aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/net/liquidio/lio_rxtx.c
blob: 376893ac8cb83dec72eb7e3f9be59ee0753f43fd (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
/*
 *   BSD LICENSE
 *
 *   Copyright(c) 2017 Cavium, Inc.. All rights reserved.
 *   All rights reserved.
 *
 *   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 Cavium, Inc. 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(S) 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 <rte_ethdev.h>
#include <rte_cycles.h>
#include <rte_malloc.h>

#include "lio_logs.h"
#include "lio_struct.h"
#include "lio_ethdev.h"
#include "lio_rxtx.h"

#define LIO_MAX_SG 12
/* Flush iq if available tx_desc fall below LIO_FLUSH_WM */
#define LIO_FLUSH_WM(_iq) ((_iq)->max_count / 2)
#define LIO_PKT_IN_DONE_CNT_MASK 0x00000000FFFFFFFFULL

static void
lio_droq_compute_max_packet_bufs(struct lio_droq *droq)
{
	uint32_t count = 0;

	do {
		count += droq->buffer_size;
	} while (count < LIO_MAX_RX_PKTLEN);
}

static void
lio_droq_reset_indices(struct lio_droq *droq)
{
	droq->read_idx	= 0;
	droq->write_idx	= 0;
	droq->refill_idx = 0;
	droq->refill_count = 0;
	rte_atomic64_set(&droq->pkts_pending, 0);
}

static void
lio_droq_destroy_ring_buffers(struct lio_droq *droq)
{
	uint32_t i;

	for (i = 0; i < droq->max_count; i++) {
		if (droq->recv_buf_list[i].buffer) {
			rte_pktmbuf_free((struct rte_mbuf *)
					 droq->recv_buf_list[i].buffer);
			droq->recv_buf_list[i].buffer = NULL;
		}
	}

	lio_droq_reset_indices(droq);
}

static int
lio_droq_setup_ring_buffers(struct lio_device *lio_dev,
			    struct lio_droq *droq)
{
	struct lio_droq_desc *desc_ring = droq->desc_ring;
	uint32_t i;
	void *buf;

	for (i = 0; i < droq->max_count; i++) {
		buf = rte_pktmbuf_alloc(droq->mpool);
		if (buf == NULL) {
			lio_dev_err(lio_dev, "buffer alloc failed\n");
			droq->stats.rx_alloc_failure++;
			lio_droq_destroy_ring_buffers(droq);
			return -ENOMEM;
		}

		droq->recv_buf_list[i].buffer = buf;
		droq->info_list[i].length = 0;

		/* map ring buffers into memory */
		desc_ring[i].info_ptr = lio_map_ring_info(droq, i);
		desc_ring[i].buffer_ptr =
			lio_map_ring(droq->recv_buf_list[i].buffer);
	}

	lio_droq_reset_indices(droq);

	lio_droq_compute_max_packet_bufs(droq);

	return 0;
}

static void
lio_dma_zone_free(struct lio_device *lio_dev, const struct rte_memzone *mz)
{
	const struct rte_memzone *mz_tmp;
	int ret = 0;

	if (mz == NULL) {
		lio_dev_err(lio_dev, "Memzone NULL\n");
		return;
	}

	mz_tmp = rte_memzone_lookup(mz->name);
	if (mz_tmp == NULL) {
		lio_dev_err(lio_dev, "Memzone %s Not Found\n", mz->name);
		return;
	}

	ret = rte_memzone_free(mz);
	if (ret)
		lio_dev_err(lio_dev, "Memzone free Failed ret %d\n", ret);
}

/**
 *  Frees the space for descriptor ring for the droq.
 *
 *  @param lio_dev	- pointer to the lio device structure
 *  @param q_no		- droq no.
 */
static void
lio_delete_droq(struct lio_device *lio_dev, uint32_t q_no)
{
	struct lio_droq *droq = lio_dev->droq[q_no];

	lio_dev_dbg(lio_dev, "OQ[%d]\n", q_no);

	lio_droq_destroy_ring_buffers(droq);
	rte_free(droq->recv_buf_list);
	droq->recv_buf_list = NULL;
	lio_dma_zone_free(lio_dev, droq->info_mz);
	lio_dma_zone_free(lio_dev, droq->desc_ring_mz);

	memset(droq, 0, LIO_DROQ_SIZE);
}

static void *
lio_alloc_info_buffer(struct lio_device *lio_dev,
		      struct lio_droq *droq, unsigned int socket_id)
{
	droq->info_mz = rte_eth_dma_zone_reserve(lio_dev->eth_dev,
						 "info_list", droq->q_no,
						 (droq->max_count *
							LIO_DROQ_INFO_SIZE),
						 RTE_CACHE_LINE_SIZE,
						 socket_id);

	if (droq->info_mz == NULL)
		return NULL;

	droq->info_list_dma = droq->info_mz->iova;
	droq->info_alloc_size = droq->info_mz->len;
	droq->info_base_addr = (size_t)droq->info_mz->addr;

	return droq->info_mz->addr;
}

/**
 *  Allocates space for the descriptor ring for the droq and
 *  sets the base addr, num desc etc in Octeon registers.
 *
 * @param lio_dev	- pointer to the lio device structure
 * @param q_no		- droq no.
 * @param app_ctx	- pointer to application context
 * @return Success: 0	Failure: -1
 */
static int
lio_init_droq(struct lio_device *lio_dev, uint32_t q_no,
	      uint32_t num_descs, uint32_t desc_size,
	      struct rte_mempool *mpool, unsigned int socket_id)
{
	uint32_t c_refill_threshold;
	uint32_t desc_ring_size;
	struct lio_droq *droq;

	lio_dev_dbg(lio_dev, "OQ[%d]\n", q_no);

	droq = lio_dev->droq[q_no];
	droq->lio_dev = lio_dev;
	droq->q_no = q_no;
	droq->mpool = mpool;

	c_refill_threshold = LIO_OQ_REFILL_THRESHOLD_CFG(lio_dev);

	droq->max_count = num_descs;
	droq->buffer_size = desc_size;

	desc_ring_size = droq->max_count * LIO_DROQ_DESC_SIZE;
	droq->desc_ring_mz = rte_eth_dma_zone_reserve(lio_dev->eth_dev,
						      "droq", q_no,
						      desc_ring_size,
						      RTE_CACHE_LINE_SIZE,
						      socket_id);

	if (droq->desc_ring_mz == NULL) {
		lio_dev_err(lio_dev,
			    "Output queue %d ring alloc failed\n", q_no);
		return -1;
	}

	droq->desc_ring_dma = droq->desc_ring_mz->iova;
	droq->desc_ring = (struct lio_droq_desc *)droq->desc_ring_mz->addr;

	lio_dev_dbg(lio_dev, "droq[%d]: desc_ring: virt: 0x%p, dma: %lx\n",
		    q_no, droq->desc_ring, (unsigned long)droq->desc_ring_dma);
	lio_dev_dbg(lio_dev, "droq[%d]: num_desc: %d\n", q_no,
		    droq->max_count);

	droq->info_list = lio_alloc_info_buffer(lio_dev, droq, socket_id);
	if (droq->info_list == NULL) {
		lio_dev_err(lio_dev, "Cannot allocate memory for info list.\n");
		goto init_droq_fail;
	}

	droq->recv_buf_list = rte_zmalloc_socket("recv_buf_list",
						 (droq->max_count *
							LIO_DROQ_RECVBUF_SIZE),
						 RTE_CACHE_LINE_SIZE,
						 socket_id);
	if (droq->recv_buf_list == NULL) {
		lio_dev_err(lio_dev,
			    "Output queue recv buf list alloc failed\n");
		goto init_droq_fail;
	}

	if (lio_droq_setup_ring_buffers(lio_dev, droq))
		goto init_droq_fail;

	droq->refill_threshold = c_refill_threshold;

	rte_spinlock_init(&droq->lock);

	lio_dev->fn_list.setup_oq_regs(lio_dev, q_no);

	lio_dev->io_qmask.oq |= (1ULL << q_no);

	return 0;

init_droq_fail:
	lio_delete_droq(lio_dev, q_no);

	return -1;
}

int
lio_setup_droq(struct lio_device *lio_dev, int oq_no, int num_descs,
	       int desc_size, struct rte_mempool *mpool, unsigned int socket_id)
{
	struct lio_droq *droq;

	PMD_INIT_FUNC_TRACE();

	if (lio_dev->droq[oq_no]) {
		lio_dev_dbg(lio_dev, "Droq %d in use\n", oq_no);
		return 0;
	}

	/* Allocate the DS for the new droq. */
	droq = rte_zmalloc_socket("ethdev RX queue", sizeof(*droq),
				  RTE_CACHE_LINE_SIZE, socket_id);
	if (droq == NULL)
		return -ENOMEM;

	lio_dev->droq[oq_no] = droq;

	/* Initialize the Droq */
	if (lio_init_droq(lio_dev, oq_no, num_descs, desc_size, mpool,
			  socket_id)) {
		lio_dev_err(lio_dev, "Droq[%u] Initialization Failed\n", oq_no);
		rte_free(lio_dev->droq[oq_no]);
		lio_dev->droq[oq_no] = NULL;
		return -ENOMEM;
	}

	lio_dev->num_oqs++;

	lio_dev_dbg(lio_dev, "Total number of OQ: %d\n", lio_dev->num_oqs);

	/* Send credit for octeon output queues. credits are always
	 * sent after the output queue is enabled.
	 */
	rte_write32(lio_dev->droq[oq_no]->max_count,
		    lio_dev->droq[oq_no]->pkts_credit_reg);
	rte_wmb();

	return 0;
}

static inline uint32_t
lio_droq_get_bufcount(uint32_t buf_size, uint32_t total_len)
{
	uint32_t buf_cnt = 0;

	while (total_len > (buf_size * buf_cnt))
		buf_cnt++;

	return buf_cnt;
}

/* If we were not able to refill all buffers, try to move around
 * the buffers that were not dispatched.
 */
static inline uint32_t
lio_droq_refill_pullup_descs(struct lio_droq *droq,
			     struct lio_droq_desc *desc_ring)
{
	uint32_t refill_index = droq->refill_idx;
	uint32_t desc_refilled = 0;

	while (refill_index != droq->read_idx) {
		if (droq->recv_buf_list[refill_index].buffer) {
			droq->recv_buf_list[droq->refill_idx].buffer =
				droq->recv_buf_list[refill_index].buffer;
			desc_ring[droq->refill_idx].buffer_ptr =
				desc_ring[refill_index].buffer_ptr;
			droq->recv_buf_list[refill_index].buffer = NULL;
			desc_ring[refill_index].buffer_ptr = 0;
			do {
				droq->refill_idx = lio_incr_index(
							droq->refill_idx, 1,
							droq->max_count);
				desc_refilled++;
				droq->refill_count--;
			} while (droq->recv_buf_list[droq->refill_idx].buffer);
		}
		refill_index = lio_incr_index(refill_index, 1,
					      droq->max_count);
	}	/* while */

	return desc_refilled;
}

/* lio_droq_refill
 *
 * @param droq		- droq in which descriptors require new buffers.
 *
 * Description:
 *  Called during normal DROQ processing in interrupt mode or by the poll
 *  thread to refill the descriptors from which buffers were dispatched
 *  to upper layers. Attempts to allocate new buffers. If that fails, moves
 *  up buffers (that were not dispatched) to form a contiguous ring.
 *
 * Returns:
 *  No of descriptors refilled.
 *
 * Locks:
 * This routine is called with droq->lock held.
 */
static uint32_t
lio_droq_refill(struct lio_droq *droq)
{
	struct lio_droq_desc *desc_ring;
	uint32_t desc_refilled = 0;
	void *buf = NULL;

	desc_ring = droq->desc_ring;

	while (droq->refill_count && (desc_refilled < droq->max_count)) {
		/* If a valid buffer exists (happens if there is no dispatch),
		 * reuse the buffer, else allocate.
		 */
		if (droq->recv_buf_list[droq->refill_idx].buffer == NULL) {
			buf = rte_pktmbuf_alloc(droq->mpool);
			/* If a buffer could not be allocated, no point in
			 * continuing
			 */
			if (buf == NULL) {
				droq->stats.rx_alloc_failure++;
				break;
			}

			droq->recv_buf_list[droq->refill_idx].buffer = buf;
		}

		desc_ring[droq->refill_idx].buffer_ptr =
		    lio_map_ring(droq->recv_buf_list[droq->refill_idx].buffer);
		/* Reset any previous values in the length field. */
		droq->info_list[droq->refill_idx].length = 0;

		droq->refill_idx = lio_incr_index(droq->refill_idx, 1,
						  droq->max_count);
		desc_refilled++;
		droq->refill_count--;
	}

	if (droq->refill_count)
		desc_refilled += lio_droq_refill_pullup_descs(droq, desc_ring);

	/* if droq->refill_count
	 * The refill count would not change in pass two. We only moved buffers
	 * to close the gap in the ring, but we would still have the same no. of
	 * buffers to refill.
	 */
	return desc_refilled;
}

static int
lio_droq_fast_process_packet(struct lio_device *lio_dev,
			     struct lio_droq *droq,
			     struct rte_mbuf **rx_pkts)
{
	struct rte_mbuf *nicbuf = NULL;
	struct lio_droq_info *info;
	uint32_t total_len = 0;
	int data_total_len = 0;
	uint32_t pkt_len = 0;
	union octeon_rh *rh;
	int data_pkts = 0;

	info = &droq->info_list[droq->read_idx];
	lio_swap_8B_data((uint64_t *)info, 2);

	if (!info->length)
		return -1;

	/* Len of resp hdr in included in the received data len. */
	info->length -= OCTEON_RH_SIZE;
	rh = &info->rh;

	total_len += (uint32_t)info->length;

	if (lio_opcode_slow_path(rh)) {
		uint32_t buf_cnt;

		buf_cnt = lio_droq_get_bufcount(droq->buffer_size,
						(uint32_t)info->length);
		droq->read_idx = lio_incr_index(droq->read_idx, buf_cnt,
						droq->max_count);
		droq->refill_count += buf_cnt;
	} else {
		if (info->length <= droq->buffer_size) {
			if (rh->r_dh.has_hash)
				pkt_len = (uint32_t)(info->length - 8);
			else
				pkt_len = (uint32_t)info->length;

			nicbuf = droq->recv_buf_list[droq->read_idx].buffer;
			droq->recv_buf_list[droq->read_idx].buffer = NULL;
			droq->read_idx = lio_incr_index(
						droq->read_idx, 1,
						droq->max_count);
			droq->refill_count++;

			if (likely(nicbuf != NULL)) {
				/* We don't have a way to pass flags yet */
				nicbuf->ol_flags = 0;
				if (rh->r_dh.has_hash) {
					uint64_t *hash_ptr;

					nicbuf->ol_flags |= PKT_RX_RSS_HASH;
					hash_ptr = rte_pktmbuf_mtod(nicbuf,
								    uint64_t *);
					lio_swap_8B_data(hash_ptr, 1);
					nicbuf->hash.rss = (uint32_t)*hash_ptr;
					nicbuf->data_off += 8;
				}

				nicbuf->pkt_len = pkt_len;
				nicbuf->data_len = pkt_len;
				nicbuf->port = lio_dev->port_id;
				/* Store the mbuf */
				rx_pkts[data_pkts++] = nicbuf;
				data_total_len += pkt_len;
			}

			/* Prefetch buffer pointers when on a cache line
			 * boundary
			 */
			if ((droq->read_idx & 3) == 0) {
				rte_prefetch0(
				    &droq->recv_buf_list[droq->read_idx]);
				rte_prefetch0(
				    &droq->info_list[droq->read_idx]);
			}
		} else {
			struct rte_mbuf *first_buf = NULL;
			struct rte_mbuf *last_buf = NULL;

			while (pkt_len < info->length) {
				int cpy_len = 0;

				cpy_len = ((pkt_len + droq->buffer_size) >
						info->length)
						? ((uint32_t)info->length -
							pkt_len)
						: droq->buffer_size;

				nicbuf =
				    droq->recv_buf_list[droq->read_idx].buffer;
				droq->recv_buf_list[droq->read_idx].buffer =
				    NULL;

				if (likely(nicbuf != NULL)) {
					/* Note the first seg */
					if (!pkt_len)
						first_buf = nicbuf;

					nicbuf->port = lio_dev->port_id;
					/* We don't have a way to pass
					 * flags yet
					 */
					nicbuf->ol_flags = 0;
					if ((!pkt_len) && (rh->r_dh.has_hash)) {
						uint64_t *hash_ptr;

						nicbuf->ol_flags |=
						    PKT_RX_RSS_HASH;
						hash_ptr = rte_pktmbuf_mtod(
						    nicbuf, uint64_t *);
						lio_swap_8B_data(hash_ptr, 1);
						nicbuf->hash.rss =
						    (uint32_t)*hash_ptr;
						nicbuf->data_off += 8;
						nicbuf->pkt_len = cpy_len - 8;
						nicbuf->data_len = cpy_len - 8;
					} else {
						nicbuf->pkt_len = cpy_len;
						nicbuf->data_len = cpy_len;
					}

					if (pkt_len)
						first_buf->nb_segs++;

					if (last_buf)
						last_buf->next = nicbuf;

					last_buf = nicbuf;
				} else {
					PMD_RX_LOG(lio_dev, ERR, "no buf\n");
				}

				pkt_len += cpy_len;
				droq->read_idx = lio_incr_index(
							droq->read_idx,
							1, droq->max_count);
				droq->refill_count++;

				/* Prefetch buffer pointers when on a
				 * cache line boundary
				 */
				if ((droq->read_idx & 3) == 0) {
					rte_prefetch0(&droq->recv_buf_list
							      [droq->read_idx]);

					rte_prefetch0(
					    &droq->info_list[droq->read_idx]);
				}
			}
			rx_pkts[data_pkts++] = first_buf;
			if (rh->r_dh.has_hash)
				data_total_len += (pkt_len - 8);
			else
				data_total_len += pkt_len;
		}

		/* Inform upper layer about packet checksum verification */
		struct rte_mbuf *m = rx_pkts[data_pkts - 1];

		if (rh->r_dh.csum_verified & LIO_IP_CSUM_VERIFIED)
			m->ol_flags |= PKT_RX_IP_CKSUM_GOOD;

		if (rh->r_dh.csum_verified & LIO_L4_CSUM_VERIFIED)
			m->ol_flags |= PKT_RX_L4_CKSUM_GOOD;
	}

	if (droq->refill_count >= droq->refill_threshold) {
		int desc_refilled = lio_droq_refill(droq);

		/* Flush the droq descriptor data to memory to be sure
		 * that when we update the credits the data in memory is
		 * accurate.
		 */
		rte_wmb();
		rte_write32(desc_refilled, droq->pkts_credit_reg);
		/* make sure mmio write completes */
		rte_wmb();
	}

	info->length = 0;
	info->rh.rh64 = 0;

	droq->stats.pkts_received++;
	droq->stats.rx_pkts_received += data_pkts;
	droq->stats.rx_bytes_received += data_total_len;
	droq->stats.bytes_received += total_len;

	return data_pkts;
}

static uint32_t
lio_droq_fast_process_packets(struct lio_device *lio_dev,
			      struct lio_droq *droq,
			      struct rte_mbuf **rx_pkts,
			      uint32_t pkts_to_process)
{
	int ret, data_pkts = 0;
	uint32_t pkt;

	for (pkt = 0; pkt < pkts_to_process; pkt++) {
		ret = lio_droq_fast_process_packet(lio_dev, droq,
						   &rx_pkts[data_pkts]);
		if (ret < 0) {
			lio_dev_err(lio_dev, "Port[%d] DROQ[%d] idx: %d len:0, pkt_cnt: %d\n",
				    lio_dev->port_id, droq->q_no,
				    droq->read_idx, pkts_to_process);
			break;
		}
		data_pkts += ret;
	}

	rte_atomic64_sub(&droq->pkts_pending, pkt);

	return data_pkts;
}

static inline uint32_t
lio_droq_check_hw_for_pkts(struct lio_droq *droq)
{
	uint32_t last_count;
	uint32_t pkt_count;

	pkt_count = rte_read32(droq->pkts_sent_reg);

	last_count = pkt_count - droq->pkt_count;
	droq->pkt_count = pkt_count;

	if (last_count)
		rte_atomic64_add(&droq->pkts_pending, last_count);

	return last_count;
}

uint16_t
lio_dev_recv_pkts(void *rx_queue,
		  struct rte_mbuf **rx_pkts,
		  uint16_t budget)
{
	struct lio_droq *droq = rx_queue;
	struct lio_device *lio_dev = droq->lio_dev;
	uint32_t pkts_processed = 0;
	uint32_t pkt_count = 0;

	lio_droq_check_hw_for_pkts(droq);

	pkt_count = rte_atomic64_read(&droq->pkts_pending);
	if (!pkt_count)
		return 0;

	if (pkt_count > budget)
		pkt_count = budget;

	/* Grab the lock */
	rte_spinlock_lock(&droq->lock);
	pkts_processed = lio_droq_fast_process_packets(lio_dev,
						       droq, rx_pkts,
						       pkt_count);

	if (droq->pkt_count) {
		rte_write32(droq->pkt_count, droq->pkts_sent_reg);
		droq->pkt_count = 0;
	}

	/* Release the spin lock */
	rte_spinlock_unlock(&droq->lock);

	return pkts_processed;
}

void
lio_delete_droq_queue(struct lio_device *lio_dev,
		      int oq_no)
{
	lio_delete_droq(lio_dev, oq_no);
	lio_dev->num_oqs--;
	rte_free(lio_dev->droq[oq_no]);
	lio_dev->droq[oq_no] = NULL;
}

/**
 *  lio_init_instr_queue()
 *  @param lio_dev	- pointer to the lio device structure.
 *  @param txpciq	- queue to be initialized.
 *
 *  Called at driver init time for each input queue. iq_conf has the
 *  configuration parameters for the queue.
 *
 *  @return  Success: 0	Failure: -1
 */
static int
lio_init_instr_queue(struct lio_device *lio_dev,
		     union octeon_txpciq txpciq,
		     uint32_t num_descs, unsigned int socket_id)
{
	uint32_t iq_no = (uint32_t)txpciq.s.q_no;
	struct lio_instr_queue *iq;
	uint32_t instr_type;
	uint32_t q_size;

	instr_type = LIO_IQ_INSTR_TYPE(lio_dev);

	q_size = instr_type * num_descs;
	iq = lio_dev->instr_queue[iq_no];
	iq->iq_mz = rte_eth_dma_zone_reserve(lio_dev->eth_dev,
					     "instr_queue", iq_no, q_size,
					     RTE_CACHE_LINE_SIZE,
					     socket_id);
	if (iq->iq_mz == NULL) {
		lio_dev_err(lio_dev, "Cannot allocate memory for instr queue %d\n",
			    iq_no);
		return -1;
	}

	iq->base_addr_dma = iq->iq_mz->iova;
	iq->base_addr = (uint8_t *)iq->iq_mz->addr;

	iq->max_count = num_descs;

	/* Initialize a list to holds requests that have been posted to Octeon
	 * but has yet to be fetched by octeon
	 */
	iq->request_list = rte_zmalloc_socket("request_list",
					      sizeof(*iq->request_list) *
							num_descs,
					      RTE_CACHE_LINE_SIZE,
					      socket_id);
	if (iq->request_list == NULL) {
		lio_dev_err(lio_dev, "Alloc failed for IQ[%d] nr free list\n",
			    iq_no);
		lio_dma_zone_free(lio_dev, iq->iq_mz);
		return -1;
	}

	lio_dev_dbg(lio_dev, "IQ[%d]: base: %p basedma: %lx count: %d\n",
		    iq_no, iq->base_addr, (unsigned long)iq->base_addr_dma,
		    iq->max_count);

	iq->lio_dev = lio_dev;
	iq->txpciq.txpciq64 = txpciq.txpciq64;
	iq->fill_cnt = 0;
	iq->host_write_index = 0;
	iq->lio_read_index = 0;
	iq->flush_index = 0;

	rte_atomic64_set(&iq->instr_pending, 0);

	/* Initialize the spinlock for this instruction queue */
	rte_spinlock_init(&iq->lock);
	rte_spinlock_init(&iq->post_lock);

	rte_atomic64_clear(&iq->iq_flush_running);

	lio_dev->io_qmask.iq |= (1ULL << iq_no);

	/* Set the 32B/64B mode for each input queue */
	lio_dev->io_qmask.iq64B |= ((instr_type == 64) << iq_no);
	iq->iqcmd_64B = (instr_type == 64);

	lio_dev->fn_list.setup_iq_regs(lio_dev, iq_no);

	return 0;
}

int
lio_setup_instr_queue0(struct lio_device *lio_dev)
{
	union octeon_txpciq txpciq;
	uint32_t num_descs = 0;
	uint32_t iq_no = 0;

	num_descs = LIO_NUM_DEF_TX_DESCS_CFG(lio_dev);

	lio_dev->num_iqs = 0;

	lio_dev->instr_queue[0] = rte_zmalloc(NULL,
					sizeof(struct lio_instr_queue), 0);
	if (lio_dev->instr_queue[0] == NULL)
		return -ENOMEM;

	lio_dev->instr_queue[0]->q_index = 0;
	lio_dev->instr_queue[0]->app_ctx = (void *)(size_t)0;
	txpciq.txpciq64 = 0;
	txpciq.s.q_no = iq_no;
	txpciq.s.pkind = lio_dev->pfvf_hsword.pkind;
	txpciq.s.use_qpg = 0;
	txpciq.s.qpg = 0;
	if (lio_init_instr_queue(lio_dev, txpciq, num_descs, SOCKET_ID_ANY)) {
		rte_free(lio_dev->instr_queue[0]);
		lio_dev->instr_queue[0] = NULL;
		return -1;
	}

	lio_dev->num_iqs++;

	return 0;
}

/**
 *  lio_delete_instr_queue()
 *  @param lio_dev	- pointer to the lio device structure.
 *  @param iq_no	- queue to be deleted.
 *
 *  Called at driver unload time for each input queue. Deletes all
 *  allocated resources for the input queue.
 */
static void
lio_delete_instr_queue(struct lio_device *lio_dev, uint32_t iq_no)
{
	struct lio_instr_queue *iq = lio_dev->instr_queue[iq_no];

	rte_free(iq->request_list);
	iq->request_list = NULL;
	lio_dma_zone_free(lio_dev, iq->iq_mz);
}

void
lio_free_instr_queue0(struct lio_device *lio_dev)
{
	lio_delete_instr_queue(lio_dev, 0);
	rte_free(lio_dev->instr_queue[0]);
	lio_dev->instr_queue[0] = NULL;
	lio_dev->num_iqs--;
}

/* Return 0 on success, -1 on failure */
int
lio_setup_iq(struct lio_device *lio_dev, int q_index,
	     union octeon_txpciq txpciq, uint32_t num_descs, void *app_ctx,
	     unsigned int socket_id)
{
	uint32_t iq_no = (uint32_t)txpciq.s.q_no;

	if (lio_dev->instr_queue[iq_no]) {
		lio_dev_dbg(lio_dev, "IQ is in use. Cannot create the IQ: %d again\n",
			    iq_no);
		lio_dev->instr_queue[iq_no]->txpciq.txpciq64 = txpciq.txpciq64;
		lio_dev->instr_queue[iq_no]->app_ctx = app_ctx;
		return 0;
	}

	lio_dev->instr_queue[iq_no] = rte_zmalloc_socket("ethdev TX queue",
						sizeof(struct lio_instr_queue),
						RTE_CACHE_LINE_SIZE, socket_id);
	if (lio_dev->instr_queue[iq_no] == NULL)
		return -1;

	lio_dev->instr_queue[iq_no]->q_index = q_index;
	lio_dev->instr_queue[iq_no]->app_ctx = app_ctx;

	if (lio_init_instr_queue(lio_dev, txpciq, num_descs, socket_id))
		goto release_lio_iq;

	lio_dev->num_iqs++;
	if (lio_dev->fn_list.enable_io_queues(lio_dev))
		goto delete_lio_iq;

	return 0;

delete_lio_iq:
	lio_delete_instr_queue(lio_dev, iq_no);
	lio_dev->num_iqs--;
release_lio_iq:
	rte_free(lio_dev->instr_queue[iq_no]);
	lio_dev->instr_queue[iq_no] = NULL;

	return -1;
}

int
lio_wait_for_instr_fetch(struct lio_device *lio_dev)
{
	int pending, instr_cnt;
	int i, retry = 1000;

	do {
		instr_cnt = 0;

		for (i = 0; i < LIO_MAX_INSTR_QUEUES(lio_dev); i++) {
			if (!(lio_dev->io_qmask.iq & (1ULL << i)))
				continue;

			if (lio_dev->instr_queue[i] == NULL)
				break;

			pending = rte_atomic64_read(
			    &lio_dev->instr_queue[i]->instr_pending);
			if (pending)
				lio_flush_iq(lio_dev, lio_dev->instr_queue[i]);

			instr_cnt += pending;
		}

		if (instr_cnt == 0)
			break;

		rte_delay_ms(1);

	} while (retry-- && instr_cnt);

	return instr_cnt;
}

static inline void
lio_ring_doorbell(struct lio_device *lio_dev,
		  struct lio_instr_queue *iq)
{
	if (rte_atomic64_read(&lio_dev->status) == LIO_DEV_RUNNING) {
		rte_write32(iq->fill_cnt, iq->doorbell_reg);
		/* make sure doorbell write goes through */
		rte_wmb();
		iq->fill_cnt = 0;
	}
}

static inline void
copy_cmd_into_iq(struct lio_instr_queue *iq, uint8_t *cmd)
{
	uint8_t *iqptr, cmdsize;

	cmdsize = ((iq->iqcmd_64B) ? 64 : 32);
	iqptr = iq->base_addr + (cmdsize * iq->host_write_index);

	rte_memcpy(iqptr, cmd, cmdsize);
}

static inline struct lio_iq_post_status
post_command2(struct lio_instr_queue *iq, uint8_t *cmd)
{
	struct lio_iq_post_status st;

	st.status = LIO_IQ_SEND_OK;

	/* This ensures that the read index does not wrap around to the same
	 * position if queue gets full before Octeon could fetch any instr.
	 */
	if (rte_atomic64_read(&iq->instr_pending) >=
			(int32_t)(iq->max_count - 1)) {
		st.status = LIO_IQ_SEND_FAILED;
		st.index = -1;
		return st;
	}

	if (rte_atomic64_read(&iq->instr_pending) >=
			(int32_t)(iq->max_count - 2))
		st.status = LIO_IQ_SEND_STOP;

	copy_cmd_into_iq(iq, cmd);

	/* "index" is returned, host_write_index is modified. */
	st.index = iq->host_write_index;
	iq->host_write_index = lio_incr_index(iq->host_write_index, 1,
					      iq->max_count);
	iq->fill_cnt++;

	/* Flush the command into memory. We need to be sure the data is in
	 * memory before indicating that the instruction is pending.
	 */
	rte_wmb();

	rte_atomic64_inc(&iq->instr_pending);

	return st;
}

static inline void
lio_add_to_request_list(struct lio_instr_queue *iq,
			int idx, void *buf, int reqtype)
{
	iq->request_list[idx].buf = buf;
	iq->request_list[idx].reqtype = reqtype;
}

static inline void
lio_free_netsgbuf(void *buf)
{
	struct lio_buf_free_info *finfo = buf;
	struct lio_device *lio_dev = finfo->lio_dev;
	struct rte_mbuf *m = finfo->mbuf;
	struct lio_gather *g = finfo->g;
	uint8_t iq = finfo->iq_no;

	/* This will take care of multiple segments also */
	rte_pktmbuf_free(m);

	rte_spinlock_lock(&lio_dev->glist_lock[iq]);
	STAILQ_INSERT_TAIL(&lio_dev->glist_head[iq], &g->list, entries);
	rte_spinlock_unlock(&lio_dev->glist_lock[iq]);
	rte_free(finfo);
}

/* Can only run in process context */
static int
lio_process_iq_request_list(struct lio_device *lio_dev,
			    struct lio_instr_queue *iq)
{
	struct octeon_instr_irh *irh = NULL;
	uint32_t old = iq->flush_index;
	struct lio_soft_command *sc;
	uint32_t inst_count = 0;
	int reqtype;
	void *buf;

	while (old != iq->lio_read_index) {
		reqtype = iq->request_list[old].reqtype;
		buf     = iq->request_list[old].buf;

		if (reqtype == LIO_REQTYPE_NONE)
			goto skip_this;

		switch (reqtype) {
		case LIO_REQTYPE_NORESP_NET:
			rte_pktmbuf_free((struct rte_mbuf *)buf);
			break;
		case LIO_REQTYPE_NORESP_NET_SG:
			lio_free_netsgbuf(buf);
			break;
		case LIO_REQTYPE_SOFT_COMMAND:
			sc = buf;
			irh = (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
			if (irh->rflag) {
				/* We're expecting a response from Octeon.
				 * It's up to lio_process_ordered_list() to
				 * process sc. Add sc to the ordered soft
				 * command response list because we expect
				 * a response from Octeon.
				 */
				rte_spinlock_lock(&lio_dev->response_list.lock);
				rte_atomic64_inc(
				    &lio_dev->response_list.pending_req_count);
				STAILQ_INSERT_TAIL(
					&lio_dev->response_list.head,
					&sc->node, entries);
				rte_spinlock_unlock(
						&lio_dev->response_list.lock);
			} else {
				if (sc->callback) {
					/* This callback must not sleep */
					sc->callback(LIO_REQUEST_DONE,
						     sc->callback_arg);
				}
			}
			break;
		default:
			lio_dev_err(lio_dev,
				    "Unknown reqtype: %d buf: %p at idx %d\n",
				    reqtype, buf, old);
		}

		iq->request_list[old].buf = NULL;
		iq->request_list[old].reqtype = 0;

skip_this:
		inst_count++;
		old = lio_incr_index(old, 1, iq->max_count);
	}

	iq->flush_index = old;

	return inst_count;
}

static void
lio_update_read_index(struct lio_instr_queue *iq)
{
	uint32_t pkt_in_done = rte_read32(iq->inst_cnt_reg);
	uint32_t last_done;

	last_done = pkt_in_done - iq->pkt_in_done;
	iq->pkt_in_done = pkt_in_done;

	/* Add last_done and modulo with the IQ size to get new index */
	iq->lio_read_index = (iq->lio_read_index +
			(uint32_t)(last_done & LIO_PKT_IN_DONE_CNT_MASK)) %
			iq->max_count;
}

int
lio_flush_iq(struct lio_device *lio_dev, struct lio_instr_queue *iq)
{
	uint32_t tot_inst_processed = 0;
	uint32_t inst_processed = 0;
	int tx_done = 1;

	if (rte_atomic64_test_and_set(&iq->iq_flush_running) == 0)
		return tx_done;

	rte_spinlock_lock(&iq->lock);

	lio_update_read_index(iq);

	do {
		/* Process any outstanding IQ packets. */
		if (iq->flush_index == iq->lio_read_index)
			break;

		inst_processed = lio_process_iq_request_list(lio_dev, iq);

		if (inst_processed) {
			rte_atomic64_sub(&iq->instr_pending, inst_processed);
			iq->stats.instr_processed += inst_processed;
		}

		tot_inst_processed += inst_processed;
		inst_processed = 0;

	} while (1);

	rte_spinlock_unlock(&iq->lock);

	rte_atomic64_clear(&iq->iq_flush_running);

	return tx_done;
}

static int
lio_send_command(struct lio_device *lio_dev, uint32_t iq_no, void *cmd,
		 void *buf, uint32_t datasize, uint32_t reqtype)
{
	struct lio_instr_queue *iq = lio_dev->instr_queue[iq_no];
	struct lio_iq_post_status st;

	rte_spinlock_lock(&iq->post_lock);

	st = post_command2(iq, cmd);

	if (st.status != LIO_IQ_SEND_FAILED) {
		lio_add_to_request_list(iq, st.index, buf, reqtype);
		LIO_INCR_INSTRQUEUE_PKT_COUNT(lio_dev, iq_no, bytes_sent,
					      datasize);
		LIO_INCR_INSTRQUEUE_PKT_COUNT(lio_dev, iq_no, instr_posted, 1);

		lio_ring_doorbell(lio_dev, iq);
	} else {
		LIO_INCR_INSTRQUEUE_PKT_COUNT(lio_dev, iq_no, instr_dropped, 1);
	}

	rte_spinlock_unlock(&iq->post_lock);

	return st.status;
}

void
lio_prepare_soft_command(struct lio_device *lio_dev,
			 struct lio_soft_command *sc, uint8_t opcode,
			 uint8_t subcode, uint32_t irh_ossp, uint64_t ossp0,
			 uint64_t ossp1)
{
	struct octeon_instr_pki_ih3 *pki_ih3;
	struct octeon_instr_ih3 *ih3;
	struct octeon_instr_irh *irh;
	struct octeon_instr_rdp *rdp;

	RTE_ASSERT(opcode <= 15);
	RTE_ASSERT(subcode <= 127);

	ih3	  = (struct octeon_instr_ih3 *)&sc->cmd.cmd3.ih3;

	ih3->pkind = lio_dev->instr_queue[sc->iq_no]->txpciq.s.pkind;

	pki_ih3 = (struct octeon_instr_pki_ih3 *)&sc->cmd.cmd3.pki_ih3;

	pki_ih3->w	= 1;
	pki_ih3->raw	= 1;
	pki_ih3->utag	= 1;
	pki_ih3->uqpg	= lio_dev->instr_queue[sc->iq_no]->txpciq.s.use_qpg;
	pki_ih3->utt	= 1;

	pki_ih3->tag	= LIO_CONTROL;
	pki_ih3->tagtype = OCTEON_ATOMIC_TAG;
	pki_ih3->qpg	= lio_dev->instr_queue[sc->iq_no]->txpciq.s.qpg;
	pki_ih3->pm	= 0x7;
	pki_ih3->sl	= 8;

	if (sc->datasize)
		ih3->dlengsz = sc->datasize;

	irh		= (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
	irh->opcode	= opcode;
	irh->subcode	= subcode;

	/* opcode/subcode specific parameters (ossp) */
	irh->ossp = irh_ossp;
	sc->cmd.cmd3.ossp[0] = ossp0;
	sc->cmd.cmd3.ossp[1] = ossp1;

	if (sc->rdatasize) {
		rdp = (struct octeon_instr_rdp *)&sc->cmd.cmd3.rdp;
		rdp->pcie_port = lio_dev->pcie_port;
		rdp->rlen      = sc->rdatasize;
		irh->rflag = 1;
		/* PKI IH3 */
		ih3->fsz    = OCTEON_SOFT_CMD_RESP_IH3;
	} else {
		irh->rflag = 0;
		/* PKI IH3 */
		ih3->fsz    = OCTEON_PCI_CMD_O3;
	}
}

int
lio_send_soft_command(struct lio_device *lio_dev,
		      struct lio_soft_command *sc)
{
	struct octeon_instr_ih3 *ih3;
	struct octeon_instr_irh *irh;
	uint32_t len = 0;

	ih3 = (struct octeon_instr_ih3 *)&sc->cmd.cmd3.ih3;
	if (ih3->dlengsz) {
		RTE_ASSERT(sc->dmadptr);
		sc->cmd.cmd3.dptr = sc->dmadptr;
	}

	irh = (struct octeon_instr_irh *)&sc->cmd.cmd3.irh;
	if (irh->rflag) {
		RTE_ASSERT(sc->dmarptr);
		RTE_ASSERT(sc->status_word != NULL);
		*sc->status_word = LIO_COMPLETION_WORD_INIT;
		sc->cmd.cmd3.rptr = sc->dmarptr;
	}

	len = (uint32_t)ih3->dlengsz;

	if (sc->wait_time)
		sc->timeout = lio_uptime + sc->wait_time;

	return lio_send_command(lio_dev, sc->iq_no, &sc->cmd, sc, len,
				LIO_REQTYPE_SOFT_COMMAND);
}

int
lio_setup_sc_buffer_pool(struct lio_device *lio_dev)
{
	char sc_pool_name[RTE_MEMPOOL_NAMESIZE];
	uint16_t buf_size;

	buf_size = LIO_SOFT_COMMAND_BUFFER_SIZE + RTE_PKTMBUF_HEADROOM;
	snprintf(sc_pool_name, sizeof(sc_pool_name),
		 "lio_sc_pool_%u", lio_dev->port_id);
	lio_dev->sc_buf_pool = rte_pktmbuf_pool_create(sc_pool_name,
						LIO_MAX_SOFT_COMMAND_BUFFERS,
						0, 0, buf_size, SOCKET_ID_ANY);
	return 0;
}

void
lio_free_sc_buffer_pool(struct lio_device *lio_dev)
{
	rte_mempool_free(lio_dev->sc_buf_pool);
}

struct lio_soft_command *
lio_alloc_soft_command(struct lio_device *lio_dev, uint32_t datasize,
		       uint32_t rdatasize, uint32_t ctxsize)
{
	uint32_t offset = sizeof(struct lio_soft_command);
	struct lio_soft_command *sc;
	struct rte_mbuf *m;
	uint64_t dma_addr;

	RTE_ASSERT((offset + datasize + rdatasize + ctxsize) <=
		   LIO_SOFT_COMMAND_BUFFER_SIZE);

	m = rte_pktmbuf_alloc(lio_dev->sc_buf_pool);
	if (m == NULL) {
		lio_dev_err(lio_dev, "Cannot allocate mbuf for sc\n");
		return NULL;
	}

	/* set rte_mbuf data size and there is only 1 segment */
	m->pkt_len = LIO_SOFT_COMMAND_BUFFER_SIZE;
	m->data_len = LIO_SOFT_COMMAND_BUFFER_SIZE;

	/* use rte_mbuf buffer for soft command */
	sc = rte_pktmbuf_mtod(m, struct lio_soft_command *);
	memset(sc, 0, LIO_SOFT_COMMAND_BUFFER_SIZE);
	sc->size = LIO_SOFT_COMMAND_BUFFER_SIZE;
	sc->dma_addr = rte_mbuf_data_iova(m);
	sc->mbuf = m;

	dma_addr = sc->dma_addr;

	if (ctxsize) {
		sc->ctxptr = (uint8_t *)sc + offset;
		sc->ctxsize = ctxsize;
	}

	/* Start data at 128 byte boundary */
	offset = (offset + ctxsize + 127) & 0xffffff80;

	if (datasize) {
		sc->virtdptr = (uint8_t *)sc + offset;
		sc->dmadptr = dma_addr + offset;
		sc->datasize = datasize;
	}

	/* Start rdata at 128 byte boundary */
	offset = (offset + datasize + 127) & 0xffffff80;

	if (rdatasize) {
		RTE_ASSERT(rdatasize >= 16);
		sc->virtrptr = (uint8_t *)sc + offset;
		sc->dmarptr = dma_addr + offset;
		sc->rdatasize = rdatasize;
		sc->status_word = (uint64_t *)((uint8_t *)(sc->virtrptr) +
					       rdatasize - 8);
	}

	return sc;
}

void
lio_free_soft_command(struct lio_soft_command *sc)
{
	rte_pktmbuf_free(sc->mbuf);
}

void
lio_setup_response_list(struct lio_device *lio_dev)
{
	STAILQ_INIT(&lio_dev->response_list.head);
	rte_spinlock_init(&lio_dev->response_list.lock);
	rte_atomic64_set(&lio_dev->response_list.pending_req_count, 0);
}

int
lio_process_ordered_list(struct lio_device *lio_dev)
{
	int resp_to_process = LIO_MAX_ORD_REQS_TO_PROCESS;
	struct lio_response_list *ordered_sc_list;
	struct lio_soft_command *sc;
	int request_complete = 0;
	uint64_t status64;
	uint32_t status;

	ordered_sc_list = &lio_dev->response_list;

	do {
		rte_spinlock_lock(&ordered_sc_list->lock);

		if (STAILQ_EMPTY(&ordered_sc_list->head)) {
			/* ordered_sc_list is empty; there is
			 * nothing to process
			 */
			rte_spinlock_unlock(&ordered_sc_list->lock);
			return -1;
		}

		sc = LIO_STQUEUE_FIRST_ENTRY(&ordered_sc_list->head,
					     struct lio_soft_command, node);

		status = LIO_REQUEST_PENDING;

		/* check if octeon has finished DMA'ing a response
		 * to where rptr is pointing to
		 */
		status64 = *sc->status_word;

		if (status64 != LIO_COMPLETION_WORD_INIT) {
			/* This logic ensures that all 64b have been written.
			 * 1. check byte 0 for non-FF
			 * 2. if non-FF, then swap result from BE to host order
			 * 3. check byte 7 (swapped to 0) for non-FF
			 * 4. if non-FF, use the low 32-bit status code
			 * 5. if either byte 0 or byte 7 is FF, don't use status
			 */
			if ((status64 & 0xff) != 0xff) {
				lio_swap_8B_data(&status64, 1);
				if (((status64 & 0xff) != 0xff)) {
					/* retrieve 16-bit firmware status */
					status = (uint32_t)(status64 &
							    0xffffULL);
					if (status) {
						status =
						LIO_FIRMWARE_STATUS_CODE(
									status);
					} else {
						/* i.e. no error */
						status = LIO_REQUEST_DONE;
					}
				}
			}
		} else if ((sc->timeout && lio_check_timeout(lio_uptime,
							     sc->timeout))) {
			lio_dev_err(lio_dev,
				    "cmd failed, timeout (%ld, %ld)\n",
				    (long)lio_uptime, (long)sc->timeout);
			status = LIO_REQUEST_TIMEOUT;
		}

		if (status != LIO_REQUEST_PENDING) {
			/* we have received a response or we have timed out.
			 * remove node from linked list
			 */
			STAILQ_REMOVE(&ordered_sc_list->head,
				      &sc->node, lio_stailq_node, entries);
			rte_atomic64_dec(
			    &lio_dev->response_list.pending_req_count);
			rte_spinlock_unlock(&ordered_sc_list->lock);

			if (sc->callback)
				sc->callback(status, sc->callback_arg);

			request_complete++;
		} else {
			/* no response yet */
			request_complete = 0;
			rte_spinlock_unlock(&ordered_sc_list->lock);
		}

		/* If we hit the Max Ordered requests to process every loop,
		 * we quit and let this function be invoked the next time
		 * the poll thread runs to process the remaining requests.
		 * This function can take up the entire CPU if there is
		 * no upper limit to the requests processed.
		 */
		if (request_complete >= resp_to_process)
			break;
	} while (request_complete);

	return 0;
}

static inline struct lio_stailq_node *
list_delete_first_node(struct lio_stailq_head *head)
{
	struct lio_stailq_node *node;

	if (STAILQ_EMPTY(head))
		node = NULL;
	else
		node = STAILQ_FIRST(head);

	if (node)
		STAILQ_REMOVE(head, node, lio_stailq_node, entries);

	return node;
}

void
lio_delete_sglist(struct lio_instr_queue *txq)
{
	struct lio_device *lio_dev = txq->lio_dev;
	int iq_no = txq->q_index;
	struct lio_gather *g;

	if (lio_dev->glist_head == NULL)
		return;

	do {
		g = (struct lio_gather *)list_delete_first_node(
						&lio_dev->glist_head[iq_no]);
		if (g) {
			if (g->sg)
				rte_free(
				    (void *)((unsigned long)g->sg - g->adjust));
			rte_free(g);
		}
	} while (g);
}

/**
 * \brief Setup gather lists
 * @param lio per-network private data
 */
int
lio_setup_sglists(struct lio_device *lio_dev, int iq_no,
		  int fw_mapped_iq, int num_descs, unsigned int socket_id)
{
	struct lio_gather *g;
	int i;

	rte_spinlock_init(&lio_dev->glist_lock[iq_no]);

	STAILQ_INIT(&lio_dev->glist_head[iq_no]);

	for (i = 0; i < num_descs; i++) {
		g = rte_zmalloc_socket(NULL, sizeof(*g), RTE_CACHE_LINE_SIZE,
				       socket_id);
		if (g == NULL) {
			lio_dev_err(lio_dev,
				    "lio_gather memory allocation failed for qno %d\n",
				    iq_no);
			break;
		}

		g->sg_size =
		    ((ROUNDUP4(LIO_MAX_SG) >> 2) * LIO_SG_ENTRY_SIZE);

		g->sg = rte_zmalloc_socket(NULL, g->sg_size + 8,
					   RTE_CACHE_LINE_SIZE, socket_id);
		if (g->sg == NULL) {
			lio_dev_err(lio_dev,
				    "sg list memory allocation failed for qno %d\n",
				    iq_no);
			rte_free(g);
			break;
		}

		/* The gather component should be aligned on 64-bit boundary */
		if (((unsigned long)g->sg) & 7) {
			g->adjust = 8 - (((unsigned long)g->sg) & 7);
			g->sg =
			    (struct lio_sg_entry *)((unsigned long)g->sg +
						       g->adjust);
		}

		STAILQ_INSERT_TAIL(&lio_dev->glist_head[iq_no], &g->list,
				   entries);
	}

	if (i != num_descs) {
		lio_delete_sglist(lio_dev->instr_queue[fw_mapped_iq]);
		return -ENOMEM;
	}

	return 0;
}

void
lio_delete_instruction_queue(struct lio_device *lio_dev, int iq_no)
{
	lio_delete_instr_queue(lio_dev, iq_no);
	rte_free(lio_dev->instr_queue[iq_no]);
	lio_dev->instr_queue[iq_no] = NULL;
	lio_dev->num_iqs--;
}

static inline uint32_t
lio_iq_get_available(struct lio_device *lio_dev, uint32_t q_no)
{
	return ((lio_dev->instr_queue[q_no]->max_count - 1) -
		(uint32_t)rte_atomic64_read(
				&lio_dev->instr_queue[q_no]->instr_pending));
}

static inline int
lio_iq_is_full(struct lio_device *lio_dev, uint32_t q_no)
{
	return ((uint32_t)rte_atomic64_read(
				&lio_dev->instr_queue[q_no]->instr_pending) >=
				(lio_dev->instr_queue[q_no]->max_count - 2));
}

static int
lio_dev_cleanup_iq(struct lio_device *lio_dev, int iq_no)
{
	struct lio_instr_queue *iq = lio_dev->instr_queue[iq_no];
	uint32_t count = 10000;

	while ((lio_iq_get_available(lio_dev, iq_no) < LIO_FLUSH_WM(iq)) &&
			--count)
		lio_flush_iq(lio_dev, iq);

	return count ? 0 : 1;
}

static void
lio_ctrl_cmd_callback(uint32_t status __rte_unused, void *sc_ptr)
{
	struct lio_soft_command *sc = sc_ptr;
	struct lio_dev_ctrl_cmd *ctrl_cmd;
	struct lio_ctrl_pkt *ctrl_pkt;

	ctrl_pkt = (struct lio_ctrl_pkt *)sc->ctxptr;
	ctrl_cmd = ctrl_pkt->ctrl_cmd;
	ctrl_cmd->cond = 1;

	lio_free_soft_command(sc);
}

static inline struct lio_soft_command *
lio_alloc_ctrl_pkt_sc(struct lio_device *lio_dev,
		      struct lio_ctrl_pkt *ctrl_pkt)
{
	struct lio_soft_command *sc = NULL;
	uint32_t uddsize, datasize;
	uint32_t rdatasize;
	uint8_t *data;

	uddsize = (uint32_t)(ctrl_pkt->ncmd.s.more * 8);

	datasize = OCTEON_CMD_SIZE + uddsize;
	rdatasize = (ctrl_pkt->wait_time) ? 16 : 0;

	sc = lio_alloc_soft_command(lio_dev, datasize,
				    rdatasize, sizeof(struct lio_ctrl_pkt));
	if (sc == NULL)
		return NULL;

	rte_memcpy(sc->ctxptr, ctrl_pkt, sizeof(struct lio_ctrl_pkt));

	data = (uint8_t *)sc->virtdptr;

	rte_memcpy(data, &ctrl_pkt->ncmd, OCTEON_CMD_SIZE);

	lio_swap_8B_data((uint64_t *)data, OCTEON_CMD_SIZE >> 3);

	if (uddsize) {
		/* Endian-Swap for UDD should have been done by caller. */
		rte_memcpy(data + OCTEON_CMD_SIZE, ctrl_pkt->udd, uddsize);
	}

	sc->iq_no = (uint32_t)ctrl_pkt->iq_no;

	lio_prepare_soft_command(lio_dev, sc,
				 LIO_OPCODE, LIO_OPCODE_CMD,
				 0, 0, 0);

	sc->callback = lio_ctrl_cmd_callback;
	sc->callback_arg = sc;
	sc->wait_time = ctrl_pkt->wait_time;

	return sc;
}

int
lio_send_ctrl_pkt(struct lio_device *lio_dev, struct lio_ctrl_pkt *ctrl_pkt)
{
	struct lio_soft_command *sc = NULL;
	int retval;

	sc = lio_alloc_ctrl_pkt_sc(lio_dev, ctrl_pkt);
	if (sc == NULL) {
		lio_dev_err(lio_dev, "soft command allocation failed\n");
		return -1;
	}

	retval = lio_send_soft_command(lio_dev, sc);
	if (retval == LIO_IQ_SEND_FAILED) {
		lio_free_soft_command(sc);
		lio_dev_err(lio_dev, "Port: %d soft command: %d send failed status: %x\n",
			    lio_dev->port_id, ctrl_pkt->ncmd.s.cmd, retval);
		return -1;
	}

	return retval;
}

/** Send data packet to the device
 *  @param lio_dev - lio device pointer
 *  @param ndata   - control structure with queueing, and buffer information
 *
 *  @returns IQ_FAILED if it failed to add to the input queue. IQ_STOP if it the
 *  queue should be stopped, and LIO_IQ_SEND_OK if it sent okay.
 */
static inline int
lio_send_data_pkt(struct lio_device *lio_dev, struct lio_data_pkt *ndata)
{
	return lio_send_command(lio_dev, ndata->q_no, &ndata->cmd,
				ndata->buf, ndata->datasize, ndata->reqtype);
}

uint16_t
lio_dev_xmit_pkts(void *tx_queue, struct rte_mbuf **pkts, uint16_t nb_pkts)
{
	struct lio_instr_queue *txq = tx_queue;
	union lio_cmd_setup cmdsetup;
	struct lio_device *lio_dev;
	struct lio_iq_stats *stats;
	struct lio_data_pkt ndata;
	int i, processed = 0;
	struct rte_mbuf *m;
	uint32_t tag = 0;
	int status = 0;
	int iq_no;

	lio_dev = txq->lio_dev;
	iq_no = txq->txpciq.s.q_no;
	stats = &lio_dev->instr_queue[iq_no]->stats;

	if (!lio_dev->intf_open || !lio_dev->linfo.link.s.link_up) {
		PMD_TX_LOG(lio_dev, ERR, "Transmit failed link_status : %d\n",
			   lio_dev->linfo.link.s.link_up);
		goto xmit_failed;
	}

	lio_dev_cleanup_iq(lio_dev, iq_no);

	for (i = 0; i < nb_pkts; i++) {
		uint32_t pkt_len = 0;

		m = pkts[i];

		/* Prepare the attributes for the data to be passed to BASE. */
		memset(&ndata, 0, sizeof(struct lio_data_pkt));

		ndata.buf = m;

		ndata.q_no = iq_no;
		if (lio_iq_is_full(lio_dev, ndata.q_no)) {
			stats->tx_iq_busy++;
			if (lio_dev_cleanup_iq(lio_dev, iq_no)) {
				PMD_TX_LOG(lio_dev, ERR,
					   "Transmit failed iq:%d full\n",
					   ndata.q_no);
				break;
			}
		}

		cmdsetup.cmd_setup64 = 0;
		cmdsetup.s.iq_no = iq_no;

		/* check checksum offload flags to form cmd */
		if (m->ol_flags & PKT_TX_IP_CKSUM)
			cmdsetup.s.ip_csum = 1;

		if (m->ol_flags & PKT_TX_OUTER_IP_CKSUM)
			cmdsetup.s.tnl_csum = 1;
		else if ((m->ol_flags & PKT_TX_TCP_CKSUM) ||
				(m->ol_flags & PKT_TX_UDP_CKSUM))
			cmdsetup.s.transport_csum = 1;

		if (m->nb_segs == 1) {
			pkt_len = rte_pktmbuf_data_len(m);
			cmdsetup.s.u.datasize = pkt_len;
			lio_prepare_pci_cmd(lio_dev, &ndata.cmd,
					    &cmdsetup, tag);
			ndata.cmd.cmd3.dptr = rte_mbuf_data_iova(m);
			ndata.reqtype = LIO_REQTYPE_NORESP_NET;
		} else {
			struct lio_buf_free_info *finfo;
			struct lio_gather *g;
			rte_iova_t phyaddr;
			int i, frags;

			finfo = (struct lio_buf_free_info *)rte_malloc(NULL,
							sizeof(*finfo), 0);
			if (finfo == NULL) {
				PMD_TX_LOG(lio_dev, ERR,
					   "free buffer alloc failed\n");
				goto xmit_failed;
			}

			rte_spinlock_lock(&lio_dev->glist_lock[iq_no]);
			g = (struct lio_gather *)list_delete_first_node(
						&lio_dev->glist_head[iq_no]);
			rte_spinlock_unlock(&lio_dev->glist_lock[iq_no]);
			if (g == NULL) {
				PMD_TX_LOG(lio_dev, ERR,
					   "Transmit scatter gather: glist null!\n");
				goto xmit_failed;
			}

			cmdsetup.s.gather = 1;
			cmdsetup.s.u.gatherptrs = m->nb_segs;
			lio_prepare_pci_cmd(lio_dev, &ndata.cmd,
					    &cmdsetup, tag);

			memset(g->sg, 0, g->sg_size);
			g->sg[0].ptr[0] = rte_mbuf_data_iova(m);
			lio_add_sg_size(&g->sg[0], m->data_len, 0);
			pkt_len = m->data_len;
			finfo->mbuf = m;

			/* First seg taken care above */
			frags = m->nb_segs - 1;
			i = 1;
			m = m->next;
			while (frags--) {
				g->sg[(i >> 2)].ptr[(i & 3)] =
						rte_mbuf_data_iova(m);
				lio_add_sg_size(&g->sg[(i >> 2)],
						m->data_len, (i & 3));
				pkt_len += m->data_len;
				i++;
				m = m->next;
			}

			phyaddr = rte_mem_virt2iova(g->sg);
			if (phyaddr == RTE_BAD_IOVA) {
				PMD_TX_LOG(lio_dev, ERR, "bad phys addr\n");
				goto xmit_failed;
			}

			ndata.cmd.cmd3.dptr = phyaddr;
			ndata.reqtype = LIO_REQTYPE_NORESP_NET_SG;

			finfo->g = g;
			finfo->lio_dev = lio_dev;
			finfo->iq_no = (uint64_t)iq_no;
			ndata.buf = finfo;
		}

		ndata.datasize = pkt_len;

		status = lio_send_data_pkt(lio_dev, &ndata);

		if (unlikely(status == LIO_IQ_SEND_FAILED)) {
			PMD_TX_LOG(lio_dev, ERR, "send failed\n");
			break;
		}

		if (unlikely(status == LIO_IQ_SEND_STOP)) {
			PMD_TX_LOG(lio_dev, DEBUG, "iq full\n");
			/* create space as iq is full */
			lio_dev_cleanup_iq(lio_dev, iq_no);
		}

		stats->tx_done++;
		stats->tx_tot_bytes += pkt_len;
		processed++;
	}

xmit_failed:
	stats->tx_dropped += (nb_pkts - processed);

	return processed;
}

void
lio_dev_clear_queues(struct rte_eth_dev *eth_dev)
{
	struct lio_instr_queue *txq;
	struct lio_droq *rxq;
	uint16_t i;

	for (i = 0; i < eth_dev->data->nb_tx_queues; i++) {
		txq = eth_dev->data->tx_queues[i];
		if (txq != NULL) {
			lio_dev_tx_queue_release(txq);
			eth_dev->data->tx_queues[i] = NULL;
		}
	}

	for (i = 0; i < eth_dev->data->nb_rx_queues; i++) {
		rxq = eth_dev->data->rx_queues[i];
		if (rxq != NULL) {
			lio_dev_rx_queue_release(rxq);
			eth_dev->data->rx_queues[i] = NULL;
		}
	}
}