summaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw/hfi1/file_ops.c
blob: 7750a9c38b066777982c8309d8436b5780235d36 (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
/*
 * Copyright(c) 2015-2017 Intel Corporation.
 *
 * This file is provided under a dual BSD/GPLv2 license.  When using or
 * redistributing this file, you may do so under either license.
 *
 * GPL LICENSE SUMMARY
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but
 * WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * BSD LICENSE
 *
 * 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.
 *  - Neither the name of Intel Corporation nor the names of its
 *    contributors may be used to endorse or promote products derived
 *    from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 */
#include <linux/poll.h>
#include <linux/cdev.h>
#include <linux/vmalloc.h>
#include <linux/io.h>
#include <linux/sched/mm.h>
#include <linux/bitmap.h>

#include <rdma/ib.h>

#include "hfi.h"
#include "pio.h"
#include "device.h"
#include "common.h"
#include "trace.h"
#include "mmu_rb.h"
#include "user_sdma.h"
#include "user_exp_rcv.h"
#include "aspm.h"

#undef pr_fmt
#define pr_fmt(fmt) DRIVER_NAME ": " fmt

#define SEND_CTXT_HALT_TIMEOUT 1000 /* msecs */

/*
 * File operation functions
 */
static int hfi1_file_open(struct inode *inode, struct file *fp);
static int hfi1_file_close(struct inode *inode, struct file *fp);
static ssize_t hfi1_write_iter(struct kiocb *kiocb, struct iov_iter *from);
static unsigned int hfi1_poll(struct file *fp, struct poll_table_struct *pt);
static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma);

static u64 kvirt_to_phys(void *addr);
static int assign_ctxt(struct hfi1_filedata *fd, unsigned long arg, u32 len);
static void init_subctxts(struct hfi1_ctxtdata *uctxt,
			  const struct hfi1_user_info *uinfo);
static int init_user_ctxt(struct hfi1_filedata *fd,
			  struct hfi1_ctxtdata *uctxt);
static void user_init(struct hfi1_ctxtdata *uctxt);
static int get_ctxt_info(struct hfi1_filedata *fd, unsigned long arg, u32 len);
static int get_base_info(struct hfi1_filedata *fd, unsigned long arg, u32 len);
static int user_exp_rcv_setup(struct hfi1_filedata *fd, unsigned long arg,
			      u32 len);
static int user_exp_rcv_clear(struct hfi1_filedata *fd, unsigned long arg,
			      u32 len);
static int user_exp_rcv_invalid(struct hfi1_filedata *fd, unsigned long arg,
				u32 len);
static int setup_base_ctxt(struct hfi1_filedata *fd,
			   struct hfi1_ctxtdata *uctxt);
static int setup_subctxt(struct hfi1_ctxtdata *uctxt);

static int find_sub_ctxt(struct hfi1_filedata *fd,
			 const struct hfi1_user_info *uinfo);
static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
			 struct hfi1_user_info *uinfo,
			 struct hfi1_ctxtdata **cd);
static void deallocate_ctxt(struct hfi1_ctxtdata *uctxt);
static unsigned int poll_urgent(struct file *fp, struct poll_table_struct *pt);
static unsigned int poll_next(struct file *fp, struct poll_table_struct *pt);
static int user_event_ack(struct hfi1_ctxtdata *uctxt, u16 subctxt,
			  unsigned long arg);
static int set_ctxt_pkey(struct hfi1_ctxtdata *uctxt, unsigned long arg);
static int ctxt_reset(struct hfi1_ctxtdata *uctxt);
static int manage_rcvq(struct hfi1_ctxtdata *uctxt, u16 subctxt,
		       unsigned long arg);
static int vma_fault(struct vm_fault *vmf);
static long hfi1_file_ioctl(struct file *fp, unsigned int cmd,
			    unsigned long arg);

static const struct file_operations hfi1_file_ops = {
	.owner = THIS_MODULE,
	.write_iter = hfi1_write_iter,
	.open = hfi1_file_open,
	.release = hfi1_file_close,
	.unlocked_ioctl = hfi1_file_ioctl,
	.poll = hfi1_poll,
	.mmap = hfi1_file_mmap,
	.llseek = noop_llseek,
};

static const struct vm_operations_struct vm_ops = {
	.fault = vma_fault,
};

/*
 * Types of memories mapped into user processes' space
 */
enum mmap_types {
	PIO_BUFS = 1,
	PIO_BUFS_SOP,
	PIO_CRED,
	RCV_HDRQ,
	RCV_EGRBUF,
	UREGS,
	EVENTS,
	STATUS,
	RTAIL,
	SUBCTXT_UREGS,
	SUBCTXT_RCV_HDRQ,
	SUBCTXT_EGRBUF,
	SDMA_COMP
};

/*
 * Masks and offsets defining the mmap tokens
 */
#define HFI1_MMAP_OFFSET_MASK   0xfffULL
#define HFI1_MMAP_OFFSET_SHIFT  0
#define HFI1_MMAP_SUBCTXT_MASK  0xfULL
#define HFI1_MMAP_SUBCTXT_SHIFT 12
#define HFI1_MMAP_CTXT_MASK     0xffULL
#define HFI1_MMAP_CTXT_SHIFT    16
#define HFI1_MMAP_TYPE_MASK     0xfULL
#define HFI1_MMAP_TYPE_SHIFT    24
#define HFI1_MMAP_MAGIC_MASK    0xffffffffULL
#define HFI1_MMAP_MAGIC_SHIFT   32

#define HFI1_MMAP_MAGIC         0xdabbad00

#define HFI1_MMAP_TOKEN_SET(field, val)	\
	(((val) & HFI1_MMAP_##field##_MASK) << HFI1_MMAP_##field##_SHIFT)
#define HFI1_MMAP_TOKEN_GET(field, token) \
	(((token) >> HFI1_MMAP_##field##_SHIFT) & HFI1_MMAP_##field##_MASK)
#define HFI1_MMAP_TOKEN(type, ctxt, subctxt, addr)   \
	(HFI1_MMAP_TOKEN_SET(MAGIC, HFI1_MMAP_MAGIC) | \
	HFI1_MMAP_TOKEN_SET(TYPE, type) | \
	HFI1_MMAP_TOKEN_SET(CTXT, ctxt) | \
	HFI1_MMAP_TOKEN_SET(SUBCTXT, subctxt) | \
	HFI1_MMAP_TOKEN_SET(OFFSET, (offset_in_page(addr))))

#define dbg(fmt, ...)				\
	pr_info(fmt, ##__VA_ARGS__)

static inline int is_valid_mmap(u64 token)
{
	return (HFI1_MMAP_TOKEN_GET(MAGIC, token) == HFI1_MMAP_MAGIC);
}

static int hfi1_file_open(struct inode *inode, struct file *fp)
{
	struct hfi1_filedata *fd;
	struct hfi1_devdata *dd = container_of(inode->i_cdev,
					       struct hfi1_devdata,
					       user_cdev);

	if (!((dd->flags & HFI1_PRESENT) && dd->kregbase1))
		return -EINVAL;

	if (!atomic_inc_not_zero(&dd->user_refcount))
		return -ENXIO;

	/* Just take a ref now. Not all opens result in a context assign */
	kobject_get(&dd->kobj);

	/* The real work is performed later in assign_ctxt() */

	fd = kzalloc(sizeof(*fd), GFP_KERNEL);

	if (fd) {
		fd->rec_cpu_num = -1; /* no cpu affinity by default */
		fd->mm = current->mm;
		mmgrab(fd->mm);
		fd->dd = dd;
		fp->private_data = fd;
	} else {
		fp->private_data = NULL;

		if (atomic_dec_and_test(&dd->user_refcount))
			complete(&dd->user_comp);

		return -ENOMEM;
	}

	return 0;
}

static long hfi1_file_ioctl(struct file *fp, unsigned int cmd,
			    unsigned long arg)
{
	struct hfi1_filedata *fd = fp->private_data;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;
	int ret = 0;
	int uval = 0;

	hfi1_cdbg(IOCTL, "IOCTL recv: 0x%x", cmd);
	if (cmd != HFI1_IOCTL_ASSIGN_CTXT &&
	    cmd != HFI1_IOCTL_GET_VERS &&
	    !uctxt)
		return -EINVAL;

	switch (cmd) {
	case HFI1_IOCTL_ASSIGN_CTXT:
		ret = assign_ctxt(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_CTXT_INFO:
		ret = get_ctxt_info(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_USER_INFO:
		ret = get_base_info(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_CREDIT_UPD:
		if (uctxt)
			sc_return_credits(uctxt->sc);
		break;

	case HFI1_IOCTL_TID_UPDATE:
		ret = user_exp_rcv_setup(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_TID_FREE:
		ret = user_exp_rcv_clear(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_TID_INVAL_READ:
		ret = user_exp_rcv_invalid(fd, arg, _IOC_SIZE(cmd));
		break;

	case HFI1_IOCTL_RECV_CTRL:
		ret = manage_rcvq(uctxt, fd->subctxt, arg);
		break;

	case HFI1_IOCTL_POLL_TYPE:
		if (get_user(uval, (int __user *)arg))
			return -EFAULT;
		uctxt->poll_type = (typeof(uctxt->poll_type))uval;
		break;

	case HFI1_IOCTL_ACK_EVENT:
		ret = user_event_ack(uctxt, fd->subctxt, arg);
		break;

	case HFI1_IOCTL_SET_PKEY:
		ret = set_ctxt_pkey(uctxt, arg);
		break;

	case HFI1_IOCTL_CTXT_RESET:
		ret = ctxt_reset(uctxt);
		break;

	case HFI1_IOCTL_GET_VERS:
		uval = HFI1_USER_SWVERSION;
		if (put_user(uval, (int __user *)arg))
			return -EFAULT;
		break;

	default:
		return -EINVAL;
	}

	return ret;
}

static ssize_t hfi1_write_iter(struct kiocb *kiocb, struct iov_iter *from)
{
	struct hfi1_filedata *fd = kiocb->ki_filp->private_data;
	struct hfi1_user_sdma_pkt_q *pq = fd->pq;
	struct hfi1_user_sdma_comp_q *cq = fd->cq;
	int done = 0, reqs = 0;
	unsigned long dim = from->nr_segs;

	if (!cq || !pq)
		return -EIO;

	if (!iter_is_iovec(from) || !dim)
		return -EINVAL;

	trace_hfi1_sdma_request(fd->dd, fd->uctxt->ctxt, fd->subctxt, dim);

	if (atomic_read(&pq->n_reqs) == pq->n_max_reqs)
		return -ENOSPC;

	while (dim) {
		int ret;
		unsigned long count = 0;

		ret = hfi1_user_sdma_process_request(
			fd, (struct iovec *)(from->iov + done),
			dim, &count);
		if (ret) {
			reqs = ret;
			break;
		}
		dim -= count;
		done += count;
		reqs++;
	}

	return reqs;
}

static int hfi1_file_mmap(struct file *fp, struct vm_area_struct *vma)
{
	struct hfi1_filedata *fd = fp->private_data;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;
	struct hfi1_devdata *dd;
	unsigned long flags;
	u64 token = vma->vm_pgoff << PAGE_SHIFT,
		memaddr = 0;
	void *memvirt = NULL;
	u8 subctxt, mapio = 0, vmf = 0, type;
	ssize_t memlen = 0;
	int ret = 0;
	u16 ctxt;

	if (!is_valid_mmap(token) || !uctxt ||
	    !(vma->vm_flags & VM_SHARED)) {
		ret = -EINVAL;
		goto done;
	}
	dd = uctxt->dd;
	ctxt = HFI1_MMAP_TOKEN_GET(CTXT, token);
	subctxt = HFI1_MMAP_TOKEN_GET(SUBCTXT, token);
	type = HFI1_MMAP_TOKEN_GET(TYPE, token);
	if (ctxt != uctxt->ctxt || subctxt != fd->subctxt) {
		ret = -EINVAL;
		goto done;
	}

	flags = vma->vm_flags;

	switch (type) {
	case PIO_BUFS:
	case PIO_BUFS_SOP:
		memaddr = ((dd->physaddr + TXE_PIO_SEND) +
				/* chip pio base */
			   (uctxt->sc->hw_context * BIT(16))) +
				/* 64K PIO space / ctxt */
			(type == PIO_BUFS_SOP ?
				(TXE_PIO_SIZE / 2) : 0); /* sop? */
		/*
		 * Map only the amount allocated to the context, not the
		 * entire available context's PIO space.
		 */
		memlen = PAGE_ALIGN(uctxt->sc->credits * PIO_BLOCK_SIZE);
		flags &= ~VM_MAYREAD;
		flags |= VM_DONTCOPY | VM_DONTEXPAND;
		vma->vm_page_prot = pgprot_writecombine(vma->vm_page_prot);
		mapio = 1;
		break;
	case PIO_CRED:
		if (flags & VM_WRITE) {
			ret = -EPERM;
			goto done;
		}
		/*
		 * The credit return location for this context could be on the
		 * second or third page allocated for credit returns (if number
		 * of enabled contexts > 64 and 128 respectively).
		 */
		memvirt = dd->cr_base[uctxt->numa_id].va;
		memaddr = virt_to_phys(memvirt) +
			(((u64)uctxt->sc->hw_free -
			  (u64)dd->cr_base[uctxt->numa_id].va) & PAGE_MASK);
		memlen = PAGE_SIZE;
		flags &= ~VM_MAYWRITE;
		flags |= VM_DONTCOPY | VM_DONTEXPAND;
		/*
		 * The driver has already allocated memory for credit
		 * returns and programmed it into the chip. Has that
		 * memory been flagged as non-cached?
		 */
		/* vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); */
		mapio = 1;
		break;
	case RCV_HDRQ:
		memlen = uctxt->rcvhdrq_size;
		memvirt = uctxt->rcvhdrq;
		break;
	case RCV_EGRBUF: {
		unsigned long addr;
		int i;
		/*
		 * The RcvEgr buffer need to be handled differently
		 * as multiple non-contiguous pages need to be mapped
		 * into the user process.
		 */
		memlen = uctxt->egrbufs.size;
		if ((vma->vm_end - vma->vm_start) != memlen) {
			dd_dev_err(dd, "Eager buffer map size invalid (%lu != %lu)\n",
				   (vma->vm_end - vma->vm_start), memlen);
			ret = -EINVAL;
			goto done;
		}
		if (vma->vm_flags & VM_WRITE) {
			ret = -EPERM;
			goto done;
		}
		vma->vm_flags &= ~VM_MAYWRITE;
		addr = vma->vm_start;
		for (i = 0 ; i < uctxt->egrbufs.numbufs; i++) {
			memlen = uctxt->egrbufs.buffers[i].len;
			memvirt = uctxt->egrbufs.buffers[i].addr;
			ret = remap_pfn_range(
				vma, addr,
				/*
				 * virt_to_pfn() does the same, but
				 * it's not available on x86_64
				 * when CONFIG_MMU is enabled.
				 */
				PFN_DOWN(__pa(memvirt)),
				memlen,
				vma->vm_page_prot);
			if (ret < 0)
				goto done;
			addr += memlen;
		}
		ret = 0;
		goto done;
	}
	case UREGS:
		/*
		 * Map only the page that contains this context's user
		 * registers.
		 */
		memaddr = (unsigned long)
			(dd->physaddr + RXE_PER_CONTEXT_USER)
			+ (uctxt->ctxt * RXE_PER_CONTEXT_SIZE);
		/*
		 * TidFlow table is on the same page as the rest of the
		 * user registers.
		 */
		memlen = PAGE_SIZE;
		flags |= VM_DONTCOPY | VM_DONTEXPAND;
		vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
		mapio = 1;
		break;
	case EVENTS:
		/*
		 * Use the page where this context's flags are. User level
		 * knows where it's own bitmap is within the page.
		 */
		memaddr = (unsigned long)
			(dd->events + uctxt_offset(uctxt)) & PAGE_MASK;
		memlen = PAGE_SIZE;
		/*
		 * v3.7 removes VM_RESERVED but the effect is kept by
		 * using VM_IO.
		 */
		flags |= VM_IO | VM_DONTEXPAND;
		vmf = 1;
		break;
	case STATUS:
		if (flags & (unsigned long)(VM_WRITE | VM_EXEC)) {
			ret = -EPERM;
			goto done;
		}
		memaddr = kvirt_to_phys((void *)dd->status);
		memlen = PAGE_SIZE;
		flags |= VM_IO | VM_DONTEXPAND;
		break;
	case RTAIL:
		if (!HFI1_CAP_IS_USET(DMA_RTAIL)) {
			/*
			 * If the memory allocation failed, the context alloc
			 * also would have failed, so we would never get here
			 */
			ret = -EINVAL;
			goto done;
		}
		if (flags & VM_WRITE) {
			ret = -EPERM;
			goto done;
		}
		memlen = PAGE_SIZE;
		memvirt = (void *)uctxt->rcvhdrtail_kvaddr;
		flags &= ~VM_MAYWRITE;
		break;
	case SUBCTXT_UREGS:
		memaddr = (u64)uctxt->subctxt_uregbase;
		memlen = PAGE_SIZE;
		flags |= VM_IO | VM_DONTEXPAND;
		vmf = 1;
		break;
	case SUBCTXT_RCV_HDRQ:
		memaddr = (u64)uctxt->subctxt_rcvhdr_base;
		memlen = uctxt->rcvhdrq_size * uctxt->subctxt_cnt;
		flags |= VM_IO | VM_DONTEXPAND;
		vmf = 1;
		break;
	case SUBCTXT_EGRBUF:
		memaddr = (u64)uctxt->subctxt_rcvegrbuf;
		memlen = uctxt->egrbufs.size * uctxt->subctxt_cnt;
		flags |= VM_IO | VM_DONTEXPAND;
		flags &= ~VM_MAYWRITE;
		vmf = 1;
		break;
	case SDMA_COMP: {
		struct hfi1_user_sdma_comp_q *cq = fd->cq;

		if (!cq) {
			ret = -EFAULT;
			goto done;
		}
		memaddr = (u64)cq->comps;
		memlen = PAGE_ALIGN(sizeof(*cq->comps) * cq->nentries);
		flags |= VM_IO | VM_DONTEXPAND;
		vmf = 1;
		break;
	}
	default:
		ret = -EINVAL;
		break;
	}

	if ((vma->vm_end - vma->vm_start) != memlen) {
		hfi1_cdbg(PROC, "%u:%u Memory size mismatch %lu:%lu",
			  uctxt->ctxt, fd->subctxt,
			  (vma->vm_end - vma->vm_start), memlen);
		ret = -EINVAL;
		goto done;
	}

	vma->vm_flags = flags;
	hfi1_cdbg(PROC,
		  "%u:%u type:%u io/vf:%d/%d, addr:0x%llx, len:%lu(%lu), flags:0x%lx\n",
		    ctxt, subctxt, type, mapio, vmf, memaddr, memlen,
		    vma->vm_end - vma->vm_start, vma->vm_flags);
	if (vmf) {
		vma->vm_pgoff = PFN_DOWN(memaddr);
		vma->vm_ops = &vm_ops;
		ret = 0;
	} else if (mapio) {
		ret = io_remap_pfn_range(vma, vma->vm_start,
					 PFN_DOWN(memaddr),
					 memlen,
					 vma->vm_page_prot);
	} else if (memvirt) {
		ret = remap_pfn_range(vma, vma->vm_start,
				      PFN_DOWN(__pa(memvirt)),
				      memlen,
				      vma->vm_page_prot);
	} else {
		ret = remap_pfn_range(vma, vma->vm_start,
				      PFN_DOWN(memaddr),
				      memlen,
				      vma->vm_page_prot);
	}
done:
	return ret;
}

/*
 * Local (non-chip) user memory is not mapped right away but as it is
 * accessed by the user-level code.
 */
static int vma_fault(struct vm_fault *vmf)
{
	struct page *page;

	page = vmalloc_to_page((void *)(vmf->pgoff << PAGE_SHIFT));
	if (!page)
		return VM_FAULT_SIGBUS;

	get_page(page);
	vmf->page = page;

	return 0;
}

static unsigned int hfi1_poll(struct file *fp, struct poll_table_struct *pt)
{
	struct hfi1_ctxtdata *uctxt;
	unsigned pollflag;

	uctxt = ((struct hfi1_filedata *)fp->private_data)->uctxt;
	if (!uctxt)
		pollflag = POLLERR;
	else if (uctxt->poll_type == HFI1_POLL_TYPE_URGENT)
		pollflag = poll_urgent(fp, pt);
	else  if (uctxt->poll_type == HFI1_POLL_TYPE_ANYRCV)
		pollflag = poll_next(fp, pt);
	else /* invalid */
		pollflag = POLLERR;

	return pollflag;
}

static int hfi1_file_close(struct inode *inode, struct file *fp)
{
	struct hfi1_filedata *fdata = fp->private_data;
	struct hfi1_ctxtdata *uctxt = fdata->uctxt;
	struct hfi1_devdata *dd = container_of(inode->i_cdev,
					       struct hfi1_devdata,
					       user_cdev);
	unsigned long flags, *ev;

	fp->private_data = NULL;

	if (!uctxt)
		goto done;

	hfi1_cdbg(PROC, "closing ctxt %u:%u", uctxt->ctxt, fdata->subctxt);

	flush_wc();
	/* drain user sdma queue */
	hfi1_user_sdma_free_queues(fdata, uctxt);

	/* release the cpu */
	hfi1_put_proc_affinity(fdata->rec_cpu_num);

	/* clean up rcv side */
	hfi1_user_exp_rcv_free(fdata);

	/*
	 * fdata->uctxt is used in the above cleanup.  It is not ready to be
	 * removed until here.
	 */
	fdata->uctxt = NULL;
	hfi1_rcd_put(uctxt);

	/*
	 * Clear any left over, unhandled events so the next process that
	 * gets this context doesn't get confused.
	 */
	ev = dd->events + uctxt_offset(uctxt) + fdata->subctxt;
	*ev = 0;

	spin_lock_irqsave(&dd->uctxt_lock, flags);
	__clear_bit(fdata->subctxt, uctxt->in_use_ctxts);
	if (!bitmap_empty(uctxt->in_use_ctxts, HFI1_MAX_SHARED_CTXTS)) {
		spin_unlock_irqrestore(&dd->uctxt_lock, flags);
		goto done;
	}
	spin_unlock_irqrestore(&dd->uctxt_lock, flags);

	/*
	 * Disable receive context and interrupt available, reset all
	 * RcvCtxtCtrl bits to default values.
	 */
	hfi1_rcvctrl(dd, HFI1_RCVCTRL_CTXT_DIS |
		     HFI1_RCVCTRL_TIDFLOW_DIS |
		     HFI1_RCVCTRL_INTRAVAIL_DIS |
		     HFI1_RCVCTRL_TAILUPD_DIS |
		     HFI1_RCVCTRL_ONE_PKT_EGR_DIS |
		     HFI1_RCVCTRL_NO_RHQ_DROP_DIS |
		     HFI1_RCVCTRL_NO_EGR_DROP_DIS, uctxt);
	/* Clear the context's J_KEY */
	hfi1_clear_ctxt_jkey(dd, uctxt);
	/*
	 * If a send context is allocated, reset context integrity
	 * checks to default and disable the send context.
	 */
	if (uctxt->sc) {
		set_pio_integrity(uctxt->sc);
		sc_disable(uctxt->sc);
	}

	hfi1_free_ctxt_rcv_groups(uctxt);
	hfi1_clear_ctxt_pkey(dd, uctxt);

	uctxt->event_flags = 0;

	deallocate_ctxt(uctxt);
done:
	mmdrop(fdata->mm);
	kobject_put(&dd->kobj);

	if (atomic_dec_and_test(&dd->user_refcount))
		complete(&dd->user_comp);

	kfree(fdata);
	return 0;
}

/*
 * Convert kernel *virtual* addresses to physical addresses.
 * This is used to vmalloc'ed addresses.
 */
static u64 kvirt_to_phys(void *addr)
{
	struct page *page;
	u64 paddr = 0;

	page = vmalloc_to_page(addr);
	if (page)
		paddr = page_to_pfn(page) << PAGE_SHIFT;

	return paddr;
}

/**
 * complete_subctxt
 * @fd: valid filedata pointer
 *
 * Sub-context info can only be set up after the base context
 * has been completed.  This is indicated by the clearing of the
 * HFI1_CTXT_BASE_UINIT bit.
 *
 * Wait for the bit to be cleared, and then complete the subcontext
 * initialization.
 *
 */
static int complete_subctxt(struct hfi1_filedata *fd)
{
	int ret;
	unsigned long flags;

	/*
	 * sub-context info can only be set up after the base context
	 * has been completed.
	 */
	ret = wait_event_interruptible(
		fd->uctxt->wait,
		!test_bit(HFI1_CTXT_BASE_UNINIT, &fd->uctxt->event_flags));

	if (test_bit(HFI1_CTXT_BASE_FAILED, &fd->uctxt->event_flags))
		ret = -ENOMEM;

	/* Finish the sub-context init */
	if (!ret) {
		fd->rec_cpu_num = hfi1_get_proc_affinity(fd->uctxt->numa_id);
		ret = init_user_ctxt(fd, fd->uctxt);
	}

	if (ret) {
		hfi1_rcd_put(fd->uctxt);
		fd->uctxt = NULL;
		spin_lock_irqsave(&fd->dd->uctxt_lock, flags);
		__clear_bit(fd->subctxt, fd->uctxt->in_use_ctxts);
		spin_unlock_irqrestore(&fd->dd->uctxt_lock, flags);
	}

	return ret;
}

static int assign_ctxt(struct hfi1_filedata *fd, unsigned long arg, u32 len)
{
	int ret;
	unsigned int swmajor;
	struct hfi1_ctxtdata *uctxt = NULL;
	struct hfi1_user_info uinfo;

	if (fd->uctxt)
		return -EINVAL;

	if (sizeof(uinfo) != len)
		return -EINVAL;

	if (copy_from_user(&uinfo, (void __user *)arg, sizeof(uinfo)))
		return -EFAULT;

	swmajor = uinfo.userversion >> 16;
	if (swmajor != HFI1_USER_SWMAJOR)
		return -ENODEV;

	if (uinfo.subctxt_cnt > HFI1_MAX_SHARED_CTXTS)
		return -EINVAL;

	/*
	 * Acquire the mutex to protect against multiple creations of what
	 * could be a shared base context.
	 */
	mutex_lock(&hfi1_mutex);
	/*
	 * Get a sub context if available  (fd->uctxt will be set).
	 * ret < 0 error, 0 no context, 1 sub-context found
	 */
	ret = find_sub_ctxt(fd, &uinfo);

	/*
	 * Allocate a base context if context sharing is not required or a
	 * sub context wasn't found.
	 */
	if (!ret)
		ret = allocate_ctxt(fd, fd->dd, &uinfo, &uctxt);

	mutex_unlock(&hfi1_mutex);

	/* Depending on the context type, finish the appropriate init */
	switch (ret) {
	case 0:
		ret = setup_base_ctxt(fd, uctxt);
		if (ret)
			deallocate_ctxt(uctxt);
		break;
	case 1:
		ret = complete_subctxt(fd);
		break;
	default:
		break;
	}

	return ret;
}

/**
 * match_ctxt
 * @fd: valid filedata pointer
 * @uinfo: user info to compare base context with
 * @uctxt: context to compare uinfo to.
 *
 * Compare the given context with the given information to see if it
 * can be used for a sub context.
 */
static int match_ctxt(struct hfi1_filedata *fd,
		      const struct hfi1_user_info *uinfo,
		      struct hfi1_ctxtdata *uctxt)
{
	struct hfi1_devdata *dd = fd->dd;
	unsigned long flags;
	u16 subctxt;

	/* Skip dynamically allocated kernel contexts */
	if (uctxt->sc && (uctxt->sc->type == SC_KERNEL))
		return 0;

	/* Skip ctxt if it doesn't match the requested one */
	if (memcmp(uctxt->uuid, uinfo->uuid, sizeof(uctxt->uuid)) ||
	    uctxt->jkey != generate_jkey(current_uid()) ||
	    uctxt->subctxt_id != uinfo->subctxt_id ||
	    uctxt->subctxt_cnt != uinfo->subctxt_cnt)
		return 0;

	/* Verify the sharing process matches the base */
	if (uctxt->userversion != uinfo->userversion)
		return -EINVAL;

	/* Find an unused sub context */
	spin_lock_irqsave(&dd->uctxt_lock, flags);
	if (bitmap_empty(uctxt->in_use_ctxts, HFI1_MAX_SHARED_CTXTS)) {
		/* context is being closed, do not use */
		spin_unlock_irqrestore(&dd->uctxt_lock, flags);
		return 0;
	}

	subctxt = find_first_zero_bit(uctxt->in_use_ctxts,
				      HFI1_MAX_SHARED_CTXTS);
	if (subctxt >= uctxt->subctxt_cnt) {
		spin_unlock_irqrestore(&dd->uctxt_lock, flags);
		return -EBUSY;
	}

	fd->subctxt = subctxt;
	__set_bit(fd->subctxt, uctxt->in_use_ctxts);
	spin_unlock_irqrestore(&dd->uctxt_lock, flags);

	fd->uctxt = uctxt;
	hfi1_rcd_get(uctxt);

	return 1;
}

/**
 * find_sub_ctxt
 * @fd: valid filedata pointer
 * @uinfo: matching info to use to find a possible context to share.
 *
 * The hfi1_mutex must be held when this function is called.  It is
 * necessary to ensure serialized creation of shared contexts.
 *
 * Return:
 *    0      No sub-context found
 *    1      Subcontext found and allocated
 *    errno  EINVAL (incorrect parameters)
 *           EBUSY (all sub contexts in use)
 */
static int find_sub_ctxt(struct hfi1_filedata *fd,
			 const struct hfi1_user_info *uinfo)
{
	struct hfi1_ctxtdata *uctxt;
	struct hfi1_devdata *dd = fd->dd;
	u16 i;
	int ret;

	if (!uinfo->subctxt_cnt)
		return 0;

	for (i = dd->first_dyn_alloc_ctxt; i < dd->num_rcv_contexts; i++) {
		uctxt = hfi1_rcd_get_by_index(dd, i);
		if (uctxt) {
			ret = match_ctxt(fd, uinfo, uctxt);
			hfi1_rcd_put(uctxt);
			/* value of != 0 will return */
			if (ret)
				return ret;
		}
	}

	return 0;
}

static int allocate_ctxt(struct hfi1_filedata *fd, struct hfi1_devdata *dd,
			 struct hfi1_user_info *uinfo,
			 struct hfi1_ctxtdata **rcd)
{
	struct hfi1_ctxtdata *uctxt;
	int ret, numa;

	if (dd->flags & HFI1_FROZEN) {
		/*
		 * Pick an error that is unique from all other errors
		 * that are returned so the user process knows that
		 * it tried to allocate while the SPC was frozen.  It
		 * it should be able to retry with success in a short
		 * while.
		 */
		return -EIO;
	}

	if (!dd->freectxts)
		return -EBUSY;

	/*
	 * If we don't have a NUMA node requested, preference is towards
	 * device NUMA node.
	 */
	fd->rec_cpu_num = hfi1_get_proc_affinity(dd->node);
	if (fd->rec_cpu_num != -1)
		numa = cpu_to_node(fd->rec_cpu_num);
	else
		numa = numa_node_id();
	ret = hfi1_create_ctxtdata(dd->pport, numa, &uctxt);
	if (ret < 0) {
		dd_dev_err(dd, "user ctxtdata allocation failed\n");
		return ret;
	}
	hfi1_cdbg(PROC, "[%u:%u] pid %u assigned to CPU %d (NUMA %u)",
		  uctxt->ctxt, fd->subctxt, current->pid, fd->rec_cpu_num,
		  uctxt->numa_id);

	/*
	 * Allocate and enable a PIO send context.
	 */
	uctxt->sc = sc_alloc(dd, SC_USER, uctxt->rcvhdrqentsize, dd->node);
	if (!uctxt->sc) {
		ret = -ENOMEM;
		goto ctxdata_free;
	}
	hfi1_cdbg(PROC, "allocated send context %u(%u)\n", uctxt->sc->sw_index,
		  uctxt->sc->hw_context);
	ret = sc_enable(uctxt->sc);
	if (ret)
		goto ctxdata_free;

	/*
	 * Setup sub context information if the user-level has requested
	 * sub contexts.
	 * This has to be done here so the rest of the sub-contexts find the
	 * proper base context.
	 */
	if (uinfo->subctxt_cnt)
		init_subctxts(uctxt, uinfo);
	uctxt->userversion = uinfo->userversion;
	uctxt->flags = hfi1_cap_mask; /* save current flag state */
	init_waitqueue_head(&uctxt->wait);
	strlcpy(uctxt->comm, current->comm, sizeof(uctxt->comm));
	memcpy(uctxt->uuid, uinfo->uuid, sizeof(uctxt->uuid));
	uctxt->jkey = generate_jkey(current_uid());
	hfi1_stats.sps_ctxts++;
	/*
	 * Disable ASPM when there are open user/PSM contexts to avoid
	 * issues with ASPM L1 exit latency
	 */
	if (dd->freectxts-- == dd->num_user_contexts)
		aspm_disable_all(dd);

	*rcd = uctxt;

	return 0;

ctxdata_free:
	hfi1_free_ctxt(uctxt);
	return ret;
}

static void deallocate_ctxt(struct hfi1_ctxtdata *uctxt)
{
	mutex_lock(&hfi1_mutex);
	hfi1_stats.sps_ctxts--;
	if (++uctxt->dd->freectxts == uctxt->dd->num_user_contexts)
		aspm_enable_all(uctxt->dd);
	mutex_unlock(&hfi1_mutex);

	hfi1_free_ctxt(uctxt);
}

static void init_subctxts(struct hfi1_ctxtdata *uctxt,
			  const struct hfi1_user_info *uinfo)
{
	uctxt->subctxt_cnt = uinfo->subctxt_cnt;
	uctxt->subctxt_id = uinfo->subctxt_id;
	set_bit(HFI1_CTXT_BASE_UNINIT, &uctxt->event_flags);
}

static int setup_subctxt(struct hfi1_ctxtdata *uctxt)
{
	int ret = 0;
	u16 num_subctxts = uctxt->subctxt_cnt;

	uctxt->subctxt_uregbase = vmalloc_user(PAGE_SIZE);
	if (!uctxt->subctxt_uregbase)
		return -ENOMEM;

	/* We can take the size of the RcvHdr Queue from the master */
	uctxt->subctxt_rcvhdr_base = vmalloc_user(uctxt->rcvhdrq_size *
						  num_subctxts);
	if (!uctxt->subctxt_rcvhdr_base) {
		ret = -ENOMEM;
		goto bail_ureg;
	}

	uctxt->subctxt_rcvegrbuf = vmalloc_user(uctxt->egrbufs.size *
						num_subctxts);
	if (!uctxt->subctxt_rcvegrbuf) {
		ret = -ENOMEM;
		goto bail_rhdr;
	}

	return 0;

bail_rhdr:
	vfree(uctxt->subctxt_rcvhdr_base);
	uctxt->subctxt_rcvhdr_base = NULL;
bail_ureg:
	vfree(uctxt->subctxt_uregbase);
	uctxt->subctxt_uregbase = NULL;

	return ret;
}

static void user_init(struct hfi1_ctxtdata *uctxt)
{
	unsigned int rcvctrl_ops = 0;

	/* initialize poll variables... */
	uctxt->urgent = 0;
	uctxt->urgent_poll = 0;

	/*
	 * Now enable the ctxt for receive.
	 * For chips that are set to DMA the tail register to memory
	 * when they change (and when the update bit transitions from
	 * 0 to 1.  So for those chips, we turn it off and then back on.
	 * This will (very briefly) affect any other open ctxts, but the
	 * duration is very short, and therefore isn't an issue.  We
	 * explicitly set the in-memory tail copy to 0 beforehand, so we
	 * don't have to wait to be sure the DMA update has happened
	 * (chip resets head/tail to 0 on transition to enable).
	 */
	if (uctxt->rcvhdrtail_kvaddr)
		clear_rcvhdrtail(uctxt);

	/* Setup J_KEY before enabling the context */
	hfi1_set_ctxt_jkey(uctxt->dd, uctxt, uctxt->jkey);

	rcvctrl_ops = HFI1_RCVCTRL_CTXT_ENB;
	if (HFI1_CAP_UGET_MASK(uctxt->flags, HDRSUPP))
		rcvctrl_ops |= HFI1_RCVCTRL_TIDFLOW_ENB;
	/*
	 * Ignore the bit in the flags for now until proper
	 * support for multiple packet per rcv array entry is
	 * added.
	 */
	if (!HFI1_CAP_UGET_MASK(uctxt->flags, MULTI_PKT_EGR))
		rcvctrl_ops |= HFI1_RCVCTRL_ONE_PKT_EGR_ENB;
	if (HFI1_CAP_UGET_MASK(uctxt->flags, NODROP_EGR_FULL))
		rcvctrl_ops |= HFI1_RCVCTRL_NO_EGR_DROP_ENB;
	if (HFI1_CAP_UGET_MASK(uctxt->flags, NODROP_RHQ_FULL))
		rcvctrl_ops |= HFI1_RCVCTRL_NO_RHQ_DROP_ENB;
	/*
	 * The RcvCtxtCtrl.TailUpd bit has to be explicitly written.
	 * We can't rely on the correct value to be set from prior
	 * uses of the chip or ctxt. Therefore, add the rcvctrl op
	 * for both cases.
	 */
	if (HFI1_CAP_UGET_MASK(uctxt->flags, DMA_RTAIL))
		rcvctrl_ops |= HFI1_RCVCTRL_TAILUPD_ENB;
	else
		rcvctrl_ops |= HFI1_RCVCTRL_TAILUPD_DIS;
	hfi1_rcvctrl(uctxt->dd, rcvctrl_ops, uctxt);
}

static int get_ctxt_info(struct hfi1_filedata *fd, unsigned long arg, u32 len)
{
	struct hfi1_ctxt_info cinfo;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;

	if (sizeof(cinfo) != len)
		return -EINVAL;

	memset(&cinfo, 0, sizeof(cinfo));
	cinfo.runtime_flags = (((uctxt->flags >> HFI1_CAP_MISC_SHIFT) &
				HFI1_CAP_MISC_MASK) << HFI1_CAP_USER_SHIFT) |
			HFI1_CAP_UGET_MASK(uctxt->flags, MASK) |
			HFI1_CAP_KGET_MASK(uctxt->flags, K2U);
	/* adjust flag if this fd is not able to cache */
	if (!fd->handler)
		cinfo.runtime_flags |= HFI1_CAP_TID_UNMAP; /* no caching */

	cinfo.num_active = hfi1_count_active_units();
	cinfo.unit = uctxt->dd->unit;
	cinfo.ctxt = uctxt->ctxt;
	cinfo.subctxt = fd->subctxt;
	cinfo.rcvtids = roundup(uctxt->egrbufs.alloced,
				uctxt->dd->rcv_entries.group_size) +
		uctxt->expected_count;
	cinfo.credits = uctxt->sc->credits;
	cinfo.numa_node = uctxt->numa_id;
	cinfo.rec_cpu = fd->rec_cpu_num;
	cinfo.send_ctxt = uctxt->sc->hw_context;

	cinfo.egrtids = uctxt->egrbufs.alloced;
	cinfo.rcvhdrq_cnt = uctxt->rcvhdrq_cnt;
	cinfo.rcvhdrq_entsize = uctxt->rcvhdrqentsize << 2;
	cinfo.sdma_ring_size = fd->cq->nentries;
	cinfo.rcvegr_size = uctxt->egrbufs.rcvtid_size;

	trace_hfi1_ctxt_info(uctxt->dd, uctxt->ctxt, fd->subctxt, cinfo);
	if (copy_to_user((void __user *)arg, &cinfo, len))
		return -EFAULT;

	return 0;
}

static int init_user_ctxt(struct hfi1_filedata *fd,
			  struct hfi1_ctxtdata *uctxt)
{
	int ret;

	ret = hfi1_user_sdma_alloc_queues(uctxt, fd);
	if (ret)
		return ret;

	ret = hfi1_user_exp_rcv_init(fd, uctxt);
	if (ret)
		hfi1_user_sdma_free_queues(fd, uctxt);

	return ret;
}

static int setup_base_ctxt(struct hfi1_filedata *fd,
			   struct hfi1_ctxtdata *uctxt)
{
	struct hfi1_devdata *dd = uctxt->dd;
	int ret = 0;

	hfi1_init_ctxt(uctxt->sc);

	/* Now allocate the RcvHdr queue and eager buffers. */
	ret = hfi1_create_rcvhdrq(dd, uctxt);
	if (ret)
		goto done;

	ret = hfi1_setup_eagerbufs(uctxt);
	if (ret)
		goto done;

	/* If sub-contexts are enabled, do the appropriate setup */
	if (uctxt->subctxt_cnt)
		ret = setup_subctxt(uctxt);
	if (ret)
		goto done;

	ret = hfi1_alloc_ctxt_rcv_groups(uctxt);
	if (ret)
		goto done;

	ret = init_user_ctxt(fd, uctxt);
	if (ret)
		goto done;

	user_init(uctxt);

	/* Now that the context is set up, the fd can get a reference. */
	fd->uctxt = uctxt;
	hfi1_rcd_get(uctxt);

done:
	if (uctxt->subctxt_cnt) {
		/*
		 * On error, set the failed bit so sub-contexts will clean up
		 * correctly.
		 */
		if (ret)
			set_bit(HFI1_CTXT_BASE_FAILED, &uctxt->event_flags);

		/*
		 * Base context is done (successfully or not), notify anybody
		 * using a sub-context that is waiting for this completion.
		 */
		clear_bit(HFI1_CTXT_BASE_UNINIT, &uctxt->event_flags);
		wake_up(&uctxt->wait);
	}

	return ret;
}

static int get_base_info(struct hfi1_filedata *fd, unsigned long arg, u32 len)
{
	struct hfi1_base_info binfo;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;
	struct hfi1_devdata *dd = uctxt->dd;
	unsigned offset;

	trace_hfi1_uctxtdata(uctxt->dd, uctxt, fd->subctxt);

	if (sizeof(binfo) != len)
		return -EINVAL;

	memset(&binfo, 0, sizeof(binfo));
	binfo.hw_version = dd->revision;
	binfo.sw_version = HFI1_KERN_SWVERSION;
	binfo.bthqp = kdeth_qp;
	binfo.jkey = uctxt->jkey;
	/*
	 * If more than 64 contexts are enabled the allocated credit
	 * return will span two or three contiguous pages. Since we only
	 * map the page containing the context's credit return address,
	 * we need to calculate the offset in the proper page.
	 */
	offset = ((u64)uctxt->sc->hw_free -
		  (u64)dd->cr_base[uctxt->numa_id].va) % PAGE_SIZE;
	binfo.sc_credits_addr = HFI1_MMAP_TOKEN(PIO_CRED, uctxt->ctxt,
						fd->subctxt, offset);
	binfo.pio_bufbase = HFI1_MMAP_TOKEN(PIO_BUFS, uctxt->ctxt,
					    fd->subctxt,
					    uctxt->sc->base_addr);
	binfo.pio_bufbase_sop = HFI1_MMAP_TOKEN(PIO_BUFS_SOP,
						uctxt->ctxt,
						fd->subctxt,
						uctxt->sc->base_addr);
	binfo.rcvhdr_bufbase = HFI1_MMAP_TOKEN(RCV_HDRQ, uctxt->ctxt,
					       fd->subctxt,
					       uctxt->rcvhdrq);
	binfo.rcvegr_bufbase = HFI1_MMAP_TOKEN(RCV_EGRBUF, uctxt->ctxt,
					       fd->subctxt,
					       uctxt->egrbufs.rcvtids[0].dma);
	binfo.sdma_comp_bufbase = HFI1_MMAP_TOKEN(SDMA_COMP, uctxt->ctxt,
						  fd->subctxt, 0);
	/*
	 * user regs are at
	 * (RXE_PER_CONTEXT_USER + (ctxt * RXE_PER_CONTEXT_SIZE))
	 */
	binfo.user_regbase = HFI1_MMAP_TOKEN(UREGS, uctxt->ctxt,
					     fd->subctxt, 0);
	offset = offset_in_page((uctxt_offset(uctxt) + fd->subctxt) *
				sizeof(*dd->events));
	binfo.events_bufbase = HFI1_MMAP_TOKEN(EVENTS, uctxt->ctxt,
					       fd->subctxt,
					       offset);
	binfo.status_bufbase = HFI1_MMAP_TOKEN(STATUS, uctxt->ctxt,
					       fd->subctxt,
					       dd->status);
	if (HFI1_CAP_IS_USET(DMA_RTAIL))
		binfo.rcvhdrtail_base = HFI1_MMAP_TOKEN(RTAIL, uctxt->ctxt,
							fd->subctxt, 0);
	if (uctxt->subctxt_cnt) {
		binfo.subctxt_uregbase = HFI1_MMAP_TOKEN(SUBCTXT_UREGS,
							 uctxt->ctxt,
							 fd->subctxt, 0);
		binfo.subctxt_rcvhdrbuf = HFI1_MMAP_TOKEN(SUBCTXT_RCV_HDRQ,
							  uctxt->ctxt,
							  fd->subctxt, 0);
		binfo.subctxt_rcvegrbuf = HFI1_MMAP_TOKEN(SUBCTXT_EGRBUF,
							  uctxt->ctxt,
							  fd->subctxt, 0);
	}

	if (copy_to_user((void __user *)arg, &binfo, len))
		return -EFAULT;

	return 0;
}

/**
 * user_exp_rcv_setup - Set up the given tid rcv list
 * @fd: file data of the current driver instance
 * @arg: ioctl argumnent for user space information
 * @len: length of data structure associated with ioctl command
 *
 * Wrapper to validate ioctl information before doing _rcv_setup.
 *
 */
static int user_exp_rcv_setup(struct hfi1_filedata *fd, unsigned long arg,
			      u32 len)
{
	int ret;
	unsigned long addr;
	struct hfi1_tid_info tinfo;

	if (sizeof(tinfo) != len)
		return -EINVAL;

	if (copy_from_user(&tinfo, (void __user *)arg, (sizeof(tinfo))))
		return -EFAULT;

	ret = hfi1_user_exp_rcv_setup(fd, &tinfo);
	if (!ret) {
		/*
		 * Copy the number of tidlist entries we used
		 * and the length of the buffer we registered.
		 */
		addr = arg + offsetof(struct hfi1_tid_info, tidcnt);
		if (copy_to_user((void __user *)addr, &tinfo.tidcnt,
				 sizeof(tinfo.tidcnt)))
			return -EFAULT;

		addr = arg + offsetof(struct hfi1_tid_info, length);
		if (copy_to_user((void __user *)addr, &tinfo.length,
				 sizeof(tinfo.length)))
			ret = -EFAULT;
	}

	return ret;
}

/**
 * user_exp_rcv_clear - Clear the given tid rcv list
 * @fd: file data of the current driver instance
 * @arg: ioctl argumnent for user space information
 * @len: length of data structure associated with ioctl command
 *
 * The hfi1_user_exp_rcv_clear() can be called from the error path.  Because
 * of this, we need to use this wrapper to copy the user space information
 * before doing the clear.
 */
static int user_exp_rcv_clear(struct hfi1_filedata *fd, unsigned long arg,
			      u32 len)
{
	int ret;
	unsigned long addr;
	struct hfi1_tid_info tinfo;

	if (sizeof(tinfo) != len)
		return -EINVAL;

	if (copy_from_user(&tinfo, (void __user *)arg, (sizeof(tinfo))))
		return -EFAULT;

	ret = hfi1_user_exp_rcv_clear(fd, &tinfo);
	if (!ret) {
		addr = arg + offsetof(struct hfi1_tid_info, tidcnt);
		if (copy_to_user((void __user *)addr, &tinfo.tidcnt,
				 sizeof(tinfo.tidcnt)))
			return -EFAULT;
	}

	return ret;
}

/**
 * user_exp_rcv_invalid - Invalidate the given tid rcv list
 * @fd: file data of the current driver instance
 * @arg: ioctl argumnent for user space information
 * @len: length of data structure associated with ioctl command
 *
 * Wrapper to validate ioctl information before doing _rcv_invalid.
 *
 */
static int user_exp_rcv_invalid(struct hfi1_filedata *fd, unsigned long arg,
				u32 len)
{
	int ret;
	unsigned long addr;
	struct hfi1_tid_info tinfo;

	if (sizeof(tinfo) != len)
		return -EINVAL;

	if (!fd->invalid_tids)
		return -EINVAL;

	if (copy_from_user(&tinfo, (void __user *)arg, (sizeof(tinfo))))
		return -EFAULT;

	ret = hfi1_user_exp_rcv_invalid(fd, &tinfo);
	if (ret)
		return ret;

	addr = arg + offsetof(struct hfi1_tid_info, tidcnt);
	if (copy_to_user((void __user *)addr, &tinfo.tidcnt,
			 sizeof(tinfo.tidcnt)))
		ret = -EFAULT;

	return ret;
}

static unsigned int poll_urgent(struct file *fp,
				struct poll_table_struct *pt)
{
	struct hfi1_filedata *fd = fp->private_data;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;
	struct hfi1_devdata *dd = uctxt->dd;
	unsigned pollflag;

	poll_wait(fp, &uctxt->wait, pt);

	spin_lock_irq(&dd->uctxt_lock);
	if (uctxt->urgent != uctxt->urgent_poll) {
		pollflag = POLLIN | POLLRDNORM;
		uctxt->urgent_poll = uctxt->urgent;
	} else {
		pollflag = 0;
		set_bit(HFI1_CTXT_WAITING_URG, &uctxt->event_flags);
	}
	spin_unlock_irq(&dd->uctxt_lock);

	return pollflag;
}

static unsigned int poll_next(struct file *fp,
			      struct poll_table_struct *pt)
{
	struct hfi1_filedata *fd = fp->private_data;
	struct hfi1_ctxtdata *uctxt = fd->uctxt;
	struct hfi1_devdata *dd = uctxt->dd;
	unsigned pollflag;

	poll_wait(fp, &uctxt->wait, pt);

	spin_lock_irq(&dd->uctxt_lock);
	if (hdrqempty(uctxt)) {
		set_bit(HFI1_CTXT_WAITING_RCV, &uctxt->event_flags);
		hfi1_rcvctrl(dd, HFI1_RCVCTRL_INTRAVAIL_ENB, uctxt);
		pollflag = 0;
	} else {
		pollflag = POLLIN | POLLRDNORM;
	}
	spin_unlock_irq(&dd->uctxt_lock);

	return pollflag;
}

/*
 * Find all user contexts in use, and set the specified bit in their
 * event mask.
 * See also find_ctxt() for a similar use, that is specific to send buffers.
 */
int hfi1_set_uevent_bits(struct hfi1_pportdata *ppd, const int evtbit)
{
	struct hfi1_ctxtdata *uctxt;
	struct hfi1_devdata *dd = ppd->dd;
	u16 ctxt;

	if (!dd->events)
		return -EINVAL;

	for (ctxt = dd->first_dyn_alloc_ctxt; ctxt < dd->num_rcv_contexts;
	     ctxt++) {
		uctxt = hfi1_rcd_get_by_index(dd, ctxt);
		if (uctxt) {
			unsigned long *evs;
			int i;
			/*
			 * subctxt_cnt is 0 if not shared, so do base
			 * separately, first, then remaining subctxt, if any
			 */
			evs = dd->events + uctxt_offset(uctxt);
			set_bit(evtbit, evs);
			for (i = 1; i < uctxt->subctxt_cnt; i++)
				set_bit(evtbit, evs + i);
			hfi1_rcd_put(uctxt);
		}
	}

	return 0;
}

/**
 * manage_rcvq - manage a context's receive queue
 * @uctxt: the context
 * @subctxt: the sub-context
 * @start_stop: action to carry out
 *
 * start_stop == 0 disables receive on the context, for use in queue
 * overflow conditions.  start_stop==1 re-enables, to be used to
 * re-init the software copy of the head register
 */
static int manage_rcvq(struct hfi1_ctxtdata *uctxt, u16 subctxt,
		       unsigned long arg)
{
	struct hfi1_devdata *dd = uctxt->dd;
	unsigned int rcvctrl_op;
	int start_stop;

	if (subctxt)
		return 0;

	if (get_user(start_stop, (int __user *)arg))
		return -EFAULT;

	/* atomically clear receive enable ctxt. */
	if (start_stop) {
		/*
		 * On enable, force in-memory copy of the tail register to
		 * 0, so that protocol code doesn't have to worry about
		 * whether or not the chip has yet updated the in-memory
		 * copy or not on return from the system call. The chip
		 * always resets it's tail register back to 0 on a
		 * transition from disabled to enabled.
		 */
		if (uctxt->rcvhdrtail_kvaddr)
			clear_rcvhdrtail(uctxt);
		rcvctrl_op = HFI1_RCVCTRL_CTXT_ENB;
	} else {
		rcvctrl_op = HFI1_RCVCTRL_CTXT_DIS;
	}
	hfi1_rcvctrl(dd, rcvctrl_op, uctxt);
	/* always; new head should be equal to new tail; see above */

	return 0;
}

/*
 * clear the event notifier events for this context.
 * User process then performs actions appropriate to bit having been
 * set, if desired, and checks again in future.
 */
static int user_event_ack(struct hfi1_ctxtdata *uctxt, u16 subctxt,
			  unsigned long arg)
{
	int i;
	struct hfi1_devdata *dd = uctxt->dd;
	unsigned long *evs;
	unsigned long events;

	if (!dd->events)
		return 0;

	if (get_user(events, (unsigned long __user *)arg))
		return -EFAULT;

	evs = dd->events + uctxt_offset(uctxt) + subctxt;

	for (i = 0; i <= _HFI1_MAX_EVENT_BIT; i++) {
		if (!test_bit(i, &events))
			continue;
		clear_bit(i, evs);
	}
	return 0;
}

static int set_ctxt_pkey(struct hfi1_ctxtdata *uctxt, unsigned long arg)
{
	int i;
	struct hfi1_pportdata *ppd = uctxt->ppd;
	struct hfi1_devdata *dd = uctxt->dd;
	u16 pkey;

	if (!HFI1_CAP_IS_USET(PKEY_CHECK))
		return -EPERM;

	if (get_user(pkey, (u16 __user *)arg))
		return -EFAULT;

	if (pkey == LIM_MGMT_P_KEY || pkey == FULL_MGMT_P_KEY)
		return -EINVAL;

	for (i = 0; i < ARRAY_SIZE(ppd->pkeys); i++)
		if (pkey == ppd->pkeys[i])
			return hfi1_set_ctxt_pkey(dd, uctxt, pkey);

	return -ENOENT;
}

/**
 * ctxt_reset - Reset the user context
 * @uctxt: valid user context
 */
static int ctxt_reset(struct hfi1_ctxtdata *uctxt)
{
	struct send_context *sc;
	struct hfi1_devdata *dd;
	int ret = 0;

	if (!uctxt || !uctxt->dd || !uctxt->sc)
		return -EINVAL;

	/*
	 * There is no protection here. User level has to guarantee that
	 * no one will be writing to the send context while it is being
	 * re-initialized.  If user level breaks that guarantee, it will
	 * break it's own context and no one else's.
	 */
	dd = uctxt->dd;
	sc = uctxt->sc;

	/*
	 * Wait until the interrupt handler has marked the context as
	 * halted or frozen. Report error if we time out.
	 */
	wait_event_interruptible_timeout(
		sc->halt_wait, (sc->flags & SCF_HALTED),
		msecs_to_jiffies(SEND_CTXT_HALT_TIMEOUT));
	if (!(sc->flags & SCF_HALTED))
		return -ENOLCK;

	/*
	 * If the send context was halted due to a Freeze, wait until the
	 * device has been "unfrozen" before resetting the context.
	 */
	if (sc->flags & SCF_FROZEN) {
		wait_event_interruptible_timeout(
			dd->event_queue,
			!(READ_ONCE(dd->flags) & HFI1_FROZEN),
			msecs_to_jiffies(SEND_CTXT_HALT_TIMEOUT));
		if (dd->flags & HFI1_FROZEN)
			return -ENOLCK;

		if (dd->flags & HFI1_FORCED_FREEZE)
			/*
			 * Don't allow context reset if we are into
			 * forced freeze
			 */
			return -ENODEV;

		sc_disable(sc);
		ret = sc_enable(sc);
		hfi1_rcvctrl(dd, HFI1_RCVCTRL_CTXT_ENB, uctxt);
	} else {
		ret = sc_restart(sc);
	}
	if (!ret)
		sc_return_credits(sc);

	return ret;
}

static void user_remove(struct hfi1_devdata *dd)
{

	hfi1_cdev_cleanup(&dd->user_cdev, &dd->user_device);
}

static int user_add(struct hfi1_devdata *dd)
{
	char name[10];
	int ret;

	snprintf(name, sizeof(name), "%s_%d", class_name(), dd->unit);
	ret = hfi1_cdev_init(dd->unit, name, &hfi1_file_ops,
			     &dd->user_cdev, &dd->user_device,
			     true, &dd->kobj);
	if (ret)
		user_remove(dd);

	return ret;
}

/*
 * Create per-unit files in /dev
 */
int hfi1_device_create(struct hfi1_devdata *dd)
{
	return user_add(dd);
}

/*
 * Remove per-unit files in /dev
 * void, core kernel returns no errors for this stuff
 */
void hfi1_device_remove(struct hfi1_devdata *dd)
{
	user_remove(dd);
}