struct socket *o2net_listen_sock = NULL;

/*
 * listen work is only queued by the listening socket callbacks on the
 * o2net_wq.  teardown detaches the callbacks before destroying the workqueue.
 * quorum work is queued as sock containers are shutdown.. stop_listening
 * tears down all the node's sock containers, preventing future shutdowns
 * and queued quroum work, before canceling delayed quorum work and
 * destroying the work queue.
 */
static struct workqueue_struct *o2net_wq;
static struct work_struct o2net_listen_work;

static struct o2hb_callback_func o2net_hb_up, o2net_hb_down;
#define O2NET_HB_PRI 0x1

static struct o2net_handshake *o2net_hand;
static struct o2net_msg *o2net_keep_req, *o2net_keep_resp;

static int o2net_sys_err_translations[O2NET_ERR_MAX] =
		{[O2NET_ERR_NONE]	= 0,
		 [O2NET_ERR_NO_HNDLR]	= -ENOPROTOOPT,
		 [O2NET_ERR_OVERFLOW]	= -EOVERFLOW,
		 [O2NET_ERR_DIED]	= -EHOSTDOWN,};

/* can't quite avoid *all* internal declarations :/ */
static void o2net_sc_connect_completed(struct work_struct *work);
static void o2net_rx_until_empty(struct work_struct *work);
static void o2net_shutdown_sc(struct work_struct *work);
static void o2net_listen_data_ready(struct sock *sk, int bytes);
static void o2net_sc_send_keep_req(struct work_struct *work);
static void o2net_idle_timer(unsigned long data);
static void o2net_sc_postpone_idle(struct o2net_sock_container *sc);
static void o2net_sc_reset_idle_timer(struct o2net_sock_container *sc);

#ifdef CONFIG_DEBUG_FS
static void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
			   u32 msgkey, struct task_struct *task, u8 node)
{
	INIT_LIST_HEAD(&nst->st_net_debug_item);
	nst->st_task = task;
	nst->st_msg_type = msgtype;
	nst->st_msg_key = msgkey;
	nst->st_node = node;
}

static void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
{
	do_gettimeofday(&nst->st_sock_time);
}

static void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
{
	do_gettimeofday(&nst->st_send_time);
}

static void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
{
	do_gettimeofday(&nst->st_status_time);
}

static void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
					 struct o2net_sock_container *sc)
{
	nst->st_sc = sc;
}

static void o2net_set_nst_msg_id(struct o2net_send_tracking *nst, u32 msg_id)
{
	nst->st_id = msg_id;
}

#else  /* CONFIG_DEBUG_FS */

static inline void o2net_init_nst(struct o2net_send_tracking *nst, u32 msgtype,
				  u32 msgkey, struct task_struct *task, u8 node)
{
}

static inline void o2net_set_nst_sock_time(struct o2net_send_tracking *nst)
{
}

static inline void o2net_set_nst_send_time(struct o2net_send_tracking *nst)
{
}

static inline void o2net_set_nst_status_time(struct o2net_send_tracking *nst)
{
}

static inline void o2net_set_nst_sock_container(struct o2net_send_tracking *nst,
						struct o2net_sock_container *sc)
{
}

static inline void o2net_set_nst_msg_id(struct o2net_send_tracking *nst,
					u32 msg_id)
{
}

#endif /* CONFIG_DEBUG_FS */

static inline int o2net_reconnect_delay(void)
{
	return o2nm_single_cluster->cl_reconnect_delay_ms;
}

static inline int o2net_keepalive_delay(void)
{
	return o2nm_single_cluster->cl_keepalive_delay_ms;
}

static inline int o2net_idle_timeout(void)
{
	return o2nm_single_cluster->cl_idle_timeout_ms;
}

static inline int o2net_sys_err_to_errno(enum o2net_system_error err)
{
	int trans;
	BUG_ON(err >= O2NET_ERR_MAX);
	trans = o2net_sys_err_translations[err];

	/* Just in case we mess up the translation table above */
	BUG_ON(err != O2NET_ERR_NONE && trans == 0);
	return trans;
}

static struct o2net_node * o2net_nn_from_num(u8 node_num)
{
	BUG_ON(node_num >= ARRAY_SIZE(o2net_nodes));
	return &o2net_nodes[node_num];
}

static u8 o2net_num_from_nn(struct o2net_node *nn)
{
	BUG_ON(nn == NULL);
	return nn - o2net_nodes;
}

/* ------------------------------------------------------------ */

static int o2net_prep_nsw(struct o2net_node *nn, struct o2net_status_wait *nsw)
{
	int ret = 0;

	do {
		if (!idr_pre_get(&nn->nn_status_idr, GFP_ATOMIC)) {
			ret = -EAGAIN;
			break;
		}
		spin_lock(&nn->nn_lock);
		ret = idr_get_new(&nn->nn_status_idr, nsw, &nsw->ns_id);
		if (ret == 0)
			list_add_tail(&nsw->ns_node_item,
				      &nn->nn_status_list);
		spin_unlock(&nn->nn_lock);
	} while (ret == -EAGAIN);

	if (ret == 0)  {
		init_waitqueue_head(&nsw->ns_wq);
		nsw->ns_sys_status = O2NET_ERR_NONE;
		nsw->ns_status = 0;
	}

	return ret;
}

static void o2net_complete_nsw_locked(struct o2net_node *nn,
				      struct o2net_status_wait *nsw,
				      enum o2net_system_error sys_status,
				      s32 status)
{
	assert_spin_locked(&nn->nn_lock);

	if (!list_empty(&nsw->ns_node_item)) {
		list_del_init(&nsw->ns_node_item);
		nsw->ns_sys_status = sys_status;
		nsw->ns_status = status;
		idr_remove(&nn->nn_status_idr, nsw->ns_id);
		wake_up(&nsw->ns_wq);
	}
}

static void o2net_complete_nsw(struct o2net_node *nn,
			       struct o2net_status_wait *nsw,
			       u64 id, enum o2net_system_error sys_status,
			       s32 status)
{
	spin_lock(&nn->nn_lock);
	if (nsw == NULL) {
		if (id > INT_MAX)
			goto out;

		nsw = idr_find(&nn->nn_status_idr, id);
		if (nsw == NULL)
			goto out;
	}

	o2net_complete_nsw_locked(nn, nsw, sys_status, status);

out:
	spin_unlock(&nn->nn_lock);
	return;
}

static void o2net_complete_nodes_nsw(struct o2net_node *nn)
{
	struct o2net_status_wait *nsw, *tmp;
	unsigned int num_kills = 0;

	assert_spin_locked(&nn->nn_lock);

	list_for_each_entry_safe(nsw, tmp, &nn->nn_status_list, ns_node_item) {
		o2net_complete_nsw_locked(nn, nsw, O2NET_ERR_DIED, 0);
		num_kills++;
	}

	mlog(0, "completed %d messages for node %u\n", num_kills,
	     o2net_num_from_nn(nn));
}

static int o2net_nsw_completed(struct o2net_node *nn,
			       struct o2net_status_wait *nsw)
{
	int completed;
	spin_lock(&nn->nn_lock);
	completed = list_empty(&nsw->ns_node_item);
	spin_unlock(&nn->nn_lock);
	return completed;
}

/* ------------------------------------------------------------ */

static void sc_kref_release(struct kref *kref)
{
	struct o2net_sock_container *sc = container_of(kref,
					struct o2net_sock_container, sc_kref);
	BUG_ON(timer_pending(&sc->sc_idle_timeout));

	sclog(sc, "releasing\n");

	if (sc->sc_sock) {
		sock_release(sc->sc_sock);
		sc->sc_sock = NULL;
	}

	o2nm_node_put(sc->sc_node);
	sc->sc_node = NULL;

	o2net_debug_del_sc(sc);
	kfree(sc);
}

static void sc_put(struct o2net_sock_container *sc)
{
	sclog(sc, "put\n");
	kref_put(&sc->sc_kref, sc_kref_release);
}
static void sc_get(struct o2net_sock_container *sc)
{
	sclog(sc, "get\n");
	kref_get(&sc->sc_kref);
}
static struct o2net_sock_container *sc_alloc(struct o2nm_node *node)
{
	struct o2net_sock_container *sc, *ret = NULL;
	struct page *page = NULL;

	page = alloc_page(GFP_NOFS);
	sc = kzalloc(sizeof(*sc), GFP_NOFS);
	if (sc == NULL || page == NULL)
		goto out;

	kref_init(&sc->sc_kref);
	o2nm_node_get(node);
	sc->sc_node = node;

	INIT_WORK(&sc->sc_connect_work, o2net_sc_connect_completed);
	INIT_WORK(&sc->sc_rx_work, o2net_rx_until_empty);
	INIT_WORK(&sc->sc_shutdown_work, o2net_shutdown_sc);
	INIT_DELAYED_WORK(&sc->sc_keepalive_work, o2net_sc_send_keep_req);

	init_timer(&sc->sc_idle_timeout);
	sc->sc_idle_timeout.function = o2net_idle_timer;
	sc->sc_idle_timeout.data = (unsigned long)sc;

	sclog(sc, "alloced\n");

	ret = sc;
	sc->sc_page = page;
	o2net_debug_add_sc(sc);
	sc = NULL;
	page = NULL;

out:
	if (page)
		__free_page(page);
	kfree(sc);

	return ret;
}

/* ------------------------------------------------------------ */

static void o2net_sc_queue_work(struct o2net_sock_container *sc,
				struct work_struct *work)
{
	sc_get(sc);
	if (!queue_work(o2net_wq, work))
		sc_put(sc);
}
static void o2net_sc_queue_delayed_work(struct o2net_sock_container *sc,
					struct delayed_work *work,
					int delay)
{
	sc_get(sc);
	if (!queue_delayed_work(o2net_wq, work, delay))
		sc_put(sc);
}
static void o2net_sc_cancel_delayed_work(struct o2net_sock_container *sc,
					 struct delayed_work *work)
{
	if (cancel_delayed_work(work))
		sc_put(sc);
}

static atomic_t o2net_connected_peers = ATOMIC_INIT(0);

int o2net_num_connected_peers(void)
{
	return atomic_read(&o2net_connected_peers);
}

static void o2net_set_nn_state(struct o2net_node *nn,
			       struct o2net_sock_container *sc,
			       unsigned valid, int err)
{
	int was_valid = nn->nn_sc_valid;
	int was_err = nn->nn_persistent_error;
	struct o2net_sock_container *old_sc = nn->nn_sc;

	assert_spin_locked(&nn->nn_lock);

	if (old_sc && !sc)
		atomic_dec(&o2net_connected_peers);
	else if (!old_sc && sc)
		atomic_inc(&o2net_connected_peers);

	/* the node num comparison and single connect/accept path should stop
	 * an non-null sc from being overwritten with another */
	BUG_ON(sc && nn->nn_sc && nn->nn_sc != sc);
	mlog_bug_on_msg(err && valid, "err %d valid %u\n", err, valid);
	mlog_bug_on_msg(valid && !sc, "valid %u sc %p\n", valid, sc);

	if (was_valid && !valid && err == 0)
		err = -ENOTCONN;

	mlog(ML_CONN, "node %u sc: %p -> %p, valid %u -> %u, err %d -> %d\n",
	     o2net_num_from_nn(nn), nn->nn_sc, sc, nn->nn_sc_valid, valid,
	     nn->nn_persistent_error, err);

	nn->nn_sc = sc;
	nn->nn_sc_valid = valid ? 1 : 0;
	nn->nn_persistent_error = err;

	/* mirrors o2net_tx_can_proceed() */
	if (nn->nn_persistent_error || nn->nn_sc_valid)
		wake_up(&nn->nn_sc_wq);

	if (!was_err && nn->nn_persistent_error) {
		o2quo_conn_err(o2net_num_from_nn(nn));
		queue_delayed_work(o2net_wq, &nn->nn_still_up,
				   msecs_to_jiffies(O2NET_QUORUM_DELAY_MS));
	}

	if (was_valid && !valid) {
		printk(KERN_NOTICE "o2net: no longer connected to "
		       SC_NODEF_FMT "\n", SC_NODEF_ARGS(old_sc));
		o2net_complete_nodes_nsw(nn);
	}

	if (!was_valid && valid) {
		o2quo_conn_up(o2net_num_from_nn(nn));
		cancel_delayed_work(&nn->nn_connect_expired);
		printk(KERN_NOTICE "o2net: %s " SC_NODEF_FMT "\n",
		       o2nm_this_node() > sc->sc_node->nd_num ?
		       		"connected to" : "accepted connection from",
		       SC_NODEF_ARGS(sc));
	}

	/* trigger the connecting worker func as long as we're not valid,
	 * it will back off if it shouldn't connect.  This can be called
	 * from node config teardown and so needs to be careful about
	 * the work queue actually being up. */
	if (!valid && o2net_wq) {
		unsigned long delay;
		/* delay if we're withing a RECONNECT_DELAY of the
		 * last attempt */
		delay = (nn->nn_last_connect_attempt +
			 msecs_to_jiffies(o2net_reconnect_delay()))
			- jiffies;
		if (delay > msecs_to_jiffies(o2net_reconnect_delay()))
			delay = 0;
		mlog(ML_CONN, "queueing conn attempt in %lu jiffies\n", delay);
		queue_delayed_work(o2net_wq, &nn->nn_connect_work, delay);

		/*
		 * Delay the expired work after idle timeout.
		 *
		 * We might have lots of failed connection attempts that run
		 * through here but we only cancel the connect_expired work when
		 * a connection attempt succeeds.  So only the first enqueue of
		 * the connect_expired work will do anything.  The rest will see
		 * that it's already queued and do nothing.
		 */
		delay += msecs_to_jiffies(o2net_idle_timeout());
		queue_delayed_work(o2net_wq, &nn->nn_connect_expired, delay);
	}

	/* keep track of the nn's sc ref for the caller */
	if ((old_sc == NULL) && sc)
		sc_get(sc);
	if (old_sc && (old_sc != sc)) {
		o2net_sc_queue_work(old_sc, &old_sc->sc_shutdown_work);
		sc_put(old_sc);
	}
}

/* see o2net_register_callbacks() */
static void o2net_data_ready(struct sock *sk, int bytes)
{
	void (*ready)(struct sock *sk, int bytes);

	read_lock(&sk->sk_callback_lock);
	if (sk->sk_user_data) {
		struct o2net_sock_container *sc = sk->sk_user_data;
		sclog(sc, "data_ready hit\n");
		do_gettimeofday(&sc->sc_tv_data_ready);
		o2net_sc_queue_work(sc, &sc->sc_rx_work);
		ready = sc->sc_data_ready;
	} else {
		ready = sk->sk_data_ready;
	}
	read_unlock(&sk->sk_callback_lock);

	ready(sk, bytes);
}

/* see o2net_register_callbacks() */
static void o2net_state_change(struct sock *sk)
{
	void (*state_change)(struct sock *sk);
	struct o2net_sock_container *sc;

	read_lock(&sk->sk_callback_lock);
	sc = sk->sk_user_data;
	if (sc == NULL) {
		state_change = sk->sk_state_change;
		goto out;
	}

	sclog(sc, "state_change to %d\n", sk->sk_state);

	state_change = sc->sc_state_change;

	switch(sk->sk_state) {
		/* ignore connecting sockets as they make progress */
		case TCP_SYN_SENT:
		case TCP_SYN_RECV:
			break;
		case TCP_ESTABLISHED:
			o2net_sc_queue_work(sc, &sc->sc_connect_work);
			break;
		default:
			printk(KERN_INFO "o2net: connection to " SC_NODEF_FMT
			      " shutdown, state %d\n",
			      SC_NODEF_ARGS(sc), sk->sk_state);
			o2net_sc_queue_work(sc, &sc->sc_shutdown_work);
			break;
	}
out:
	read_unlock(&sk->sk_callback_lock);
	state_change(sk);
}

/*
 * we register callbacks so we can queue work on events before calling
 * the original callbacks.  our callbacks our careful to test user_data
 * to discover when they've reaced with o2net_unregister_callbacks().
 */
static void o2net_register_callbacks(struct sock *sk,
				     struct o2net_sock_container *sc)
{
	write_lock_bh(&sk->sk_callback_lock);

	/* accepted sockets inherit the old listen socket data ready */
	if (sk->sk_data_ready == o2net_listen_data_ready) {
		sk->sk_data_ready = sk->sk_user_data;
		sk->sk_user_data = NULL;
	}

	BUG_ON(sk->sk_user_data != NULL);
	sk->sk_user_data = sc;
	sc_get(sc);

	sc->sc_data_ready = sk->sk_data_ready;
	sc->sc_state_change = sk->sk_state_change;
	sk->sk_data_ready = o2net_data_ready;
	sk->sk_state_change = o2net_state_change;

	mutex_init(&sc->sc_send_lock);

	write_unlock_bh(&sk->sk_callback_lock);
}

static int o2net_unregister_callbacks(struct sock *sk,
			           struct o2net_sock_container *sc)
{
	int ret = 0;

	write_lock_bh(&sk->sk_callback_lock);
	if (sk->sk_user_data == sc) {
		ret = 1;
		sk->sk_user_data = NULL;
		sk->sk_data_ready = sc->sc_data_ready;
		sk->sk_state_change = sc->sc_state_change;
	}
	write_unlock_bh(&sk->sk_callback_lock);

	return ret;
}

/*
 * this is a little helper that is called by callers who have seen a problem
 * with an sc and want to detach it from the nn if someone already hasn't beat
 * them to it.  if an error is given then the shutdown will be persistent
 * and pending transmits will be canceled.
 */
static void o2net_ensure_shutdown(struct o2net_node *nn,
			           struct o2net_sock_container *sc,
				   int err)
{
	spin_lock(&nn->nn_lock);
	if (nn->nn_sc == sc)
		o2net_set_nn_state(nn, NULL, 0, err);
	spin_unlock(&nn->nn_lock);
}

/*
 * This work queue function performs the blocking parts of socket shutdown.  A
 * few paths lead here.  set_nn_state will trigger this callback if it sees an
 * sc detached from the nn.  state_change will also trigger this callback
 * directly when it sees errors.  In that case we need to call set_nn_state
 * ourselves as state_change couldn't get the nn_lock and call set_nn_state
 * itself.
 */
static void o2net_shutdown_sc(struct work_struct *work)
{
	struct o2net_sock_container *sc =
		container_of(work, struct o2net_sock_container,
			     sc_shutdown_work);
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);

	sclog(sc, "shutting down\n");

	/* drop the callbacks ref and call shutdown only once */
	if (o2net_unregister_callbacks(sc->sc_sock->sk, sc)) {
		/* we shouldn't flush as we're in the thread, the
		 * races with pending sc work structs are harmless */
		del_timer_sync(&sc->sc_idle_timeout);
		o2net_sc_cancel_delayed_work(sc, &sc->sc_keepalive_work);
		sc_put(sc);
		kernel_sock_shutdown(sc->sc_sock, SHUT_RDWR);
	}

	/* not fatal so failed connects before the other guy has our
	 * heartbeat can be retried */
	o2net_ensure_shutdown(nn, sc, 0);
	sc_put(sc);
}

/* ------------------------------------------------------------ */

static int o2net_handler_cmp(struct o2net_msg_handler *nmh, u32 msg_type,
			     u32 key)
{
	int ret = memcmp(&nmh->nh_key, &key, sizeof(key));

	if (ret == 0)
		ret = memcmp(&nmh->nh_msg_type, &msg_type, sizeof(msg_type));

	return ret;
}

static struct o2net_msg_handler *
o2net_handler_tree_lookup(u32 msg_type, u32 key, struct rb_node ***ret_p,
			  struct rb_node **ret_parent)
{
        struct rb_node **p = &o2net_handler_tree.rb_node;
        struct rb_node *parent = NULL;
	struct o2net_msg_handler *nmh, *ret = NULL;
	int cmp;

        while (*p) {
                parent = *p;
                nmh = rb_entry(parent, struct o2net_msg_handler, nh_node);
		cmp = o2net_handler_cmp(nmh, msg_type, key);

                if (cmp < 0)
                        p = &(*p)->rb_left;
                else if (cmp > 0)
                        p = &(*p)->rb_right;
                else {
			ret = nmh;
                        break;
		}
        }

        if (ret_p != NULL)
                *ret_p = p;
        if (ret_parent != NULL)
                *ret_parent = parent;

        return ret;
}

static void o2net_handler_kref_release(struct kref *kref)
{
	struct o2net_msg_handler *nmh;
	nmh = container_of(kref, struct o2net_msg_handler, nh_kref);

	kfree(nmh);
}

static void o2net_handler_put(struct o2net_msg_handler *nmh)
{
	kref_put(&nmh->nh_kref, o2net_handler_kref_release);
}

/* max_len is protection for the handler func.  incoming messages won't
 * be given to the handler if their payload is longer than the max. */
int o2net_register_handler(u32 msg_type, u32 key, u32 max_len,
			   o2net_msg_handler_func *func, void *data,
			   o2net_post_msg_handler_func *post_func,
			   struct list_head *unreg_list)
{
	struct o2net_msg_handler *nmh = NULL;
	struct rb_node **p, *parent;
	int ret = 0;

	if (max_len > O2NET_MAX_PAYLOAD_BYTES) {
		mlog(0, "max_len for message handler out of range: %u\n",
			max_len);
		ret = -EINVAL;
		goto out;
	}

	if (!msg_type) {
		mlog(0, "no message type provided: %u, %p\n", msg_type, func);
		ret = -EINVAL;
		goto out;

	}
	if (!func) {
		mlog(0, "no message handler provided: %u, %p\n",
		       msg_type, func);
		ret = -EINVAL;
		goto out;
	}

       	nmh = kzalloc(sizeof(struct o2net_msg_handler), GFP_NOFS);
	if (nmh == NULL) {
		ret = -ENOMEM;
		goto out;
	}

	nmh->nh_func = func;
	nmh->nh_func_data = data;
	nmh->nh_post_func = post_func;
	nmh->nh_msg_type = msg_type;
	nmh->nh_max_len = max_len;
	nmh->nh_key = key;
	/* the tree and list get this ref.. they're both removed in
	 * unregister when this ref is dropped */
	kref_init(&nmh->nh_kref);
	INIT_LIST_HEAD(&nmh->nh_unregister_item);

	write_lock(&o2net_handler_lock);
	if (o2net_handler_tree_lookup(msg_type, key, &p, &parent))
		ret = -EEXIST;
	else {
	        rb_link_node(&nmh->nh_node, parent, p);
		rb_insert_color(&nmh->nh_node, &o2net_handler_tree);
		list_add_tail(&nmh->nh_unregister_item, unreg_list);

		mlog(ML_TCP, "registered handler func %p type %u key %08x\n",
		     func, msg_type, key);
		/* we've had some trouble with handlers seemingly vanishing. */
		mlog_bug_on_msg(o2net_handler_tree_lookup(msg_type, key, &p,
							  &parent) == NULL,
			        "couldn't find handler we *just* registerd "
				"for type %u key %08x\n", msg_type, key);
	}
	write_unlock(&o2net_handler_lock);
	if (ret)
		goto out;

out:
	if (ret)
		kfree(nmh);

	return ret;
}
EXPORT_SYMBOL_GPL(o2net_register_handler);

void o2net_unregister_handler_list(struct list_head *list)
{
	struct o2net_msg_handler *nmh, *n;

	write_lock(&o2net_handler_lock);
	list_for_each_entry_safe(nmh, n, list, nh_unregister_item) {
		mlog(ML_TCP, "unregistering handler func %p type %u key %08x\n",
		     nmh->nh_func, nmh->nh_msg_type, nmh->nh_key);
		rb_erase(&nmh->nh_node, &o2net_handler_tree);
		list_del_init(&nmh->nh_unregister_item);
		kref_put(&nmh->nh_kref, o2net_handler_kref_release);
	}
	write_unlock(&o2net_handler_lock);
}
EXPORT_SYMBOL_GPL(o2net_unregister_handler_list);

static struct o2net_msg_handler *o2net_handler_get(u32 msg_type, u32 key)
{
	struct o2net_msg_handler *nmh;

	read_lock(&o2net_handler_lock);
	nmh = o2net_handler_tree_lookup(msg_type, key, NULL, NULL);
	if (nmh)
		kref_get(&nmh->nh_kref);
	read_unlock(&o2net_handler_lock);

	return nmh;
}

/* ------------------------------------------------------------ */

static int o2net_recv_tcp_msg(struct socket *sock, void *data, size_t len)
{
	int ret;
	mm_segment_t oldfs;
	struct kvec vec = {
		.iov_len = len,
		.iov_base = data,
	};
	struct msghdr msg = {
		.msg_iovlen = 1,
		.msg_iov = (struct iovec *)&vec,
       		.msg_flags = MSG_DONTWAIT,
	};

	oldfs = get_fs();
	set_fs(get_ds());
	ret = sock_recvmsg(sock, &msg, len, msg.msg_flags);
	set_fs(oldfs);

	return ret;
}

static int o2net_send_tcp_msg(struct socket *sock, struct kvec *vec,
			      size_t veclen, size_t total)
{
	int ret;
	mm_segment_t oldfs;
	struct msghdr msg = {
		.msg_iov = (struct iovec *)vec,
		.msg_iovlen = veclen,
	};

	if (sock == NULL) {
		ret = -EINVAL;
		goto out;
	}

	oldfs = get_fs();
	set_fs(get_ds());
	ret = sock_sendmsg(sock, &msg, total);
	set_fs(oldfs);
	if (ret != total) {
		mlog(ML_ERROR, "sendmsg returned %d instead of %zu\n", ret,
		     total);
		if (ret >= 0)
			ret = -EPIPE; /* should be smarter, I bet */
		goto out;
	}

	ret = 0;
out:
	if (ret < 0)
		mlog(0, "returning error: %d\n", ret);
	return ret;
}

static void o2net_sendpage(struct o2net_sock_container *sc,
			   void *kmalloced_virt,
			   size_t size)
{
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
	ssize_t ret;

	while (1) {
		mutex_lock(&sc->sc_send_lock);
		ret = sc->sc_sock->ops->sendpage(sc->sc_sock,
						 virt_to_page(kmalloced_virt),
						 (long)kmalloced_virt & ~PAGE_MASK,
						 size, MSG_DONTWAIT);
		mutex_unlock(&sc->sc_send_lock);
		if (ret == size)
			break;
		if (ret == (ssize_t)-EAGAIN) {
			mlog(0, "sendpage of size %zu to " SC_NODEF_FMT
			     " returned EAGAIN\n", size, SC_NODEF_ARGS(sc));
			cond_resched();
			continue;
		}
		mlog(ML_ERROR, "sendpage of size %zu to " SC_NODEF_FMT
		     " failed with %zd\n", size, SC_NODEF_ARGS(sc), ret);
		o2net_ensure_shutdown(nn, sc, 0);
		break;
	}
}

static void o2net_init_msg(struct o2net_msg *msg, u16 data_len, u16 msg_type, u32 key)
{
	memset(msg, 0, sizeof(struct o2net_msg));
	msg->magic = cpu_to_be16(O2NET_MSG_MAGIC);
	msg->data_len = cpu_to_be16(data_len);
	msg->msg_type = cpu_to_be16(msg_type);
	msg->sys_status = cpu_to_be32(O2NET_ERR_NONE);
	msg->status = 0;
	msg->key = cpu_to_be32(key);
}

static int o2net_tx_can_proceed(struct o2net_node *nn,
			        struct o2net_sock_container **sc_ret,
				int *error)
{
	int ret = 0;

	spin_lock(&nn->nn_lock);
	if (nn->nn_persistent_error) {
		ret = 1;
		*sc_ret = NULL;
		*error = nn->nn_persistent_error;
	} else if (nn->nn_sc_valid) {
		kref_get(&nn->nn_sc->sc_kref);

		ret = 1;
		*sc_ret = nn->nn_sc;
		*error = 0;
	}
	spin_unlock(&nn->nn_lock);

	return ret;
}

int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec,
			   size_t caller_veclen, u8 target_node, int *status)
{
	int ret = 0;
	struct o2net_msg *msg = NULL;
	size_t veclen, caller_bytes = 0;
	struct kvec *vec = NULL;
	struct o2net_sock_container *sc = NULL;
	struct o2net_node *nn = o2net_nn_from_num(target_node);
	struct o2net_status_wait nsw = {
		.ns_node_item = LIST_HEAD_INIT(nsw.ns_node_item),
	};
	struct o2net_send_tracking nst;

	o2net_init_nst(&nst, msg_type, key, current, target_node);

	if (o2net_wq == NULL) {
		mlog(0, "attempt to tx without o2netd running\n");
		ret = -ESRCH;
		goto out;
	}

	if (caller_veclen == 0) {
		mlog(0, "bad kvec array length\n");
		ret = -EINVAL;
		goto out;
	}

	caller_bytes = iov_length((struct iovec *)caller_vec, caller_veclen);
	if (caller_bytes > O2NET_MAX_PAYLOAD_BYTES) {
		mlog(0, "total payload len %zu too large\n", caller_bytes);
		ret = -EINVAL;
		goto out;
	}

	if (target_node == o2nm_this_node()) {
		ret = -ELOOP;
		goto out;
	}

	o2net_debug_add_nst(&nst);

	o2net_set_nst_sock_time(&nst);

	wait_event(nn->nn_sc_wq, o2net_tx_can_proceed(nn, &sc, &ret));
	if (ret)
		goto out;

	o2net_set_nst_sock_container(&nst, sc);

	veclen = caller_veclen + 1;
	vec = kmalloc(sizeof(struct kvec) * veclen, GFP_ATOMIC);
	if (vec == NULL) {
		mlog(0, "failed to %zu element kvec!\n", veclen);
		ret = -ENOMEM;
		goto out;
	}

	msg = kmalloc(sizeof(struct o2net_msg), GFP_ATOMIC);
	if (!msg) {
		mlog(0, "failed to allocate a o2net_msg!\n");
		ret = -ENOMEM;
		goto out;
	}

	o2net_init_msg(msg, caller_bytes, msg_type, key);

	vec[0].iov_len = sizeof(struct o2net_msg);
	vec[0].iov_base = msg;
	memcpy(&vec[1], caller_vec, caller_veclen * sizeof(struct kvec));

	ret = o2net_prep_nsw(nn, &nsw);
	if (ret)
		goto out;

	msg->msg_num = cpu_to_be32(nsw.ns_id);
	o2net_set_nst_msg_id(&nst, nsw.ns_id);

	o2net_set_nst_send_time(&nst);

	/* finally, convert the message header to network byte-order
	 * and send */
	mutex_lock(&sc->sc_send_lock);
	ret = o2net_send_tcp_msg(sc->sc_sock, vec, veclen,
				 sizeof(struct o2net_msg) + caller_bytes);
	mutex_unlock(&sc->sc_send_lock);
	msglog(msg, "sending returned %d\n", ret);
	if (ret < 0) {
		mlog(0, "error returned from o2net_send_tcp_msg=%d\n", ret);
		goto out;
	}

	/* wait on other node's handler */
	o2net_set_nst_status_time(&nst);
	wait_event(nsw.ns_wq, o2net_nsw_completed(nn, &nsw));

	/* Note that we avoid overwriting the callers status return
	 * variable if a system error was reported on the other
	 * side. Callers beware. */
	ret = o2net_sys_err_to_errno(nsw.ns_sys_status);
	if (status && !ret)
		*status = nsw.ns_status;

	mlog(0, "woken, returning system status %d, user status %d\n",
	     ret, nsw.ns_status);
out:
	o2net_debug_del_nst(&nst); /* must be before dropping sc and node */
	if (sc)
		sc_put(sc);
	if (vec)
		kfree(vec);
	if (msg)
		kfree(msg);
	o2net_complete_nsw(nn, &nsw, 0, 0, 0);
	return ret;
}
EXPORT_SYMBOL_GPL(o2net_send_message_vec);

int o2net_send_message(u32 msg_type, u32 key, void *data, u32 len,
		       u8 target_node, int *status)
{
	struct kvec vec = {
		.iov_base = data,
		.iov_len = len,
	};
	return o2net_send_message_vec(msg_type, key, &vec, 1,
				      target_node, status);
}
EXPORT_SYMBOL_GPL(o2net_send_message);

static int o2net_send_status_magic(struct socket *sock, struct o2net_msg *hdr,
				   enum o2net_system_error syserr, int err)
{
	struct kvec vec = {
		.iov_base = hdr,
		.iov_len = sizeof(struct o2net_msg),
	};

	BUG_ON(syserr >= O2NET_ERR_MAX);

	/* leave other fields intact from the incoming message, msg_num
	 * in particular */
	hdr->sys_status = cpu_to_be32(syserr);
	hdr->status = cpu_to_be32(err);
	hdr->magic = cpu_to_be16(O2NET_MSG_STATUS_MAGIC);  // twiddle the magic
	hdr->data_len = 0;

	msglog(hdr, "about to send status magic %d\n", err);
	/* hdr has been in host byteorder this whole time */
	return o2net_send_tcp_msg(sock, &vec, 1, sizeof(struct o2net_msg));
}

/* this returns -errno if the header was unknown or too large, etc.
 * after this is called the buffer us reused for the next message */
static int o2net_process_message(struct o2net_sock_container *sc,
				 struct o2net_msg *hdr)
{
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
	int ret = 0, handler_status;
	enum  o2net_system_error syserr;
	struct o2net_msg_handler *nmh = NULL;
	void *ret_data = NULL;

	msglog(hdr, "processing message\n");

	o2net_sc_postpone_idle(sc);

	switch(be16_to_cpu(hdr->magic)) {
		case O2NET_MSG_STATUS_MAGIC:
			/* special type for returning message status */
			o2net_complete_nsw(nn, NULL,
					   be32_to_cpu(hdr->msg_num),
					   be32_to_cpu(hdr->sys_status),
					   be32_to_cpu(hdr->status));
			goto out;
		case O2NET_MSG_KEEP_REQ_MAGIC:
			o2net_sendpage(sc, o2net_keep_resp,
				       sizeof(*o2net_keep_resp));
			goto out;
		case O2NET_MSG_KEEP_RESP_MAGIC:
			goto out;
		case O2NET_MSG_MAGIC:
			break;
		default:
			msglog(hdr, "bad magic\n");
			ret = -EINVAL;
			goto out;
			break;
	}

	/* find a handler for it */
	handler_status = 0;
	nmh = o2net_handler_get(be16_to_cpu(hdr->msg_type),
				be32_to_cpu(hdr->key));
	if (!nmh) {
		mlog(ML_TCP, "couldn't find handler for type %u key %08x\n",
		     be16_to_cpu(hdr->msg_type), be32_to_cpu(hdr->key));
		syserr = O2NET_ERR_NO_HNDLR;
		goto out_respond;
	}

	syserr = O2NET_ERR_NONE;

	if (be16_to_cpu(hdr->data_len) > nmh->nh_max_len)
		syserr = O2NET_ERR_OVERFLOW;

	if (syserr != O2NET_ERR_NONE)
		goto out_respond;

	do_gettimeofday(&sc->sc_tv_func_start);
	sc->sc_msg_key = be32_to_cpu(hdr->key);
	sc->sc_msg_type = be16_to_cpu(hdr->msg_type);
	handler_status = (nmh->nh_func)(hdr, sizeof(struct o2net_msg) +
					     be16_to_cpu(hdr->data_len),
					nmh->nh_func_data, &ret_data);
	do_gettimeofday(&sc->sc_tv_func_stop);

out_respond:
	/* this destroys the hdr, so don't use it after this */
	mutex_lock(&sc->sc_send_lock);
	ret = o2net_send_status_magic(sc->sc_sock, hdr, syserr,
				      handler_status);
	mutex_unlock(&sc->sc_send_lock);
	hdr = NULL;
	mlog(0, "sending handler status %d, syserr %d returned %d\n",
	     handler_status, syserr, ret);

	if (nmh) {
		BUG_ON(ret_data != NULL && nmh->nh_post_func == NULL);
		if (nmh->nh_post_func)
			(nmh->nh_post_func)(handler_status, nmh->nh_func_data,
					    ret_data);
	}

out:
	if (nmh)
		o2net_handler_put(nmh);
	return ret;
}

static int o2net_check_handshake(struct o2net_sock_container *sc)
{
	struct o2net_handshake *hand = page_address(sc->sc_page);
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);

	if (hand->protocol_version != cpu_to_be64(O2NET_PROTOCOL_VERSION)) {
		mlog(ML_NOTICE, SC_NODEF_FMT " advertised net protocol "
		     "version %llu but %llu is required, disconnecting\n",
		     SC_NODEF_ARGS(sc),
		     (unsigned long long)be64_to_cpu(hand->protocol_version),
		     O2NET_PROTOCOL_VERSION);

		/* don't bother reconnecting if its the wrong version. */
		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
		return -1;
	}

	/*
	 * Ensure timeouts are consistent with other nodes, otherwise
	 * we can end up with one node thinking that the other must be down,
	 * but isn't. This can ultimately cause corruption.
	 */
	if (be32_to_cpu(hand->o2net_idle_timeout_ms) !=
				o2net_idle_timeout()) {
		mlog(ML_NOTICE, SC_NODEF_FMT " uses a network idle timeout of "
		     "%u ms, but we use %u ms locally.  disconnecting\n",
		     SC_NODEF_ARGS(sc),
		     be32_to_cpu(hand->o2net_idle_timeout_ms),
		     o2net_idle_timeout());
		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
		return -1;
	}

	if (be32_to_cpu(hand->o2net_keepalive_delay_ms) !=
			o2net_keepalive_delay()) {
		mlog(ML_NOTICE, SC_NODEF_FMT " uses a keepalive delay of "
		     "%u ms, but we use %u ms locally.  disconnecting\n",
		     SC_NODEF_ARGS(sc),
		     be32_to_cpu(hand->o2net_keepalive_delay_ms),
		     o2net_keepalive_delay());
		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
		return -1;
	}

	if (be32_to_cpu(hand->o2hb_heartbeat_timeout_ms) !=
			O2HB_MAX_WRITE_TIMEOUT_MS) {
		mlog(ML_NOTICE, SC_NODEF_FMT " uses a heartbeat timeout of "
		     "%u ms, but we use %u ms locally.  disconnecting\n",
		     SC_NODEF_ARGS(sc),
		     be32_to_cpu(hand->o2hb_heartbeat_timeout_ms),
		     O2HB_MAX_WRITE_TIMEOUT_MS);
		o2net_ensure_shutdown(nn, sc, -ENOTCONN);
		return -1;
	}

	sc->sc_handshake_ok = 1;

	spin_lock(&nn->nn_lock);
	/* set valid and queue the idle timers only if it hasn't been
	 * shut down already */
	if (nn->nn_sc == sc) {
		o2net_sc_reset_idle_timer(sc);
		atomic_set(&nn->nn_timeout, 0);
		o2net_set_nn_state(nn, sc, 1, 0);
	}
	spin_unlock(&nn->nn_lock);

	/* shift everything up as though it wasn't there */
	sc->sc_page_off -= sizeof(struct o2net_handshake);
	if (sc->sc_page_off)
		memmove(hand, hand + 1, sc->sc_page_off);

	return 0;
}

/* this demuxes the queued rx bytes into header or payload bits and calls
 * handlers as each full message is read off the socket.  it returns -error,
 * == 0 eof, or > 0 for progress made.*/
static int o2net_advance_rx(struct o2net_sock_container *sc)
{
	struct o2net_msg *hdr;
	int ret = 0;
	void *data;
	size_t datalen;

	sclog(sc, "receiving\n");
	do_gettimeofday(&sc->sc_tv_advance_start);

	if (unlikely(sc->sc_handshake_ok == 0)) {
		if(sc->sc_page_off < sizeof(struct o2net_handshake)) {
			data = page_address(sc->sc_page) + sc->sc_page_off;
			datalen = sizeof(struct o2net_handshake) - sc->sc_page_off;
			ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
			if (ret > 0)
				sc->sc_page_off += ret;
		}

		if (sc->sc_page_off == sizeof(struct o2net_handshake)) {
			o2net_check_handshake(sc);
			if (unlikely(sc->sc_handshake_ok == 0))
				ret = -EPROTO;
		}
		goto out;
	}

	/* do we need more header? */
	if (sc->sc_page_off < sizeof(struct o2net_msg)) {
		data = page_address(sc->sc_page) + sc->sc_page_off;
		datalen = sizeof(struct o2net_msg) - sc->sc_page_off;
		ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
		if (ret > 0) {
			sc->sc_page_off += ret;
			/* only swab incoming here.. we can
			 * only get here once as we cross from
			 * being under to over */
			if (sc->sc_page_off == sizeof(struct o2net_msg)) {
				hdr = page_address(sc->sc_page);
				if (be16_to_cpu(hdr->data_len) >
				    O2NET_MAX_PAYLOAD_BYTES)
					ret = -EOVERFLOW;
			}
		}
		if (ret <= 0)
			goto out;
	}

	if (sc->sc_page_off < sizeof(struct o2net_msg)) {
		/* oof, still don't have a header */
		goto out;
	}

	/* this was swabbed above when we first read it */
	hdr = page_address(sc->sc_page);

	msglog(hdr, "at page_off %zu\n", sc->sc_page_off);

	/* do we need more payload? */
	if (sc->sc_page_off - sizeof(struct o2net_msg) < be16_to_cpu(hdr->data_len)) {
		/* need more payload */
		data = page_address(sc->sc_page) + sc->sc_page_off;
		datalen = (sizeof(struct o2net_msg) + be16_to_cpu(hdr->data_len)) -
			  sc->sc_page_off;
		ret = o2net_recv_tcp_msg(sc->sc_sock, data, datalen);
		if (ret > 0)
			sc->sc_page_off += ret;
		if (ret <= 0)
			goto out;
	}

	if (sc->sc_page_off - sizeof(struct o2net_msg) == be16_to_cpu(hdr->data_len)) {
		/* we can only get here once, the first time we read
		 * the payload.. so set ret to progress if the handler
		 * works out. after calling this the message is toast */
		ret = o2net_process_message(sc, hdr);
		if (ret == 0)
			ret = 1;
		sc->sc_page_off = 0;
	}

out:
	sclog(sc, "ret = %d\n", ret);
	do_gettimeofday(&sc->sc_tv_advance_stop);
	return ret;
}

/* this work func is triggerd by data ready.  it reads until it can read no
 * more.  it interprets 0, eof, as fatal.  if data_ready hits while we're doing
 * our work the work struct will be marked and we'll be called again. */
static void o2net_rx_until_empty(struct work_struct *work)
{
	struct o2net_sock_container *sc =
		container_of(work, struct o2net_sock_container, sc_rx_work);
	int ret;

	do {
		ret = o2net_advance_rx(sc);
	} while (ret > 0);

	if (ret <= 0 && ret != -EAGAIN) {
		struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
		sclog(sc, "saw error %d, closing\n", ret);
		/* not permanent so read failed handshake can retry */
		o2net_ensure_shutdown(nn, sc, 0);
	}

	sc_put(sc);
}

static int o2net_set_nodelay(struct socket *sock)
{
	int ret, val = 1;
	mm_segment_t oldfs;

	oldfs = get_fs();
	set_fs(KERNEL_DS);

	/*
	 * Dear unsuspecting programmer,
	 *
	 * Don't use sock_setsockopt() for SOL_TCP.  It doesn't check its level
	 * argument and assumes SOL_SOCKET so, say, your TCP_NODELAY will
	 * silently turn into SO_DEBUG.
	 *
	 * Yours,
	 * Keeper of hilariously fragile interfaces.
	 */
	ret = sock->ops->setsockopt(sock, SOL_TCP, TCP_NODELAY,
				    (char __user *)&val, sizeof(val));

	set_fs(oldfs);
	return ret;
}

static void o2net_initialize_handshake(void)
{
	o2net_hand->o2hb_heartbeat_timeout_ms = cpu_to_be32(
		O2HB_MAX_WRITE_TIMEOUT_MS);
	o2net_hand->o2net_idle_timeout_ms = cpu_to_be32(o2net_idle_timeout());
	o2net_hand->o2net_keepalive_delay_ms = cpu_to_be32(
		o2net_keepalive_delay());
	o2net_hand->o2net_reconnect_delay_ms = cpu_to_be32(
		o2net_reconnect_delay());
}

/* ------------------------------------------------------------ */

/* called when a connect completes and after a sock is accepted.  the
 * rx path will see the response and mark the sc valid */
static void o2net_sc_connect_completed(struct work_struct *work)
{
	struct o2net_sock_container *sc =
		container_of(work, struct o2net_sock_container,
			     sc_connect_work);

	mlog(ML_MSG, "sc sending handshake with ver %llu id %llx\n",
              (unsigned long long)O2NET_PROTOCOL_VERSION,
	      (unsigned long long)be64_to_cpu(o2net_hand->connector_id));

	o2net_initialize_handshake();
	o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));
	sc_put(sc);
}

/* this is called as a work_struct func. */
static void o2net_sc_send_keep_req(struct work_struct *work)
{
	struct o2net_sock_container *sc =
		container_of(work, struct o2net_sock_container,
			     sc_keepalive_work.work);

	o2net_sendpage(sc, o2net_keep_req, sizeof(*o2net_keep_req));
	sc_put(sc);
}

/* socket shutdown does a del_timer_sync against this as it tears down.
 * we can't start this timer until we've got to the point in sc buildup
 * where shutdown is going to be involved */
static void o2net_idle_timer(unsigned long data)
{
	struct o2net_sock_container *sc = (struct o2net_sock_container *)data;
	struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num);
	struct timeval now;

	do_gettimeofday(&now);

	printk(KERN_NOTICE "o2net: connection to " SC_NODEF_FMT " has been idle for %u.%u "
	     "seconds, shutting it down.\n", SC_NODEF_ARGS(sc),
		     o2net_idle_timeout() / 1000,
		     o2net_idle_timeout() % 1000);
	mlog(ML_NOTICE, "here are some times that might help debug the "
	     "situation: (tmr %ld.%ld now %ld.%ld dr %ld.%ld adv "
	     "%ld.%ld:%ld.%ld func (%08x:%u) %ld.%ld:%ld.%ld)\n",
	     sc->sc_tv_timer.tv_sec, (long) sc->sc_tv_timer.tv_usec,
	     now.tv_sec, (long) now.tv_usec,
	     sc->sc_tv_data_ready.tv_sec, (long) sc->sc_tv_data_ready.tv_usec,
	     sc->sc_tv_advance_start.tv_sec,
	     (long) sc->sc_tv_advance_start.tv_usec,
	     sc->sc_tv_advance_stop.tv_sec,
	     (long) sc->sc_tv_advance_stop.tv_usec,
	     sc->sc_msg_key, sc->sc_msg_type,
	     sc->sc_tv_func_start.tv_sec, (long) sc->sc_tv_func_start.tv_usec,
	     sc->sc_tv_func_stop.tv_sec, (long) sc->sc_tv_func_stop.tv_usec);

	/*
	 * Initialize the nn_timeout so that the next connection attempt
	 * will continue in o2net_start_connect.
	 */
	atomic_set(&nn->nn_timeout, 1);

	o2net_sc_queue_work(sc, &sc->sc_shutdown_work);
}

static void o2net_sc_reset_idle_timer(struct o2net_sock_container *sc)
{
	o2net_sc_cancel_delayed_work(sc, &sc->sc_keepalive_work);
	o2net_sc_queue_delayed_work(sc, &sc->sc_keepalive_work,
		      msecs_to_jiffies(o2net_keepalive_delay()));
	do_gettimeofday(&sc->sc_tv_timer);
	mod_timer(&sc->sc_idle_timeout,
	       jiffies + msecs_to_jiffies(o2net_idle_timeout()));
}

static void o2net_sc_postpone_idle(struct o2net_sock_container *sc)
{
	/* Only push out an existing timer */
	if (timer_pending(&sc->sc_idle_timeout))
		o2net_sc_reset_idle_timer(sc);
}

/* this work func is kicked whenever a path sets the nn state which doesn't
 * have valid set.  This includes seeing hb come up, losing a connection,
 * having a connect attempt fail, etc. This centralizes the logic which decides
 * if a connect attempt should be made or if we should give up and all future
 * transmit attempts should fail */
static void o2net_start_connect(struct work_struct *work)
{
	struct o2net_node *nn =
		container_of(work, struct o2net_node, nn_connect_work.work);
	struct o2net_sock_container *sc = NULL;
	struct o2nm_node *node = NULL, *mynode = NULL;
	struct socket *sock = NULL;
	struct sockaddr_in myaddr = {0, }, remoteaddr = {0, };
	int ret = 0, stop;
	unsigned int timeout;

	/* if we're greater we initiate tx, otherwise we accept */
	if (o2nm_this_node() <= o2net_num_from_nn(nn))
		goto out;

	/* watch for racing with tearing a node down */
	node = o2nm_get_node_by_num(o2net_num_from_nn(nn));
	if (node == NULL) {
		ret = 0;
		goto out;
	}

	mynode = o2nm_get_node_by_num(o2nm_this_node());
	if (mynode == NULL) {
		ret = 0;
		goto out;
	}

	spin_lock(&nn->nn_lock);
	/*
	 * see if we already have one pending or have given up.
	 * For nn_timeout, it is set when we close the connection
	 * because of the idle time out. So it means that we have
	 * at least connected to that node successfully once,
	 * now try to connect to it again.
	 */
	timeout = atomic_read(&nn->nn_timeout);
	stop = (nn->nn_sc ||
		(nn->nn_persistent_error &&
		(nn->nn_persistent_error != -ENOTCONN || timeout == 0)));
	spin_unlock(&nn->nn_lock);
	if (stop)
		goto out;

	nn->nn_last_connect_attempt = jiffies;

	sc = sc_alloc(node);
	if (sc == NULL) {
		mlog(0, "couldn't allocate sc\n");
		ret = -ENOMEM;
		goto out;
	}

	ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
	if (ret < 0) {
		mlog(0, "can't create socket: %d\n", ret);
		goto out;
	}
	sc->sc_sock = sock; /* freed by sc_kref_release */

	sock->sk->sk_allocation = GFP_ATOMIC;

	myaddr.sin_family = AF_INET;
	myaddr.sin_addr.s_addr = mynode->nd_ipv4_address;
	myaddr.sin_port = htons(0); /* any port */

	ret = sock->ops->bind(sock, (struct sockaddr *)&myaddr,
			      sizeof(myaddr));
	if (ret) {
		mlog(ML_ERROR, "bind failed with %d at address %pI4\n",
		     ret, &mynode->nd_ipv4_address);
		goto out;
	}

	ret = o2net_set_nodelay(sc->sc_sock);
	if (ret) {
		mlog(ML_ERROR, "setting TCP_NODELAY failed with %d\n", ret);
		goto out;
	}

	o2net_register_callbacks(sc->sc_sock->sk, sc);

	spin_lock(&nn->nn_lock);
	/* handshake completion will set nn->nn_sc_valid */
	o2net_set_nn_state(nn, sc, 0, 0);
	spin_unlock(&nn->nn_lock);

	remoteaddr.sin_family = AF_INET;
	remoteaddr.sin_addr.s_addr = node->nd_ipv4_address;
	remoteaddr.sin_port = node->nd_ipv4_port;

	ret = sc->sc_sock->ops->connect(sc->sc_sock,
					(struct sockaddr *)&remoteaddr,
					sizeof(remoteaddr),
					O_NONBLOCK);
	if (ret == -EINPROGRESS)
		ret = 0;

out:
	if (ret) {
		mlog(ML_NOTICE, "connect attempt to " SC_NODEF_FMT " failed "
		     "with errno %d\n", SC_NODEF_ARGS(sc), ret);
		/* 0 err so that another will be queued and attempted
		 * from set_nn_state */
		if (sc)
			o2net_ensure_shutdown(nn, sc, 0);
	}
	if (sc)
		sc_put(sc);
	if (node)
		o2nm_node_put(node);
	if (mynode)
		o2nm_node_put(mynode);

	return;
}

static void o2net_connect_expired(struct work_struct *work)
{
	struct o2net_node *nn =
		container_of(work, struct o2net_node, nn_connect_expired.work);

	spin_lock(&nn->nn_lock);
	if (!nn->nn_sc_valid) {
		mlog(ML_ERROR, "no connection established with node %u after "
		     "%u.%u seconds, giving up and returning errors.\n",
		     o2net_num_from_nn(nn),
		     o2net_idle_timeout() / 1000,
		     o2net_idle_timeout() % 1000);

		o2net_set_nn_state(nn, NULL, 0, -ENOTCONN);
	}
	spin_unlock(&nn->nn_lock);
}

static void o2net_still_up(struct work_struct *work)
{
	struct o2net_node *nn =
		container_of(work, struct o2net_node, nn_still_up.work);

	o2quo_hb_still_up(o2net_num_from_nn(nn));
}

/* ------------------------------------------------------------ */

void o2net_disconnect_node(struct o2nm_node *node)
{
	struct o2net_node *nn = o2net_nn_from_num(node->nd_num);

	/* don't reconnect until it's heartbeating again */
	spin_lock(&nn->nn_lock);
	atomic_set(&nn->nn_timeout, 0);
	o2net_set_nn_state(nn, NULL, 0, -ENOTCONN);
	spin_unlock(&nn->nn_lock);

	if (o2net_wq) {
		cancel_delayed_work(&nn->nn_connect_expired);
		cancel_delayed_work(&nn->nn_connect_work);
		cancel_delayed_work(&nn->nn_still_up);
		flush_workqueue(o2net_wq);
	}
}

static void o2net_hb_node_down_cb(struct o2nm_node *node, int node_num,
				  void *data)
{
	o2quo_hb_down(node_num);

	if (!node)
		return;

	if (node_num != o2nm_this_node())
		o2net_disconnect_node(node);

	BUG_ON(atomic_read(&o2net_connected_peers) < 0);
}

static void o2net_hb_node_up_cb(struct o2nm_node *node, int node_num,
				void *data)
{
	struct o2net_node *nn = o2net_nn_from_num(node_num);

	o2quo_hb_up(node_num);

	BUG_ON(!node);

	/* ensure an immediate connect attempt */
	nn->nn_last_connect_attempt = jiffies -
		(msecs_to_jiffies(o2net_reconnect_delay()) + 1);

	if (node_num != o2nm_this_node()) {
		/* believe it or not, accept and node hearbeating testing
		 * can succeed for this node before we got here.. so
		 * only use set_nn_state to clear the persistent error
		 * if that hasn't already happened */
		spin_lock(&nn->nn_lock);
		atomic_set(&nn->nn_timeout, 0);
		if (nn->nn_persistent_error)
			o2net_set_nn_state(nn, NULL, 0, 0);
		spin_unlock(&nn->nn_lock);
	}
}

void o2net_unregister_hb_callbacks(void)
{
	o2hb_unregister_callback(NULL, &o2net_hb_up);
	o2hb_unregister_callback(NULL, &o2net_hb_down);
}

int o2net_register_hb_callbacks(void)
{
	int ret;

	o2hb_setup_callback(&o2net_hb_down, O2HB_NODE_DOWN_CB,
			    o2net_hb_node_down_cb, NULL, O2NET_HB_PRI);
	o2hb_setup_callback(&o2net_hb_up, O2HB_NODE_UP_CB,
			    o2net_hb_node_up_cb, NULL, O2NET_HB_PRI);

	ret = o2hb_register_callback(NULL, &o2net_hb_up);
	if (ret == 0)
		ret = o2hb_register_callback(NULL, &o2net_hb_down);

	if (ret)
		o2net_unregister_hb_callbacks();

	return ret;
}

/* ------------------------------------------------------------ */

static int o2net_accept_one(struct socket *sock)
{
	int ret, slen;
	struct sockaddr_in sin;
	struct socket *new_sock = NULL;
	struct o2nm_node *node = NULL;
	struct o2nm_node *local_node = NULL;
	struct o2net_sock_container *sc = NULL;
	struct o2net_node *nn;

	BUG_ON(sock == NULL);
	ret = sock_create_lite(sock->sk->sk_family, sock->sk->sk_type,
			       sock->sk->sk_protocol, &new_sock);
	if (ret)
		goto out;

	new_sock->type = sock->type;
	new_sock->ops = sock->ops;
	ret = sock->ops->accept(sock, new_sock, O_NONBLOCK);
	if (ret < 0)
		goto out;

	new_sock->sk->sk_allocation = GFP_ATOMIC;

	ret = o2net_set_nodelay(new_sock);
	if (ret) {
		mlog(ML_ERROR, "setting TCP_NODELAY failed with %d\n", ret);
		goto out;
	}

	slen = sizeof(sin);
	ret = new_sock->ops->getname(new_sock, (struct sockaddr *) &sin,
				       &slen, 1);
	if (ret < 0)
		goto out;

	node = o2nm_get_node_by_ip(sin.sin_addr.s_addr);
	if (node == NULL) {
		mlog(ML_NOTICE, "attempt to connect from unknown node at %pI4:%d\n",
		     &sin.sin_addr.s_addr, ntohs(sin.sin_port));
		ret = -EINVAL;
		goto out;
	}

	if (o2nm_this_node() >= node->nd_num) {
		local_node = o2nm_get_node_by_num(o2nm_this_node());
		mlog(ML_NOTICE, "unexpected connect attempt seen at node '%s' ("
		     "%u, %pI4:%d) from node '%s' (%u, %pI4:%d)\n",
		     local_node->nd_name, local_node->nd_num,
		     &(local_node->nd_ipv4_address),
		     ntohs(local_node->nd_ipv4_port),
		     node->nd_name, node->nd_num, &sin.sin_addr.s_addr,
		     ntohs(sin.sin_port));
		ret = -EINVAL;
		goto out;
	}

	/* this happens all the time when the other node sees our heartbeat
	 * and tries to connect before we see their heartbeat */
	if (!o2hb_check_node_heartbeating_from_callback(node->nd_num)) {
		mlog(ML_CONN, "attempt to connect from node '%s' at "
		     "%pI4:%d but it isn't heartbeating\n",
		     node->nd_name, &sin.sin_addr.s_addr,
		     ntohs(sin.sin_port));
		ret = -EINVAL;
		goto out;
	}

	nn = o2net_nn_from_num(node->nd_num);

	spin_lock(&nn->nn_lock);
	if (nn->nn_sc)
		ret = -EBUSY;
	else
		ret = 0;
	spin_unlock(&nn->nn_lock);
	if (ret) {
		mlog(ML_NOTICE, "attempt to connect from node '%s' at "
		     "%pI4:%d but it already has an open connection\n",
		     node->nd_name, &sin.sin_addr.s_addr,
		     ntohs(sin.sin_port));
		goto out;
	}

	sc = sc_alloc(node);
	if (sc == NULL) {
		ret = -ENOMEM;
		goto out;
	}

	sc->sc_sock = new_sock;
	new_sock = NULL;

	spin_lock(&nn->nn_lock);
	atomic_set(&nn->nn_timeout, 0);
	o2net_set_nn_state(nn, sc, 0, 0);
	spin_unlock(&nn->nn_lock);

	o2net_register_callbacks(sc->sc_sock->sk, sc);
	o2net_sc_queue_work(sc, &sc->sc_rx_work);

	o2net_initialize_handshake();
	o2net_sendpage(sc, o2net_hand, sizeof(*o2net_hand));

out:
	if (new_sock)
		sock_release(new_sock);
	if (node)
		o2nm_node_put(node);
	if (local_node)
		o2nm_node_put(local_node);
	if (sc)
		sc_put(sc);
	return ret;
}

static void o2net_accept_many(struct work_struct *work)
{
	struct socket *sock = o2net_listen_sock;
	while (o2net_accept_one(sock) == 0)
		cond_resched();
}

static void o2net_listen_data_ready(struct sock *sk, int bytes)
{
	void (*ready)(struct sock *sk, int bytes);

	read_lock(&sk->sk_callback_lock);
	ready = sk->sk_user_data;
	if (ready == NULL) { /* check for teardown race */
		ready = sk->sk_data_ready;
		goto out;
	}

	/* ->sk_data_ready is also called for a newly established child socket
	 * before it has been accepted and the acceptor has set up their
	 * data_ready.. we only want to queue listen work for our listening
	 * socket */
	if (sk->sk_state == TCP_LISTEN) {
		mlog(ML_TCP, "bytes: %d\n", bytes);
		queue_work(o2net_wq, &o2net_listen_work);
	}

out:
	read_unlock(&sk->sk_callback_lock);
	ready(sk, bytes);
}

static int o2net_open_listening_sock(__be32 addr, __be16 port)
{
	struct socket *sock = NULL;
	int ret;
	struct sockaddr_in sin = {
		.sin_family = PF_INET,
		.sin_addr = { .s_addr = addr },
		.sin_port = port,
	};

	ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock);
	if (ret < 0) {
		mlog(ML_ERROR, "unable to create socket, ret=%d\n", ret);
		goto out;
	}

	sock->sk->sk_allocation = GFP_ATOMIC;

	write_lock_bh(&sock->sk->sk_callback_lock);
	sock->sk->sk_user_data = sock->sk->sk_data_ready;
	sock->sk->sk_data_ready = o2net_listen_data_ready;
	write_unlock_bh(&sock->sk->sk_callback_lock);

	o2net_listen_sock = sock;
	INIT_WORK(&o2net_listen_work, o2net_accept_many);

	sock->sk->sk_reuse = 1;
	ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin));
	if (ret < 0) {
		mlog(ML_ERROR, "unable to bind socket at %pI4:%u, "
		     "ret=%d\n", &addr, ntohs(port), ret);
		goto out;
	}

	ret = sock->ops->listen(sock, 64);
	if (ret < 0) {
		mlog(ML_ERROR, "unable to listen on %pI4:%u, ret=%d\n",
		     &addr, ntohs(port), ret);
	}

out:
	if (ret) {
		o2net_listen_sock = NULL;
		if (sock)
			sock_release(sock);
	}
	return ret;
}

/*
 * called from node manager when we should bring up our network listening
 * socket.  node manager handles all the serialization to only call this
 * once and to match it with o2net_stop_listening().  note,
 * o2nm_this_node() doesn't work yet as we're being called while it
 * is being set up.
 */
int o2net_start_listening(struct o2nm_node *node)
{
	int ret = 0;

	BUG_ON(o2net_wq != NULL);
	BUG_ON(o2net_listen_sock != NULL);

	mlog(ML_KTHREAD, "starting o2net thread...\n");
	o2net_wq = create_singlethread_workqueue("o2net");
	if (o2net_wq == NULL) {
		mlog(ML_ERROR, "unable to launch o2net thread\n");
		return -ENOMEM; /* ? */
	}

	ret = o2net_open_listening_sock(node->nd_ipv4_address,
					node->nd_ipv4_port);
	if (ret) {
		destroy_workqueue(o2net_wq);
		o2net_wq = NULL;
	} else
		o2quo_conn_up(node->nd_num);

	return ret;
}

/* again, o2nm_this_node() doesn't work here as we're involved in
 * tearing it down */
void o2net_stop_listening(struct o2nm_node *node)
{
	struct socket *sock = o2net_listen_sock;
	size_t i;

	BUG_ON(o2net_wq == NULL);
	BUG_ON(o2net_listen_sock == NULL);

	/* stop the listening socket from generating work */
	write_lock_bh(&sock->sk->sk_callback_lock);
	sock->sk->sk_data_ready = sock->sk->sk_user_data;
	sock->sk->sk_user_data = NULL;
	write_unlock_bh(&sock->sk->sk_callback_lock);

	for (i = 0; i < ARRAY_SIZE(o2net_nodes); i++) {
		struct o2nm_node *node = o2nm_get_node_by_num(i);
		if (node) {
			o2net_disconnect_node(node);
			o2nm_node_put(node);
		}
	}

	/* finish all work and tear down the work queue */
	mlog(ML_KTHREAD, "waiting for o2net thread to exit....\n");
	destroy_workqueue(o2net_wq);
	o2net_wq = NULL;

	sock_release(o2net_listen_sock);
	o2net_listen_sock = NULL;

	o2quo_conn_err(node->nd_num);
}

/* ------------------------------------------------------------ */

int o2net_init(void)
{
	unsigned long i;

	o2quo_init();

	if (o2net_debugfs_init())
		return -ENOMEM;

	o2net_hand = kzalloc(sizeof(struct o2net_handshake), GFP_KERNEL);
	o2net_keep_req = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
	o2net_keep_resp = kzalloc(sizeof(struct o2net_msg), GFP_KERNEL);
	if (!o2net_hand || !o2net_keep_req || !o2net_keep_resp) {
		kfree(o2net_hand);
		kfree(o2net_keep_req);
		kfree(o2net_keep_resp);
		return -ENOMEM;
	}

	o2net_hand->protocol_version = cpu_to_be64(O2NET_PROTOCOL_VERSION);
	o2net_hand->connector_id = cpu_to_be64(1);

	o2net_keep_req->magic = cpu_to_be16(O2NET_MSG_KEEP_REQ_MAGIC);
	o2net_keep_resp->magic = cpu_to_be16(O2NET_MSG_KEEP_RESP_MAGIC);

	for (i = 0; i < ARRAY_SIZE(o2net_nodes); i++) {
		struct o2net_node *nn = o2net_nn_from_num(i);

		atomic_set(&nn->nn_timeout, 0);
		spin_lock_init(&nn->nn_lock);
		INIT_DELAYED_WORK(&nn->nn_connect_work, o2net_start_connect);
		INIT_DELAYED_WORK(&nn->nn_connect_expired,
				  o2net_connect_expired);
		INIT_DELAYED_WORK(&nn->nn_still_up, o2net_still_up);
		/* until we see hb from a node we'll return einval */
		nn->nn_persistent_error = -ENOTCONN;
		init_waitqueue_head(&nn->nn_sc_wq);
		idr_init(&nn->nn_status_idr);
		INIT_LIST_HEAD(&nn->nn_status_list);
	}

	return 0;
}

void o2net_exit(void)
{
	o2quo_exit();
	kfree(o2net_hand);
	kfree(o2net_keep_req);
	kfree(o2net_keep_resp);
	o2net_debugfs_exit();
}