summaryrefslogtreecommitdiffstats
path: root/drivers/infiniband/hw/hfi1/verbs.c
blob: 26bea51869bf058e2d8fdecefd83962b20f97df7 (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
// SPDX-License-Identifier: GPL-2.0 or BSD-3-Clause
/*
 * Copyright(c) 2015 - 2020 Intel Corporation.
 */

#include <rdma/ib_mad.h>
#include <rdma/ib_user_verbs.h>
#include <linux/io.h>
#include <linux/module.h>
#include <linux/utsname.h>
#include <linux/rculist.h>
#include <linux/mm.h>
#include <linux/vmalloc.h>
#include <rdma/opa_addr.h>
#include <linux/nospec.h>

#include "hfi.h"
#include "common.h"
#include "device.h"
#include "trace.h"
#include "qp.h"
#include "verbs_txreq.h"
#include "debugfs.h"
#include "vnic.h"
#include "fault.h"
#include "affinity.h"
#include "ipoib.h"

static unsigned int hfi1_lkey_table_size = 16;
module_param_named(lkey_table_size, hfi1_lkey_table_size, uint,
		   S_IRUGO);
MODULE_PARM_DESC(lkey_table_size,
		 "LKEY table size in bits (2^n, 1 <= n <= 23)");

static unsigned int hfi1_max_pds = 0xFFFF;
module_param_named(max_pds, hfi1_max_pds, uint, S_IRUGO);
MODULE_PARM_DESC(max_pds,
		 "Maximum number of protection domains to support");

static unsigned int hfi1_max_ahs = 0xFFFF;
module_param_named(max_ahs, hfi1_max_ahs, uint, S_IRUGO);
MODULE_PARM_DESC(max_ahs, "Maximum number of address handles to support");

unsigned int hfi1_max_cqes = 0x2FFFFF;
module_param_named(max_cqes, hfi1_max_cqes, uint, S_IRUGO);
MODULE_PARM_DESC(max_cqes,
		 "Maximum number of completion queue entries to support");

unsigned int hfi1_max_cqs = 0x1FFFF;
module_param_named(max_cqs, hfi1_max_cqs, uint, S_IRUGO);
MODULE_PARM_DESC(max_cqs, "Maximum number of completion queues to support");

unsigned int hfi1_max_qp_wrs = 0x3FFF;
module_param_named(max_qp_wrs, hfi1_max_qp_wrs, uint, S_IRUGO);
MODULE_PARM_DESC(max_qp_wrs, "Maximum number of QP WRs to support");

unsigned int hfi1_max_qps = 32768;
module_param_named(max_qps, hfi1_max_qps, uint, S_IRUGO);
MODULE_PARM_DESC(max_qps, "Maximum number of QPs to support");

unsigned int hfi1_max_sges = 0x60;
module_param_named(max_sges, hfi1_max_sges, uint, S_IRUGO);
MODULE_PARM_DESC(max_sges, "Maximum number of SGEs to support");

unsigned int hfi1_max_mcast_grps = 16384;
module_param_named(max_mcast_grps, hfi1_max_mcast_grps, uint, S_IRUGO);
MODULE_PARM_DESC(max_mcast_grps,
		 "Maximum number of multicast groups to support");

unsigned int hfi1_max_mcast_qp_attached = 16;
module_param_named(max_mcast_qp_attached, hfi1_max_mcast_qp_attached,
		   uint, S_IRUGO);
MODULE_PARM_DESC(max_mcast_qp_attached,
		 "Maximum number of attached QPs to support");

unsigned int hfi1_max_srqs = 1024;
module_param_named(max_srqs, hfi1_max_srqs, uint, S_IRUGO);
MODULE_PARM_DESC(max_srqs, "Maximum number of SRQs to support");

unsigned int hfi1_max_srq_sges = 128;
module_param_named(max_srq_sges, hfi1_max_srq_sges, uint, S_IRUGO);
MODULE_PARM_DESC(max_srq_sges, "Maximum number of SRQ SGEs to support");

unsigned int hfi1_max_srq_wrs = 0x1FFFF;
module_param_named(max_srq_wrs, hfi1_max_srq_wrs, uint, S_IRUGO);
MODULE_PARM_DESC(max_srq_wrs, "Maximum number of SRQ WRs support");

unsigned short piothreshold = 256;
module_param(piothreshold, ushort, S_IRUGO);
MODULE_PARM_DESC(piothreshold, "size used to determine sdma vs. pio");

static unsigned int sge_copy_mode;
module_param(sge_copy_mode, uint, S_IRUGO);
MODULE_PARM_DESC(sge_copy_mode,
		 "Verbs copy mode: 0 use memcpy, 1 use cacheless copy, 2 adapt based on WSS");

static void verbs_sdma_complete(
	struct sdma_txreq *cookie,
	int status);

static int pio_wait(struct rvt_qp *qp,
		    struct send_context *sc,
		    struct hfi1_pkt_state *ps,
		    u32 flag);

/* Length of buffer to create verbs txreq cache name */
#define TXREQ_NAME_LEN 24

static uint wss_threshold = 80;
module_param(wss_threshold, uint, S_IRUGO);
MODULE_PARM_DESC(wss_threshold, "Percentage (1-100) of LLC to use as a threshold for a cacheless copy");
static uint wss_clean_period = 256;
module_param(wss_clean_period, uint, S_IRUGO);
MODULE_PARM_DESC(wss_clean_period, "Count of verbs copies before an entry in the page copy table is cleaned");

/*
 * Translate ib_wr_opcode into ib_wc_opcode.
 */
const enum ib_wc_opcode ib_hfi1_wc_opcode[] = {
	[IB_WR_RDMA_WRITE] = IB_WC_RDMA_WRITE,
	[IB_WR_TID_RDMA_WRITE] = IB_WC_RDMA_WRITE,
	[IB_WR_RDMA_WRITE_WITH_IMM] = IB_WC_RDMA_WRITE,
	[IB_WR_SEND] = IB_WC_SEND,
	[IB_WR_SEND_WITH_IMM] = IB_WC_SEND,
	[IB_WR_RDMA_READ] = IB_WC_RDMA_READ,
	[IB_WR_TID_RDMA_READ] = IB_WC_RDMA_READ,
	[IB_WR_ATOMIC_CMP_AND_SWP] = IB_WC_COMP_SWAP,
	[IB_WR_ATOMIC_FETCH_AND_ADD] = IB_WC_FETCH_ADD,
	[IB_WR_SEND_WITH_INV] = IB_WC_SEND,
	[IB_WR_LOCAL_INV] = IB_WC_LOCAL_INV,
	[IB_WR_REG_MR] = IB_WC_REG_MR
};

/*
 * Length of header by opcode, 0 --> not supported
 */
const u8 hdr_len_by_opcode[256] = {
	/* RC */
	[IB_OPCODE_RC_SEND_FIRST]                     = 12 + 8,
	[IB_OPCODE_RC_SEND_MIDDLE]                    = 12 + 8,
	[IB_OPCODE_RC_SEND_LAST]                      = 12 + 8,
	[IB_OPCODE_RC_SEND_LAST_WITH_IMMEDIATE]       = 12 + 8 + 4,
	[IB_OPCODE_RC_SEND_ONLY]                      = 12 + 8,
	[IB_OPCODE_RC_SEND_ONLY_WITH_IMMEDIATE]       = 12 + 8 + 4,
	[IB_OPCODE_RC_RDMA_WRITE_FIRST]               = 12 + 8 + 16,
	[IB_OPCODE_RC_RDMA_WRITE_MIDDLE]              = 12 + 8,
	[IB_OPCODE_RC_RDMA_WRITE_LAST]                = 12 + 8,
	[IB_OPCODE_RC_RDMA_WRITE_LAST_WITH_IMMEDIATE] = 12 + 8 + 4,
	[IB_OPCODE_RC_RDMA_WRITE_ONLY]                = 12 + 8 + 16,
	[IB_OPCODE_RC_RDMA_WRITE_ONLY_WITH_IMMEDIATE] = 12 + 8 + 20,
	[IB_OPCODE_RC_RDMA_READ_REQUEST]              = 12 + 8 + 16,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST]       = 12 + 8 + 4,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE]      = 12 + 8,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_LAST]        = 12 + 8 + 4,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_ONLY]        = 12 + 8 + 4,
	[IB_OPCODE_RC_ACKNOWLEDGE]                    = 12 + 8 + 4,
	[IB_OPCODE_RC_ATOMIC_ACKNOWLEDGE]             = 12 + 8 + 4 + 8,
	[IB_OPCODE_RC_COMPARE_SWAP]                   = 12 + 8 + 28,
	[IB_OPCODE_RC_FETCH_ADD]                      = 12 + 8 + 28,
	[IB_OPCODE_RC_SEND_LAST_WITH_INVALIDATE]      = 12 + 8 + 4,
	[IB_OPCODE_RC_SEND_ONLY_WITH_INVALIDATE]      = 12 + 8 + 4,
	[IB_OPCODE_TID_RDMA_READ_REQ]                 = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_READ_RESP]                = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_WRITE_REQ]                = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_WRITE_RESP]               = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_WRITE_DATA]               = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_WRITE_DATA_LAST]          = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_ACK]                      = 12 + 8 + 36,
	[IB_OPCODE_TID_RDMA_RESYNC]                   = 12 + 8 + 36,
	/* UC */
	[IB_OPCODE_UC_SEND_FIRST]                     = 12 + 8,
	[IB_OPCODE_UC_SEND_MIDDLE]                    = 12 + 8,
	[IB_OPCODE_UC_SEND_LAST]                      = 12 + 8,
	[IB_OPCODE_UC_SEND_LAST_WITH_IMMEDIATE]       = 12 + 8 + 4,
	[IB_OPCODE_UC_SEND_ONLY]                      = 12 + 8,
	[IB_OPCODE_UC_SEND_ONLY_WITH_IMMEDIATE]       = 12 + 8 + 4,
	[IB_OPCODE_UC_RDMA_WRITE_FIRST]               = 12 + 8 + 16,
	[IB_OPCODE_UC_RDMA_WRITE_MIDDLE]              = 12 + 8,
	[IB_OPCODE_UC_RDMA_WRITE_LAST]                = 12 + 8,
	[IB_OPCODE_UC_RDMA_WRITE_LAST_WITH_IMMEDIATE] = 12 + 8 + 4,
	[IB_OPCODE_UC_RDMA_WRITE_ONLY]                = 12 + 8 + 16,
	[IB_OPCODE_UC_RDMA_WRITE_ONLY_WITH_IMMEDIATE] = 12 + 8 + 20,
	/* UD */
	[IB_OPCODE_UD_SEND_ONLY]                      = 12 + 8 + 8,
	[IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE]       = 12 + 8 + 12
};

static const opcode_handler opcode_handler_tbl[256] = {
	/* RC */
	[IB_OPCODE_RC_SEND_FIRST]                     = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_MIDDLE]                    = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_LAST]                      = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_LAST_WITH_IMMEDIATE]       = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_ONLY]                      = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_ONLY_WITH_IMMEDIATE]       = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_FIRST]               = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_MIDDLE]              = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_LAST]                = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_LAST_WITH_IMMEDIATE] = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_ONLY]                = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_WRITE_ONLY_WITH_IMMEDIATE] = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_READ_REQUEST]              = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_FIRST]       = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_MIDDLE]      = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_LAST]        = &hfi1_rc_rcv,
	[IB_OPCODE_RC_RDMA_READ_RESPONSE_ONLY]        = &hfi1_rc_rcv,
	[IB_OPCODE_RC_ACKNOWLEDGE]                    = &hfi1_rc_rcv,
	[IB_OPCODE_RC_ATOMIC_ACKNOWLEDGE]             = &hfi1_rc_rcv,
	[IB_OPCODE_RC_COMPARE_SWAP]                   = &hfi1_rc_rcv,
	[IB_OPCODE_RC_FETCH_ADD]                      = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_LAST_WITH_INVALIDATE]      = &hfi1_rc_rcv,
	[IB_OPCODE_RC_SEND_ONLY_WITH_INVALIDATE]      = &hfi1_rc_rcv,

	/* TID RDMA has separate handlers for different opcodes.*/
	[IB_OPCODE_TID_RDMA_WRITE_REQ]       = &hfi1_rc_rcv_tid_rdma_write_req,
	[IB_OPCODE_TID_RDMA_WRITE_RESP]      = &hfi1_rc_rcv_tid_rdma_write_resp,
	[IB_OPCODE_TID_RDMA_WRITE_DATA]      = &hfi1_rc_rcv_tid_rdma_write_data,
	[IB_OPCODE_TID_RDMA_WRITE_DATA_LAST] = &hfi1_rc_rcv_tid_rdma_write_data,
	[IB_OPCODE_TID_RDMA_READ_REQ]        = &hfi1_rc_rcv_tid_rdma_read_req,
	[IB_OPCODE_TID_RDMA_READ_RESP]       = &hfi1_rc_rcv_tid_rdma_read_resp,
	[IB_OPCODE_TID_RDMA_RESYNC]          = &hfi1_rc_rcv_tid_rdma_resync,
	[IB_OPCODE_TID_RDMA_ACK]             = &hfi1_rc_rcv_tid_rdma_ack,

	/* UC */
	[IB_OPCODE_UC_SEND_FIRST]                     = &hfi1_uc_rcv,
	[IB_OPCODE_UC_SEND_MIDDLE]                    = &hfi1_uc_rcv,
	[IB_OPCODE_UC_SEND_LAST]                      = &hfi1_uc_rcv,
	[IB_OPCODE_UC_SEND_LAST_WITH_IMMEDIATE]       = &hfi1_uc_rcv,
	[IB_OPCODE_UC_SEND_ONLY]                      = &hfi1_uc_rcv,
	[IB_OPCODE_UC_SEND_ONLY_WITH_IMMEDIATE]       = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_FIRST]               = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_MIDDLE]              = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_LAST]                = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_LAST_WITH_IMMEDIATE] = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_ONLY]                = &hfi1_uc_rcv,
	[IB_OPCODE_UC_RDMA_WRITE_ONLY_WITH_IMMEDIATE] = &hfi1_uc_rcv,
	/* UD */
	[IB_OPCODE_UD_SEND_ONLY]                      = &hfi1_ud_rcv,
	[IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE]       = &hfi1_ud_rcv,
	/* CNP */
	[IB_OPCODE_CNP]				      = &hfi1_cnp_rcv
};

#define OPMASK 0x1f

static const u32 pio_opmask[BIT(3)] = {
	/* RC */
	[IB_OPCODE_RC >> 5] =
		BIT(RC_OP(SEND_ONLY) & OPMASK) |
		BIT(RC_OP(SEND_ONLY_WITH_IMMEDIATE) & OPMASK) |
		BIT(RC_OP(RDMA_WRITE_ONLY) & OPMASK) |
		BIT(RC_OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE) & OPMASK) |
		BIT(RC_OP(RDMA_READ_REQUEST) & OPMASK) |
		BIT(RC_OP(ACKNOWLEDGE) & OPMASK) |
		BIT(RC_OP(ATOMIC_ACKNOWLEDGE) & OPMASK) |
		BIT(RC_OP(COMPARE_SWAP) & OPMASK) |
		BIT(RC_OP(FETCH_ADD) & OPMASK),
	/* UC */
	[IB_OPCODE_UC >> 5] =
		BIT(UC_OP(SEND_ONLY) & OPMASK) |
		BIT(UC_OP(SEND_ONLY_WITH_IMMEDIATE) & OPMASK) |
		BIT(UC_OP(RDMA_WRITE_ONLY) & OPMASK) |
		BIT(UC_OP(RDMA_WRITE_ONLY_WITH_IMMEDIATE) & OPMASK),
};

/*
 * System image GUID.
 */
__be64 ib_hfi1_sys_image_guid;

/*
 * Make sure the QP is ready and able to accept the given opcode.
 */
static inline opcode_handler qp_ok(struct hfi1_packet *packet)
{
	if (!(ib_rvt_state_ops[packet->qp->state] & RVT_PROCESS_RECV_OK))
		return NULL;
	if (((packet->opcode & RVT_OPCODE_QP_MASK) ==
	     packet->qp->allowed_ops) ||
	    (packet->opcode == IB_OPCODE_CNP))
		return opcode_handler_tbl[packet->opcode];

	return NULL;
}

static u64 hfi1_fault_tx(struct rvt_qp *qp, u8 opcode, u64 pbc)
{
#ifdef CONFIG_FAULT_INJECTION
	if ((opcode & IB_OPCODE_MSP) == IB_OPCODE_MSP) {
		/*
		 * In order to drop non-IB traffic we
		 * set PbcInsertHrc to NONE (0x2).
		 * The packet will still be delivered
		 * to the receiving node but a
		 * KHdrHCRCErr (KDETH packet with a bad
		 * HCRC) will be triggered and the
		 * packet will not be delivered to the
		 * correct context.
		 */
		pbc &= ~PBC_INSERT_HCRC_SMASK;
		pbc |= (u64)PBC_IHCRC_NONE << PBC_INSERT_HCRC_SHIFT;
	} else {
		/*
		 * In order to drop regular verbs
		 * traffic we set the PbcTestEbp
		 * flag. The packet will still be
		 * delivered to the receiving node but
		 * a 'late ebp error' will be
		 * triggered and will be dropped.
		 */
		pbc |= PBC_TEST_EBP;
	}
#endif
	return pbc;
}

static opcode_handler tid_qp_ok(int opcode, struct hfi1_packet *packet)
{
	if (packet->qp->ibqp.qp_type != IB_QPT_RC ||
	    !(ib_rvt_state_ops[packet->qp->state] & RVT_PROCESS_RECV_OK))
		return NULL;
	if ((opcode & RVT_OPCODE_QP_MASK) == IB_OPCODE_TID_RDMA)
		return opcode_handler_tbl[opcode];
	return NULL;
}

void hfi1_kdeth_eager_rcv(struct hfi1_packet *packet)
{
	struct hfi1_ctxtdata *rcd = packet->rcd;
	struct ib_header *hdr = packet->hdr;
	u32 tlen = packet->tlen;
	struct hfi1_pportdata *ppd = rcd->ppd;
	struct hfi1_ibport *ibp = &ppd->ibport_data;
	struct rvt_dev_info *rdi = &ppd->dd->verbs_dev.rdi;
	opcode_handler opcode_handler;
	unsigned long flags;
	u32 qp_num;
	int lnh;
	u8 opcode;

	/* DW == LRH (2) + BTH (3) + KDETH (9) + CRC (1) */
	if (unlikely(tlen < 15 * sizeof(u32)))
		goto drop;

	lnh = be16_to_cpu(hdr->lrh[0]) & 3;
	if (lnh != HFI1_LRH_BTH)
		goto drop;

	packet->ohdr = &hdr->u.oth;
	trace_input_ibhdr(rcd->dd, packet, !!(rhf_dc_info(packet->rhf)));

	opcode = (be32_to_cpu(packet->ohdr->bth[0]) >> 24);
	inc_opstats(tlen, &rcd->opstats->stats[opcode]);

	/* verbs_qp can be picked up from any tid_rdma header struct */
	qp_num = be32_to_cpu(packet->ohdr->u.tid_rdma.r_req.verbs_qp) &
		RVT_QPN_MASK;

	rcu_read_lock();
	packet->qp = rvt_lookup_qpn(rdi, &ibp->rvp, qp_num);
	if (!packet->qp)
		goto drop_rcu;
	spin_lock_irqsave(&packet->qp->r_lock, flags);
	opcode_handler = tid_qp_ok(opcode, packet);
	if (likely(opcode_handler))
		opcode_handler(packet);
	else
		goto drop_unlock;
	spin_unlock_irqrestore(&packet->qp->r_lock, flags);
	rcu_read_unlock();

	return;
drop_unlock:
	spin_unlock_irqrestore(&packet->qp->r_lock, flags);
drop_rcu:
	rcu_read_unlock();
drop:
	ibp->rvp.n_pkt_drops++;
}

void hfi1_kdeth_expected_rcv(struct hfi1_packet *packet)
{
	struct hfi1_ctxtdata *rcd = packet->rcd;
	struct ib_header *hdr = packet->hdr;
	u32 tlen = packet->tlen;
	struct hfi1_pportdata *ppd = rcd->ppd;
	struct hfi1_ibport *ibp = &ppd->ibport_data;
	struct rvt_dev_info *rdi = &ppd->dd->verbs_dev.rdi;
	opcode_handler opcode_handler;
	unsigned long flags;
	u32 qp_num;
	int lnh;
	u8 opcode;

	/* DW == LRH (2) + BTH (3) + KDETH (9) + CRC (1) */
	if (unlikely(tlen < 15 * sizeof(u32)))
		goto drop;

	lnh = be16_to_cpu(hdr->lrh[0]) & 3;
	if (lnh != HFI1_LRH_BTH)
		goto drop;

	packet->ohdr = &hdr->u.oth;
	trace_input_ibhdr(rcd->dd, packet, !!(rhf_dc_info(packet->rhf)));

	opcode = (be32_to_cpu(packet->ohdr->bth[0]) >> 24);
	inc_opstats(tlen, &rcd->opstats->stats[opcode]);

	/* verbs_qp can be picked up from any tid_rdma header struct */
	qp_num = be32_to_cpu(packet->ohdr->u.tid_rdma.r_rsp.verbs_qp) &
		RVT_QPN_MASK;

	rcu_read_lock();
	packet->qp = rvt_lookup_qpn(rdi, &ibp->rvp, qp_num);
	if (!packet->qp)
		goto drop_rcu;
	spin_lock_irqsave(&packet->qp->r_lock, flags);
	opcode_handler = tid_qp_ok(opcode, packet);
	if (likely(opcode_handler))
		opcode_handler(packet);
	else
		goto drop_unlock;
	spin_unlock_irqrestore(&packet->qp->r_lock, flags);
	rcu_read_unlock();

	return;
drop_unlock:
	spin_unlock_irqrestore(&packet->qp->r_lock, flags);
drop_rcu:
	rcu_read_unlock();
drop:
	ibp->rvp.n_pkt_drops++;
}

static int hfi1_do_pkey_check(struct hfi1_packet *packet)
{
	struct hfi1_ctxtdata *rcd = packet->rcd;
	struct hfi1_pportdata *ppd = rcd->ppd;
	struct hfi1_16b_header *hdr = packet->hdr;
	u16 pkey;

	/* Pkey check needed only for bypass packets */
	if (packet->etype != RHF_RCV_TYPE_BYPASS)
		return 0;

	/* Perform pkey check */
	pkey = hfi1_16B_get_pkey(hdr);
	return ingress_pkey_check(ppd, pkey, packet->sc,
				  packet->qp->s_pkey_index,
				  packet->slid, true);
}

static inline void hfi1_handle_packet(struct hfi1_packet *packet,
				      bool is_mcast)
{
	u32 qp_num;
	struct hfi1_ctxtdata *rcd = packet->rcd;
	struct hfi1_pportdata *ppd = rcd->ppd;
	struct hfi1_ibport *ibp = rcd_to_iport(rcd);
	struct rvt_dev_info *rdi = &ppd->dd->verbs_dev.rdi;
	opcode_handler packet_handler;
	unsigned long flags;

	inc_opstats(packet->tlen, &rcd->opstats->stats[packet->opcode]);

	if (unlikely(is_mcast)) {
		struct rvt_mcast *mcast;
		struct rvt_mcast_qp *p;

		if (!packet->grh)
			goto drop;
		mcast = rvt_mcast_find(&ibp->rvp,
				       &packet->grh->dgid,
				       opa_get_lid(packet->dlid, 9B));
		if (!mcast)
			goto drop;
		rcu_read_lock();
		list_for_each_entry_rcu(p, &mcast->qp_list, list) {
			packet->qp = p->qp;
			if (hfi1_do_pkey_check(packet))
				goto unlock_drop;
			spin_lock_irqsave(&packet->qp->r_lock, flags);
			packet_handler = qp_ok(packet);
			if (likely(packet_handler))
				packet_handler(packet);
			else
				ibp->rvp.n_pkt_drops++;
			spin_unlock_irqrestore(&packet->qp->r_lock, flags);
		}
		rcu_read_unlock();
		/*
		 * Notify rvt_multicast_detach() if it is waiting for us
		 * to finish.
		 */
		if (atomic_dec_return(&mcast->refcount) <= 1)
			wake_up(&mcast->wait);
	} else {
		/* Get the destination QP number. */
		if (packet->etype == RHF_RCV_TYPE_BYPASS &&
		    hfi1_16B_get_l4(packet->hdr) == OPA_16B_L4_FM)
			qp_num = hfi1_16B_get_dest_qpn(packet->mgmt);
		else
			qp_num = ib_bth_get_qpn(packet->ohdr);

		rcu_read_lock();
		packet->qp = rvt_lookup_qpn(rdi, &ibp->rvp, qp_num);
		if (!packet->qp)
			goto unlock_drop;

		if (hfi1_do_pkey_check(packet))
			goto unlock_drop;

		spin_lock_irqsave(&packet->qp->r_lock, flags);
		packet_handler = qp_ok(packet);
		if (likely(packet_handler))
			packet_handler(packet);
		else
			ibp->rvp.n_pkt_drops++;
		spin_unlock_irqrestore(&packet->qp->r_lock, flags);
		rcu_read_unlock();
	}
	return;
unlock_drop:
	rcu_read_unlock();
drop:
	ibp->rvp.n_pkt_drops++;
}

/**
 * hfi1_ib_rcv - process an incoming packet
 * @packet: data packet information
 *
 * This is called to process an incoming packet at interrupt level.
 */
void hfi1_ib_rcv(struct hfi1_packet *packet)
{
	struct hfi1_ctxtdata *rcd = packet->rcd;

	trace_input_ibhdr(rcd->dd, packet, !!(rhf_dc_info(packet->rhf)));
	hfi1_handle_packet(packet, hfi1_check_mcast(packet->dlid));
}

void hfi1_16B_rcv(struct hfi1_packet *packet)
{
	struct hfi1_ctxtdata *rcd = packet->rcd;

	trace_input_ibhdr(rcd->dd, packet, false);
	hfi1_handle_packet(packet, hfi1_check_mcast(packet->dlid));
}

/*
 * This is called from a timer to check for QPs
 * which need kernel memory in order to send a packet.
 */
static void mem_timer(struct timer_list *t)
{
	struct hfi1_ibdev *dev = from_timer(dev, t, mem_timer);
	struct list_head *list = &dev->memwait;
	struct rvt_qp *qp = NULL;
	struct iowait *wait;
	unsigned long flags;
	struct hfi1_qp_priv *priv;

	write_seqlock_irqsave(&dev->iowait_lock, flags);
	if (!list_empty(list)) {
		wait = list_first_entry(list, struct iowait, list);
		qp = iowait_to_qp(wait);
		priv = qp->priv;
		list_del_init(&priv->s_iowait.list);
		priv->s_iowait.lock = NULL;
		/* refcount held until actual wake up */
		if (!list_empty(list))
			mod_timer(&dev->mem_timer, jiffies + 1);
	}
	write_sequnlock_irqrestore(&dev->iowait_lock, flags);

	if (qp)
		hfi1_qp_wakeup(qp, RVT_S_WAIT_KMEM);
}

/*
 * This is called with progress side lock held.
 */
/* New API */
static void verbs_sdma_complete(
	struct sdma_txreq *cookie,
	int status)
{
	struct verbs_txreq *tx =
		container_of(cookie, struct verbs_txreq, txreq);
	struct rvt_qp *qp = tx->qp;

	spin_lock(&qp->s_lock);
	if (tx->wqe) {
		rvt_send_complete(qp, tx->wqe, IB_WC_SUCCESS);
	} else if (qp->ibqp.qp_type == IB_QPT_RC) {
		struct hfi1_opa_header *hdr;

		hdr = &tx->phdr.hdr;
		if (unlikely(status == SDMA_TXREQ_S_ABORTED))
			hfi1_rc_verbs_aborted(qp, hdr);
		hfi1_rc_send_complete(qp, hdr);
	}
	spin_unlock(&qp->s_lock);

	hfi1_put_txreq(tx);
}

void hfi1_wait_kmem(struct rvt_qp *qp)
{
	struct hfi1_qp_priv *priv = qp->priv;
	struct ib_qp *ibqp = &qp->ibqp;
	struct ib_device *ibdev = ibqp->device;
	struct hfi1_ibdev *dev = to_idev(ibdev);

	if (list_empty(&priv->s_iowait.list)) {
		if (list_empty(&dev->memwait))
			mod_timer(&dev->mem_timer, jiffies + 1);
		qp->s_flags |= RVT_S_WAIT_KMEM;
		list_add_tail(&priv->s_iowait.list, &dev->memwait);
		priv->s_iowait.lock = &dev->iowait_lock;
		trace_hfi1_qpsleep(qp, RVT_S_WAIT_KMEM);
		rvt_get_qp(qp);
	}
}

static int wait_kmem(struct hfi1_ibdev *dev,
		     struct rvt_qp *qp,
		     struct hfi1_pkt_state *ps)
{
	unsigned long flags;
	int ret = 0;

	spin_lock_irqsave(&qp->s_lock, flags);
	if (ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) {
		write_seqlock(&dev->iowait_lock);
		list_add_tail(&ps->s_txreq->txreq.list,
			      &ps->wait->tx_head);
		hfi1_wait_kmem(qp);
		write_sequnlock(&dev->iowait_lock);
		hfi1_qp_unbusy(qp, ps->wait);
		ret = -EBUSY;
	}
	spin_unlock_irqrestore(&qp->s_lock, flags);

	return ret;
}

/*
 * This routine calls txadds for each sg entry.
 *
 * Add failures will revert the sge cursor
 */
static noinline int build_verbs_ulp_payload(
	struct sdma_engine *sde,
	u32 length,
	struct verbs_txreq *tx)
{
	struct rvt_sge_state *ss = tx->ss;
	struct rvt_sge *sg_list = ss->sg_list;
	struct rvt_sge sge = ss->sge;
	u8 num_sge = ss->num_sge;
	u32 len;
	int ret = 0;

	while (length) {
		len = rvt_get_sge_length(&ss->sge, length);
		WARN_ON_ONCE(len == 0);
		ret = sdma_txadd_kvaddr(
			sde->dd,
			&tx->txreq,
			ss->sge.vaddr,
			len);
		if (ret)
			goto bail_txadd;
		rvt_update_sge(ss, len, false);
		length -= len;
	}
	return ret;
bail_txadd:
	/* unwind cursor */
	ss->sge = sge;
	ss->num_sge = num_sge;
	ss->sg_list = sg_list;
	return ret;
}

/**
 * update_tx_opstats - record stats by opcode
 * @qp: the qp
 * @ps: transmit packet state
 * @plen: the plen in dwords
 *
 * This is a routine to record the tx opstats after a
 * packet has been presented to the egress mechanism.
 */
static void update_tx_opstats(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
			      u32 plen)
{
#ifdef CONFIG_DEBUG_FS
	struct hfi1_devdata *dd = dd_from_ibdev(qp->ibqp.device);
	struct hfi1_opcode_stats_perctx *s = get_cpu_ptr(dd->tx_opstats);

	inc_opstats(plen * 4, &s->stats[ps->opcode]);
	put_cpu_ptr(s);
#endif
}

/*
 * Build the number of DMA descriptors needed to send length bytes of data.
 *
 * NOTE: DMA mapping is held in the tx until completed in the ring or
 *       the tx desc is freed without having been submitted to the ring
 *
 * This routine ensures all the helper routine calls succeed.
 */
/* New API */
static int build_verbs_tx_desc(
	struct sdma_engine *sde,
	u32 length,
	struct verbs_txreq *tx,
	struct hfi1_ahg_info *ahg_info,
	u64 pbc)
{
	int ret = 0;
	struct hfi1_sdma_header *phdr = &tx->phdr;
	u16 hdrbytes = (tx->hdr_dwords + sizeof(pbc) / 4) << 2;
	u8 extra_bytes = 0;

	if (tx->phdr.hdr.hdr_type) {
		/*
		 * hdrbytes accounts for PBC. Need to subtract 8 bytes
		 * before calculating padding.
		 */
		extra_bytes = hfi1_get_16b_padding(hdrbytes - 8, length) +
			      (SIZE_OF_CRC << 2) + SIZE_OF_LT;
	}
	if (!ahg_info->ahgcount) {
		ret = sdma_txinit_ahg(
			&tx->txreq,
			ahg_info->tx_flags,
			hdrbytes + length +
			extra_bytes,
			ahg_info->ahgidx,
			0,
			NULL,
			0,
			verbs_sdma_complete);
		if (ret)
			goto bail_txadd;
		phdr->pbc = cpu_to_le64(pbc);
		ret = sdma_txadd_kvaddr(
			sde->dd,
			&tx->txreq,
			phdr,
			hdrbytes);
		if (ret)
			goto bail_txadd;
	} else {
		ret = sdma_txinit_ahg(
			&tx->txreq,
			ahg_info->tx_flags,
			length,
			ahg_info->ahgidx,
			ahg_info->ahgcount,
			ahg_info->ahgdesc,
			hdrbytes,
			verbs_sdma_complete);
		if (ret)
			goto bail_txadd;
	}
	/* add the ulp payload - if any. tx->ss can be NULL for acks */
	if (tx->ss) {
		ret = build_verbs_ulp_payload(sde, length, tx);
		if (ret)
			goto bail_txadd;
	}

	/* add icrc, lt byte, and padding to flit */
	if (extra_bytes)
		ret = sdma_txadd_daddr(sde->dd, &tx->txreq,
				       sde->dd->sdma_pad_phys, extra_bytes);

bail_txadd:
	return ret;
}

static u64 update_hcrc(u8 opcode, u64 pbc)
{
	if ((opcode & IB_OPCODE_TID_RDMA) == IB_OPCODE_TID_RDMA) {
		pbc &= ~PBC_INSERT_HCRC_SMASK;
		pbc |= (u64)PBC_IHCRC_LKDETH << PBC_INSERT_HCRC_SHIFT;
	}
	return pbc;
}

int hfi1_verbs_send_dma(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
			u64 pbc)
{
	struct hfi1_qp_priv *priv = qp->priv;
	struct hfi1_ahg_info *ahg_info = priv->s_ahg;
	u32 hdrwords = ps->s_txreq->hdr_dwords;
	u32 len = ps->s_txreq->s_cur_size;
	u32 plen;
	struct hfi1_ibdev *dev = ps->dev;
	struct hfi1_pportdata *ppd = ps->ppd;
	struct verbs_txreq *tx;
	u8 sc5 = priv->s_sc;
	int ret;
	u32 dwords;

	if (ps->s_txreq->phdr.hdr.hdr_type) {
		u8 extra_bytes = hfi1_get_16b_padding((hdrwords << 2), len);

		dwords = (len + extra_bytes + (SIZE_OF_CRC << 2) +
			  SIZE_OF_LT) >> 2;
	} else {
		dwords = (len + 3) >> 2;
	}
	plen = hdrwords + dwords + sizeof(pbc) / 4;

	tx = ps->s_txreq;
	if (!sdma_txreq_built(&tx->txreq)) {
		if (likely(pbc == 0)) {
			u32 vl = sc_to_vlt(dd_from_ibdev(qp->ibqp.device), sc5);

			/* No vl15 here */
			/* set PBC_DC_INFO bit (aka SC[4]) in pbc */
			if (ps->s_txreq->phdr.hdr.hdr_type)
				pbc |= PBC_PACKET_BYPASS |
				       PBC_INSERT_BYPASS_ICRC;
			else
				pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);

			pbc = create_pbc(ppd,
					 pbc,
					 qp->srate_mbps,
					 vl,
					 plen);

			if (unlikely(hfi1_dbg_should_fault_tx(qp, ps->opcode)))
				pbc = hfi1_fault_tx(qp, ps->opcode, pbc);
			else
				/* Update HCRC based on packet opcode */
				pbc = update_hcrc(ps->opcode, pbc);
		}
		tx->wqe = qp->s_wqe;
		ret = build_verbs_tx_desc(tx->sde, len, tx, ahg_info, pbc);
		if (unlikely(ret))
			goto bail_build;
	}
	ret =  sdma_send_txreq(tx->sde, ps->wait, &tx->txreq, ps->pkts_sent);
	if (unlikely(ret < 0)) {
		if (ret == -ECOMM)
			goto bail_ecomm;
		return ret;
	}

	update_tx_opstats(qp, ps, plen);
	trace_sdma_output_ibhdr(dd_from_ibdev(qp->ibqp.device),
				&ps->s_txreq->phdr.hdr, ib_is_sc5(sc5));
	return ret;

bail_ecomm:
	/* The current one got "sent" */
	return 0;
bail_build:
	ret = wait_kmem(dev, qp, ps);
	if (!ret) {
		/* free txreq - bad state */
		hfi1_put_txreq(ps->s_txreq);
		ps->s_txreq = NULL;
	}
	return ret;
}

/*
 * If we are now in the error state, return zero to flush the
 * send work request.
 */
static int pio_wait(struct rvt_qp *qp,
		    struct send_context *sc,
		    struct hfi1_pkt_state *ps,
		    u32 flag)
{
	struct hfi1_qp_priv *priv = qp->priv;
	struct hfi1_devdata *dd = sc->dd;
	unsigned long flags;
	int ret = 0;

	/*
	 * Note that as soon as want_buffer() is called and
	 * possibly before it returns, sc_piobufavail()
	 * could be called. Therefore, put QP on the I/O wait list before
	 * enabling the PIO avail interrupt.
	 */
	spin_lock_irqsave(&qp->s_lock, flags);
	if (ib_rvt_state_ops[qp->state] & RVT_PROCESS_RECV_OK) {
		write_seqlock(&sc->waitlock);
		list_add_tail(&ps->s_txreq->txreq.list,
			      &ps->wait->tx_head);
		if (list_empty(&priv->s_iowait.list)) {
			struct hfi1_ibdev *dev = &dd->verbs_dev;
			int was_empty;

			dev->n_piowait += !!(flag & RVT_S_WAIT_PIO);
			dev->n_piodrain += !!(flag & HFI1_S_WAIT_PIO_DRAIN);
			qp->s_flags |= flag;
			was_empty = list_empty(&sc->piowait);
			iowait_get_priority(&priv->s_iowait);
			iowait_queue(ps->pkts_sent, &priv->s_iowait,
				     &sc->piowait);
			priv->s_iowait.lock = &sc->waitlock;
			trace_hfi1_qpsleep(qp, RVT_S_WAIT_PIO);
			rvt_get_qp(qp);
			/* counting: only call wantpiobuf_intr if first user */
			if (was_empty)
				hfi1_sc_wantpiobuf_intr(sc, 1);
		}
		write_sequnlock(&sc->waitlock);
		hfi1_qp_unbusy(qp, ps->wait);
		ret = -EBUSY;
	}
	spin_unlock_irqrestore(&qp->s_lock, flags);
	return ret;
}

static void verbs_pio_complete(void *arg, int code)
{
	struct rvt_qp *qp = (struct rvt_qp *)arg;
	struct hfi1_qp_priv *priv = qp->priv;

	if (iowait_pio_dec(&priv->s_iowait))
		iowait_drain_wakeup(&priv->s_iowait);
}

int hfi1_verbs_send_pio(struct rvt_qp *qp, struct hfi1_pkt_state *ps,
			u64 pbc)
{
	struct hfi1_qp_priv *priv = qp->priv;
	u32 hdrwords = ps->s_txreq->hdr_dwords;
	struct rvt_sge_state *ss = ps->s_txreq->ss;
	u32 len = ps->s_txreq->s_cur_size;
	u32 dwords;
	u32 plen;
	struct hfi1_pportdata *ppd = ps->ppd;
	u32 *hdr;
	u8 sc5;
	unsigned long flags = 0;
	struct send_context *sc;
	struct pio_buf *pbuf;
	int wc_status = IB_WC_SUCCESS;
	int ret = 0;
	pio_release_cb cb = NULL;
	u8 extra_bytes = 0;

	if (ps->s_txreq->phdr.hdr.hdr_type) {
		u8 pad_size = hfi1_get_16b_padding((hdrwords << 2), len);

		extra_bytes = pad_size + (SIZE_OF_CRC << 2) + SIZE_OF_LT;
		dwords = (len + extra_bytes) >> 2;
		hdr = (u32 *)&ps->s_txreq->phdr.hdr.opah;
	} else {
		dwords = (len + 3) >> 2;
		hdr = (u32 *)&ps->s_txreq->phdr.hdr.ibh;
	}
	plen = hdrwords + dwords + sizeof(pbc) / 4;

	/* only RC/UC use complete */
	switch (qp->ibqp.qp_type) {
	case IB_QPT_RC:
	case IB_QPT_UC:
		cb = verbs_pio_complete;
		break;
	default:
		break;
	}

	/* vl15 special case taken care of in ud.c */
	sc5 = priv->s_sc;
	sc = ps->s_txreq->psc;

	if (likely(pbc == 0)) {
		u8 vl = sc_to_vlt(dd_from_ibdev(qp->ibqp.device), sc5);

		/* set PBC_DC_INFO bit (aka SC[4]) in pbc */
		if (ps->s_txreq->phdr.hdr.hdr_type)
			pbc |= PBC_PACKET_BYPASS | PBC_INSERT_BYPASS_ICRC;
		else
			pbc |= (ib_is_sc5(sc5) << PBC_DC_INFO_SHIFT);

		pbc = create_pbc(ppd, pbc, qp->srate_mbps, vl, plen);
		if (unlikely(hfi1_dbg_should_fault_tx(qp, ps->opcode)))
			pbc = hfi1_fault_tx(qp, ps->opcode, pbc);
		else
			/* Update HCRC based on packet opcode */
			pbc = update_hcrc(ps->opcode, pbc);
	}
	if (cb)
		iowait_pio_inc(&priv->s_iowait);
	pbuf = sc_buffer_alloc(sc, plen, cb, qp);
	if (IS_ERR_OR_NULL(pbuf)) {
		if (cb)
			verbs_pio_complete(qp, 0);
		if (IS_ERR(pbuf)) {
			/*
			 * If we have filled the PIO buffers to capacity and are
			 * not in an active state this request is not going to
			 * go out to so just complete it with an error or else a
			 * ULP or the core may be stuck waiting.
			 */
			hfi1_cdbg(
				PIO,
				"alloc failed. state not active, completing");
			wc_status = IB_WC_GENERAL_ERR;
			goto pio_bail;
		} else {
			/*
			 * This is a normal occurrence. The PIO buffs are full
			 * up but we are still happily sending, well we could be
			 * so lets continue to queue the request.
			 */
			hfi1_cdbg(PIO, "alloc failed. state active, queuing");
			ret = pio_wait(qp, sc, ps, RVT_S_WAIT_PIO);
			if (!ret)
				/* txreq not queued - free */
				goto bail;
			/* tx consumed in wait */
			return ret;
		}
	}

	if (dwords == 0) {
		pio_copy(ppd->dd, pbuf, pbc, hdr, hdrwords);
	} else {
		seg_pio_copy_start(pbuf, pbc,
				   hdr, hdrwords * 4);
		if (ss) {
			while (len) {
				void *addr = ss->sge.vaddr;
				u32 slen = rvt_get_sge_length(&ss->sge, len);

				rvt_update_sge(ss, slen, false);
				seg_pio_copy_mid(pbuf, addr, slen);
				len -= slen;
			}
		}
		/* add icrc, lt byte, and padding to flit */
		if (extra_bytes)
			seg_pio_copy_mid(pbuf, ppd->dd->sdma_pad_dma,
					 extra_bytes);

		seg_pio_copy_end(pbuf);
	}

	update_tx_opstats(qp, ps, plen);
	trace_pio_output_ibhdr(dd_from_ibdev(qp->ibqp.device),
			       &ps->s_txreq->phdr.hdr, ib_is_sc5(sc5));

pio_bail:
	spin_lock_irqsave(&qp->s_lock, flags);
	if (qp->s_wqe) {
		rvt_send_complete(qp, qp->s_wqe, wc_status);
	} else if (qp->ibqp.qp_type == IB_QPT_RC) {
		if (unlikely(wc_status == IB_WC_GENERAL_ERR))
			hfi1_rc_verbs_aborted(qp, &ps->s_txreq->phdr.hdr);
		hfi1_rc_send_complete(qp, &ps->s_txreq->phdr.hdr);
	}
	spin_unlock_irqrestore(&qp->s_lock, flags);

	ret = 0;

bail:
	hfi1_put_txreq(ps->s_txreq);
	return ret;
}

/*
 * egress_pkey_matches_entry - return 1 if the pkey matches ent (ent
 * being an entry from the partition key table), return 0
 * otherwise. Use the matching criteria for egress partition keys
 * specified in the OPAv1 spec., section 9.1l.7.
 */
static inline int egress_pkey_matches_entry(u16 pkey, u16 ent)
{
	u16 mkey = pkey & PKEY_LOW_15_MASK;
	u16 mentry = ent & PKEY_LOW_15_MASK;

	if (mkey == mentry) {
		/*
		 * If pkey[15] is set (full partition member),
		 * is bit 15 in the corresponding table element
		 * clear (limited member)?
		 */
		if (pkey & PKEY_MEMBER_MASK)
			return !!(ent & PKEY_MEMBER_MASK);
		return 1;
	}
	return 0;
}

/**
 * egress_pkey_check - check P_KEY of a packet
 * @ppd:  Physical IB port data
 * @slid: SLID for packet
 * @pkey: PKEY for header
 * @sc5:  SC for packet
 * @s_pkey_index: It will be used for look up optimization for kernel contexts
 * only. If it is negative value, then it means user contexts is calling this
 * function.
 *
 * It checks if hdr's pkey is valid.
 *
 * Return: 0 on success, otherwise, 1
 */
int egress_pkey_check(struct hfi1_pportdata *ppd, u32 slid, u16 pkey,
		      u8 sc5, int8_t s_pkey_index)
{
	struct hfi1_devdata *dd;
	int i;
	int is_user_ctxt_mechanism = (s_pkey_index < 0);

	if (!(ppd->part_enforce & HFI1_PART_ENFORCE_OUT))
		return 0;

	/* If SC15, pkey[0:14] must be 0x7fff */
	if ((sc5 == 0xf) && ((pkey & PKEY_LOW_15_MASK) != PKEY_LOW_15_MASK))
		goto bad;

	/* Is the pkey = 0x0, or 0x8000? */
	if ((pkey & PKEY_LOW_15_MASK) == 0)
		goto bad;

	/*
	 * For the kernel contexts only, if a qp is passed into the function,
	 * the most likely matching pkey has index qp->s_pkey_index
	 */
	if (!is_user_ctxt_mechanism &&
	    egress_pkey_matches_entry(pkey, ppd->pkeys[s_pkey_index])) {
		return 0;
	}

	for (i = 0; i < MAX_PKEY_VALUES; i++) {
		if (egress_pkey_matches_entry(pkey, ppd->pkeys[i]))
			return 0;
	}
bad:
	/*
	 * For the user-context mechanism, the P_KEY check would only happen
	 * once per SDMA request, not once per packet.  Therefore, there's no
	 * need to increment the counter for the user-context mechanism.
	 */
	if (!is_user_ctxt_mechanism) {
		incr_cntr64(&ppd->port_xmit_constraint_errors);
		dd = ppd->dd;
		if (!(dd->err_info_xmit_constraint.status &
		      OPA_EI_STATUS_SMASK)) {
			dd->err_info_xmit_constraint.status |=
				OPA_EI_STATUS_SMASK;
			dd->err_info_xmit_constraint.slid = slid;
			dd->err_info_xmit_constraint.pkey = pkey;
		}
	}
	return 1;
}

/*
 * get_send_routine - choose an egress routine
 *
 * Choose an egress routine based on QP type
 * and size
 */
static inline send_routine get_send_routine(struct rvt_qp *qp,
					    struct hfi1_pkt_state *ps)
{
	struct hfi1_devdata *dd = dd_from_ibdev(qp->ibqp.device);
	struct hfi1_qp_priv *priv = qp->priv;
	struct verbs_txreq *tx = ps->s_txreq;

	if (unlikely(!(dd->flags & HFI1_HAS_SEND_DMA)))
		return dd->process_pio_send;
	switch (qp->ibqp.qp_type) {
	case IB_QPT_SMI:
		return dd->process_pio_send;
	case IB_QPT_GSI:
	case IB_QPT_UD:
		break;
	case IB_QPT_UC:
	case IB_QPT_RC:
		priv->s_running_pkt_size =
			(tx->s_cur_size + priv->s_running_pkt_size) / 2;
		if (piothreshold &&
		    priv->s_running_pkt_size <= min(piothreshold, qp->pmtu) &&
		    (BIT(ps->opcode & OPMASK) & pio_opmask[ps->opcode >> 5]) &&
		    iowait_sdma_pending(&priv->s_iowait) == 0 &&
		    !sdma_txreq_built(&tx->txreq))
			return dd->process_pio_send;
		break;
	default:
		break;
	}
	return dd->process_dma_send;
}

/**
 * hfi1_verbs_send - send a packet
 * @qp: the QP to send on
 * @ps: the state of the packet to send
 *
 * Return zero if packet is sent or queued OK.
 * Return non-zero and clear qp->s_flags RVT_S_BUSY otherwise.
 */
int hfi1_verbs_send(struct rvt_qp *qp, struct hfi1_pkt_state *ps)
{
	struct hfi1_devdata *dd = dd_from_ibdev(qp->ibqp.device);
	struct hfi1_qp_priv *priv = qp->priv;
	struct ib_other_headers *ohdr = NULL;
	send_routine sr;
	int ret;
	u16 pkey;
	u32 slid;
	u8 l4 = 0;

	/* locate the pkey within the headers */
	if (ps->s_txreq->phdr.hdr.hdr_type) {
		struct hfi1_16b_header *hdr = &ps->s_txreq->phdr.hdr.opah;

		l4 = hfi1_16B_get_l4(hdr);
		if (l4 == OPA_16B_L4_IB_LOCAL)
			ohdr = &hdr->u.oth;
		else if (l4 == OPA_16B_L4_IB_GLOBAL)
			ohdr = &hdr->u.l.oth;

		slid = hfi1_16B_get_slid(hdr);
		pkey = hfi1_16B_get_pkey(hdr);
	} else {
		struct ib_header *hdr = &ps->s_txreq->phdr.hdr.ibh;
		u8 lnh = ib_get_lnh(hdr);

		if (lnh == HFI1_LRH_GRH)
			ohdr = &hdr->u.l.oth;
		else
			ohdr = &hdr->u.oth;
		slid = ib_get_slid(hdr);
		pkey = ib_bth_get_pkey(ohdr);
	}

	if (likely(l4 != OPA_16B_L4_FM))
		ps->opcode = ib_bth_get_opcode(ohdr);
	else
		ps->opcode = IB_OPCODE_UD_SEND_ONLY;

	sr = get_send_routine(qp, ps);
	ret = egress_pkey_check(dd->pport, slid, pkey,
				priv->s_sc, qp->s_pkey_index);
	if (unlikely(ret)) {
		/*
		 * The value we are returning here does not get propagated to
		 * the verbs caller. Thus we need to complete the request with
		 * error otherwise the caller could be sitting waiting on the
		 * completion event. Only do this for PIO. SDMA has its own
		 * mechanism for handling the errors. So for SDMA we can just
		 * return.
		 */
		if (sr == dd->process_pio_send) {
			unsigned long flags;

			hfi1_cdbg(PIO, "%s() Failed. Completing with err",
				  __func__);
			spin_lock_irqsave(&qp->s_lock, flags);
			rvt_send_complete(qp, qp->s_wqe, IB_WC_GENERAL_ERR);
			spin_unlock_irqrestore(&qp->s_lock, flags);
		}
		return -EINVAL;
	}
	if (sr == dd->process_dma_send && iowait_pio_pending(&priv->s_iowait))
		return pio_wait(qp,
				ps->s_txreq->psc,
				ps,
				HFI1_S_WAIT_PIO_DRAIN);
	return sr(qp, ps, 0);
}

/**
 * hfi1_fill_device_attr - Fill in rvt dev info device attributes.
 * @dd: the device data structure
 */
static void hfi1_fill_device_attr(struct hfi1_devdata *dd)
{
	struct rvt_dev_info *rdi = &dd->verbs_dev.rdi;
	u32 ver = dd->dc8051_ver;

	memset(&rdi->dparms.props, 0, sizeof(rdi->dparms.props));

	rdi->dparms.props.fw_ver = ((u64)(dc8051_ver_maj(ver)) << 32) |
		((u64)(dc8051_ver_min(ver)) << 16) |
		(u64)dc8051_ver_patch(ver);

	rdi->dparms.props.device_cap_flags = IB_DEVICE_BAD_PKEY_CNTR |
			IB_DEVICE_BAD_QKEY_CNTR | IB_DEVICE_SHUTDOWN_PORT |
			IB_DEVICE_SYS_IMAGE_GUID | IB_DEVICE_RC_RNR_NAK_GEN |
			IB_DEVICE_PORT_ACTIVE_EVENT | IB_DEVICE_SRQ_RESIZE |
			IB_DEVICE_MEM_MGT_EXTENSIONS |
			IB_DEVICE_RDMA_NETDEV_OPA;
	rdi->dparms.props.page_size_cap = PAGE_SIZE;
	rdi->dparms.props.vendor_id = dd->oui1 << 16 | dd->oui2 << 8 | dd->oui3;
	rdi->dparms.props.vendor_part_id = dd->pcidev->device;
	rdi->dparms.props.hw_ver = dd->minrev;
	rdi->dparms.props.sys_image_guid = ib_hfi1_sys_image_guid;
	rdi->dparms.props.max_mr_size = U64_MAX;
	rdi->dparms.props.max_fast_reg_page_list_len = UINT_MAX;
	rdi->dparms.props.max_qp = hfi1_max_qps;
	rdi->dparms.props.max_qp_wr =
		(hfi1_max_qp_wrs >= HFI1_QP_WQE_INVALID ?
		 HFI1_QP_WQE_INVALID - 1 : hfi1_max_qp_wrs);
	rdi->dparms.props.max_send_sge = hfi1_max_sges;
	rdi->dparms.props.max_recv_sge = hfi1_max_sges;
	rdi->dparms.props.max_sge_rd = hfi1_max_sges;
	rdi->dparms.props.max_cq = hfi1_max_cqs;
	rdi->dparms.props.max_ah = hfi1_max_ahs;
	rdi->dparms.props.max_cqe = hfi1_max_cqes;
	rdi->dparms.props.max_pd = hfi1_max_pds;
	rdi->dparms.props.max_qp_rd_atom = HFI1_MAX_RDMA_ATOMIC;
	rdi->dparms.props.max_qp_init_rd_atom = 255;
	rdi->dparms.props.max_srq = hfi1_max_srqs;
	rdi->dparms.props.max_srq_wr = hfi1_max_srq_wrs;
	rdi->dparms.props.max_srq_sge = hfi1_max_srq_sges;
	rdi->dparms.props.atomic_cap = IB_ATOMIC_GLOB;
	rdi->dparms.props.max_pkeys = hfi1_get_npkeys(dd);
	rdi->dparms.props.max_mcast_grp = hfi1_max_mcast_grps;
	rdi->dparms.props.max_mcast_qp_attach = hfi1_max_mcast_qp_attached;
	rdi->dparms.props.max_total_mcast_qp_attach =
					rdi->dparms.props.max_mcast_qp_attach *
					rdi->dparms.props.max_mcast_grp;
}

static inline u16 opa_speed_to_ib(u16 in)
{
	u16 out = 0;

	if (in & OPA_LINK_SPEED_25G)
		out |= IB_SPEED_EDR;
	if (in & OPA_LINK_SPEED_12_5G)
		out |= IB_SPEED_FDR;

	return out;
}

/*
 * Convert a single OPA link width (no multiple flags) to an IB value.
 * A zero OPA link width means link down, which means the IB width value
 * is a don't care.
 */
static inline u16 opa_width_to_ib(u16 in)
{
	switch (in) {
	case OPA_LINK_WIDTH_1X:
	/* map 2x and 3x to 1x as they don't exist in IB */
	case OPA_LINK_WIDTH_2X:
	case OPA_LINK_WIDTH_3X:
		return IB_WIDTH_1X;
	default: /* link down or unknown, return our largest width */
	case OPA_LINK_WIDTH_4X:
		return IB_WIDTH_4X;
	}
}

static int query_port(struct rvt_dev_info *rdi, u32 port_num,
		      struct ib_port_attr *props)
{
	struct hfi1_ibdev *verbs_dev = dev_from_rdi(rdi);
	struct hfi1_devdata *dd = dd_from_dev(verbs_dev);
	struct hfi1_pportdata *ppd = &dd->pport[port_num - 1];
	u32 lid = ppd->lid;

	/* props being zeroed by the caller, avoid zeroing it here */
	props->lid = lid ? lid : 0;
	props->lmc = ppd->lmc;
	/* OPA logical states match IB logical states */
	props->state = driver_lstate(ppd);
	props->phys_state = driver_pstate(ppd);
	props->gid_tbl_len = HFI1_GUIDS_PER_PORT;
	props->active_width = (u8)opa_width_to_ib(ppd->link_width_active);
	/* see rate_show() in ib core/sysfs.c */
	props->active_speed = opa_speed_to_ib(ppd->link_speed_active);
	props->max_vl_num = ppd->vls_supported;

	/* Once we are a "first class" citizen and have added the OPA MTUs to
	 * the core we can advertise the larger MTU enum to the ULPs, for now
	 * advertise only 4K.
	 *
	 * Those applications which are either OPA aware or pass the MTU enum
	 * from the Path Records to us will get the new 8k MTU.  Those that
	 * attempt to process the MTU enum may fail in various ways.
	 */
	props->max_mtu = mtu_to_enum((!valid_ib_mtu(hfi1_max_mtu) ?
				      4096 : hfi1_max_mtu), IB_MTU_4096);
	props->active_mtu = !valid_ib_mtu(ppd->ibmtu) ? props->max_mtu :
		mtu_to_enum(ppd->ibmtu, IB_MTU_4096);
	props->phys_mtu = HFI1_CAP_IS_KSET(AIP) ? hfi1_max_mtu :
				ib_mtu_enum_to_int(props->max_mtu);

	return 0;
}

static int modify_device(struct ib_device *device,
			 int device_modify_mask,
			 struct ib_device_modify *device_modify)
{
	struct hfi1_devdata *dd = dd_from_ibdev(device);
	unsigned i;
	int ret;

	if (device_modify_mask & ~(IB_DEVICE_MODIFY_SYS_IMAGE_GUID |
				   IB_DEVICE_MODIFY_NODE_DESC)) {
		ret = -EOPNOTSUPP;
		goto bail;
	}

	if (device_modify_mask & IB_DEVICE_MODIFY_NODE_DESC) {
		memcpy(device->node_desc, device_modify->node_desc,
		       IB_DEVICE_NODE_DESC_MAX);
		for (i = 0; i < dd->num_pports; i++) {
			struct hfi1_ibport *ibp = &dd->pport[i].ibport_data;

			hfi1_node_desc_chg(ibp);
		}
	}

	if (device_modify_mask & IB_DEVICE_MODIFY_SYS_IMAGE_GUID) {
		ib_hfi1_sys_image_guid =
			cpu_to_be64(device_modify->sys_image_guid);
		for (i = 0; i < dd->num_pports; i++) {
			struct hfi1_ibport *ibp = &dd->pport[i].ibport_data;

			hfi1_sys_guid_chg(ibp);
		}
	}

	ret = 0;

bail:
	return ret;
}

static int shut_down_port(struct rvt_dev_info *rdi, u32 port_num)
{
	struct hfi1_ibdev *verbs_dev = dev_from_rdi(rdi);
	struct hfi1_devdata *dd = dd_from_dev(verbs_dev);
	struct hfi1_pportdata *ppd = &dd->pport[port_num - 1];
	int ret;

	set_link_down_reason(ppd, OPA_LINKDOWN_REASON_UNKNOWN, 0,
			     OPA_LINKDOWN_REASON_UNKNOWN);
	ret = set_link_state(ppd, HLS_DN_DOWNDEF);
	return ret;
}

static int hfi1_get_guid_be(struct rvt_dev_info *rdi, struct rvt_ibport *rvp,
			    int guid_index, __be64 *guid)
{
	struct hfi1_ibport *ibp = container_of(rvp, struct hfi1_ibport, rvp);

	if (guid_index >= HFI1_GUIDS_PER_PORT)
		return -EINVAL;

	*guid = get_sguid(ibp, guid_index);
	return 0;
}

/*
 * convert ah port,sl to sc
 */
u8 ah_to_sc(struct ib_device *ibdev, struct rdma_ah_attr *ah)
{
	struct hfi1_ibport *ibp = to_iport(ibdev, rdma_ah_get_port_num(ah));

	return ibp->sl_to_sc[rdma_ah_get_sl(ah)];
}

static int hfi1_check_ah(struct ib_device *ibdev, struct rdma_ah_attr *ah_attr)
{
	struct hfi1_ibport *ibp;
	struct hfi1_pportdata *ppd;
	struct hfi1_devdata *dd;
	u8 sc5;
	u8 sl;

	if (hfi1_check_mcast(rdma_ah_get_dlid(ah_attr)) &&
	    !(rdma_ah_get_ah_flags(ah_attr) & IB_AH_GRH))
		return -EINVAL;

	/* test the mapping for validity */
	ibp = to_iport(ibdev, rdma_ah_get_port_num(ah_attr));
	ppd = ppd_from_ibp(ibp);
	dd = dd_from_ppd(ppd);

	sl = rdma_ah_get_sl(ah_attr);
	if (sl >= ARRAY_SIZE(ibp->sl_to_sc))
		return -EINVAL;
	sl = array_index_nospec(sl, ARRAY_SIZE(ibp->sl_to_sc));

	sc5 = ibp->sl_to_sc[sl];
	if (sc_to_vlt(dd, sc5) > num_vls && sc_to_vlt(dd, sc5) != 0xf)
		return -EINVAL;
	return 0;
}

static void hfi1_notify_new_ah(struct ib_device *ibdev,
			       struct rdma_ah_attr *ah_attr,
			       struct rvt_ah *ah)
{
	struct hfi1_ibport *ibp;
	struct hfi1_pportdata *ppd;
	struct hfi1_devdata *dd;
	u8 sc5;
	struct rdma_ah_attr *attr = &ah->attr;

	/*
	 * Do not trust reading anything from rvt_ah at this point as it is not
	 * done being setup. We can however modify things which we need to set.
	 */

	ibp = to_iport(ibdev, rdma_ah_get_port_num(ah_attr));
	ppd = ppd_from_ibp(ibp);
	sc5 = ibp->sl_to_sc[rdma_ah_get_sl(&ah->attr)];
	hfi1_update_ah_attr(ibdev, attr);
	hfi1_make_opa_lid(attr);
	dd = dd_from_ppd(ppd);
	ah->vl = sc_to_vlt(dd, sc5);
	if (ah->vl < num_vls || ah->vl == 15)
		ah->log_pmtu = ilog2(dd->vld[ah->vl].mtu);
}

/**
 * hfi1_get_npkeys - return the size of the PKEY table for context 0
 * @dd: the hfi1_ib device
 */
unsigned hfi1_get_npkeys(struct hfi1_devdata *dd)
{
	return ARRAY_SIZE(dd->pport[0].pkeys);
}

static void init_ibport(struct hfi1_pportdata *ppd)
{
	struct hfi1_ibport *ibp = &ppd->ibport_data;
	size_t sz = ARRAY_SIZE(ibp->sl_to_sc);
	int i;

	for (i = 0; i < sz; i++) {
		ibp->sl_to_sc[i] = i;
		ibp->sc_to_sl[i] = i;
	}

	for (i = 0; i < RVT_MAX_TRAP_LISTS ; i++)
		INIT_LIST_HEAD(&ibp->rvp.trap_lists[i].list);
	timer_setup(&ibp->rvp.trap_timer, hfi1_handle_trap_timer, 0);

	spin_lock_init(&ibp->rvp.lock);
	/* Set the prefix to the default value (see ch. 4.1.1) */
	ibp->rvp.gid_prefix = IB_DEFAULT_GID_PREFIX;
	ibp->rvp.sm_lid = 0;
	/*
	 * Below should only set bits defined in OPA PortInfo.CapabilityMask
	 * and PortInfo.CapabilityMask3
	 */
	ibp->rvp.port_cap_flags = IB_PORT_AUTO_MIGR_SUP |
		IB_PORT_CAP_MASK_NOTICE_SUP;
	ibp->rvp.port_cap3_flags = OPA_CAP_MASK3_IsSharedSpaceSupported;
	ibp->rvp.pma_counter_select[0] = IB_PMA_PORT_XMIT_DATA;
	ibp->rvp.pma_counter_select[1] = IB_PMA_PORT_RCV_DATA;
	ibp->rvp.pma_counter_select[2] = IB_PMA_PORT_XMIT_PKTS;
	ibp->rvp.pma_counter_select[3] = IB_PMA_PORT_RCV_PKTS;
	ibp->rvp.pma_counter_select[4] = IB_PMA_PORT_XMIT_WAIT;

	RCU_INIT_POINTER(ibp->rvp.qp[0], NULL);
	RCU_INIT_POINTER(ibp->rvp.qp[1], NULL);
}

static void hfi1_get_dev_fw_str(struct ib_device *ibdev, char *str)
{
	struct rvt_dev_info *rdi = ib_to_rvt(ibdev);
	struct hfi1_ibdev *dev = dev_from_rdi(rdi);
	u32 ver = dd_from_dev(dev)->dc8051_ver;

	snprintf(str, IB_FW_VERSION_NAME_MAX, "%u.%u.%u", dc8051_ver_maj(ver),
		 dc8051_ver_min(ver), dc8051_ver_patch(ver));
}

static const char * const driver_cntr_names[] = {
	/* must be element 0*/
	"DRIVER_KernIntr",
	"DRIVER_ErrorIntr",
	"DRIVER_Tx_Errs",
	"DRIVER_Rcv_Errs",
	"DRIVER_HW_Errs",
	"DRIVER_NoPIOBufs",
	"DRIVER_CtxtsOpen",
	"DRIVER_RcvLen_Errs",
	"DRIVER_EgrBufFull",
	"DRIVER_EgrHdrFull"
};

static DEFINE_MUTEX(cntr_names_lock); /* protects the *_cntr_names bufers */
static const char **dev_cntr_names;
static const char **port_cntr_names;
int num_driver_cntrs = ARRAY_SIZE(driver_cntr_names);
static int num_dev_cntrs;
static int num_port_cntrs;
static int cntr_names_initialized;

/*
 * Convert a list of names separated by '\n' into an array of NULL terminated
 * strings. Optionally some entries can be reserved in the array to hold extra
 * external strings.
 */
static int init_cntr_names(const char *names_in,
			   const size_t names_len,
			   int num_extra_names,
			   int *num_cntrs,
			   const char ***cntr_names)
{
	char *names_out, *p, **q;
	int i, n;

	n = 0;
	for (i = 0; i < names_len; i++)
		if (names_in[i] == '\n')
			n++;

	names_out = kmalloc((n + num_extra_names) * sizeof(char *) + names_len,
			    GFP_KERNEL);
	if (!names_out) {
		*num_cntrs = 0;
		*cntr_names = NULL;
		return -ENOMEM;
	}

	p = names_out + (n + num_extra_names) * sizeof(char *);
	memcpy(p, names_in, names_len);

	q = (char **)names_out;
	for (i = 0; i < n; i++) {
		q[i] = p;
		p = strchr(p, '\n');
		*p++ = '\0';
	}

	*num_cntrs = n;
	*cntr_names = (const char **)names_out;
	return 0;
}

static int init_counters(struct ib_device *ibdev)
{
	struct hfi1_devdata *dd = dd_from_ibdev(ibdev);
	int i, err = 0;

	mutex_lock(&cntr_names_lock);
	if (cntr_names_initialized)
		goto out_unlock;

	err = init_cntr_names(dd->cntrnames, dd->cntrnameslen, num_driver_cntrs,
			      &num_dev_cntrs, &dev_cntr_names);
	if (err)
		goto out_unlock;

	for (i = 0; i < num_driver_cntrs; i++)
		dev_cntr_names[num_dev_cntrs + i] = driver_cntr_names[i];

	err = init_cntr_names(dd->portcntrnames, dd->portcntrnameslen, 0,
			      &num_port_cntrs, &port_cntr_names);
	if (err) {
		kfree(dev_cntr_names);
		dev_cntr_names = NULL;
		goto out_unlock;
	}
	cntr_names_initialized = 1;

out_unlock:
	mutex_unlock(&cntr_names_lock);
	return err;
}

static struct rdma_hw_stats *hfi1_alloc_hw_device_stats(struct ib_device *ibdev)
{
	if (init_counters(ibdev))
		return NULL;
	return rdma_alloc_hw_stats_struct(dev_cntr_names,
					  num_dev_cntrs + num_driver_cntrs,
					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
}

static struct rdma_hw_stats *hfi_alloc_hw_port_stats(struct ib_device *ibdev,
						     u32 port_num)
{
	if (init_counters(ibdev))
		return NULL;
	return rdma_alloc_hw_stats_struct(port_cntr_names, num_port_cntrs,
					  RDMA_HW_STATS_DEFAULT_LIFESPAN);
}

static u64 hfi1_sps_ints(void)
{
	unsigned long index, flags;
	struct hfi1_devdata *dd;
	u64 sps_ints = 0;

	xa_lock_irqsave(&hfi1_dev_table, flags);
	xa_for_each(&hfi1_dev_table, index, dd) {
		sps_ints += get_all_cpu_total(dd->int_counter);
	}
	xa_unlock_irqrestore(&hfi1_dev_table, flags);
	return sps_ints;
}

static int get_hw_stats(struct ib_device *ibdev, struct rdma_hw_stats *stats,
			u32 port, int index)
{
	u64 *values;
	int count;

	if (!port) {
		u64 *stats = (u64 *)&hfi1_stats;
		int i;

		hfi1_read_cntrs(dd_from_ibdev(ibdev), NULL, &values);
		values[num_dev_cntrs] = hfi1_sps_ints();
		for (i = 1; i < num_driver_cntrs; i++)
			values[num_dev_cntrs + i] = stats[i];
		count = num_dev_cntrs + num_driver_cntrs;
	} else {
		struct hfi1_ibport *ibp = to_iport(ibdev, port);

		hfi1_read_portcntrs(ppd_from_ibp(ibp), NULL, &values);
		count = num_port_cntrs;
	}

	memcpy(stats->value, values, count * sizeof(u64));
	return count;
}

static const struct ib_device_ops hfi1_dev_ops = {
	.owner = THIS_MODULE,
	.driver_id = RDMA_DRIVER_HFI1,

	.alloc_hw_device_stats = hfi1_alloc_hw_device_stats,
	.alloc_hw_port_stats = hfi_alloc_hw_port_stats,
	.alloc_rdma_netdev = hfi1_vnic_alloc_rn,
	.device_group = &ib_hfi1_attr_group,
	.get_dev_fw_str = hfi1_get_dev_fw_str,
	.get_hw_stats = get_hw_stats,
	.modify_device = modify_device,
	.port_groups = hfi1_attr_port_groups,
	/* keep process mad in the driver */
	.process_mad = hfi1_process_mad,
	.rdma_netdev_get_params = hfi1_ipoib_rn_get_params,
};

/**
 * hfi1_register_ib_device - register our device with the infiniband core
 * @dd: the device data structure
 * Return 0 if successful, errno if unsuccessful.
 */
int hfi1_register_ib_device(struct hfi1_devdata *dd)
{
	struct hfi1_ibdev *dev = &dd->verbs_dev;
	struct ib_device *ibdev = &dev->rdi.ibdev;
	struct hfi1_pportdata *ppd = dd->pport;
	struct hfi1_ibport *ibp = &ppd->ibport_data;
	unsigned i;
	int ret;

	for (i = 0; i < dd->num_pports; i++)
		init_ibport(ppd + i);

	/* Only need to initialize non-zero fields. */

	timer_setup(&dev->mem_timer, mem_timer, 0);

	seqlock_init(&dev->iowait_lock);
	seqlock_init(&dev->txwait_lock);
	INIT_LIST_HEAD(&dev->txwait);
	INIT_LIST_HEAD(&dev->memwait);

	ret = verbs_txreq_init(dev);
	if (ret)
		goto err_verbs_txreq;

	/* Use first-port GUID as node guid */
	ibdev->node_guid = get_sguid(ibp, HFI1_PORT_GUID_INDEX);

	/*
	 * The system image GUID is supposed to be the same for all
	 * HFIs in a single system but since there can be other
	 * device types in the system, we can't be sure this is unique.
	 */
	if (!ib_hfi1_sys_image_guid)
		ib_hfi1_sys_image_guid = ibdev->node_guid;
	ibdev->phys_port_cnt = dd->num_pports;
	ibdev->dev.parent = &dd->pcidev->dev;

	ib_set_device_ops(ibdev, &hfi1_dev_ops);

	strlcpy(ibdev->node_desc, init_utsname()->nodename,
		sizeof(ibdev->node_desc));

	/*
	 * Fill in rvt info object.
	 */
	dd->verbs_dev.rdi.driver_f.get_pci_dev = get_pci_dev;
	dd->verbs_dev.rdi.driver_f.check_ah = hfi1_check_ah;
	dd->verbs_dev.rdi.driver_f.notify_new_ah = hfi1_notify_new_ah;
	dd->verbs_dev.rdi.driver_f.get_guid_be = hfi1_get_guid_be;
	dd->verbs_dev.rdi.driver_f.query_port_state = query_port;
	dd->verbs_dev.rdi.driver_f.shut_down_port = shut_down_port;
	dd->verbs_dev.rdi.driver_f.cap_mask_chg = hfi1_cap_mask_chg;
	/*
	 * Fill in rvt info device attributes.
	 */
	hfi1_fill_device_attr(dd);

	/* queue pair */
	dd->verbs_dev.rdi.dparms.qp_table_size = hfi1_qp_table_size;
	dd->verbs_dev.rdi.dparms.qpn_start = 0;
	dd->verbs_dev.rdi.dparms.qpn_inc = 1;
	dd->verbs_dev.rdi.dparms.qos_shift = dd->qos_shift;
	dd->verbs_dev.rdi.dparms.qpn_res_start = RVT_KDETH_QP_BASE;
	dd->verbs_dev.rdi.dparms.qpn_res_end = RVT_AIP_QP_MAX;
	dd->verbs_dev.rdi.dparms.max_rdma_atomic = HFI1_MAX_RDMA_ATOMIC;
	dd->verbs_dev.rdi.dparms.psn_mask = PSN_MASK;
	dd->verbs_dev.rdi.dparms.psn_shift = PSN_SHIFT;
	dd->verbs_dev.rdi.dparms.psn_modify_mask = PSN_MODIFY_MASK;
	dd->verbs_dev.rdi.dparms.core_cap_flags = RDMA_CORE_PORT_INTEL_OPA |
						RDMA_CORE_CAP_OPA_AH;
	dd->verbs_dev.rdi.dparms.max_mad_size = OPA_MGMT_MAD_SIZE;

	dd->verbs_dev.rdi.driver_f.qp_priv_alloc = qp_priv_alloc;
	dd->verbs_dev.rdi.driver_f.qp_priv_init = hfi1_qp_priv_init;
	dd->verbs_dev.rdi.driver_f.qp_priv_free = qp_priv_free;
	dd->verbs_dev.rdi.driver_f.free_all_qps = free_all_qps;
	dd->verbs_dev.rdi.driver_f.notify_qp_reset = notify_qp_reset;
	dd->verbs_dev.rdi.driver_f.do_send = hfi1_do_send_from_rvt;
	dd->verbs_dev.rdi.driver_f.schedule_send = hfi1_schedule_send;
	dd->verbs_dev.rdi.driver_f.schedule_send_no_lock = _hfi1_schedule_send;
	dd->verbs_dev.rdi.driver_f.get_pmtu_from_attr = get_pmtu_from_attr;
	dd->verbs_dev.rdi.driver_f.notify_error_qp = notify_error_qp;
	dd->verbs_dev.rdi.driver_f.flush_qp_waiters = flush_qp_waiters;
	dd->verbs_dev.rdi.driver_f.stop_send_queue = stop_send_queue;
	dd->verbs_dev.rdi.driver_f.quiesce_qp = quiesce_qp;
	dd->verbs_dev.rdi.driver_f.notify_error_qp = notify_error_qp;
	dd->verbs_dev.rdi.driver_f.mtu_from_qp = mtu_from_qp;
	dd->verbs_dev.rdi.driver_f.mtu_to_path_mtu = mtu_to_path_mtu;
	dd->verbs_dev.rdi.driver_f.check_modify_qp = hfi1_check_modify_qp;
	dd->verbs_dev.rdi.driver_f.modify_qp = hfi1_modify_qp;
	dd->verbs_dev.rdi.driver_f.notify_restart_rc = hfi1_restart_rc;
	dd->verbs_dev.rdi.driver_f.setup_wqe = hfi1_setup_wqe;
	dd->verbs_dev.rdi.driver_f.comp_vect_cpu_lookup =
						hfi1_comp_vect_mappings_lookup;

	/* completeion queue */
	dd->verbs_dev.rdi.ibdev.num_comp_vectors = dd->comp_vect_possible_cpus;
	dd->verbs_dev.rdi.dparms.node = dd->node;

	/* misc settings */
	dd->verbs_dev.rdi.flags = 0; /* Let rdmavt handle it all */
	dd->verbs_dev.rdi.dparms.lkey_table_size = hfi1_lkey_table_size;
	dd->verbs_dev.rdi.dparms.nports = dd->num_pports;
	dd->verbs_dev.rdi.dparms.npkeys = hfi1_get_npkeys(dd);
	dd->verbs_dev.rdi.dparms.sge_copy_mode = sge_copy_mode;
	dd->verbs_dev.rdi.dparms.wss_threshold = wss_threshold;
	dd->verbs_dev.rdi.dparms.wss_clean_period = wss_clean_period;
	dd->verbs_dev.rdi.dparms.reserved_operations = 1;
	dd->verbs_dev.rdi.dparms.extra_rdma_atomic = HFI1_TID_RDMA_WRITE_CNT;

	/* post send table */
	dd->verbs_dev.rdi.post_parms = hfi1_post_parms;

	/* opcode translation table */
	dd->verbs_dev.rdi.wc_opcode = ib_hfi1_wc_opcode;

	ppd = dd->pport;
	for (i = 0; i < dd->num_pports; i++, ppd++)
		rvt_init_port(&dd->verbs_dev.rdi,
			      &ppd->ibport_data.rvp,
			      i,
			      ppd->pkeys);

	ret = rvt_register_device(&dd->verbs_dev.rdi);
	if (ret)
		goto err_verbs_txreq;

	ret = hfi1_verbs_register_sysfs(dd);
	if (ret)
		goto err_class;

	return ret;

err_class:
	rvt_unregister_device(&dd->verbs_dev.rdi);
err_verbs_txreq:
	verbs_txreq_exit(dev);
	dd_dev_err(dd, "cannot register verbs: %d!\n", -ret);
	return ret;
}

void hfi1_unregister_ib_device(struct hfi1_devdata *dd)
{
	struct hfi1_ibdev *dev = &dd->verbs_dev;

	hfi1_verbs_unregister_sysfs(dd);

	rvt_unregister_device(&dd->verbs_dev.rdi);

	if (!list_empty(&dev->txwait))
		dd_dev_err(dd, "txwait list not empty!\n");
	if (!list_empty(&dev->memwait))
		dd_dev_err(dd, "memwait list not empty!\n");

	del_timer_sync(&dev->mem_timer);
	verbs_txreq_exit(dev);

	mutex_lock(&cntr_names_lock);
	kfree(dev_cntr_names);
	kfree(port_cntr_names);
	dev_cntr_names = NULL;
	port_cntr_names = NULL;
	cntr_names_initialized = 0;
	mutex_unlock(&cntr_names_lock);
}

void hfi1_cnp_rcv(struct hfi1_packet *packet)
{
	struct hfi1_ibport *ibp = rcd_to_iport(packet->rcd);
	struct hfi1_pportdata *ppd = ppd_from_ibp(ibp);
	struct ib_header *hdr = packet->hdr;
	struct rvt_qp *qp = packet->qp;
	u32 lqpn, rqpn = 0;
	u16 rlid = 0;
	u8 sl, sc5, svc_type;

	switch (packet->qp->ibqp.qp_type) {
	case IB_QPT_UC:
		rlid = rdma_ah_get_dlid(&qp->remote_ah_attr);
		rqpn = qp->remote_qpn;
		svc_type = IB_CC_SVCTYPE_UC;
		break;
	case IB_QPT_RC:
		rlid = rdma_ah_get_dlid(&qp->remote_ah_attr);
		rqpn = qp->remote_qpn;
		svc_type = IB_CC_SVCTYPE_RC;
		break;
	case IB_QPT_SMI:
	case IB_QPT_GSI:
	case IB_QPT_UD:
		svc_type = IB_CC_SVCTYPE_UD;
		break;
	default:
		ibp->rvp.n_pkt_drops++;
		return;
	}

	sc5 = hfi1_9B_get_sc5(hdr, packet->rhf);
	sl = ibp->sc_to_sl[sc5];
	lqpn = qp->ibqp.qp_num;

	process_becn(ppd, sl, rlid, lqpn, rqpn, svc_type);
}