Free Electrons

Embedded Linux Experts

   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
/*
 * Copyright (c) 2006 Mellanox Technologies. All rights reserved
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <rdma/ib_cm.h>
#include <net/dst.h>
#include <net/icmp.h>
#include <linux/icmpv6.h>
#include <linux/delay.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/moduleparam.h>

#include "ipoib.h"

int ipoib_max_conn_qp = 128;

module_param_named(max_nonsrq_conn_qp, ipoib_max_conn_qp, int, 0444);
MODULE_PARM_DESC(max_nonsrq_conn_qp,
		 "Max number of connected-mode QPs per interface "
		 "(applied only if shared receive queue is not available)");

#ifdef CONFIG_INFINIBAND_IPOIB_DEBUG_DATA
static int data_debug_level;

module_param_named(cm_data_debug_level, data_debug_level, int, 0644);
MODULE_PARM_DESC(cm_data_debug_level,
		 "Enable data path debug tracing for connected mode if > 0");
#endif

#define IPOIB_CM_IETF_ID 0x1000000000000000ULL

#define IPOIB_CM_RX_UPDATE_TIME (256 * HZ)
#define IPOIB_CM_RX_TIMEOUT     (2 * 256 * HZ)
#define IPOIB_CM_RX_DELAY       (3 * 256 * HZ)
#define IPOIB_CM_RX_UPDATE_MASK (0x3)

static struct ib_qp_attr ipoib_cm_err_attr = {
	.qp_state = IB_QPS_ERR
};

#define IPOIB_CM_RX_DRAIN_WRID 0xffffffff

static struct ib_send_wr ipoib_cm_rx_drain_wr = {
	.wr_id = IPOIB_CM_RX_DRAIN_WRID,
	.opcode = IB_WR_SEND,
};

static int ipoib_cm_tx_handler(struct ib_cm_id *cm_id,
			       struct ib_cm_event *event);

static void ipoib_cm_dma_unmap_rx(struct ipoib_dev_priv *priv, int frags,
				  u64 mapping[IPOIB_CM_RX_SG])
{
	int i;

	ib_dma_unmap_single(priv->ca, mapping[0], IPOIB_CM_HEAD_SIZE, DMA_FROM_DEVICE);

	for (i = 0; i < frags; ++i)
		ib_dma_unmap_page(priv->ca, mapping[i + 1], PAGE_SIZE, DMA_FROM_DEVICE);
}

static int ipoib_cm_post_receive_srq(struct net_device *dev, int id)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_recv_wr *bad_wr;
	int i, ret;

	priv->cm.rx_wr.wr_id = id | IPOIB_OP_CM | IPOIB_OP_RECV;

	for (i = 0; i < priv->cm.num_frags; ++i)
		priv->cm.rx_sge[i].addr = priv->cm.srq_ring[id].mapping[i];

	ret = ib_post_srq_recv(priv->cm.srq, &priv->cm.rx_wr, &bad_wr);
	if (unlikely(ret)) {
		ipoib_warn(priv, "post srq failed for buf %d (%d)\n", id, ret);
		ipoib_cm_dma_unmap_rx(priv, priv->cm.num_frags - 1,
				      priv->cm.srq_ring[id].mapping);
		dev_kfree_skb_any(priv->cm.srq_ring[id].skb);
		priv->cm.srq_ring[id].skb = NULL;
	}

	return ret;
}

static int ipoib_cm_post_receive_nonsrq(struct net_device *dev,
					struct ipoib_cm_rx *rx,
					struct ib_recv_wr *wr,
					struct ib_sge *sge, int id)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_recv_wr *bad_wr;
	int i, ret;

	wr->wr_id = id | IPOIB_OP_CM | IPOIB_OP_RECV;

	for (i = 0; i < IPOIB_CM_RX_SG; ++i)
		sge[i].addr = rx->rx_ring[id].mapping[i];

	ret = ib_post_recv(rx->qp, wr, &bad_wr);
	if (unlikely(ret)) {
		ipoib_warn(priv, "post recv failed for buf %d (%d)\n", id, ret);
		ipoib_cm_dma_unmap_rx(priv, IPOIB_CM_RX_SG - 1,
				      rx->rx_ring[id].mapping);
		dev_kfree_skb_any(rx->rx_ring[id].skb);
		rx->rx_ring[id].skb = NULL;
	}

	return ret;
}

static struct sk_buff *ipoib_cm_alloc_rx_skb(struct net_device *dev,
					     struct ipoib_cm_rx_buf *rx_ring,
					     int id, int frags,
					     u64 mapping[IPOIB_CM_RX_SG])
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct sk_buff *skb;
	int i;

	skb = dev_alloc_skb(IPOIB_CM_HEAD_SIZE + 12);
	if (unlikely(!skb))
		return NULL;

	/*
	 * IPoIB adds a 4 byte header. So we need 12 more bytes to align the
	 * IP header to a multiple of 16.
	 */
	skb_reserve(skb, 12);

	mapping[0] = ib_dma_map_single(priv->ca, skb->data, IPOIB_CM_HEAD_SIZE,
				       DMA_FROM_DEVICE);
	if (unlikely(ib_dma_mapping_error(priv->ca, mapping[0]))) {
		dev_kfree_skb_any(skb);
		return NULL;
	}

	for (i = 0; i < frags; i++) {
		struct page *page = alloc_page(GFP_ATOMIC);

		if (!page)
			goto partial_error;
		skb_fill_page_desc(skb, i, page, 0, PAGE_SIZE);

		mapping[i + 1] = ib_dma_map_page(priv->ca, page,
						 0, PAGE_SIZE, DMA_FROM_DEVICE);
		if (unlikely(ib_dma_mapping_error(priv->ca, mapping[i + 1])))
			goto partial_error;
	}

	rx_ring[id].skb = skb;
	return skb;

partial_error:

	ib_dma_unmap_single(priv->ca, mapping[0], IPOIB_CM_HEAD_SIZE, DMA_FROM_DEVICE);

	for (; i > 0; --i)
		ib_dma_unmap_page(priv->ca, mapping[i], PAGE_SIZE, DMA_FROM_DEVICE);

	dev_kfree_skb_any(skb);
	return NULL;
}

static void ipoib_cm_free_rx_ring(struct net_device *dev,
				  struct ipoib_cm_rx_buf *rx_ring)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int i;

	for (i = 0; i < ipoib_recvq_size; ++i)
		if (rx_ring[i].skb) {
			ipoib_cm_dma_unmap_rx(priv, IPOIB_CM_RX_SG - 1,
					      rx_ring[i].mapping);
			dev_kfree_skb_any(rx_ring[i].skb);
		}

	vfree(rx_ring);
}

static void ipoib_cm_start_rx_drain(struct ipoib_dev_priv *priv)
{
	struct ib_send_wr *bad_wr;
	struct ipoib_cm_rx *p;

	/* We only reserved 1 extra slot in CQ for drain WRs, so
	 * make sure we have at most 1 outstanding WR. */
	if (list_empty(&priv->cm.rx_flush_list) ||
	    !list_empty(&priv->cm.rx_drain_list))
		return;

	/*
	 * QPs on flush list are error state.  This way, a "flush
	 * error" WC will be immediately generated for each WR we post.
	 */
	p = list_entry(priv->cm.rx_flush_list.next, typeof(*p), list);
	if (ib_post_send(p->qp, &ipoib_cm_rx_drain_wr, &bad_wr))
		ipoib_warn(priv, "failed to post drain wr\n");

	list_splice_init(&priv->cm.rx_flush_list, &priv->cm.rx_drain_list);
}

static void ipoib_cm_rx_event_handler(struct ib_event *event, void *ctx)
{
	struct ipoib_cm_rx *p = ctx;
	struct ipoib_dev_priv *priv = netdev_priv(p->dev);
	unsigned long flags;

	if (event->event != IB_EVENT_QP_LAST_WQE_REACHED)
		return;

	spin_lock_irqsave(&priv->lock, flags);
	list_move(&p->list, &priv->cm.rx_flush_list);
	p->state = IPOIB_CM_RX_FLUSH;
	ipoib_cm_start_rx_drain(priv);
	spin_unlock_irqrestore(&priv->lock, flags);
}

static struct ib_qp *ipoib_cm_create_rx_qp(struct net_device *dev,
					   struct ipoib_cm_rx *p)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_qp_init_attr attr = {
		.event_handler = ipoib_cm_rx_event_handler,
		.send_cq = priv->recv_cq, /* For drain WR */
		.recv_cq = priv->recv_cq,
		.srq = priv->cm.srq,
		.cap.max_send_wr = 1, /* For drain WR */
		.cap.max_send_sge = 1, /* FIXME: 0 Seems not to work */
		.sq_sig_type = IB_SIGNAL_ALL_WR,
		.qp_type = IB_QPT_RC,
		.qp_context = p,
	};

	if (!ipoib_cm_has_srq(dev)) {
		attr.cap.max_recv_wr  = ipoib_recvq_size;
		attr.cap.max_recv_sge = IPOIB_CM_RX_SG;
	}

	return ib_create_qp(priv->pd, &attr);
}

static int ipoib_cm_modify_rx_qp(struct net_device *dev,
				 struct ib_cm_id *cm_id, struct ib_qp *qp,
				 unsigned psn)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;

	qp_attr.qp_state = IB_QPS_INIT;
	ret = ib_cm_init_qp_attr(cm_id, &qp_attr, &qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to init QP attr for INIT: %d\n", ret);
		return ret;
	}
	ret = ib_modify_qp(qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify QP to INIT: %d\n", ret);
		return ret;
	}
	qp_attr.qp_state = IB_QPS_RTR;
	ret = ib_cm_init_qp_attr(cm_id, &qp_attr, &qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to init QP attr for RTR: %d\n", ret);
		return ret;
	}
	qp_attr.rq_psn = psn;
	ret = ib_modify_qp(qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify QP to RTR: %d\n", ret);
		return ret;
	}

	/*
	 * Current Mellanox HCA firmware won't generate completions
	 * with error for drain WRs unless the QP has been moved to
	 * RTS first. This work-around leaves a window where a QP has
	 * moved to error asynchronously, but this will eventually get
	 * fixed in firmware, so let's not error out if modify QP
	 * fails.
	 */
	qp_attr.qp_state = IB_QPS_RTS;
	ret = ib_cm_init_qp_attr(cm_id, &qp_attr, &qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to init QP attr for RTS: %d\n", ret);
		return 0;
	}
	ret = ib_modify_qp(qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify QP to RTS: %d\n", ret);
		return 0;
	}

	return 0;
}

static void ipoib_cm_init_rx_wr(struct net_device *dev,
				struct ib_recv_wr *wr,
				struct ib_sge *sge)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int i;

	for (i = 0; i < priv->cm.num_frags; ++i)
		sge[i].lkey = priv->mr->lkey;

	sge[0].length = IPOIB_CM_HEAD_SIZE;
	for (i = 1; i < priv->cm.num_frags; ++i)
		sge[i].length = PAGE_SIZE;

	wr->next    = NULL;
	wr->sg_list = sge;
	wr->num_sge = priv->cm.num_frags;
}

static int ipoib_cm_nonsrq_init_rx(struct net_device *dev, struct ib_cm_id *cm_id,
				   struct ipoib_cm_rx *rx)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct {
		struct ib_recv_wr wr;
		struct ib_sge sge[IPOIB_CM_RX_SG];
	} *t;
	int ret;
	int i;

	rx->rx_ring = vzalloc(ipoib_recvq_size * sizeof *rx->rx_ring);
	if (!rx->rx_ring) {
		printk(KERN_WARNING "%s: failed to allocate CM non-SRQ ring (%d entries)\n",
		       priv->ca->name, ipoib_recvq_size);
		return -ENOMEM;
	}

	t = kmalloc(sizeof *t, GFP_KERNEL);
	if (!t) {
		ret = -ENOMEM;
		goto err_free;
	}

	ipoib_cm_init_rx_wr(dev, &t->wr, t->sge);

	spin_lock_irq(&priv->lock);

	if (priv->cm.nonsrq_conn_qp >= ipoib_max_conn_qp) {
		spin_unlock_irq(&priv->lock);
		ib_send_cm_rej(cm_id, IB_CM_REJ_NO_QP, NULL, 0, NULL, 0);
		ret = -EINVAL;
		goto err_free;
	} else
		++priv->cm.nonsrq_conn_qp;

	spin_unlock_irq(&priv->lock);

	for (i = 0; i < ipoib_recvq_size; ++i) {
		if (!ipoib_cm_alloc_rx_skb(dev, rx->rx_ring, i, IPOIB_CM_RX_SG - 1,
					   rx->rx_ring[i].mapping)) {
			ipoib_warn(priv, "failed to allocate receive buffer %d\n", i);
				ret = -ENOMEM;
				goto err_count;
		}
		ret = ipoib_cm_post_receive_nonsrq(dev, rx, &t->wr, t->sge, i);
		if (ret) {
			ipoib_warn(priv, "ipoib_cm_post_receive_nonsrq "
				   "failed for buf %d\n", i);
			ret = -EIO;
			goto err_count;
		}
	}

	rx->recv_count = ipoib_recvq_size;

	kfree(t);

	return 0;

err_count:
	spin_lock_irq(&priv->lock);
	--priv->cm.nonsrq_conn_qp;
	spin_unlock_irq(&priv->lock);

err_free:
	kfree(t);
	ipoib_cm_free_rx_ring(dev, rx->rx_ring);

	return ret;
}

static int ipoib_cm_send_rep(struct net_device *dev, struct ib_cm_id *cm_id,
			     struct ib_qp *qp, struct ib_cm_req_event_param *req,
			     unsigned psn)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_data data = {};
	struct ib_cm_rep_param rep = {};

	data.qpn = cpu_to_be32(priv->qp->qp_num);
	data.mtu = cpu_to_be32(IPOIB_CM_BUF_SIZE);

	rep.private_data = &data;
	rep.private_data_len = sizeof data;
	rep.flow_control = 0;
	rep.rnr_retry_count = req->rnr_retry_count;
	rep.srq = ipoib_cm_has_srq(dev);
	rep.qp_num = qp->qp_num;
	rep.starting_psn = psn;
	return ib_send_cm_rep(cm_id, &rep);
}

static int ipoib_cm_req_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
{
	struct net_device *dev = cm_id->context;
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_rx *p;
	unsigned psn;
	int ret;

	ipoib_dbg(priv, "REQ arrived\n");
	p = kzalloc(sizeof *p, GFP_KERNEL);
	if (!p)
		return -ENOMEM;
	p->dev = dev;
	p->id = cm_id;
	cm_id->context = p;
	p->state = IPOIB_CM_RX_LIVE;
	p->jiffies = jiffies;
	INIT_LIST_HEAD(&p->list);

	p->qp = ipoib_cm_create_rx_qp(dev, p);
	if (IS_ERR(p->qp)) {
		ret = PTR_ERR(p->qp);
		goto err_qp;
	}

	psn = random32() & 0xffffff;
	ret = ipoib_cm_modify_rx_qp(dev, cm_id, p->qp, psn);
	if (ret)
		goto err_modify;

	if (!ipoib_cm_has_srq(dev)) {
		ret = ipoib_cm_nonsrq_init_rx(dev, cm_id, p);
		if (ret)
			goto err_modify;
	}

	spin_lock_irq(&priv->lock);
	queue_delayed_work(ipoib_workqueue,
			   &priv->cm.stale_task, IPOIB_CM_RX_DELAY);
	/* Add this entry to passive ids list head, but do not re-add it
	 * if IB_EVENT_QP_LAST_WQE_REACHED has moved it to flush list. */
	p->jiffies = jiffies;
	if (p->state == IPOIB_CM_RX_LIVE)
		list_move(&p->list, &priv->cm.passive_ids);
	spin_unlock_irq(&priv->lock);

	ret = ipoib_cm_send_rep(dev, cm_id, p->qp, &event->param.req_rcvd, psn);
	if (ret) {
		ipoib_warn(priv, "failed to send REP: %d\n", ret);
		if (ib_modify_qp(p->qp, &ipoib_cm_err_attr, IB_QP_STATE))
			ipoib_warn(priv, "unable to move qp to error state\n");
	}
	return 0;

err_modify:
	ib_destroy_qp(p->qp);
err_qp:
	kfree(p);
	return ret;
}

static int ipoib_cm_rx_handler(struct ib_cm_id *cm_id,
			       struct ib_cm_event *event)
{
	struct ipoib_cm_rx *p;
	struct ipoib_dev_priv *priv;

	switch (event->event) {
	case IB_CM_REQ_RECEIVED:
		return ipoib_cm_req_handler(cm_id, event);
	case IB_CM_DREQ_RECEIVED:
		p = cm_id->context;
		ib_send_cm_drep(cm_id, NULL, 0);
		/* Fall through */
	case IB_CM_REJ_RECEIVED:
		p = cm_id->context;
		priv = netdev_priv(p->dev);
		if (ib_modify_qp(p->qp, &ipoib_cm_err_attr, IB_QP_STATE))
			ipoib_warn(priv, "unable to move qp to error state\n");
		/* Fall through */
	default:
		return 0;
	}
}
/* Adjust length of skb with fragments to match received data */
static void skb_put_frags(struct sk_buff *skb, unsigned int hdr_space,
			  unsigned int length, struct sk_buff *toskb)
{
	int i, num_frags;
	unsigned int size;

	/* put header into skb */
	size = min(length, hdr_space);
	skb->tail += size;
	skb->len += size;
	length -= size;

	num_frags = skb_shinfo(skb)->nr_frags;
	for (i = 0; i < num_frags; i++) {
		skb_frag_t *frag = &skb_shinfo(skb)->frags[i];

		if (length == 0) {
			/* don't need this page */
			skb_fill_page_desc(toskb, i, skb_frag_page(frag),
					   0, PAGE_SIZE);
			--skb_shinfo(skb)->nr_frags;
		} else {
			size = min(length, (unsigned) PAGE_SIZE);

			skb_frag_size_set(frag, size);
			skb->data_len += size;
			skb->truesize += size;
			skb->len += size;
			length -= size;
		}
	}
}

void ipoib_cm_handle_rx_wc(struct net_device *dev, struct ib_wc *wc)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_rx_buf *rx_ring;
	unsigned int wr_id = wc->wr_id & ~(IPOIB_OP_CM | IPOIB_OP_RECV);
	struct sk_buff *skb, *newskb;
	struct ipoib_cm_rx *p;
	unsigned long flags;
	u64 mapping[IPOIB_CM_RX_SG];
	int frags;
	int has_srq;
	struct sk_buff *small_skb;

	ipoib_dbg_data(priv, "cm recv completion: id %d, status: %d\n",
		       wr_id, wc->status);

	if (unlikely(wr_id >= ipoib_recvq_size)) {
		if (wr_id == (IPOIB_CM_RX_DRAIN_WRID & ~(IPOIB_OP_CM | IPOIB_OP_RECV))) {
			spin_lock_irqsave(&priv->lock, flags);
			list_splice_init(&priv->cm.rx_drain_list, &priv->cm.rx_reap_list);
			ipoib_cm_start_rx_drain(priv);
			queue_work(ipoib_workqueue, &priv->cm.rx_reap_task);
			spin_unlock_irqrestore(&priv->lock, flags);
		} else
			ipoib_warn(priv, "cm recv completion event with wrid %d (> %d)\n",
				   wr_id, ipoib_recvq_size);
		return;
	}

	p = wc->qp->qp_context;

	has_srq = ipoib_cm_has_srq(dev);
	rx_ring = has_srq ? priv->cm.srq_ring : p->rx_ring;

	skb = rx_ring[wr_id].skb;

	if (unlikely(wc->status != IB_WC_SUCCESS)) {
		ipoib_dbg(priv, "cm recv error "
			   "(status=%d, wrid=%d vend_err %x)\n",
			   wc->status, wr_id, wc->vendor_err);
		++dev->stats.rx_dropped;
		if (has_srq)
			goto repost;
		else {
			if (!--p->recv_count) {
				spin_lock_irqsave(&priv->lock, flags);
				list_move(&p->list, &priv->cm.rx_reap_list);
				spin_unlock_irqrestore(&priv->lock, flags);
				queue_work(ipoib_workqueue, &priv->cm.rx_reap_task);
			}
			return;
		}
	}

	if (unlikely(!(wr_id & IPOIB_CM_RX_UPDATE_MASK))) {
		if (p && time_after_eq(jiffies, p->jiffies + IPOIB_CM_RX_UPDATE_TIME)) {
			spin_lock_irqsave(&priv->lock, flags);
			p->jiffies = jiffies;
			/* Move this entry to list head, but do not re-add it
			 * if it has been moved out of list. */
			if (p->state == IPOIB_CM_RX_LIVE)
				list_move(&p->list, &priv->cm.passive_ids);
			spin_unlock_irqrestore(&priv->lock, flags);
		}
	}

	if (wc->byte_len < IPOIB_CM_COPYBREAK) {
		int dlen = wc->byte_len;

		small_skb = dev_alloc_skb(dlen + 12);
		if (small_skb) {
			skb_reserve(small_skb, 12);
			ib_dma_sync_single_for_cpu(priv->ca, rx_ring[wr_id].mapping[0],
						   dlen, DMA_FROM_DEVICE);
			skb_copy_from_linear_data(skb, small_skb->data, dlen);
			ib_dma_sync_single_for_device(priv->ca, rx_ring[wr_id].mapping[0],
						      dlen, DMA_FROM_DEVICE);
			skb_put(small_skb, dlen);
			skb = small_skb;
			goto copied;
		}
	}

	frags = PAGE_ALIGN(wc->byte_len - min(wc->byte_len,
					      (unsigned)IPOIB_CM_HEAD_SIZE)) / PAGE_SIZE;

	newskb = ipoib_cm_alloc_rx_skb(dev, rx_ring, wr_id, frags, mapping);
	if (unlikely(!newskb)) {
		/*
		 * If we can't allocate a new RX buffer, dump
		 * this packet and reuse the old buffer.
		 */
		ipoib_dbg(priv, "failed to allocate receive buffer %d\n", wr_id);
		++dev->stats.rx_dropped;
		goto repost;
	}

	ipoib_cm_dma_unmap_rx(priv, frags, rx_ring[wr_id].mapping);
	memcpy(rx_ring[wr_id].mapping, mapping, (frags + 1) * sizeof *mapping);

	ipoib_dbg_data(priv, "received %d bytes, SLID 0x%04x\n",
		       wc->byte_len, wc->slid);

	skb_put_frags(skb, IPOIB_CM_HEAD_SIZE, wc->byte_len, newskb);

copied:
	skb->protocol = ((struct ipoib_header *) skb->data)->proto;
	skb_reset_mac_header(skb);
	skb_pull(skb, IPOIB_ENCAP_LEN);

	++dev->stats.rx_packets;
	dev->stats.rx_bytes += skb->len;

	skb->dev = dev;
	/* XXX get correct PACKET_ type here */
	skb->pkt_type = PACKET_HOST;
	netif_receive_skb(skb);

repost:
	if (has_srq) {
		if (unlikely(ipoib_cm_post_receive_srq(dev, wr_id)))
			ipoib_warn(priv, "ipoib_cm_post_receive_srq failed "
				   "for buf %d\n", wr_id);
	} else {
		if (unlikely(ipoib_cm_post_receive_nonsrq(dev, p,
							  &priv->cm.rx_wr,
							  priv->cm.rx_sge,
							  wr_id))) {
			--p->recv_count;
			ipoib_warn(priv, "ipoib_cm_post_receive_nonsrq failed "
				   "for buf %d\n", wr_id);
		}
	}
}

static inline int post_send(struct ipoib_dev_priv *priv,
			    struct ipoib_cm_tx *tx,
			    unsigned int wr_id,
			    u64 addr, int len)
{
	struct ib_send_wr *bad_wr;

	priv->tx_sge[0].addr          = addr;
	priv->tx_sge[0].length        = len;

	priv->tx_wr.num_sge	= 1;
	priv->tx_wr.wr_id	= wr_id | IPOIB_OP_CM;

	return ib_post_send(tx->qp, &priv->tx_wr, &bad_wr);
}

void ipoib_cm_send(struct net_device *dev, struct sk_buff *skb, struct ipoib_cm_tx *tx)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_tx_buf *tx_req;
	u64 addr;
	int rc;

	if (unlikely(skb->len > tx->mtu)) {
		ipoib_warn(priv, "packet len %d (> %d) too long to send, dropping\n",
			   skb->len, tx->mtu);
		++dev->stats.tx_dropped;
		++dev->stats.tx_errors;
		ipoib_cm_skb_too_long(dev, skb, tx->mtu - IPOIB_ENCAP_LEN);
		return;
	}

	ipoib_dbg_data(priv, "sending packet: head 0x%x length %d connection 0x%x\n",
		       tx->tx_head, skb->len, tx->qp->qp_num);

	/*
	 * We put the skb into the tx_ring _before_ we call post_send()
	 * because it's entirely possible that the completion handler will
	 * run before we execute anything after the post_send().  That
	 * means we have to make sure everything is properly recorded and
	 * our state is consistent before we call post_send().
	 */
	tx_req = &tx->tx_ring[tx->tx_head & (ipoib_sendq_size - 1)];
	tx_req->skb = skb;
	addr = ib_dma_map_single(priv->ca, skb->data, skb->len, DMA_TO_DEVICE);
	if (unlikely(ib_dma_mapping_error(priv->ca, addr))) {
		++dev->stats.tx_errors;
		dev_kfree_skb_any(skb);
		return;
	}

	tx_req->mapping = addr;

	rc = post_send(priv, tx, tx->tx_head & (ipoib_sendq_size - 1),
		       addr, skb->len);
	if (unlikely(rc)) {
		ipoib_warn(priv, "post_send failed, error %d\n", rc);
		++dev->stats.tx_errors;
		ib_dma_unmap_single(priv->ca, addr, skb->len, DMA_TO_DEVICE);
		dev_kfree_skb_any(skb);
	} else {
		dev->trans_start = jiffies;
		++tx->tx_head;

		if (++priv->tx_outstanding == ipoib_sendq_size) {
			ipoib_dbg(priv, "TX ring 0x%x full, stopping kernel net queue\n",
				  tx->qp->qp_num);
			if (ib_req_notify_cq(priv->send_cq, IB_CQ_NEXT_COMP))
				ipoib_warn(priv, "request notify on send CQ failed\n");
			netif_stop_queue(dev);
		}
	}
}

void ipoib_cm_handle_tx_wc(struct net_device *dev, struct ib_wc *wc)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_tx *tx = wc->qp->qp_context;
	unsigned int wr_id = wc->wr_id & ~IPOIB_OP_CM;
	struct ipoib_cm_tx_buf *tx_req;
	unsigned long flags;

	ipoib_dbg_data(priv, "cm send completion: id %d, status: %d\n",
		       wr_id, wc->status);

	if (unlikely(wr_id >= ipoib_sendq_size)) {
		ipoib_warn(priv, "cm send completion event with wrid %d (> %d)\n",
			   wr_id, ipoib_sendq_size);
		return;
	}

	tx_req = &tx->tx_ring[wr_id];

	ib_dma_unmap_single(priv->ca, tx_req->mapping, tx_req->skb->len, DMA_TO_DEVICE);

	/* FIXME: is this right? Shouldn't we only increment on success? */
	++dev->stats.tx_packets;
	dev->stats.tx_bytes += tx_req->skb->len;

	dev_kfree_skb_any(tx_req->skb);

	netif_tx_lock(dev);

	++tx->tx_tail;
	if (unlikely(--priv->tx_outstanding == ipoib_sendq_size >> 1) &&
	    netif_queue_stopped(dev) &&
	    test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
		netif_wake_queue(dev);

	if (wc->status != IB_WC_SUCCESS &&
	    wc->status != IB_WC_WR_FLUSH_ERR) {
		struct ipoib_neigh *neigh;

		ipoib_dbg(priv, "failed cm send event "
			   "(status=%d, wrid=%d vend_err %x)\n",
			   wc->status, wr_id, wc->vendor_err);

		spin_lock_irqsave(&priv->lock, flags);
		neigh = tx->neigh;

		if (neigh) {
			neigh->cm = NULL;
			list_del(&neigh->list);
			if (neigh->ah)
				ipoib_put_ah(neigh->ah);
			ipoib_neigh_free(dev, neigh);

			tx->neigh = NULL;
		}

		if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) {
			list_move(&tx->list, &priv->cm.reap_list);
			queue_work(ipoib_workqueue, &priv->cm.reap_task);
		}

		clear_bit(IPOIB_FLAG_OPER_UP, &tx->flags);

		spin_unlock_irqrestore(&priv->lock, flags);
	}

	netif_tx_unlock(dev);
}

int ipoib_cm_dev_open(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int ret;

	if (!IPOIB_CM_SUPPORTED(dev->dev_addr))
		return 0;

	priv->cm.id = ib_create_cm_id(priv->ca, ipoib_cm_rx_handler, dev);
	if (IS_ERR(priv->cm.id)) {
		printk(KERN_WARNING "%s: failed to create CM ID\n", priv->ca->name);
		ret = PTR_ERR(priv->cm.id);
		goto err_cm;
	}

	ret = ib_cm_listen(priv->cm.id, cpu_to_be64(IPOIB_CM_IETF_ID | priv->qp->qp_num),
			   0, NULL);
	if (ret) {
		printk(KERN_WARNING "%s: failed to listen on ID 0x%llx\n", priv->ca->name,
		       IPOIB_CM_IETF_ID | priv->qp->qp_num);
		goto err_listen;
	}

	return 0;

err_listen:
	ib_destroy_cm_id(priv->cm.id);
err_cm:
	priv->cm.id = NULL;
	return ret;
}

static void ipoib_cm_free_rx_reap_list(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_rx *rx, *n;
	LIST_HEAD(list);

	spin_lock_irq(&priv->lock);
	list_splice_init(&priv->cm.rx_reap_list, &list);
	spin_unlock_irq(&priv->lock);

	list_for_each_entry_safe(rx, n, &list, list) {
		ib_destroy_cm_id(rx->id);
		ib_destroy_qp(rx->qp);
		if (!ipoib_cm_has_srq(dev)) {
			ipoib_cm_free_rx_ring(priv->dev, rx->rx_ring);
			spin_lock_irq(&priv->lock);
			--priv->cm.nonsrq_conn_qp;
			spin_unlock_irq(&priv->lock);
		}
		kfree(rx);
	}
}

void ipoib_cm_dev_stop(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_rx *p;
	unsigned long begin;
	int ret;

	if (!IPOIB_CM_SUPPORTED(dev->dev_addr) || !priv->cm.id)
		return;

	ib_destroy_cm_id(priv->cm.id);
	priv->cm.id = NULL;

	spin_lock_irq(&priv->lock);
	while (!list_empty(&priv->cm.passive_ids)) {
		p = list_entry(priv->cm.passive_ids.next, typeof(*p), list);
		list_move(&p->list, &priv->cm.rx_error_list);
		p->state = IPOIB_CM_RX_ERROR;
		spin_unlock_irq(&priv->lock);
		ret = ib_modify_qp(p->qp, &ipoib_cm_err_attr, IB_QP_STATE);
		if (ret)
			ipoib_warn(priv, "unable to move qp to error state: %d\n", ret);
		spin_lock_irq(&priv->lock);
	}

	/* Wait for all RX to be drained */
	begin = jiffies;

	while (!list_empty(&priv->cm.rx_error_list) ||
	       !list_empty(&priv->cm.rx_flush_list) ||
	       !list_empty(&priv->cm.rx_drain_list)) {
		if (time_after(jiffies, begin + 5 * HZ)) {
			ipoib_warn(priv, "RX drain timing out\n");

			/*
			 * assume the HW is wedged and just free up everything.
			 */
			list_splice_init(&priv->cm.rx_flush_list,
					 &priv->cm.rx_reap_list);
			list_splice_init(&priv->cm.rx_error_list,
					 &priv->cm.rx_reap_list);
			list_splice_init(&priv->cm.rx_drain_list,
					 &priv->cm.rx_reap_list);
			break;
		}
		spin_unlock_irq(&priv->lock);
		msleep(1);
		ipoib_drain_cq(dev);
		spin_lock_irq(&priv->lock);
	}

	spin_unlock_irq(&priv->lock);

	ipoib_cm_free_rx_reap_list(dev);

	cancel_delayed_work(&priv->cm.stale_task);
}

static int ipoib_cm_rep_handler(struct ib_cm_id *cm_id, struct ib_cm_event *event)
{
	struct ipoib_cm_tx *p = cm_id->context;
	struct ipoib_dev_priv *priv = netdev_priv(p->dev);
	struct ipoib_cm_data *data = event->private_data;
	struct sk_buff_head skqueue;
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;
	struct sk_buff *skb;

	p->mtu = be32_to_cpu(data->mtu);

	if (p->mtu <= IPOIB_ENCAP_LEN) {
		ipoib_warn(priv, "Rejecting connection: mtu %d <= %d\n",
			   p->mtu, IPOIB_ENCAP_LEN);
		return -EINVAL;
	}

	qp_attr.qp_state = IB_QPS_RTR;
	ret = ib_cm_init_qp_attr(cm_id, &qp_attr, &qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to init QP attr for RTR: %d\n", ret);
		return ret;
	}

	qp_attr.rq_psn = 0 /* FIXME */;
	ret = ib_modify_qp(p->qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify QP to RTR: %d\n", ret);
		return ret;
	}

	qp_attr.qp_state = IB_QPS_RTS;
	ret = ib_cm_init_qp_attr(cm_id, &qp_attr, &qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to init QP attr for RTS: %d\n", ret);
		return ret;
	}
	ret = ib_modify_qp(p->qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify QP to RTS: %d\n", ret);
		return ret;
	}

	skb_queue_head_init(&skqueue);

	spin_lock_irq(&priv->lock);
	set_bit(IPOIB_FLAG_OPER_UP, &p->flags);
	if (p->neigh)
		while ((skb = __skb_dequeue(&p->neigh->queue)))
			__skb_queue_tail(&skqueue, skb);
	spin_unlock_irq(&priv->lock);

	while ((skb = __skb_dequeue(&skqueue))) {
		skb->dev = p->dev;
		if (dev_queue_xmit(skb))
			ipoib_warn(priv, "dev_queue_xmit failed "
				   "to requeue packet\n");
	}

	ret = ib_send_cm_rtu(cm_id, NULL, 0);
	if (ret) {
		ipoib_warn(priv, "failed to send RTU: %d\n", ret);
		return ret;
	}
	return 0;
}

static struct ib_qp *ipoib_cm_create_tx_qp(struct net_device *dev, struct ipoib_cm_tx *tx)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_qp_init_attr attr = {
		.send_cq		= priv->recv_cq,
		.recv_cq		= priv->recv_cq,
		.srq			= priv->cm.srq,
		.cap.max_send_wr	= ipoib_sendq_size,
		.cap.max_send_sge	= 1,
		.sq_sig_type		= IB_SIGNAL_ALL_WR,
		.qp_type		= IB_QPT_RC,
		.qp_context		= tx
	};

	return ib_create_qp(priv->pd, &attr);
}

static int ipoib_cm_send_req(struct net_device *dev,
			     struct ib_cm_id *id, struct ib_qp *qp,
			     u32 qpn,
			     struct ib_sa_path_rec *pathrec)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_data data = {};
	struct ib_cm_req_param req = {};

	data.qpn = cpu_to_be32(priv->qp->qp_num);
	data.mtu = cpu_to_be32(IPOIB_CM_BUF_SIZE);

	req.primary_path		= pathrec;
	req.alternate_path		= NULL;
	req.service_id			= cpu_to_be64(IPOIB_CM_IETF_ID | qpn);
	req.qp_num			= qp->qp_num;
	req.qp_type			= qp->qp_type;
	req.private_data		= &data;
	req.private_data_len		= sizeof data;
	req.flow_control		= 0;

	req.starting_psn		= 0; /* FIXME */

	/*
	 * Pick some arbitrary defaults here; we could make these
	 * module parameters if anyone cared about setting them.
	 */
	req.responder_resources		= 4;
	req.remote_cm_response_timeout	= 20;
	req.local_cm_response_timeout	= 20;
	req.retry_count			= 0; /* RFC draft warns against retries */
	req.rnr_retry_count		= 0; /* RFC draft warns against retries */
	req.max_cm_retries		= 15;
	req.srq				= ipoib_cm_has_srq(dev);
	return ib_send_cm_req(id, &req);
}

static int ipoib_cm_modify_tx_init(struct net_device *dev,
				  struct ib_cm_id *cm_id, struct ib_qp *qp)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_qp_attr qp_attr;
	int qp_attr_mask, ret;
	ret = ib_find_pkey(priv->ca, priv->port, priv->pkey, &qp_attr.pkey_index);
	if (ret) {
		ipoib_warn(priv, "pkey 0x%x not found: %d\n", priv->pkey, ret);
		return ret;
	}

	qp_attr.qp_state = IB_QPS_INIT;
	qp_attr.qp_access_flags = IB_ACCESS_LOCAL_WRITE;
	qp_attr.port_num = priv->port;
	qp_attr_mask = IB_QP_STATE | IB_QP_ACCESS_FLAGS | IB_QP_PKEY_INDEX | IB_QP_PORT;

	ret = ib_modify_qp(qp, &qp_attr, qp_attr_mask);
	if (ret) {
		ipoib_warn(priv, "failed to modify tx QP to INIT: %d\n", ret);
		return ret;
	}
	return 0;
}

static int ipoib_cm_tx_init(struct ipoib_cm_tx *p, u32 qpn,
			    struct ib_sa_path_rec *pathrec)
{
	struct ipoib_dev_priv *priv = netdev_priv(p->dev);
	int ret;

	p->tx_ring = vzalloc(ipoib_sendq_size * sizeof *p->tx_ring);
	if (!p->tx_ring) {
		ipoib_warn(priv, "failed to allocate tx ring\n");
		ret = -ENOMEM;
		goto err_tx;
	}

	p->qp = ipoib_cm_create_tx_qp(p->dev, p);
	if (IS_ERR(p->qp)) {
		ret = PTR_ERR(p->qp);
		ipoib_warn(priv, "failed to allocate tx qp: %d\n", ret);
		goto err_qp;
	}

	p->id = ib_create_cm_id(priv->ca, ipoib_cm_tx_handler, p);
	if (IS_ERR(p->id)) {
		ret = PTR_ERR(p->id);
		ipoib_warn(priv, "failed to create tx cm id: %d\n", ret);
		goto err_id;
	}

	ret = ipoib_cm_modify_tx_init(p->dev, p->id,  p->qp);
	if (ret) {
		ipoib_warn(priv, "failed to modify tx qp to rtr: %d\n", ret);
		goto err_modify;
	}

	ret = ipoib_cm_send_req(p->dev, p->id, p->qp, qpn, pathrec);
	if (ret) {
		ipoib_warn(priv, "failed to send cm req: %d\n", ret);
		goto err_send_cm;
	}

	ipoib_dbg(priv, "Request connection 0x%x for gid %pI6 qpn 0x%x\n",
		  p->qp->qp_num, pathrec->dgid.raw, qpn);

	return 0;

err_send_cm:
err_modify:
	ib_destroy_cm_id(p->id);
err_id:
	p->id = NULL;
	ib_destroy_qp(p->qp);
err_qp:
	p->qp = NULL;
	vfree(p->tx_ring);
err_tx:
	return ret;
}

static void ipoib_cm_tx_destroy(struct ipoib_cm_tx *p)
{
	struct ipoib_dev_priv *priv = netdev_priv(p->dev);
	struct ipoib_cm_tx_buf *tx_req;
	unsigned long begin;

	ipoib_dbg(priv, "Destroy active connection 0x%x head 0x%x tail 0x%x\n",
		  p->qp ? p->qp->qp_num : 0, p->tx_head, p->tx_tail);

	if (p->id)
		ib_destroy_cm_id(p->id);

	if (p->tx_ring) {
		/* Wait for all sends to complete */
		begin = jiffies;
		while ((int) p->tx_tail - (int) p->tx_head < 0) {
			if (time_after(jiffies, begin + 5 * HZ)) {
				ipoib_warn(priv, "timing out; %d sends not completed\n",
					   p->tx_head - p->tx_tail);
				goto timeout;
			}

			msleep(1);
		}
	}

timeout:

	while ((int) p->tx_tail - (int) p->tx_head < 0) {
		tx_req = &p->tx_ring[p->tx_tail & (ipoib_sendq_size - 1)];
		ib_dma_unmap_single(priv->ca, tx_req->mapping, tx_req->skb->len,
				    DMA_TO_DEVICE);
		dev_kfree_skb_any(tx_req->skb);
		++p->tx_tail;
		netif_tx_lock_bh(p->dev);
		if (unlikely(--priv->tx_outstanding == ipoib_sendq_size >> 1) &&
		    netif_queue_stopped(p->dev) &&
		    test_bit(IPOIB_FLAG_ADMIN_UP, &priv->flags))
			netif_wake_queue(p->dev);
		netif_tx_unlock_bh(p->dev);
	}

	if (p->qp)
		ib_destroy_qp(p->qp);

	vfree(p->tx_ring);
	kfree(p);
}

static int ipoib_cm_tx_handler(struct ib_cm_id *cm_id,
			       struct ib_cm_event *event)
{
	struct ipoib_cm_tx *tx = cm_id->context;
	struct ipoib_dev_priv *priv = netdev_priv(tx->dev);
	struct net_device *dev = priv->dev;
	struct ipoib_neigh *neigh;
	unsigned long flags;
	int ret;

	switch (event->event) {
	case IB_CM_DREQ_RECEIVED:
		ipoib_dbg(priv, "DREQ received.\n");
		ib_send_cm_drep(cm_id, NULL, 0);
		break;
	case IB_CM_REP_RECEIVED:
		ipoib_dbg(priv, "REP received.\n");
		ret = ipoib_cm_rep_handler(cm_id, event);
		if (ret)
			ib_send_cm_rej(cm_id, IB_CM_REJ_CONSUMER_DEFINED,
				       NULL, 0, NULL, 0);
		break;
	case IB_CM_REQ_ERROR:
	case IB_CM_REJ_RECEIVED:
	case IB_CM_TIMEWAIT_EXIT:
		ipoib_dbg(priv, "CM error %d.\n", event->event);
		netif_tx_lock_bh(dev);
		spin_lock_irqsave(&priv->lock, flags);
		neigh = tx->neigh;

		if (neigh) {
			neigh->cm = NULL;
			list_del(&neigh->list);
			if (neigh->ah)
				ipoib_put_ah(neigh->ah);
			ipoib_neigh_free(dev, neigh);

			tx->neigh = NULL;
		}

		if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) {
			list_move(&tx->list, &priv->cm.reap_list);
			queue_work(ipoib_workqueue, &priv->cm.reap_task);
		}

		spin_unlock_irqrestore(&priv->lock, flags);
		netif_tx_unlock_bh(dev);
		break;
	default:
		break;
	}

	return 0;
}

struct ipoib_cm_tx *ipoib_cm_create_tx(struct net_device *dev, struct ipoib_path *path,
				       struct ipoib_neigh *neigh)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ipoib_cm_tx *tx;

	tx = kzalloc(sizeof *tx, GFP_ATOMIC);
	if (!tx)
		return NULL;

	neigh->cm = tx;
	tx->neigh = neigh;
	tx->path = path;
	tx->dev = dev;
	list_add(&tx->list, &priv->cm.start_list);
	set_bit(IPOIB_FLAG_INITIALIZED, &tx->flags);
	queue_work(ipoib_workqueue, &priv->cm.start_task);
	return tx;
}

void ipoib_cm_destroy_tx(struct ipoib_cm_tx *tx)
{
	struct ipoib_dev_priv *priv = netdev_priv(tx->dev);
	if (test_and_clear_bit(IPOIB_FLAG_INITIALIZED, &tx->flags)) {
		list_move(&tx->list, &priv->cm.reap_list);
		queue_work(ipoib_workqueue, &priv->cm.reap_task);
		ipoib_dbg(priv, "Reap connection for gid %pI6\n",
			  tx->neigh->dgid.raw);
		tx->neigh = NULL;
	}
}

static void ipoib_cm_tx_start(struct work_struct *work)
{
	struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv,
						   cm.start_task);
	struct net_device *dev = priv->dev;
	struct ipoib_neigh *neigh;
	struct ipoib_cm_tx *p;
	unsigned long flags;
	int ret;

	struct ib_sa_path_rec pathrec;
	u32 qpn;

	netif_tx_lock_bh(dev);
	spin_lock_irqsave(&priv->lock, flags);

	while (!list_empty(&priv->cm.start_list)) {
		p = list_entry(priv->cm.start_list.next, typeof(*p), list);
		list_del_init(&p->list);
		neigh = p->neigh;
		qpn = IPOIB_QPN(neigh->neighbour->ha);
		memcpy(&pathrec, &p->path->pathrec, sizeof pathrec);

		spin_unlock_irqrestore(&priv->lock, flags);
		netif_tx_unlock_bh(dev);

		ret = ipoib_cm_tx_init(p, qpn, &pathrec);

		netif_tx_lock_bh(dev);
		spin_lock_irqsave(&priv->lock, flags);

		if (ret) {
			neigh = p->neigh;
			if (neigh) {
				neigh->cm = NULL;
				list_del(&neigh->list);
				if (neigh->ah)
					ipoib_put_ah(neigh->ah);
				ipoib_neigh_free(dev, neigh);
			}
			list_del(&p->list);
			kfree(p);
		}
	}

	spin_unlock_irqrestore(&priv->lock, flags);
	netif_tx_unlock_bh(dev);
}

static void ipoib_cm_tx_reap(struct work_struct *work)
{
	struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv,
						   cm.reap_task);
	struct net_device *dev = priv->dev;
	struct ipoib_cm_tx *p;
	unsigned long flags;

	netif_tx_lock_bh(dev);
	spin_lock_irqsave(&priv->lock, flags);

	while (!list_empty(&priv->cm.reap_list)) {
		p = list_entry(priv->cm.reap_list.next, typeof(*p), list);
		list_del(&p->list);
		spin_unlock_irqrestore(&priv->lock, flags);
		netif_tx_unlock_bh(dev);
		ipoib_cm_tx_destroy(p);
		netif_tx_lock_bh(dev);
		spin_lock_irqsave(&priv->lock, flags);
	}

	spin_unlock_irqrestore(&priv->lock, flags);
	netif_tx_unlock_bh(dev);
}

static void ipoib_cm_skb_reap(struct work_struct *work)
{
	struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv,
						   cm.skb_task);
	struct net_device *dev = priv->dev;
	struct sk_buff *skb;
	unsigned long flags;
	unsigned mtu = priv->mcast_mtu;

	netif_tx_lock_bh(dev);
	spin_lock_irqsave(&priv->lock, flags);

	while ((skb = skb_dequeue(&priv->cm.skb_queue))) {
		spin_unlock_irqrestore(&priv->lock, flags);
		netif_tx_unlock_bh(dev);

		if (skb->protocol == htons(ETH_P_IP))
			icmp_send(skb, ICMP_DEST_UNREACH, ICMP_FRAG_NEEDED, htonl(mtu));
#if defined(CONFIG_IPV6) || defined(CONFIG_IPV6_MODULE)
		else if (skb->protocol == htons(ETH_P_IPV6))
			icmpv6_send(skb, ICMPV6_PKT_TOOBIG, 0, mtu);
#endif
		dev_kfree_skb_any(skb);

		netif_tx_lock_bh(dev);
		spin_lock_irqsave(&priv->lock, flags);
	}

	spin_unlock_irqrestore(&priv->lock, flags);
	netif_tx_unlock_bh(dev);
}

void ipoib_cm_skb_too_long(struct net_device *dev, struct sk_buff *skb,
			   unsigned int mtu)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int e = skb_queue_empty(&priv->cm.skb_queue);

	if (skb_dst(skb))
		skb_dst(skb)->ops->update_pmtu(skb_dst(skb), mtu);

	skb_queue_tail(&priv->cm.skb_queue, skb);
	if (e)
		queue_work(ipoib_workqueue, &priv->cm.skb_task);
}

static void ipoib_cm_rx_reap(struct work_struct *work)
{
	ipoib_cm_free_rx_reap_list(container_of(work, struct ipoib_dev_priv,
						cm.rx_reap_task)->dev);
}

static void ipoib_cm_stale_task(struct work_struct *work)
{
	struct ipoib_dev_priv *priv = container_of(work, struct ipoib_dev_priv,
						   cm.stale_task.work);
	struct ipoib_cm_rx *p;
	int ret;

	spin_lock_irq(&priv->lock);
	while (!list_empty(&priv->cm.passive_ids)) {
		/* List is sorted by LRU, start from tail,
		 * stop when we see a recently used entry */
		p = list_entry(priv->cm.passive_ids.prev, typeof(*p), list);
		if (time_before_eq(jiffies, p->jiffies + IPOIB_CM_RX_TIMEOUT))
			break;
		list_move(&p->list, &priv->cm.rx_error_list);
		p->state = IPOIB_CM_RX_ERROR;
		spin_unlock_irq(&priv->lock);
		ret = ib_modify_qp(p->qp, &ipoib_cm_err_attr, IB_QP_STATE);
		if (ret)
			ipoib_warn(priv, "unable to move qp to error state: %d\n", ret);
		spin_lock_irq(&priv->lock);
	}

	if (!list_empty(&priv->cm.passive_ids))
		queue_delayed_work(ipoib_workqueue,
				   &priv->cm.stale_task, IPOIB_CM_RX_DELAY);
	spin_unlock_irq(&priv->lock);
}


static ssize_t show_mode(struct device *d, struct device_attribute *attr,
			 char *buf)
{
	struct ipoib_dev_priv *priv = netdev_priv(to_net_dev(d));

	if (test_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags))
		return sprintf(buf, "connected\n");
	else
		return sprintf(buf, "datagram\n");
}

static ssize_t set_mode(struct device *d, struct device_attribute *attr,
			const char *buf, size_t count)
{
	struct net_device *dev = to_net_dev(d);
	struct ipoib_dev_priv *priv = netdev_priv(dev);

	if (!rtnl_trylock())
		return restart_syscall();

	/* flush paths if we switch modes so that connections are restarted */
	if (IPOIB_CM_SUPPORTED(dev->dev_addr) && !strcmp(buf, "connected\n")) {
		set_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags);
		ipoib_warn(priv, "enabling connected mode "
			   "will cause multicast packet drops\n");
		netdev_update_features(dev);
		rtnl_unlock();
		priv->tx_wr.send_flags &= ~IB_SEND_IP_CSUM;

		ipoib_flush_paths(dev);
		return count;
	}

	if (!strcmp(buf, "datagram\n")) {
		clear_bit(IPOIB_FLAG_ADMIN_CM, &priv->flags);
		netdev_update_features(dev);
		dev_set_mtu(dev, min(priv->mcast_mtu, dev->mtu));
		rtnl_unlock();
		ipoib_flush_paths(dev);

		return count;
	}
	rtnl_unlock();

	return -EINVAL;
}

static DEVICE_ATTR(mode, S_IWUSR | S_IRUGO, show_mode, set_mode);

int ipoib_cm_add_mode_attr(struct net_device *dev)
{
	return device_create_file(&dev->dev, &dev_attr_mode);
}

static void ipoib_cm_create_srq(struct net_device *dev, int max_sge)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	struct ib_srq_init_attr srq_init_attr = {
		.srq_type = IB_SRQT_BASIC,
		.attr = {
			.max_wr  = ipoib_recvq_size,
			.max_sge = max_sge
		}
	};

	priv->cm.srq = ib_create_srq(priv->pd, &srq_init_attr);
	if (IS_ERR(priv->cm.srq)) {
		if (PTR_ERR(priv->cm.srq) != -ENOSYS)
			printk(KERN_WARNING "%s: failed to allocate SRQ, error %ld\n",
			       priv->ca->name, PTR_ERR(priv->cm.srq));
		priv->cm.srq = NULL;
		return;
	}

	priv->cm.srq_ring = vzalloc(ipoib_recvq_size * sizeof *priv->cm.srq_ring);
	if (!priv->cm.srq_ring) {
		printk(KERN_WARNING "%s: failed to allocate CM SRQ ring (%d entries)\n",
		       priv->ca->name, ipoib_recvq_size);
		ib_destroy_srq(priv->cm.srq);
		priv->cm.srq = NULL;
		return;
	}

}

int ipoib_cm_dev_init(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int i, ret;
	struct ib_device_attr attr;

	INIT_LIST_HEAD(&priv->cm.passive_ids);
	INIT_LIST_HEAD(&priv->cm.reap_list);
	INIT_LIST_HEAD(&priv->cm.start_list);
	INIT_LIST_HEAD(&priv->cm.rx_error_list);
	INIT_LIST_HEAD(&priv->cm.rx_flush_list);
	INIT_LIST_HEAD(&priv->cm.rx_drain_list);
	INIT_LIST_HEAD(&priv->cm.rx_reap_list);
	INIT_WORK(&priv->cm.start_task, ipoib_cm_tx_start);
	INIT_WORK(&priv->cm.reap_task, ipoib_cm_tx_reap);
	INIT_WORK(&priv->cm.skb_task, ipoib_cm_skb_reap);
	INIT_WORK(&priv->cm.rx_reap_task, ipoib_cm_rx_reap);
	INIT_DELAYED_WORK(&priv->cm.stale_task, ipoib_cm_stale_task);

	skb_queue_head_init(&priv->cm.skb_queue);

	ret = ib_query_device(priv->ca, &attr);
	if (ret) {
		printk(KERN_WARNING "ib_query_device() failed with %d\n", ret);
		return ret;
	}

	ipoib_dbg(priv, "max_srq_sge=%d\n", attr.max_srq_sge);

	attr.max_srq_sge = min_t(int, IPOIB_CM_RX_SG, attr.max_srq_sge);
	ipoib_cm_create_srq(dev, attr.max_srq_sge);
	if (ipoib_cm_has_srq(dev)) {
		priv->cm.max_cm_mtu = attr.max_srq_sge * PAGE_SIZE - 0x10;
		priv->cm.num_frags  = attr.max_srq_sge;
		ipoib_dbg(priv, "max_cm_mtu = 0x%x, num_frags=%d\n",
			  priv->cm.max_cm_mtu, priv->cm.num_frags);
	} else {
		priv->cm.max_cm_mtu = IPOIB_CM_MTU;
		priv->cm.num_frags  = IPOIB_CM_RX_SG;
	}

	ipoib_cm_init_rx_wr(dev, &priv->cm.rx_wr, priv->cm.rx_sge);

	if (ipoib_cm_has_srq(dev)) {
		for (i = 0; i < ipoib_recvq_size; ++i) {
			if (!ipoib_cm_alloc_rx_skb(dev, priv->cm.srq_ring, i,
						   priv->cm.num_frags - 1,
						   priv->cm.srq_ring[i].mapping)) {
				ipoib_warn(priv, "failed to allocate "
					   "receive buffer %d\n", i);
				ipoib_cm_dev_cleanup(dev);
				return -ENOMEM;
			}

			if (ipoib_cm_post_receive_srq(dev, i)) {
				ipoib_warn(priv, "ipoib_cm_post_receive_srq "
					   "failed for buf %d\n", i);
				ipoib_cm_dev_cleanup(dev);
				return -EIO;
			}
		}
	}

	priv->dev->dev_addr[0] = IPOIB_FLAGS_RC;
	return 0;
}

void ipoib_cm_dev_cleanup(struct net_device *dev)
{
	struct ipoib_dev_priv *priv = netdev_priv(dev);
	int ret;

	if (!priv->cm.srq)
		return;

	ipoib_dbg(priv, "Cleanup ipoib connected mode.\n");

	ret = ib_destroy_srq(priv->cm.srq);
	if (ret)
		ipoib_warn(priv, "ib_destroy_srq failed: %d\n", ret);

	priv->cm.srq = NULL;
	if (!priv->cm.srq_ring)
		return;

	ipoib_cm_free_rx_ring(dev, priv->cm.srq_ring);
	priv->cm.srq_ring = NULL;
}