summaryrefslogtreecommitdiff
path: root/drivers/scsi/csiostor/csio_wr.c
blob: 713e77d12cb41ffe8288b045bce2063ecd1abd5e (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
/*
 * This file is part of the Chelsio FCoE driver for Linux.
 *
 * Copyright (c) 2008-2012 Chelsio Communications, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - Redistributions in binary form must reproduce the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer in the documentation and/or other materials
 *        provided with the distribution.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/compiler.h>
#include <linux/slab.h>
#include <asm/page.h>
#include <linux/cache.h>

#include "csio_hw.h"
#include "csio_wr.h"
#include "csio_mb.h"
#include "csio_defs.h"

int csio_intr_coalesce_cnt;		/* value:SGE_INGRESS_RX_THRESHOLD[0] */
static int csio_sge_thresh_reg;		/* SGE_INGRESS_RX_THRESHOLD[0] */

int csio_intr_coalesce_time = 10;	/* value:SGE_TIMER_VALUE_1 */
static int csio_sge_timer_reg = 1;

#define CSIO_SET_FLBUF_SIZE(_hw, _reg, _val)				\
	csio_wr_reg32((_hw), (_val), SGE_FL_BUFFER_SIZE##_reg)

static void
csio_get_flbuf_size(struct csio_hw *hw, struct csio_sge *sge, uint32_t reg)
{
	sge->sge_fl_buf_size[reg] = csio_rd_reg32(hw, SGE_FL_BUFFER_SIZE0 +
							reg * sizeof(uint32_t));
}

/* Free list buffer size */
static inline uint32_t
csio_wr_fl_bufsz(struct csio_sge *sge, struct csio_dma_buf *buf)
{
	return sge->sge_fl_buf_size[buf->paddr & 0xF];
}

/* Size of the egress queue status page */
static inline uint32_t
csio_wr_qstat_pgsz(struct csio_hw *hw)
{
	return (hw->wrm.sge.sge_control & EGRSTATUSPAGESIZE(1)) ?  128 : 64;
}

/* Ring freelist doorbell */
static inline void
csio_wr_ring_fldb(struct csio_hw *hw, struct csio_q *flq)
{
	/*
	 * Ring the doorbell only when we have atleast CSIO_QCREDIT_SZ
	 * number of bytes in the freelist queue. This translates to atleast
	 * 8 freelist buffer pointers (since each pointer is 8 bytes).
	 */
	if (flq->inc_idx >= 8) {
		csio_wr_reg32(hw, DBPRIO(1) | QID(flq->un.fl.flid) |
			      PIDX(flq->inc_idx / 8),
			      MYPF_REG(SGE_PF_KDOORBELL));
		flq->inc_idx &= 7;
	}
}

/* Write a 0 cidx increment value to enable SGE interrupts for this queue */
static void
csio_wr_sge_intr_enable(struct csio_hw *hw, uint16_t iqid)
{
	csio_wr_reg32(hw, CIDXINC(0)		|
			  INGRESSQID(iqid)	|
			  TIMERREG(X_TIMERREG_RESTART_COUNTER),
			  MYPF_REG(SGE_PF_GTS));
}

/*
 * csio_wr_fill_fl - Populate the FL buffers of a FL queue.
 * @hw: HW module.
 * @flq: Freelist queue.
 *
 * Fill up freelist buffer entries with buffers of size specified
 * in the size register.
 *
 */
static int
csio_wr_fill_fl(struct csio_hw *hw, struct csio_q *flq)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;
	__be64 *d = (__be64 *)(flq->vstart);
	struct csio_dma_buf *buf = &flq->un.fl.bufs[0];
	uint64_t paddr;
	int sreg = flq->un.fl.sreg;
	int n = flq->credits;

	while (n--) {
		buf->len = sge->sge_fl_buf_size[sreg];
		buf->vaddr = pci_alloc_consistent(hw->pdev, buf->len,
						  &buf->paddr);
		if (!buf->vaddr) {
			csio_err(hw, "Could only fill %d buffers!\n", n + 1);
			return -ENOMEM;
		}

		paddr = buf->paddr | (sreg & 0xF);

		*d++ = cpu_to_be64(paddr);
		buf++;
	}

	return 0;
}

/*
 * csio_wr_update_fl -
 * @hw: HW module.
 * @flq: Freelist queue.
 *
 *
 */
static inline void
csio_wr_update_fl(struct csio_hw *hw, struct csio_q *flq, uint16_t n)
{

	flq->inc_idx += n;
	flq->pidx += n;
	if (unlikely(flq->pidx >= flq->credits))
		flq->pidx -= (uint16_t)flq->credits;

	CSIO_INC_STATS(flq, n_flq_refill);
}

/*
 * csio_wr_alloc_q - Allocate a WR queue and initialize it.
 * @hw: HW module
 * @qsize: Size of the queue in bytes
 * @wrsize: Since of WR in this queue, if fixed.
 * @type: Type of queue (Ingress/Egress/Freelist)
 * @owner: Module that owns this queue.
 * @nflb: Number of freelist buffers for FL.
 * @sreg: What is the FL buffer size register?
 * @iq_int_handler: Ingress queue handler in INTx mode.
 *
 * This function allocates and sets up a queue for the caller
 * of size qsize, aligned at the required boundary. This is subject to
 * be free entries being available in the queue array. If one is found,
 * it is initialized with the allocated queue, marked as being used (owner),
 * and a handle returned to the caller in form of the queue's index
 * into the q_arr array.
 * If user has indicated a freelist (by specifying nflb > 0), create
 * another queue (with its own index into q_arr) for the freelist. Allocate
 * memory for DMA buffer metadata (vaddr, len etc). Save off the freelist
 * idx in the ingress queue's flq.idx. This is how a Freelist is associated
 * with its owning ingress queue.
 */
int
csio_wr_alloc_q(struct csio_hw *hw, uint32_t qsize, uint32_t wrsize,
		uint16_t type, void *owner, uint32_t nflb, int sreg,
		iq_handler_t iq_intx_handler)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_q	*q, *flq;
	int		free_idx = wrm->free_qidx;
	int		ret_idx = free_idx;
	uint32_t	qsz;
	int flq_idx;

	if (free_idx >= wrm->num_q) {
		csio_err(hw, "No more free queues.\n");
		return -1;
	}

	switch (type) {
	case CSIO_EGRESS:
		qsz = ALIGN(qsize, CSIO_QCREDIT_SZ) + csio_wr_qstat_pgsz(hw);
		break;
	case CSIO_INGRESS:
		switch (wrsize) {
		case 16:
		case 32:
		case 64:
		case 128:
			break;
		default:
			csio_err(hw, "Invalid Ingress queue WR size:%d\n",
				    wrsize);
			return -1;
		}

		/*
		 * Number of elements must be a multiple of 16
		 * So this includes status page size
		 */
		qsz = ALIGN(qsize/wrsize, 16) * wrsize;

		break;
	case CSIO_FREELIST:
		qsz = ALIGN(qsize/wrsize, 8) * wrsize + csio_wr_qstat_pgsz(hw);
		break;
	default:
		csio_err(hw, "Invalid queue type: 0x%x\n", type);
		return -1;
	}

	q = wrm->q_arr[free_idx];

	q->vstart = pci_alloc_consistent(hw->pdev, qsz, &q->pstart);
	if (!q->vstart) {
		csio_err(hw,
			 "Failed to allocate DMA memory for "
			 "queue at id: %d size: %d\n", free_idx, qsize);
		return -1;
	}

	/*
	 * We need to zero out the contents, importantly for ingress,
	 * since we start with a generatiom bit of 1 for ingress.
	 */
	memset(q->vstart, 0, qsz);

	q->type		= type;
	q->owner	= owner;
	q->pidx		= q->cidx = q->inc_idx = 0;
	q->size		= qsz;
	q->wr_sz	= wrsize;	/* If using fixed size WRs */

	wrm->free_qidx++;

	if (type == CSIO_INGRESS) {
		/* Since queue area is set to zero */
		q->un.iq.genbit	= 1;

		/*
		 * Ingress queue status page size is always the size of
		 * the ingress queue entry.
		 */
		q->credits	= (qsz - q->wr_sz) / q->wr_sz;
		q->vwrap	= (void *)((uintptr_t)(q->vstart) + qsz
							- q->wr_sz);

		/* Allocate memory for FL if requested */
		if (nflb > 0) {
			flq_idx = csio_wr_alloc_q(hw, nflb * sizeof(__be64),
						  sizeof(__be64), CSIO_FREELIST,
						  owner, 0, sreg, NULL);
			if (flq_idx == -1) {
				csio_err(hw,
					 "Failed to allocate FL queue"
					 " for IQ idx:%d\n", free_idx);
				return -1;
			}

			/* Associate the new FL with the Ingress quue */
			q->un.iq.flq_idx = flq_idx;

			flq = wrm->q_arr[q->un.iq.flq_idx];
			flq->un.fl.bufs = kzalloc(flq->credits *
						  sizeof(struct csio_dma_buf),
						  GFP_KERNEL);
			if (!flq->un.fl.bufs) {
				csio_err(hw,
					 "Failed to allocate FL queue bufs"
					 " for IQ idx:%d\n", free_idx);
				return -1;
			}

			flq->un.fl.packen = 0;
			flq->un.fl.offset = 0;
			flq->un.fl.sreg = sreg;

			/* Fill up the free list buffers */
			if (csio_wr_fill_fl(hw, flq))
				return -1;

			/*
			 * Make sure in a FLQ, atleast 1 credit (8 FL buffers)
			 * remains unpopulated,otherwise HW thinks
			 * FLQ is empty.
			 */
			flq->pidx = flq->inc_idx = flq->credits - 8;
		} else {
			q->un.iq.flq_idx = -1;
		}

		/* Associate the IQ INTx handler. */
		q->un.iq.iq_intx_handler = iq_intx_handler;

		csio_q_iqid(hw, ret_idx) = CSIO_MAX_QID;

	} else if (type == CSIO_EGRESS) {
		q->credits = (qsz - csio_wr_qstat_pgsz(hw)) / CSIO_QCREDIT_SZ;
		q->vwrap   = (void *)((uintptr_t)(q->vstart) + qsz
						- csio_wr_qstat_pgsz(hw));
		csio_q_eqid(hw, ret_idx) = CSIO_MAX_QID;
	} else { /* Freelist */
		q->credits = (qsz - csio_wr_qstat_pgsz(hw)) / sizeof(__be64);
		q->vwrap   = (void *)((uintptr_t)(q->vstart) + qsz
						- csio_wr_qstat_pgsz(hw));
		csio_q_flid(hw, ret_idx) = CSIO_MAX_QID;
	}

	return ret_idx;
}

/*
 * csio_wr_iq_create_rsp - Response handler for IQ creation.
 * @hw: The HW module.
 * @mbp: Mailbox.
 * @iq_idx: Ingress queue that got created.
 *
 * Handle FW_IQ_CMD mailbox completion. Save off the assigned IQ/FL ids.
 */
static int
csio_wr_iq_create_rsp(struct csio_hw *hw, struct csio_mb *mbp, int iq_idx)
{
	struct csio_iq_params iqp;
	enum fw_retval retval;
	uint32_t iq_id;
	int flq_idx;

	memset(&iqp, 0, sizeof(struct csio_iq_params));

	csio_mb_iq_alloc_write_rsp(hw, mbp, &retval, &iqp);

	if (retval != FW_SUCCESS) {
		csio_err(hw, "IQ cmd returned 0x%x!\n", retval);
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}

	csio_q_iqid(hw, iq_idx)		= iqp.iqid;
	csio_q_physiqid(hw, iq_idx)	= iqp.physiqid;
	csio_q_pidx(hw, iq_idx)		= csio_q_cidx(hw, iq_idx) = 0;
	csio_q_inc_idx(hw, iq_idx)	= 0;

	/* Actual iq-id. */
	iq_id = iqp.iqid - hw->wrm.fw_iq_start;

	/* Set the iq-id to iq map table. */
	if (iq_id >= CSIO_MAX_IQ) {
		csio_err(hw,
			 "Exceeding MAX_IQ(%d) supported!"
			 " iqid:%d rel_iqid:%d FW iq_start:%d\n",
			 CSIO_MAX_IQ, iq_id, iqp.iqid, hw->wrm.fw_iq_start);
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}
	csio_q_set_intr_map(hw, iq_idx, iq_id);

	/*
	 * During FW_IQ_CMD, FW sets interrupt_sent bit to 1 in the SGE
	 * ingress context of this queue. This will block interrupts to
	 * this queue until the next GTS write. Therefore, we do a
	 * 0-cidx increment GTS write for this queue just to clear the
	 * interrupt_sent bit. This will re-enable interrupts to this
	 * queue.
	 */
	csio_wr_sge_intr_enable(hw, iqp.physiqid);

	flq_idx = csio_q_iq_flq_idx(hw, iq_idx);
	if (flq_idx != -1) {
		struct csio_q *flq = hw->wrm.q_arr[flq_idx];

		csio_q_flid(hw, flq_idx) = iqp.fl0id;
		csio_q_cidx(hw, flq_idx) = 0;
		csio_q_pidx(hw, flq_idx)    = csio_q_credits(hw, flq_idx) - 8;
		csio_q_inc_idx(hw, flq_idx) = csio_q_credits(hw, flq_idx) - 8;

		/* Now update SGE about the buffers allocated during init */
		csio_wr_ring_fldb(hw, flq);
	}

	mempool_free(mbp, hw->mb_mempool);

	return 0;
}

/*
 * csio_wr_iq_create - Configure an Ingress queue with FW.
 * @hw: The HW module.
 * @priv: Private data object.
 * @iq_idx: Ingress queue index in the WR module.
 * @vec: MSIX vector.
 * @portid: PCIE Channel to be associated with this queue.
 * @async: Is this a FW asynchronous message handling queue?
 * @cbfn: Completion callback.
 *
 * This API configures an ingress queue with FW by issuing a FW_IQ_CMD mailbox
 * with alloc/write bits set.
 */
int
csio_wr_iq_create(struct csio_hw *hw, void *priv, int iq_idx,
		  uint32_t vec, uint8_t portid, bool async,
		  void (*cbfn) (struct csio_hw *, struct csio_mb *))
{
	struct csio_mb  *mbp;
	struct csio_iq_params iqp;
	int flq_idx;

	memset(&iqp, 0, sizeof(struct csio_iq_params));
	csio_q_portid(hw, iq_idx) = portid;

	mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
	if (!mbp) {
		csio_err(hw, "IQ command out of memory!\n");
		return -ENOMEM;
	}

	switch (hw->intr_mode) {
	case CSIO_IM_INTX:
	case CSIO_IM_MSI:
		/* For interrupt forwarding queue only */
		if (hw->intr_iq_idx == iq_idx)
			iqp.iqandst	= X_INTERRUPTDESTINATION_PCIE;
		else
			iqp.iqandst	= X_INTERRUPTDESTINATION_IQ;
		iqp.iqandstindex	=
			csio_q_physiqid(hw, hw->intr_iq_idx);
		break;
	case CSIO_IM_MSIX:
		iqp.iqandst		= X_INTERRUPTDESTINATION_PCIE;
		iqp.iqandstindex	= (uint16_t)vec;
		break;
	case CSIO_IM_NONE:
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}

	/* Pass in the ingress queue cmd parameters */
	iqp.pfn			= hw->pfn;
	iqp.vfn			= 0;
	iqp.iq_start		= 1;
	iqp.viid		= 0;
	iqp.type		= FW_IQ_TYPE_FL_INT_CAP;
	iqp.iqasynch		= async;
	if (csio_intr_coalesce_cnt)
		iqp.iqanus	= X_UPDATESCHEDULING_COUNTER_OPTTIMER;
	else
		iqp.iqanus	= X_UPDATESCHEDULING_TIMER;
	iqp.iqanud		= X_UPDATEDELIVERY_INTERRUPT;
	iqp.iqpciech		= portid;
	iqp.iqintcntthresh	= (uint8_t)csio_sge_thresh_reg;

	switch (csio_q_wr_sz(hw, iq_idx)) {
	case 16:
		iqp.iqesize = 0; break;
	case 32:
		iqp.iqesize = 1; break;
	case 64:
		iqp.iqesize = 2; break;
	case 128:
		iqp.iqesize = 3; break;
	}

	iqp.iqsize		= csio_q_size(hw, iq_idx) /
						csio_q_wr_sz(hw, iq_idx);
	iqp.iqaddr		= csio_q_pstart(hw, iq_idx);

	flq_idx = csio_q_iq_flq_idx(hw, iq_idx);
	if (flq_idx != -1) {
		struct csio_q *flq = hw->wrm.q_arr[flq_idx];

		iqp.fl0paden	= 1;
		iqp.fl0packen	= flq->un.fl.packen ? 1 : 0;
		iqp.fl0fbmin	= X_FETCHBURSTMIN_64B;
		iqp.fl0fbmax	= X_FETCHBURSTMAX_512B;
		iqp.fl0size	= csio_q_size(hw, flq_idx) / CSIO_QCREDIT_SZ;
		iqp.fl0addr	= csio_q_pstart(hw, flq_idx);
	}

	csio_mb_iq_alloc_write(hw, mbp, priv, CSIO_MB_DEFAULT_TMO, &iqp, cbfn);

	if (csio_mb_issue(hw, mbp)) {
		csio_err(hw, "Issue of IQ cmd failed!\n");
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}

	if (cbfn != NULL)
		return 0;

	return csio_wr_iq_create_rsp(hw, mbp, iq_idx);
}

/*
 * csio_wr_eq_create_rsp - Response handler for EQ creation.
 * @hw: The HW module.
 * @mbp: Mailbox.
 * @eq_idx: Egress queue that got created.
 *
 * Handle FW_EQ_OFLD_CMD mailbox completion. Save off the assigned EQ ids.
 */
static int
csio_wr_eq_cfg_rsp(struct csio_hw *hw, struct csio_mb *mbp, int eq_idx)
{
	struct csio_eq_params eqp;
	enum fw_retval retval;

	memset(&eqp, 0, sizeof(struct csio_eq_params));

	csio_mb_eq_ofld_alloc_write_rsp(hw, mbp, &retval, &eqp);

	if (retval != FW_SUCCESS) {
		csio_err(hw, "EQ OFLD cmd returned 0x%x!\n", retval);
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}

	csio_q_eqid(hw, eq_idx)	= (uint16_t)eqp.eqid;
	csio_q_physeqid(hw, eq_idx) = (uint16_t)eqp.physeqid;
	csio_q_pidx(hw, eq_idx)	= csio_q_cidx(hw, eq_idx) = 0;
	csio_q_inc_idx(hw, eq_idx) = 0;

	mempool_free(mbp, hw->mb_mempool);

	return 0;
}

/*
 * csio_wr_eq_create - Configure an Egress queue with FW.
 * @hw: HW module.
 * @priv: Private data.
 * @eq_idx: Egress queue index in the WR module.
 * @iq_idx: Associated ingress queue index.
 * @cbfn: Completion callback.
 *
 * This API configures a offload egress queue with FW by issuing a
 * FW_EQ_OFLD_CMD  (with alloc + write ) mailbox.
 */
int
csio_wr_eq_create(struct csio_hw *hw, void *priv, int eq_idx,
		  int iq_idx, uint8_t portid,
		  void (*cbfn) (struct csio_hw *, struct csio_mb *))
{
	struct csio_mb  *mbp;
	struct csio_eq_params eqp;

	memset(&eqp, 0, sizeof(struct csio_eq_params));

	mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
	if (!mbp) {
		csio_err(hw, "EQ command out of memory!\n");
		return -ENOMEM;
	}

	eqp.pfn			= hw->pfn;
	eqp.vfn			= 0;
	eqp.eqstart		= 1;
	eqp.hostfcmode		= X_HOSTFCMODE_STATUS_PAGE;
	eqp.iqid		= csio_q_iqid(hw, iq_idx);
	eqp.fbmin		= X_FETCHBURSTMIN_64B;
	eqp.fbmax		= X_FETCHBURSTMAX_512B;
	eqp.cidxfthresh		= 0;
	eqp.pciechn		= portid;
	eqp.eqsize		= csio_q_size(hw, eq_idx) / CSIO_QCREDIT_SZ;
	eqp.eqaddr		= csio_q_pstart(hw, eq_idx);

	csio_mb_eq_ofld_alloc_write(hw, mbp, priv, CSIO_MB_DEFAULT_TMO,
				    &eqp, cbfn);

	if (csio_mb_issue(hw, mbp)) {
		csio_err(hw, "Issue of EQ OFLD cmd failed!\n");
		mempool_free(mbp, hw->mb_mempool);
		return -EINVAL;
	}

	if (cbfn != NULL)
		return 0;

	return csio_wr_eq_cfg_rsp(hw, mbp, eq_idx);
}

/*
 * csio_wr_iq_destroy_rsp - Response handler for IQ removal.
 * @hw: The HW module.
 * @mbp: Mailbox.
 * @iq_idx: Ingress queue that was freed.
 *
 * Handle FW_IQ_CMD (free) mailbox completion.
 */
static int
csio_wr_iq_destroy_rsp(struct csio_hw *hw, struct csio_mb *mbp, int iq_idx)
{
	enum fw_retval retval = csio_mb_fw_retval(mbp);
	int rv = 0;

	if (retval != FW_SUCCESS)
		rv = -EINVAL;

	mempool_free(mbp, hw->mb_mempool);

	return rv;
}

/*
 * csio_wr_iq_destroy - Free an ingress queue.
 * @hw: The HW module.
 * @priv: Private data object.
 * @iq_idx: Ingress queue index to destroy
 * @cbfn: Completion callback.
 *
 * This API frees an ingress queue by issuing the FW_IQ_CMD
 * with the free bit set.
 */
static int
csio_wr_iq_destroy(struct csio_hw *hw, void *priv, int iq_idx,
		   void (*cbfn)(struct csio_hw *, struct csio_mb *))
{
	int rv = 0;
	struct csio_mb  *mbp;
	struct csio_iq_params iqp;
	int flq_idx;

	memset(&iqp, 0, sizeof(struct csio_iq_params));

	mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
	if (!mbp)
		return -ENOMEM;

	iqp.pfn		= hw->pfn;
	iqp.vfn		= 0;
	iqp.iqid	= csio_q_iqid(hw, iq_idx);
	iqp.type	= FW_IQ_TYPE_FL_INT_CAP;

	flq_idx = csio_q_iq_flq_idx(hw, iq_idx);
	if (flq_idx != -1)
		iqp.fl0id = csio_q_flid(hw, flq_idx);
	else
		iqp.fl0id = 0xFFFF;

	iqp.fl1id = 0xFFFF;

	csio_mb_iq_free(hw, mbp, priv, CSIO_MB_DEFAULT_TMO, &iqp, cbfn);

	rv = csio_mb_issue(hw, mbp);
	if (rv != 0) {
		mempool_free(mbp, hw->mb_mempool);
		return rv;
	}

	if (cbfn != NULL)
		return 0;

	return csio_wr_iq_destroy_rsp(hw, mbp, iq_idx);
}

/*
 * csio_wr_eq_destroy_rsp - Response handler for OFLD EQ creation.
 * @hw: The HW module.
 * @mbp: Mailbox.
 * @eq_idx: Egress queue that was freed.
 *
 * Handle FW_OFLD_EQ_CMD (free) mailbox completion.
 */
static int
csio_wr_eq_destroy_rsp(struct csio_hw *hw, struct csio_mb *mbp, int eq_idx)
{
	enum fw_retval retval = csio_mb_fw_retval(mbp);
	int rv = 0;

	if (retval != FW_SUCCESS)
		rv = -EINVAL;

	mempool_free(mbp, hw->mb_mempool);

	return rv;
}

/*
 * csio_wr_eq_destroy - Free an Egress queue.
 * @hw: The HW module.
 * @priv: Private data object.
 * @eq_idx: Egress queue index to destroy
 * @cbfn: Completion callback.
 *
 * This API frees an Egress queue by issuing the FW_EQ_OFLD_CMD
 * with the free bit set.
 */
static int
csio_wr_eq_destroy(struct csio_hw *hw, void *priv, int eq_idx,
		   void (*cbfn) (struct csio_hw *, struct csio_mb *))
{
	int rv = 0;
	struct csio_mb  *mbp;
	struct csio_eq_params eqp;

	memset(&eqp, 0, sizeof(struct csio_eq_params));

	mbp = mempool_alloc(hw->mb_mempool, GFP_ATOMIC);
	if (!mbp)
		return -ENOMEM;

	eqp.pfn		= hw->pfn;
	eqp.vfn		= 0;
	eqp.eqid	= csio_q_eqid(hw, eq_idx);

	csio_mb_eq_ofld_free(hw, mbp, priv, CSIO_MB_DEFAULT_TMO, &eqp, cbfn);

	rv = csio_mb_issue(hw, mbp);
	if (rv != 0) {
		mempool_free(mbp, hw->mb_mempool);
		return rv;
	}

	if (cbfn != NULL)
		return 0;

	return csio_wr_eq_destroy_rsp(hw, mbp, eq_idx);
}

/*
 * csio_wr_cleanup_eq_stpg - Cleanup Egress queue status page
 * @hw: HW module
 * @qidx: Egress queue index
 *
 * Cleanup the Egress queue status page.
 */
static void
csio_wr_cleanup_eq_stpg(struct csio_hw *hw, int qidx)
{
	struct csio_q	*q = csio_hw_to_wrm(hw)->q_arr[qidx];
	struct csio_qstatus_page *stp = (struct csio_qstatus_page *)q->vwrap;

	memset(stp, 0, sizeof(*stp));
}

/*
 * csio_wr_cleanup_iq_ftr - Cleanup Footer entries in IQ
 * @hw: HW module
 * @qidx: Ingress queue index
 *
 * Cleanup the footer entries in the given ingress queue,
 * set to 1 the internal copy of genbit.
 */
static void
csio_wr_cleanup_iq_ftr(struct csio_hw *hw, int qidx)
{
	struct csio_wrm *wrm	= csio_hw_to_wrm(hw);
	struct csio_q	*q	= wrm->q_arr[qidx];
	void *wr;
	struct csio_iqwr_footer *ftr;
	uint32_t i = 0;

	/* set to 1 since we are just about zero out genbit */
	q->un.iq.genbit = 1;

	for (i = 0; i < q->credits; i++) {
		/* Get the WR */
		wr = (void *)((uintptr_t)q->vstart +
					   (i * q->wr_sz));
		/* Get the footer */
		ftr = (struct csio_iqwr_footer *)((uintptr_t)wr +
					  (q->wr_sz - sizeof(*ftr)));
		/* Zero out footer */
		memset(ftr, 0, sizeof(*ftr));
	}
}

int
csio_wr_destroy_queues(struct csio_hw *hw, bool cmd)
{
	int i, flq_idx;
	struct csio_q *q;
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	int rv;

	for (i = 0; i < wrm->free_qidx; i++) {
		q = wrm->q_arr[i];

		switch (q->type) {
		case CSIO_EGRESS:
			if (csio_q_eqid(hw, i) != CSIO_MAX_QID) {
				csio_wr_cleanup_eq_stpg(hw, i);
				if (!cmd) {
					csio_q_eqid(hw, i) = CSIO_MAX_QID;
					continue;
				}

				rv = csio_wr_eq_destroy(hw, NULL, i, NULL);
				if ((rv == -EBUSY) || (rv == -ETIMEDOUT))
					cmd = false;

				csio_q_eqid(hw, i) = CSIO_MAX_QID;
			}
		case CSIO_INGRESS:
			if (csio_q_iqid(hw, i) != CSIO_MAX_QID) {
				csio_wr_cleanup_iq_ftr(hw, i);
				if (!cmd) {
					csio_q_iqid(hw, i) = CSIO_MAX_QID;
					flq_idx = csio_q_iq_flq_idx(hw, i);
					if (flq_idx != -1)
						csio_q_flid(hw, flq_idx) =
								CSIO_MAX_QID;
					continue;
				}

				rv = csio_wr_iq_destroy(hw, NULL, i, NULL);
				if ((rv == -EBUSY) || (rv == -ETIMEDOUT))
					cmd = false;

				csio_q_iqid(hw, i) = CSIO_MAX_QID;
				flq_idx = csio_q_iq_flq_idx(hw, i);
				if (flq_idx != -1)
					csio_q_flid(hw, flq_idx) = CSIO_MAX_QID;
			}
		default:
			break;
		}
	}

	hw->flags &= ~CSIO_HWF_Q_FW_ALLOCED;

	return 0;
}

/*
 * csio_wr_get - Get requested size of WR entry/entries from queue.
 * @hw: HW module.
 * @qidx: Index of queue.
 * @size: Cumulative size of Work request(s).
 * @wrp: Work request pair.
 *
 * If requested credits are available, return the start address of the
 * work request in the work request pair. Set pidx accordingly and
 * return.
 *
 * NOTE about WR pair:
 * ==================
 * A WR can start towards the end of a queue, and then continue at the
 * beginning, since the queue is considered to be circular. This will
 * require a pair of address/size to be passed back to the caller -
 * hence Work request pair format.
 */
int
csio_wr_get(struct csio_hw *hw, int qidx, uint32_t size,
	    struct csio_wr_pair *wrp)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_q *q = wrm->q_arr[qidx];
	void *cwr = (void *)((uintptr_t)(q->vstart) +
						(q->pidx * CSIO_QCREDIT_SZ));
	struct csio_qstatus_page *stp = (struct csio_qstatus_page *)q->vwrap;
	uint16_t cidx = q->cidx = ntohs(stp->cidx);
	uint16_t pidx = q->pidx;
	uint32_t req_sz	= ALIGN(size, CSIO_QCREDIT_SZ);
	int req_credits	= req_sz / CSIO_QCREDIT_SZ;
	int credits;

	CSIO_DB_ASSERT(q->owner != NULL);
	CSIO_DB_ASSERT((qidx >= 0) && (qidx < wrm->free_qidx));
	CSIO_DB_ASSERT(cidx <= q->credits);

	/* Calculate credits */
	if (pidx > cidx) {
		credits = q->credits - (pidx - cidx) - 1;
	} else if (cidx > pidx) {
		credits = cidx - pidx - 1;
	} else {
		/* cidx == pidx, empty queue */
		credits = q->credits;
		CSIO_INC_STATS(q, n_qempty);
	}

	/*
	 * Check if we have enough credits.
	 * credits = 1 implies queue is full.
	 */
	if (!credits || (req_credits > credits)) {
		CSIO_INC_STATS(q, n_qfull);
		return -EBUSY;
	}

	/*
	 * If we are here, we have enough credits to satisfy the
	 * request. Check if we are near the end of q, and if WR spills over.
	 * If it does, use the first addr/size to cover the queue until
	 * the end. Fit the remainder portion of the request at the top
	 * of queue and return it in the second addr/len. Set pidx
	 * accordingly.
	 */
	if (unlikely(((uintptr_t)cwr + req_sz) > (uintptr_t)(q->vwrap))) {
		wrp->addr1 = cwr;
		wrp->size1 = (uint32_t)((uintptr_t)q->vwrap - (uintptr_t)cwr);
		wrp->addr2 = q->vstart;
		wrp->size2 = req_sz - wrp->size1;
		q->pidx	= (uint16_t)(ALIGN(wrp->size2, CSIO_QCREDIT_SZ) /
							CSIO_QCREDIT_SZ);
		CSIO_INC_STATS(q, n_qwrap);
		CSIO_INC_STATS(q, n_eq_wr_split);
	} else {
		wrp->addr1 = cwr;
		wrp->size1 = req_sz;
		wrp->addr2 = NULL;
		wrp->size2 = 0;
		q->pidx	+= (uint16_t)req_credits;

		/* We are the end of queue, roll back pidx to top of queue */
		if (unlikely(q->pidx == q->credits)) {
			q->pidx = 0;
			CSIO_INC_STATS(q, n_qwrap);
		}
	}

	q->inc_idx = (uint16_t)req_credits;

	CSIO_INC_STATS(q, n_tot_reqs);

	return 0;
}

/*
 * csio_wr_copy_to_wrp - Copies given data into WR.
 * @data_buf - Data buffer
 * @wrp - Work request pair.
 * @wr_off - Work request offset.
 * @data_len - Data length.
 *
 * Copies the given data in Work Request. Work request pair(wrp) specifies
 * address information of Work request.
 * Returns: none
 */
void
csio_wr_copy_to_wrp(void *data_buf, struct csio_wr_pair *wrp,
		   uint32_t wr_off, uint32_t data_len)
{
	uint32_t nbytes;

	/* Number of space available in buffer addr1 of WRP */
	nbytes = ((wrp->size1 - wr_off) >= data_len) ?
					data_len : (wrp->size1 - wr_off);

	memcpy((uint8_t *) wrp->addr1 + wr_off, data_buf, nbytes);
	data_len -= nbytes;

	/* Write the remaining data from the begining of circular buffer */
	if (data_len) {
		CSIO_DB_ASSERT(data_len <= wrp->size2);
		CSIO_DB_ASSERT(wrp->addr2 != NULL);
		memcpy(wrp->addr2, (uint8_t *) data_buf + nbytes, data_len);
	}
}

/*
 * csio_wr_issue - Notify chip of Work request.
 * @hw: HW module.
 * @qidx: Index of queue.
 * @prio: 0: Low priority, 1: High priority
 *
 * Rings the SGE Doorbell by writing the current producer index of the passed
 * in queue into the register.
 *
 */
int
csio_wr_issue(struct csio_hw *hw, int qidx, bool prio)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_q *q = wrm->q_arr[qidx];

	CSIO_DB_ASSERT((qidx >= 0) && (qidx < wrm->free_qidx));

	wmb();
	/* Ring SGE Doorbell writing q->pidx into it */
	csio_wr_reg32(hw, DBPRIO(prio) | QID(q->un.eq.physeqid) |
		      PIDX(q->inc_idx), MYPF_REG(SGE_PF_KDOORBELL));
	q->inc_idx = 0;

	return 0;
}

static inline uint32_t
csio_wr_avail_qcredits(struct csio_q *q)
{
	if (q->pidx > q->cidx)
		return q->pidx - q->cidx;
	else if (q->cidx > q->pidx)
		return q->credits - (q->cidx - q->pidx);
	else
		return 0;	/* cidx == pidx, empty queue */
}

/*
 * csio_wr_inval_flq_buf - Invalidate a free list buffer entry.
 * @hw: HW module.
 * @flq: The freelist queue.
 *
 * Invalidate the driver's version of a freelist buffer entry,
 * without freeing the associated the DMA memory. The entry
 * to be invalidated is picked up from the current Free list
 * queue cidx.
 *
 */
static inline void
csio_wr_inval_flq_buf(struct csio_hw *hw, struct csio_q *flq)
{
	flq->cidx++;
	if (flq->cidx == flq->credits) {
		flq->cidx = 0;
		CSIO_INC_STATS(flq, n_qwrap);
	}
}

/*
 * csio_wr_process_fl - Process a freelist completion.
 * @hw: HW module.
 * @q: The ingress queue attached to the Freelist.
 * @wr: The freelist completion WR in the ingress queue.
 * @len_to_qid: The lower 32-bits of the first flit of the RSP footer
 * @iq_handler: Caller's handler for this completion.
 * @priv: Private pointer of caller
 *
 */
static inline void
csio_wr_process_fl(struct csio_hw *hw, struct csio_q *q,
		   void *wr, uint32_t len_to_qid,
		   void (*iq_handler)(struct csio_hw *, void *,
				      uint32_t, struct csio_fl_dma_buf *,
				      void *),
		   void *priv)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;
	struct csio_fl_dma_buf flb;
	struct csio_dma_buf *buf, *fbuf;
	uint32_t bufsz, len, lastlen = 0;
	int n;
	struct csio_q *flq = hw->wrm.q_arr[q->un.iq.flq_idx];

	CSIO_DB_ASSERT(flq != NULL);

	len = len_to_qid;

	if (len & IQWRF_NEWBUF) {
		if (flq->un.fl.offset > 0) {
			csio_wr_inval_flq_buf(hw, flq);
			flq->un.fl.offset = 0;
		}
		len = IQWRF_LEN_GET(len);
	}

	CSIO_DB_ASSERT(len != 0);

	flb.totlen = len;

	/* Consume all freelist buffers used for len bytes */
	for (n = 0, fbuf = flb.flbufs; ; n++, fbuf++) {
		buf = &flq->un.fl.bufs[flq->cidx];
		bufsz = csio_wr_fl_bufsz(sge, buf);

		fbuf->paddr	= buf->paddr;
		fbuf->vaddr	= buf->vaddr;

		flb.offset	= flq->un.fl.offset;
		lastlen		= min(bufsz, len);
		fbuf->len	= lastlen;

		len -= lastlen;
		if (!len)
			break;
		csio_wr_inval_flq_buf(hw, flq);
	}

	flb.defer_free = flq->un.fl.packen ? 0 : 1;

	iq_handler(hw, wr, q->wr_sz - sizeof(struct csio_iqwr_footer),
		   &flb, priv);

	if (flq->un.fl.packen)
		flq->un.fl.offset += ALIGN(lastlen, sge->csio_fl_align);
	else
		csio_wr_inval_flq_buf(hw, flq);

}

/*
 * csio_is_new_iqwr - Is this a new Ingress queue entry ?
 * @q: Ingress quueue.
 * @ftr: Ingress queue WR SGE footer.
 *
 * The entry is new if our generation bit matches the corresponding
 * bit in the footer of the current WR.
 */
static inline bool
csio_is_new_iqwr(struct csio_q *q, struct csio_iqwr_footer *ftr)
{
	return (q->un.iq.genbit == (ftr->u.type_gen >> IQWRF_GEN_SHIFT));
}

/*
 * csio_wr_process_iq - Process elements in Ingress queue.
 * @hw:  HW pointer
 * @qidx: Index of queue
 * @iq_handler: Handler for this queue
 * @priv: Caller's private pointer
 *
 * This routine walks through every entry of the ingress queue, calling
 * the provided iq_handler with the entry, until the generation bit
 * flips.
 */
int
csio_wr_process_iq(struct csio_hw *hw, struct csio_q *q,
		   void (*iq_handler)(struct csio_hw *, void *,
				      uint32_t, struct csio_fl_dma_buf *,
				      void *),
		   void *priv)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	void *wr = (void *)((uintptr_t)q->vstart + (q->cidx * q->wr_sz));
	struct csio_iqwr_footer *ftr;
	uint32_t wr_type, fw_qid, qid;
	struct csio_q *q_completed;
	struct csio_q *flq = csio_iq_has_fl(q) ?
					wrm->q_arr[q->un.iq.flq_idx] : NULL;
	int rv = 0;

	/* Get the footer */
	ftr = (struct csio_iqwr_footer *)((uintptr_t)wr +
					  (q->wr_sz - sizeof(*ftr)));

	/*
	 * When q wrapped around last time, driver should have inverted
	 * ic.genbit as well.
	 */
	while (csio_is_new_iqwr(q, ftr)) {

		CSIO_DB_ASSERT(((uintptr_t)wr + q->wr_sz) <=
						(uintptr_t)q->vwrap);
		rmb();
		wr_type = IQWRF_TYPE_GET(ftr->u.type_gen);

		switch (wr_type) {
		case X_RSPD_TYPE_CPL:
			/* Subtract footer from WR len */
			iq_handler(hw, wr, q->wr_sz - sizeof(*ftr), NULL, priv);
			break;
		case X_RSPD_TYPE_FLBUF:
			csio_wr_process_fl(hw, q, wr,
					   ntohl(ftr->pldbuflen_qid),
					   iq_handler, priv);
			break;
		case X_RSPD_TYPE_INTR:
			fw_qid = ntohl(ftr->pldbuflen_qid);
			qid = fw_qid - wrm->fw_iq_start;
			q_completed = hw->wrm.intr_map[qid];

			if (unlikely(qid ==
					csio_q_physiqid(hw, hw->intr_iq_idx))) {
				/*
				 * We are already in the Forward Interrupt
				 * Interrupt Queue Service! Do-not service
				 * again!
				 *
				 */
			} else {
				CSIO_DB_ASSERT(q_completed);
				CSIO_DB_ASSERT(
					q_completed->un.iq.iq_intx_handler);

				/* Call the queue handler. */
				q_completed->un.iq.iq_intx_handler(hw, NULL,
						0, NULL, (void *)q_completed);
			}
			break;
		default:
			csio_warn(hw, "Unknown resp type 0x%x received\n",
				 wr_type);
			CSIO_INC_STATS(q, n_rsp_unknown);
			break;
		}

		/*
		 * Ingress *always* has fixed size WR entries. Therefore,
		 * there should always be complete WRs towards the end of
		 * queue.
		 */
		if (((uintptr_t)wr + q->wr_sz) == (uintptr_t)q->vwrap) {

			/* Roll over to start of queue */
			q->cidx = 0;
			wr	= q->vstart;

			/* Toggle genbit */
			q->un.iq.genbit ^= 0x1;

			CSIO_INC_STATS(q, n_qwrap);
		} else {
			q->cidx++;
			wr	= (void *)((uintptr_t)(q->vstart) +
					   (q->cidx * q->wr_sz));
		}

		ftr = (struct csio_iqwr_footer *)((uintptr_t)wr +
						  (q->wr_sz - sizeof(*ftr)));
		q->inc_idx++;

	} /* while (q->un.iq.genbit == hdr->genbit) */

	/*
	 * We need to re-arm SGE interrupts in case we got a stray interrupt,
	 * especially in msix mode. With INTx, this may be a common occurence.
	 */
	if (unlikely(!q->inc_idx)) {
		CSIO_INC_STATS(q, n_stray_comp);
		rv = -EINVAL;
		goto restart;
	}

	/* Replenish free list buffers if pending falls below low water mark */
	if (flq) {
		uint32_t avail  = csio_wr_avail_qcredits(flq);
		if (avail <= 16) {
			/* Make sure in FLQ, atleast 1 credit (8 FL buffers)
			 * remains unpopulated otherwise HW thinks
			 * FLQ is empty.
			 */
			csio_wr_update_fl(hw, flq, (flq->credits - 8) - avail);
			csio_wr_ring_fldb(hw, flq);
		}
	}

restart:
	/* Now inform SGE about our incremental index value */
	csio_wr_reg32(hw, CIDXINC(q->inc_idx)		|
			  INGRESSQID(q->un.iq.physiqid)	|
			  TIMERREG(csio_sge_timer_reg),
			  MYPF_REG(SGE_PF_GTS));
	q->stats.n_tot_rsps += q->inc_idx;

	q->inc_idx = 0;

	return rv;
}

int
csio_wr_process_iq_idx(struct csio_hw *hw, int qidx,
		   void (*iq_handler)(struct csio_hw *, void *,
				      uint32_t, struct csio_fl_dma_buf *,
				      void *),
		   void *priv)
{
	struct csio_wrm *wrm	= csio_hw_to_wrm(hw);
	struct csio_q	*iq	= wrm->q_arr[qidx];

	return csio_wr_process_iq(hw, iq, iq_handler, priv);
}

static int
csio_closest_timer(struct csio_sge *s, int time)
{
	int i, delta, match = 0, min_delta = INT_MAX;

	for (i = 0; i < ARRAY_SIZE(s->timer_val); i++) {
		delta = time - s->timer_val[i];
		if (delta < 0)
			delta = -delta;
		if (delta < min_delta) {
			min_delta = delta;
			match = i;
		}
	}
	return match;
}

static int
csio_closest_thresh(struct csio_sge *s, int cnt)
{
	int i, delta, match = 0, min_delta = INT_MAX;

	for (i = 0; i < ARRAY_SIZE(s->counter_val); i++) {
		delta = cnt - s->counter_val[i];
		if (delta < 0)
			delta = -delta;
		if (delta < min_delta) {
			min_delta = delta;
			match = i;
		}
	}
	return match;
}

static void
csio_wr_fixup_host_params(struct csio_hw *hw)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;
	uint32_t clsz = L1_CACHE_BYTES;
	uint32_t s_hps = PAGE_SHIFT - 10;
	uint32_t ingpad = 0;
	uint32_t stat_len = clsz > 64 ? 128 : 64;

	csio_wr_reg32(hw, HOSTPAGESIZEPF0(s_hps) | HOSTPAGESIZEPF1(s_hps) |
		      HOSTPAGESIZEPF2(s_hps) | HOSTPAGESIZEPF3(s_hps) |
		      HOSTPAGESIZEPF4(s_hps) | HOSTPAGESIZEPF5(s_hps) |
		      HOSTPAGESIZEPF6(s_hps) | HOSTPAGESIZEPF7(s_hps),
		      SGE_HOST_PAGE_SIZE);

	sge->csio_fl_align = clsz < 32 ? 32 : clsz;
	ingpad = ilog2(sge->csio_fl_align) - 5;

	csio_set_reg_field(hw, SGE_CONTROL, INGPADBOUNDARY_MASK |
					    EGRSTATUSPAGESIZE(1),
			   INGPADBOUNDARY(ingpad) |
			   EGRSTATUSPAGESIZE(stat_len != 64));

	/* FL BUFFER SIZE#0 is Page size i,e already aligned to cache line */
	csio_wr_reg32(hw, PAGE_SIZE, SGE_FL_BUFFER_SIZE0);

	/*
	 * If using hard params, the following will get set correctly
	 * in csio_wr_set_sge().
	 */
	if (hw->flags & CSIO_HWF_USING_SOFT_PARAMS) {
		csio_wr_reg32(hw,
			(csio_rd_reg32(hw, SGE_FL_BUFFER_SIZE2) +
			sge->csio_fl_align - 1) & ~(sge->csio_fl_align - 1),
			SGE_FL_BUFFER_SIZE2);
		csio_wr_reg32(hw,
			(csio_rd_reg32(hw, SGE_FL_BUFFER_SIZE3) +
			sge->csio_fl_align - 1) & ~(sge->csio_fl_align - 1),
			SGE_FL_BUFFER_SIZE3);
	}

	csio_wr_reg32(hw, HPZ0(PAGE_SHIFT - 12), ULP_RX_TDDP_PSZ);

	/* default value of rx_dma_offset of the NIC driver */
	csio_set_reg_field(hw, SGE_CONTROL, PKTSHIFT_MASK,
			   PKTSHIFT(CSIO_SGE_RX_DMA_OFFSET));
}

static void
csio_init_intr_coalesce_parms(struct csio_hw *hw)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;

	csio_sge_thresh_reg = csio_closest_thresh(sge, csio_intr_coalesce_cnt);
	if (csio_intr_coalesce_cnt) {
		csio_sge_thresh_reg = 0;
		csio_sge_timer_reg = X_TIMERREG_RESTART_COUNTER;
		return;
	}

	csio_sge_timer_reg = csio_closest_timer(sge, csio_intr_coalesce_time);
}

/*
 * csio_wr_get_sge - Get SGE register values.
 * @hw: HW module.
 *
 * Used by non-master functions and by master-functions relying on config file.
 */
static void
csio_wr_get_sge(struct csio_hw *hw)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;
	uint32_t ingpad;
	int i;
	u32 timer_value_0_and_1, timer_value_2_and_3, timer_value_4_and_5;
	u32 ingress_rx_threshold;

	sge->sge_control = csio_rd_reg32(hw, SGE_CONTROL);

	ingpad = INGPADBOUNDARY_GET(sge->sge_control);

	switch (ingpad) {
	case X_INGPCIEBOUNDARY_32B:
		sge->csio_fl_align = 32; break;
	case X_INGPCIEBOUNDARY_64B:
		sge->csio_fl_align = 64; break;
	case X_INGPCIEBOUNDARY_128B:
		sge->csio_fl_align = 128; break;
	case X_INGPCIEBOUNDARY_256B:
		sge->csio_fl_align = 256; break;
	case X_INGPCIEBOUNDARY_512B:
		sge->csio_fl_align = 512; break;
	case X_INGPCIEBOUNDARY_1024B:
		sge->csio_fl_align = 1024; break;
	case X_INGPCIEBOUNDARY_2048B:
		sge->csio_fl_align = 2048; break;
	case X_INGPCIEBOUNDARY_4096B:
		sge->csio_fl_align = 4096; break;
	}

	for (i = 0; i < CSIO_SGE_FL_SIZE_REGS; i++)
		csio_get_flbuf_size(hw, sge, i);

	timer_value_0_and_1 = csio_rd_reg32(hw, SGE_TIMER_VALUE_0_AND_1);
	timer_value_2_and_3 = csio_rd_reg32(hw, SGE_TIMER_VALUE_2_AND_3);
	timer_value_4_and_5 = csio_rd_reg32(hw, SGE_TIMER_VALUE_4_AND_5);

	sge->timer_val[0] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE0_GET(timer_value_0_and_1));
	sge->timer_val[1] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE1_GET(timer_value_0_and_1));
	sge->timer_val[2] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE2_GET(timer_value_2_and_3));
	sge->timer_val[3] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE3_GET(timer_value_2_and_3));
	sge->timer_val[4] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE4_GET(timer_value_4_and_5));
	sge->timer_val[5] = (uint16_t)csio_core_ticks_to_us(hw,
					TIMERVALUE5_GET(timer_value_4_and_5));

	ingress_rx_threshold = csio_rd_reg32(hw, SGE_INGRESS_RX_THRESHOLD);
	sge->counter_val[0] = THRESHOLD_0_GET(ingress_rx_threshold);
	sge->counter_val[1] = THRESHOLD_1_GET(ingress_rx_threshold);
	sge->counter_val[2] = THRESHOLD_2_GET(ingress_rx_threshold);
	sge->counter_val[3] = THRESHOLD_3_GET(ingress_rx_threshold);

	csio_init_intr_coalesce_parms(hw);
}

/*
 * csio_wr_set_sge - Initialize SGE registers
 * @hw: HW module.
 *
 * Used by Master function to initialize SGE registers in the absence
 * of a config file.
 */
static void
csio_wr_set_sge(struct csio_hw *hw)
{
	struct csio_wrm *wrm = csio_hw_to_wrm(hw);
	struct csio_sge *sge = &wrm->sge;
	int i;

	/*
	 * Set up our basic SGE mode to deliver CPL messages to our Ingress
	 * Queue and Packet Date to the Free List.
	 */
	csio_set_reg_field(hw, SGE_CONTROL, RXPKTCPLMODE(1), RXPKTCPLMODE(1));

	sge->sge_control = csio_rd_reg32(hw, SGE_CONTROL);

	/* sge->csio_fl_align is set up by csio_wr_fixup_host_params(). */

	/*
	 * Set up to drop DOORBELL writes when the DOORBELL FIFO overflows
	 * and generate an interrupt when this occurs so we can recover.
	 */
	csio_set_reg_field(hw, SGE_DBFIFO_STATUS,
			   HP_INT_THRESH(HP_INT_THRESH_MASK) |
			   LP_INT_THRESH(LP_INT_THRESH_MASK),
			   HP_INT_THRESH(CSIO_SGE_DBFIFO_INT_THRESH) |
			   LP_INT_THRESH(CSIO_SGE_DBFIFO_INT_THRESH));
	csio_set_reg_field(hw, SGE_DOORBELL_CONTROL, ENABLE_DROP,
			   ENABLE_DROP);

	/* SGE_FL_BUFFER_SIZE0 is set up by csio_wr_fixup_host_params(). */

	CSIO_SET_FLBUF_SIZE(hw, 1, CSIO_SGE_FLBUF_SIZE1);
	csio_wr_reg32(hw, (CSIO_SGE_FLBUF_SIZE2 + sge->csio_fl_align - 1)
		      & ~(sge->csio_fl_align - 1), SGE_FL_BUFFER_SIZE2);
	csio_wr_reg32(hw, (CSIO_SGE_FLBUF_SIZE3 + sge->csio_fl_align - 1)
		      & ~(sge->csio_fl_align - 1), SGE_FL_BUFFER_SIZE3);
	CSIO_SET_FLBUF_SIZE(hw, 4, CSIO_SGE_FLBUF_SIZE4);
	CSIO_SET_FLBUF_SIZE(hw, 5, CSIO_SGE_FLBUF_SIZE5);
	CSIO_SET_FLBUF_SIZE(hw, 6, CSIO_SGE_FLBUF_SIZE6);
	CSIO_SET_FLBUF_SIZE(hw, 7, CSIO_SGE_FLBUF_SIZE7);
	CSIO_SET_FLBUF_SIZE(hw, 8, CSIO_SGE_FLBUF_SIZE8);

	for (i = 0; i < CSIO_SGE_FL_SIZE_REGS; i++)
		csio_get_flbuf_size(hw, sge, i);

	/* Initialize interrupt coalescing attributes */
	sge->timer_val[0] = CSIO_SGE_TIMER_VAL_0;
	sge->timer_val[1] = CSIO_SGE_TIMER_VAL_1;
	sge->timer_val[2] = CSIO_SGE_TIMER_VAL_2;
	sge->timer_val[3] = CSIO_SGE_TIMER_VAL_3;
	sge->timer_val[4] = CSIO_SGE_TIMER_VAL_4;
	sge->timer_val[5] = CSIO_SGE_TIMER_VAL_5;

	sge->counter_val[0] = CSIO_SGE_INT_CNT_VAL_0;
	sge->counter_val[1] = CSIO_SGE_INT_CNT_VAL_1;
	sge->counter_val[2] = CSIO_SGE_INT_CNT_VAL_2;
	sge->counter_val[3] = CSIO_SGE_INT_CNT_VAL_3;

	csio_wr_reg32(hw, THRESHOLD_0(sge->counter_val[0]) |
		      THRESHOLD_1(sge->counter_val[1]) |
		      THRESHOLD_2(sge->counter_val[2]) |
		      THRESHOLD_3(sge->counter_val[3]),
		      SGE_INGRESS_RX_THRESHOLD);

	csio_wr_reg32(hw,
		   TIMERVALUE0(csio_us_to_core_ticks(hw, sge->timer_val[0])) |
		   TIMERVALUE1(csio_us_to_core_ticks(hw, sge->timer_val[1])),
		   SGE_TIMER_VALUE_0_AND_1);

	csio_wr_reg32(hw,
		   TIMERVALUE2(csio_us_to_core_ticks(hw, sge->timer_val[2])) |
		   TIMERVALUE3(csio_us_to_core_ticks(hw, sge->timer_val[3])),
		   SGE_TIMER_VALUE_2_AND_3);

	csio_wr_reg32(hw,
		   TIMERVALUE4(csio_us_to_core_ticks(hw, sge->timer_val[4])) |
		   TIMERVALUE5(csio_us_to_core_ticks(hw, sge->timer_val[5])),
		   SGE_TIMER_VALUE_4_AND_5);

	csio_init_intr_coalesce_parms(hw);
}

void
csio_wr_sge_init(struct csio_hw *hw)
{
	/*
	 * If we are master and chip is not initialized:
	 *    - If we plan to use the config file, we need to fixup some
	 *      host specific registers, and read the rest of the SGE
	 *      configuration.
	 *    - If we dont plan to use the config file, we need to initialize
	 *      SGE entirely, including fixing the host specific registers.
	 * If we are master and chip is initialized, just read and work off of
	 *	the already initialized SGE values.
	 * If we arent the master, we are only allowed to read and work off of
	 *      the already initialized SGE values.
	 *
	 * Therefore, before calling this function, we assume that the master-
	 * ship of the card, state and whether to use config file or not, have
	 * already been decided.
	 */
	if (csio_is_hw_master(hw)) {
		if (hw->fw_state != CSIO_DEV_STATE_INIT)
			csio_wr_fixup_host_params(hw);

		if (hw->flags & CSIO_HWF_USING_SOFT_PARAMS)
			csio_wr_get_sge(hw);
		else
			csio_wr_set_sge(hw);
	} else
		csio_wr_get_sge(hw);
}

/*
 * csio_wrm_init - Initialize Work request module.
 * @wrm: WR module
 * @hw: HW pointer
 *
 * Allocates memory for an array of queue pointers starting at q_arr.
 */
int
csio_wrm_init(struct csio_wrm *wrm, struct csio_hw *hw)
{
	int i;

	if (!wrm->num_q) {
		csio_err(hw, "Num queues is not set\n");
		return -EINVAL;
	}

	wrm->q_arr = kzalloc(sizeof(struct csio_q *) * wrm->num_q, GFP_KERNEL);
	if (!wrm->q_arr)
		goto err;

	for (i = 0; i < wrm->num_q; i++) {
		wrm->q_arr[i] = kzalloc(sizeof(struct csio_q), GFP_KERNEL);
		if (!wrm->q_arr[i]) {
			while (--i >= 0)
				kfree(wrm->q_arr[i]);
			goto err_free_arr;
		}
	}
	wrm->free_qidx	= 0;

	return 0;

err_free_arr:
	kfree(wrm->q_arr);
err:
	return -ENOMEM;
}

/*
 * csio_wrm_exit - Initialize Work request module.
 * @wrm: WR module
 * @hw: HW module
 *
 * Uninitialize WR module. Free q_arr and pointers in it.
 * We have the additional job of freeing the DMA memory associated
 * with the queues.
 */
void
csio_wrm_exit(struct csio_wrm *wrm, struct csio_hw *hw)
{
	int i;
	uint32_t j;
	struct csio_q *q;
	struct csio_dma_buf *buf;

	for (i = 0; i < wrm->num_q; i++) {
		q = wrm->q_arr[i];

		if (wrm->free_qidx && (i < wrm->free_qidx)) {
			if (q->type == CSIO_FREELIST) {
				if (!q->un.fl.bufs)
					continue;
				for (j = 0; j < q->credits; j++) {
					buf = &q->un.fl.bufs[j];
					if (!buf->vaddr)
						continue;
					pci_free_consistent(hw->pdev, buf->len,
							    buf->vaddr,
							    buf->paddr);
				}
				kfree(q->un.fl.bufs);
			}
			pci_free_consistent(hw->pdev, q->size,
					    q->vstart, q->pstart);
		}
		kfree(q);
	}

	hw->flags &= ~CSIO_HWF_Q_MEM_ALLOCED;

	kfree(wrm->q_arr);
}