summaryrefslogtreecommitdiffstats
path: root/drivers/net/wireless/brcm80211/brcmsmac/dma.c
blob: 6ebec8f42846977439c07844cba9a6f7e429801c (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
/*
 * Copyright (c) 2010 Broadcom Corporation
 *
 * Permission to use, copy, modify, and/or distribute this software for any
 * purpose with or without fee is hereby granted, provided that the above
 * copyright notice and this permission notice appear in all copies.
 *
 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY
 * SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN ACTION
 * OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF OR IN
 * CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
 */
#include <linux/slab.h>
#include <linux/skbuff.h>
#include <linux/delay.h>
#include <linux/pci.h>

#include <brcmu_utils.h>
#include <aiutils.h>
#include "types.h"
#include "dma.h"

/*
 * DMA hardware requires each descriptor ring to be 8kB aligned, and fit within
 * a contiguous 8kB physical address.
 */
#define D64RINGALIGN_BITS	13
#define	D64MAXRINGSZ		(1 << D64RINGALIGN_BITS)
#define	D64RINGALIGN		(1 << D64RINGALIGN_BITS)

#define	D64MAXDD	(D64MAXRINGSZ / sizeof(struct dma64desc))

/* transmit channel control */
#define	D64_XC_XE		0x00000001	/* transmit enable */
#define	D64_XC_SE		0x00000002	/* transmit suspend request */
#define	D64_XC_LE		0x00000004	/* loopback enable */
#define	D64_XC_FL		0x00000010	/* flush request */
#define	D64_XC_PD		0x00000800	/* parity check disable */
#define	D64_XC_AE		0x00030000	/* address extension bits */
#define	D64_XC_AE_SHIFT		16

/* transmit descriptor table pointer */
#define	D64_XP_LD_MASK		0x00000fff	/* last valid descriptor */

/* transmit channel status */
#define	D64_XS0_CD_MASK		0x00001fff	/* current descriptor pointer */
#define	D64_XS0_XS_MASK		0xf0000000	/* transmit state */
#define	D64_XS0_XS_SHIFT		28
#define	D64_XS0_XS_DISABLED	0x00000000	/* disabled */
#define	D64_XS0_XS_ACTIVE	0x10000000	/* active */
#define	D64_XS0_XS_IDLE		0x20000000	/* idle wait */
#define	D64_XS0_XS_STOPPED	0x30000000	/* stopped */
#define	D64_XS0_XS_SUSP		0x40000000	/* suspend pending */

#define	D64_XS1_AD_MASK		0x00001fff	/* active descriptor */
#define	D64_XS1_XE_MASK		0xf0000000	/* transmit errors */
#define	D64_XS1_XE_SHIFT		28
#define	D64_XS1_XE_NOERR	0x00000000	/* no error */
#define	D64_XS1_XE_DPE		0x10000000	/* descriptor protocol error */
#define	D64_XS1_XE_DFU		0x20000000	/* data fifo underrun */
#define	D64_XS1_XE_DTE		0x30000000	/* data transfer error */
#define	D64_XS1_XE_DESRE	0x40000000	/* descriptor read error */
#define	D64_XS1_XE_COREE	0x50000000	/* core error */

/* receive channel control */
/* receive enable */
#define	D64_RC_RE		0x00000001
/* receive frame offset */
#define	D64_RC_RO_MASK		0x000000fe
#define	D64_RC_RO_SHIFT		1
/* direct fifo receive (pio) mode */
#define	D64_RC_FM		0x00000100
/* separate rx header descriptor enable */
#define	D64_RC_SH		0x00000200
/* overflow continue */
#define	D64_RC_OC		0x00000400
/* parity check disable */
#define	D64_RC_PD		0x00000800
/* address extension bits */
#define	D64_RC_AE		0x00030000
#define	D64_RC_AE_SHIFT		16

/* flags for dma controller */
/* partity enable */
#define DMA_CTRL_PEN		(1 << 0)
/* rx overflow continue */
#define DMA_CTRL_ROC		(1 << 1)
/* allow rx scatter to multiple descriptors */
#define DMA_CTRL_RXMULTI	(1 << 2)
/* Unframed Rx/Tx data */
#define DMA_CTRL_UNFRAMED	(1 << 3)

/* receive descriptor table pointer */
#define	D64_RP_LD_MASK		0x00000fff	/* last valid descriptor */

/* receive channel status */
#define	D64_RS0_CD_MASK		0x00001fff	/* current descriptor pointer */
#define	D64_RS0_RS_MASK		0xf0000000	/* receive state */
#define	D64_RS0_RS_SHIFT		28
#define	D64_RS0_RS_DISABLED	0x00000000	/* disabled */
#define	D64_RS0_RS_ACTIVE	0x10000000	/* active */
#define	D64_RS0_RS_IDLE		0x20000000	/* idle wait */
#define	D64_RS0_RS_STOPPED	0x30000000	/* stopped */
#define	D64_RS0_RS_SUSP		0x40000000	/* suspend pending */

#define	D64_RS1_AD_MASK		0x0001ffff	/* active descriptor */
#define	D64_RS1_RE_MASK		0xf0000000	/* receive errors */
#define	D64_RS1_RE_SHIFT		28
#define	D64_RS1_RE_NOERR	0x00000000	/* no error */
#define	D64_RS1_RE_DPO		0x10000000	/* descriptor protocol error */
#define	D64_RS1_RE_DFU		0x20000000	/* data fifo overflow */
#define	D64_RS1_RE_DTE		0x30000000	/* data transfer error */
#define	D64_RS1_RE_DESRE	0x40000000	/* descriptor read error */
#define	D64_RS1_RE_COREE	0x50000000	/* core error */

/* fifoaddr */
#define	D64_FA_OFF_MASK		0xffff	/* offset */
#define	D64_FA_SEL_MASK		0xf0000	/* select */
#define	D64_FA_SEL_SHIFT	16
#define	D64_FA_SEL_XDD		0x00000	/* transmit dma data */
#define	D64_FA_SEL_XDP		0x10000	/* transmit dma pointers */
#define	D64_FA_SEL_RDD		0x40000	/* receive dma data */
#define	D64_FA_SEL_RDP		0x50000	/* receive dma pointers */
#define	D64_FA_SEL_XFD		0x80000	/* transmit fifo data */
#define	D64_FA_SEL_XFP		0x90000	/* transmit fifo pointers */
#define	D64_FA_SEL_RFD		0xc0000	/* receive fifo data */
#define	D64_FA_SEL_RFP		0xd0000	/* receive fifo pointers */
#define	D64_FA_SEL_RSD		0xe0000	/* receive frame status data */
#define	D64_FA_SEL_RSP		0xf0000	/* receive frame status pointers */

/* descriptor control flags 1 */
#define D64_CTRL_COREFLAGS	0x0ff00000	/* core specific flags */
#define	D64_CTRL1_EOT		((u32)1 << 28)	/* end of descriptor table */
#define	D64_CTRL1_IOC		((u32)1 << 29)	/* interrupt on completion */
#define	D64_CTRL1_EOF		((u32)1 << 30)	/* end of frame */
#define	D64_CTRL1_SOF		((u32)1 << 31)	/* start of frame */

/* descriptor control flags 2 */
/* buffer byte count. real data len must <= 16KB */
#define	D64_CTRL2_BC_MASK	0x00007fff
/* address extension bits */
#define	D64_CTRL2_AE		0x00030000
#define	D64_CTRL2_AE_SHIFT	16
/* parity bit */
#define D64_CTRL2_PARITY	0x00040000

/* control flags in the range [27:20] are core-specific and not defined here */
#define	D64_CTRL_CORE_MASK	0x0ff00000

#define D64_RX_FRM_STS_LEN	0x0000ffff	/* frame length mask */
#define D64_RX_FRM_STS_OVFL	0x00800000	/* RxOverFlow */
#define D64_RX_FRM_STS_DSCRCNT	0x0f000000  /* no. of descriptors used - 1 */
#define D64_RX_FRM_STS_DATATYPE	0xf0000000	/* core-dependent data type */

/*
 * packet headroom necessary to accommodate the largest header
 * in the system, (i.e TXOFF). By doing, we avoid the need to
 * allocate an extra buffer for the header when bridging to WL.
 * There is a compile time check in wlc.c which ensure that this
 * value is at least as big as TXOFF. This value is used in
 * dma_rxfill().
 */

#define BCMEXTRAHDROOM 172

/* debug/trace */
#ifdef BCMDBG
#define	DMA_ERROR(args) \
	do { \
		if (!(*di->msg_level & 1)) \
			; \
		else \
			printk args; \
	} while (0)
#define	DMA_TRACE(args) \
	do { \
		if (!(*di->msg_level & 2)) \
			; \
		else \
			printk args; \
	} while (0)
#else
#define	DMA_ERROR(args)
#define	DMA_TRACE(args)
#endif				/* BCMDBG */

#define	DMA_NONE(args)

#define	MAXNAMEL	8	/* 8 char names */

/* macros to convert between byte offsets and indexes */
#define	B2I(bytes, type)	((bytes) / sizeof(type))
#define	I2B(index, type)	((index) * sizeof(type))

#define	PCI32ADDR_HIGH		0xc0000000	/* address[31:30] */
#define	PCI32ADDR_HIGH_SHIFT	30	/* address[31:30] */

#define	PCI64ADDR_HIGH		0x80000000	/* address[63] */
#define	PCI64ADDR_HIGH_SHIFT	31	/* address[63] */

/*
 * DMA Descriptor
 * Descriptors are only read by the hardware, never written back.
 */
struct dma64desc {
	__le32 ctrl1;	/* misc control bits & bufcount */
	__le32 ctrl2;	/* buffer count and address extension */
	__le32 addrlow;	/* memory address of the date buffer, bits 31:0 */
	__le32 addrhigh; /* memory address of the date buffer, bits 63:32 */
};

/* dma engine software state */
struct dma_info {
	struct dma_pub dma; /* exported structure */
	uint *msg_level;	/* message level pointer */
	char name[MAXNAMEL];	/* callers name for diag msgs */

	struct pci_dev *pbus;		/* bus handle */

	bool dma64;	/* this dma engine is operating in 64-bit mode */
	bool addrext;	/* this dma engine supports DmaExtendedAddrChanges */

	/* 64-bit dma tx engine registers */
	struct dma64regs __iomem *d64txregs;
	/* 64-bit dma rx engine registers */
	struct dma64regs __iomem *d64rxregs;
	/* pointer to dma64 tx descriptor ring */
	struct dma64desc *txd64;
	/* pointer to dma64 rx descriptor ring */
	struct dma64desc *rxd64;

	u16 dmadesc_align;	/* alignment requirement for dma descriptors */

	u16 ntxd;		/* # tx descriptors tunable */
	u16 txin;		/* index of next descriptor to reclaim */
	u16 txout;		/* index of next descriptor to post */
	/* pointer to parallel array of pointers to packets */
	struct sk_buff **txp;
	/* Aligned physical address of descriptor ring */
	dma_addr_t txdpa;
	/* Original physical address of descriptor ring */
	dma_addr_t txdpaorig;
	u16 txdalign;	/* #bytes added to alloc'd mem to align txd */
	u32 txdalloc;	/* #bytes allocated for the ring */
	u32 xmtptrbase;	/* When using unaligned descriptors, the ptr register
			 * is not just an index, it needs all 13 bits to be
			 * an offset from the addr register.
			 */

	u16 nrxd;	/* # rx descriptors tunable */
	u16 rxin;	/* index of next descriptor to reclaim */
	u16 rxout;	/* index of next descriptor to post */
	/* pointer to parallel array of pointers to packets */
	struct sk_buff **rxp;
	/* Aligned physical address of descriptor ring */
	dma_addr_t rxdpa;
	/* Original physical address of descriptor ring */
	dma_addr_t rxdpaorig;
	u16 rxdalign;	/* #bytes added to alloc'd mem to align rxd */
	u32 rxdalloc;	/* #bytes allocated for the ring */
	u32 rcvptrbase;	/* Base for ptr reg when using unaligned descriptors */

	/* tunables */
	unsigned int rxbufsize;	/* rx buffer size in bytes, not including
				 * the extra headroom
				 */
	uint rxextrahdrroom;	/* extra rx headroom, reverseved to assist upper
				 * stack, e.g. some rx pkt buffers will be
				 * bridged to tx side without byte copying.
				 * The extra headroom needs to be large enough
				 * to fit txheader needs. Some dongle driver may
				 * not need it.
				 */
	uint nrxpost;		/* # rx buffers to keep posted */
	unsigned int rxoffset;	/* rxcontrol offset */
	/* add to get dma address of descriptor ring, low 32 bits */
	uint ddoffsetlow;
	/*   high 32 bits */
	uint ddoffsethigh;
	/* add to get dma address of data buffer, low 32 bits */
	uint dataoffsetlow;
	/*   high 32 bits */
	uint dataoffsethigh;
	/* descriptor base need to be aligned or not */
	bool aligndesc_4k;
};

/*
 * default dma message level (if input msg_level
 * pointer is null in dma_attach())
 */
static uint dma_msg_level;

/* Check for odd number of 1's */
static u32 parity32(__le32 data)
{
	/* no swap needed for counting 1's */
	u32 par_data = *(u32 *)&data;

	par_data ^= par_data >> 16;
	par_data ^= par_data >> 8;
	par_data ^= par_data >> 4;
	par_data ^= par_data >> 2;
	par_data ^= par_data >> 1;

	return par_data & 1;
}

static bool dma64_dd_parity(struct dma64desc *dd)
{
	return parity32(dd->addrlow ^ dd->addrhigh ^ dd->ctrl1 ^ dd->ctrl2);
}

/* descriptor bumping functions */

static uint xxd(uint x, uint n)
{
	return x & (n - 1); /* faster than %, but n must be power of 2 */
}

static uint txd(struct dma_info *di, uint x)
{
	return xxd(x, di->ntxd);
}

static uint rxd(struct dma_info *di, uint x)
{
	return xxd(x, di->nrxd);
}

static uint nexttxd(struct dma_info *di, uint i)
{
	return txd(di, i + 1);
}

static uint prevtxd(struct dma_info *di, uint i)
{
	return txd(di, i - 1);
}

static uint nextrxd(struct dma_info *di, uint i)
{
	return txd(di, i + 1);
}

static uint ntxdactive(struct dma_info *di, uint h, uint t)
{
	return txd(di, t-h);
}

static uint nrxdactive(struct dma_info *di, uint h, uint t)
{
	return rxd(di, t-h);
}

static uint _dma_ctrlflags(struct dma_info *di, uint mask, uint flags)
{
	uint dmactrlflags;

	if (di == NULL) {
		DMA_ERROR(("_dma_ctrlflags: NULL dma handle\n"));
		return 0;
	}

	dmactrlflags = di->dma.dmactrlflags;
	dmactrlflags &= ~mask;
	dmactrlflags |= flags;

	/* If trying to enable parity, check if parity is actually supported */
	if (dmactrlflags & DMA_CTRL_PEN) {
		u32 control;

		control = R_REG(&di->d64txregs->control);
		W_REG(&di->d64txregs->control,
		      control | D64_XC_PD);
		if (R_REG(&di->d64txregs->control) & D64_XC_PD)
			/* We *can* disable it so it is supported,
			 * restore control register
			 */
			W_REG(&di->d64txregs->control,
			control);
		else
			/* Not supported, don't allow it to be enabled */
			dmactrlflags &= ~DMA_CTRL_PEN;
	}

	di->dma.dmactrlflags = dmactrlflags;

	return dmactrlflags;
}

static bool _dma64_addrext(struct dma64regs __iomem *dma64regs)
{
	u32 w;
	OR_REG(&dma64regs->control, D64_XC_AE);
	w = R_REG(&dma64regs->control);
	AND_REG(&dma64regs->control, ~D64_XC_AE);
	return (w & D64_XC_AE) == D64_XC_AE;
}

/*
 * return true if this dma engine supports DmaExtendedAddrChanges,
 * otherwise false
 */
static bool _dma_isaddrext(struct dma_info *di)
{
	/* DMA64 supports full 32- or 64-bit operation. AE is always valid */

	/* not all tx or rx channel are available */
	if (di->d64txregs != NULL) {
		if (!_dma64_addrext(di->d64txregs))
			DMA_ERROR(("%s: _dma_isaddrext: DMA64 tx doesn't have "
				   "AE set\n", di->name));
		return true;
	} else if (di->d64rxregs != NULL) {
		if (!_dma64_addrext(di->d64rxregs))
			DMA_ERROR(("%s: _dma_isaddrext: DMA64 rx doesn't have "
				   "AE set\n", di->name));
		return true;
	}

	return false;
}

static bool _dma_descriptor_align(struct dma_info *di)
{
	u32 addrl;

	/* Check to see if the descriptors need to be aligned on 4K/8K or not */
	if (di->d64txregs != NULL) {
		W_REG(&di->d64txregs->addrlow, 0xff0);
		addrl = R_REG(&di->d64txregs->addrlow);
		if (addrl != 0)
			return false;
	} else if (di->d64rxregs != NULL) {
		W_REG(&di->d64rxregs->addrlow, 0xff0);
		addrl = R_REG(&di->d64rxregs->addrlow);
		if (addrl != 0)
			return false;
	}
	return true;
}

/*
 * Descriptor table must start at the DMA hardware dictated alignment, so
 * allocated memory must be large enough to support this requirement.
 */
static void *dma_alloc_consistent(struct pci_dev *pdev, uint size,
				  u16 align_bits, uint *alloced,
				  dma_addr_t *pap)
{
	if (align_bits) {
		u16 align = (1 << align_bits);
		if (!IS_ALIGNED(PAGE_SIZE, align))
			size += align;
		*alloced = size;
	}
	return pci_alloc_consistent(pdev, size, pap);
}

static
u8 dma_align_sizetobits(uint size)
{
	u8 bitpos = 0;
	while (size >>= 1)
		bitpos++;
	return bitpos;
}

/* This function ensures that the DMA descriptor ring will not get allocated
 * across Page boundary. If the allocation is done across the page boundary
 * at the first time, then it is freed and the allocation is done at
 * descriptor ring size aligned location. This will ensure that the ring will
 * not cross page boundary
 */
static void *dma_ringalloc(struct dma_info *di, u32 boundary, uint size,
			   u16 *alignbits, uint *alloced,
			   dma_addr_t *descpa)
{
	void *va;
	u32 desc_strtaddr;
	u32 alignbytes = 1 << *alignbits;

	va = dma_alloc_consistent(di->pbus, size, *alignbits, alloced, descpa);

	if (NULL == va)
		return NULL;

	desc_strtaddr = (u32) roundup((unsigned long)va, alignbytes);
	if (((desc_strtaddr + size - 1) & boundary) != (desc_strtaddr
							& boundary)) {
		*alignbits = dma_align_sizetobits(size);
		pci_free_consistent(di->pbus, size, va, *descpa);
		va = dma_alloc_consistent(di->pbus, size, *alignbits,
			alloced, descpa);
	}
	return va;
}

static bool dma64_alloc(struct dma_info *di, uint direction)
{
	u16 size;
	uint ddlen;
	void *va;
	uint alloced = 0;
	u16 align;
	u16 align_bits;

	ddlen = sizeof(struct dma64desc);

	size = (direction == DMA_TX) ? (di->ntxd * ddlen) : (di->nrxd * ddlen);
	align_bits = di->dmadesc_align;
	align = (1 << align_bits);

	if (direction == DMA_TX) {
		va = dma_ringalloc(di, D64RINGALIGN, size, &align_bits,
			&alloced, &di->txdpaorig);
		if (va == NULL) {
			DMA_ERROR(("%s: dma64_alloc: DMA_ALLOC_CONSISTENT(ntxd)"
				   " failed\n", di->name));
			return false;
		}
		align = (1 << align_bits);
		di->txd64 = (struct dma64desc *)
					roundup((unsigned long)va, align);
		di->txdalign = (uint) ((s8 *)di->txd64 - (s8 *) va);
		di->txdpa = di->txdpaorig + di->txdalign;
		di->txdalloc = alloced;
	} else {
		va = dma_ringalloc(di, D64RINGALIGN, size, &align_bits,
			&alloced, &di->rxdpaorig);
		if (va == NULL) {
			DMA_ERROR(("%s: dma64_alloc: DMA_ALLOC_CONSISTENT(nrxd)"
				   " failed\n", di->name));
			return false;
		}
		align = (1 << align_bits);
		di->rxd64 = (struct dma64desc *)
					roundup((unsigned long)va, align);
		di->rxdalign = (uint) ((s8 *)di->rxd64 - (s8 *) va);
		di->rxdpa = di->rxdpaorig + di->rxdalign;
		di->rxdalloc = alloced;
	}

	return true;
}

static bool _dma_alloc(struct dma_info *di, uint direction)
{
	return dma64_alloc(di, direction);
}

struct dma_pub *dma_attach(char *name, struct si_pub *sih,
		     void __iomem *dmaregstx, void __iomem *dmaregsrx,
		     uint ntxd, uint nrxd,
		     uint rxbufsize, int rxextheadroom,
		     uint nrxpost, uint rxoffset, uint *msg_level)
{
	struct dma_info *di;
	uint size;

	/* allocate private info structure */
	di = kzalloc(sizeof(struct dma_info), GFP_ATOMIC);
	if (di == NULL)
		return NULL;

	di->msg_level = msg_level ? msg_level : &dma_msg_level;


	di->dma64 = ((ai_core_sflags(sih, 0, 0) & SISF_DMA64) == SISF_DMA64);

	/* init dma reg pointer */
	di->d64txregs = (struct dma64regs __iomem *) dmaregstx;
	di->d64rxregs = (struct dma64regs __iomem *) dmaregsrx;

	/*
	 * Default flags (which can be changed by the driver calling
	 * dma_ctrlflags before enable): For backwards compatibility
	 * both Rx Overflow Continue and Parity are DISABLED.
	 */
	_dma_ctrlflags(di, DMA_CTRL_ROC | DMA_CTRL_PEN, 0);

	DMA_TRACE(("%s: dma_attach: %s flags 0x%x ntxd %d nrxd %d "
		   "rxbufsize %d rxextheadroom %d nrxpost %d rxoffset %d "
		   "dmaregstx %p dmaregsrx %p\n", name, "DMA64",
		   di->dma.dmactrlflags, ntxd, nrxd, rxbufsize,
		   rxextheadroom, nrxpost, rxoffset, dmaregstx, dmaregsrx));

	/* make a private copy of our callers name */
	strncpy(di->name, name, MAXNAMEL);
	di->name[MAXNAMEL - 1] = '\0';

	di->pbus = ((struct si_info *)sih)->pbus;

	/* save tunables */
	di->ntxd = (u16) ntxd;
	di->nrxd = (u16) nrxd;

	/* the actual dma size doesn't include the extra headroom */
	di->rxextrahdrroom =
	    (rxextheadroom == -1) ? BCMEXTRAHDROOM : rxextheadroom;
	if (rxbufsize > BCMEXTRAHDROOM)
		di->rxbufsize = (u16) (rxbufsize - di->rxextrahdrroom);
	else
		di->rxbufsize = (u16) rxbufsize;

	di->nrxpost = (u16) nrxpost;
	di->rxoffset = (u8) rxoffset;

	/*
	 * figure out the DMA physical address offset for dd and data
	 *     PCI/PCIE: they map silicon backplace address to zero
	 *     based memory, need offset
	 *     Other bus: use zero SI_BUS BIGENDIAN kludge: use sdram
	 *     swapped region for data buffer, not descriptor
	 */
	di->ddoffsetlow = 0;
	di->dataoffsetlow = 0;
	/* add offset for pcie with DMA64 bus */
	di->ddoffsetlow = 0;
	di->ddoffsethigh = SI_PCIE_DMA_H32;
	di->dataoffsetlow = di->ddoffsetlow;
	di->dataoffsethigh = di->ddoffsethigh;
	/* WAR64450 : DMACtl.Addr ext fields are not supported in SDIOD core. */
	if ((ai_coreid(sih) == SDIOD_CORE_ID)
	    && ((ai_corerev(sih) > 0) && (ai_corerev(sih) <= 2)))
		di->addrext = 0;
	else if ((ai_coreid(sih) == I2S_CORE_ID) &&
		 ((ai_corerev(sih) == 0) || (ai_corerev(sih) == 1)))
		di->addrext = 0;
	else
		di->addrext = _dma_isaddrext(di);

	/* does the descriptor need to be aligned and if yes, on 4K/8K or not */
	di->aligndesc_4k = _dma_descriptor_align(di);
	if (di->aligndesc_4k) {
		di->dmadesc_align = D64RINGALIGN_BITS;
		if ((ntxd < D64MAXDD / 2) && (nrxd < D64MAXDD / 2))
			/* for smaller dd table, HW relax alignment reqmnt */
			di->dmadesc_align = D64RINGALIGN_BITS - 1;
	} else {
		di->dmadesc_align = 4;	/* 16 byte alignment */
	}

	DMA_NONE(("DMA descriptor align_needed %d, align %d\n",
		  di->aligndesc_4k, di->dmadesc_align));

	/* allocate tx packet pointer vector */
	if (ntxd) {
		size = ntxd * sizeof(void *);
		di->txp = kzalloc(size, GFP_ATOMIC);
		if (di->txp == NULL)
			goto fail;
	}

	/* allocate rx packet pointer vector */
	if (nrxd) {
		size = nrxd * sizeof(void *);
		di->rxp = kzalloc(size, GFP_ATOMIC);
		if (di->rxp == NULL)
			goto fail;
	}

	/*
	 * allocate transmit descriptor ring, only need ntxd descriptors
	 * but it must be aligned
	 */
	if (ntxd) {
		if (!_dma_alloc(di, DMA_TX))
			goto fail;
	}

	/*
	 * allocate receive descriptor ring, only need nrxd descriptors
	 * but it must be aligned
	 */
	if (nrxd) {
		if (!_dma_alloc(di, DMA_RX))
			goto fail;
	}

	if ((di->ddoffsetlow != 0) && !di->addrext) {
		if (di->txdpa > SI_PCI_DMA_SZ) {
			DMA_ERROR(("%s: dma_attach: txdpa 0x%x: addrext not "
				   "supported\n", di->name, (u32)di->txdpa));
			goto fail;
		}
		if (di->rxdpa > SI_PCI_DMA_SZ) {
			DMA_ERROR(("%s: dma_attach: rxdpa 0x%x: addrext not "
				   "supported\n", di->name, (u32)di->rxdpa));
			goto fail;
		}
	}

	DMA_TRACE(("ddoffsetlow 0x%x ddoffsethigh 0x%x dataoffsetlow 0x%x "
		   "dataoffsethigh " "0x%x addrext %d\n", di->ddoffsetlow,
		   di->ddoffsethigh, di->dataoffsetlow, di->dataoffsethigh,
		   di->addrext));

	return (struct dma_pub *) di;

 fail:
	dma_detach((struct dma_pub *)di);
	return NULL;
}

static inline void
dma64_dd_upd(struct dma_info *di, struct dma64desc *ddring,
	     dma_addr_t pa, uint outidx, u32 *flags, u32 bufcount)
{
	u32 ctrl2 = bufcount & D64_CTRL2_BC_MASK;

	/* PCI bus with big(>1G) physical address, use address extension */
	if ((di->dataoffsetlow == 0) || !(pa & PCI32ADDR_HIGH)) {
		ddring[outidx].addrlow = cpu_to_le32(pa + di->dataoffsetlow);
		ddring[outidx].addrhigh = cpu_to_le32(di->dataoffsethigh);
		ddring[outidx].ctrl1 = cpu_to_le32(*flags);
		ddring[outidx].ctrl2 = cpu_to_le32(ctrl2);
	} else {
		/* address extension for 32-bit PCI */
		u32 ae;

		ae = (pa & PCI32ADDR_HIGH) >> PCI32ADDR_HIGH_SHIFT;
		pa &= ~PCI32ADDR_HIGH;

		ctrl2 |= (ae << D64_CTRL2_AE_SHIFT) & D64_CTRL2_AE;
		ddring[outidx].addrlow = cpu_to_le32(pa + di->dataoffsetlow);
		ddring[outidx].addrhigh = cpu_to_le32(di->dataoffsethigh);
		ddring[outidx].ctrl1 = cpu_to_le32(*flags);
		ddring[outidx].ctrl2 = cpu_to_le32(ctrl2);
	}
	if (di->dma.dmactrlflags & DMA_CTRL_PEN) {
		if (dma64_dd_parity(&ddring[outidx]))
			ddring[outidx].ctrl2 =
			     cpu_to_le32(ctrl2 | D64_CTRL2_PARITY);
	}
}

/* !! may be called with core in reset */
void dma_detach(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;

	DMA_TRACE(("%s: dma_detach\n", di->name));

	/* free dma descriptor rings */
	if (di->txd64)
		pci_free_consistent(di->pbus, di->txdalloc,
				    ((s8 *)di->txd64 - di->txdalign),
				    (di->txdpaorig));
	if (di->rxd64)
		pci_free_consistent(di->pbus, di->rxdalloc,
				    ((s8 *)di->rxd64 - di->rxdalign),
				    (di->rxdpaorig));

	/* free packet pointer vectors */
	kfree(di->txp);
	kfree(di->rxp);

	/* free our private info structure */
	kfree(di);

}

/* initialize descriptor table base address */
static void
_dma_ddtable_init(struct dma_info *di, uint direction, dma_addr_t pa)
{
	if (!di->aligndesc_4k) {
		if (direction == DMA_TX)
			di->xmtptrbase = pa;
		else
			di->rcvptrbase = pa;
	}

	if ((di->ddoffsetlow == 0)
	    || !(pa & PCI32ADDR_HIGH)) {
		if (direction == DMA_TX) {
			W_REG(&di->d64txregs->addrlow, pa + di->ddoffsetlow);
			W_REG(&di->d64txregs->addrhigh, di->ddoffsethigh);
		} else {
			W_REG(&di->d64rxregs->addrlow, pa + di->ddoffsetlow);
			W_REG(&di->d64rxregs->addrhigh, di->ddoffsethigh);
		}
	} else {
		/* DMA64 32bits address extension */
		u32 ae;

		/* shift the high bit(s) from pa to ae */
		ae = (pa & PCI32ADDR_HIGH) >> PCI32ADDR_HIGH_SHIFT;
		pa &= ~PCI32ADDR_HIGH;

		if (direction == DMA_TX) {
			W_REG(&di->d64txregs->addrlow, pa + di->ddoffsetlow);
			W_REG(&di->d64txregs->addrhigh, di->ddoffsethigh);
			SET_REG(&di->d64txregs->control,
				D64_XC_AE, (ae << D64_XC_AE_SHIFT));
		} else {
			W_REG(&di->d64rxregs->addrlow, pa + di->ddoffsetlow);
			W_REG(&di->d64rxregs->addrhigh, di->ddoffsethigh);
			SET_REG(&di->d64rxregs->control,
				D64_RC_AE, (ae << D64_RC_AE_SHIFT));
		}
	}
}

static void _dma_rxenable(struct dma_info *di)
{
	uint dmactrlflags = di->dma.dmactrlflags;
	u32 control;

	DMA_TRACE(("%s: dma_rxenable\n", di->name));

	control =
	    (R_REG(&di->d64rxregs->control) & D64_RC_AE) |
	    D64_RC_RE;

	if ((dmactrlflags & DMA_CTRL_PEN) == 0)
		control |= D64_RC_PD;

	if (dmactrlflags & DMA_CTRL_ROC)
		control |= D64_RC_OC;

	W_REG(&di->d64rxregs->control,
		((di->rxoffset << D64_RC_RO_SHIFT) | control));
}

void dma_rxinit(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;

	DMA_TRACE(("%s: dma_rxinit\n", di->name));

	if (di->nrxd == 0)
		return;

	di->rxin = di->rxout = 0;

	/* clear rx descriptor ring */
	memset(di->rxd64, '\0', di->nrxd * sizeof(struct dma64desc));

	/* DMA engine with out alignment requirement requires table to be inited
	 * before enabling the engine
	 */
	if (!di->aligndesc_4k)
		_dma_ddtable_init(di, DMA_RX, di->rxdpa);

	_dma_rxenable(di);

	if (di->aligndesc_4k)
		_dma_ddtable_init(di, DMA_RX, di->rxdpa);
}

static struct sk_buff *dma64_getnextrxp(struct dma_info *di, bool forceall)
{
	uint i, curr;
	struct sk_buff *rxp;
	dma_addr_t pa;

	i = di->rxin;

	/* return if no packets posted */
	if (i == di->rxout)
		return NULL;

	curr =
	    B2I(((R_REG(&di->d64rxregs->status0) & D64_RS0_CD_MASK) -
		 di->rcvptrbase) & D64_RS0_CD_MASK, struct dma64desc);

	/* ignore curr if forceall */
	if (!forceall && (i == curr))
		return NULL;

	/* get the packet pointer that corresponds to the rx descriptor */
	rxp = di->rxp[i];
	di->rxp[i] = NULL;

	pa = le32_to_cpu(di->rxd64[i].addrlow) - di->dataoffsetlow;

	/* clear this packet from the descriptor ring */
	pci_unmap_single(di->pbus, pa, di->rxbufsize, PCI_DMA_FROMDEVICE);

	di->rxd64[i].addrlow = cpu_to_le32(0xdeadbeef);
	di->rxd64[i].addrhigh = cpu_to_le32(0xdeadbeef);

	di->rxin = nextrxd(di, i);

	return rxp;
}

static struct sk_buff *_dma_getnextrxp(struct dma_info *di, bool forceall)
{
	if (di->nrxd == 0)
		return NULL;

	return dma64_getnextrxp(di, forceall);
}

/*
 * !! rx entry routine
 * returns a pointer to the next frame received, or NULL if there are no more
 *   if DMA_CTRL_RXMULTI is defined, DMA scattering(multiple buffers) is
 *   supported with pkts chain
 *   otherwise, it's treated as giant pkt and will be tossed.
 *   The DMA scattering starts with normal DMA header, followed by first
 *   buffer data. After it reaches the max size of buffer, the data continues
 *   in next DMA descriptor buffer WITHOUT DMA header
 */
struct sk_buff *dma_rx(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	struct sk_buff *p, *head, *tail;
	uint len;
	uint pkt_len;
	int resid = 0;

 next_frame:
	head = _dma_getnextrxp(di, false);
	if (head == NULL)
		return NULL;

	len = le16_to_cpu(*(__le16 *) (head->data));
	DMA_TRACE(("%s: dma_rx len %d\n", di->name, len));
	dma_spin_for_len(len, head);

	/* set actual length */
	pkt_len = min((di->rxoffset + len), di->rxbufsize);
	__skb_trim(head, pkt_len);
	resid = len - (di->rxbufsize - di->rxoffset);

	/* check for single or multi-buffer rx */
	if (resid > 0) {
		tail = head;
		while ((resid > 0) && (p = _dma_getnextrxp(di, false))) {
			tail->next = p;
			pkt_len = min_t(uint, resid, di->rxbufsize);
			__skb_trim(p, pkt_len);

			tail = p;
			resid -= di->rxbufsize;
		}

#ifdef BCMDBG
		if (resid > 0) {
			uint cur;
			cur =
			    B2I(((R_REG(&di->d64rxregs->status0) &
				  D64_RS0_CD_MASK) -
				 di->rcvptrbase) & D64_RS0_CD_MASK,
				struct dma64desc);
			DMA_ERROR(("dma_rx, rxin %d rxout %d, hw_curr %d\n",
				   di->rxin, di->rxout, cur));
		}
#endif				/* BCMDBG */

		if ((di->dma.dmactrlflags & DMA_CTRL_RXMULTI) == 0) {
			DMA_ERROR(("%s: dma_rx: bad frame length (%d)\n",
				   di->name, len));
			brcmu_pkt_buf_free_skb(head);
			di->dma.rxgiants++;
			goto next_frame;
		}
	}

	return head;
}

static bool dma64_rxidle(struct dma_info *di)
{
	DMA_TRACE(("%s: dma_rxidle\n", di->name));

	if (di->nrxd == 0)
		return true;

	return ((R_REG(&di->d64rxregs->status0) & D64_RS0_CD_MASK) ==
		(R_REG(&di->d64rxregs->ptr) & D64_RS0_CD_MASK));
}

/*
 * post receive buffers
 *  return false is refill failed completely and ring is empty this will stall
 *  the rx dma and user might want to call rxfill again asap. This unlikely
 *  happens on memory-rich NIC, but often on memory-constrained dongle
 */
bool dma_rxfill(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	struct sk_buff *p;
	u16 rxin, rxout;
	u32 flags = 0;
	uint n;
	uint i;
	dma_addr_t pa;
	uint extra_offset = 0;
	bool ring_empty;

	ring_empty = false;

	/*
	 * Determine how many receive buffers we're lacking
	 * from the full complement, allocate, initialize,
	 * and post them, then update the chip rx lastdscr.
	 */

	rxin = di->rxin;
	rxout = di->rxout;

	n = di->nrxpost - nrxdactive(di, rxin, rxout);

	DMA_TRACE(("%s: dma_rxfill: post %d\n", di->name, n));

	if (di->rxbufsize > BCMEXTRAHDROOM)
		extra_offset = di->rxextrahdrroom;

	for (i = 0; i < n; i++) {
		/*
		 * the di->rxbufsize doesn't include the extra headroom,
		 * we need to add it to the size to be allocated
		 */
		p = brcmu_pkt_buf_get_skb(di->rxbufsize + extra_offset);

		if (p == NULL) {
			DMA_ERROR(("%s: dma_rxfill: out of rxbufs\n",
				   di->name));
			if (i == 0 && dma64_rxidle(di)) {
				DMA_ERROR(("%s: rxfill64: ring is empty !\n",
					   di->name));
				ring_empty = true;
			}
			di->dma.rxnobuf++;
			break;
		}
		/* reserve an extra headroom, if applicable */
		if (extra_offset)
			skb_pull(p, extra_offset);

		/* Do a cached write instead of uncached write since DMA_MAP
		 * will flush the cache.
		 */
		*(u32 *) (p->data) = 0;

		pa = pci_map_single(di->pbus, p->data,
			di->rxbufsize, PCI_DMA_FROMDEVICE);

		/* save the free packet pointer */
		di->rxp[rxout] = p;

		/* reset flags for each descriptor */
		flags = 0;
		if (rxout == (di->nrxd - 1))
			flags = D64_CTRL1_EOT;

		dma64_dd_upd(di, di->rxd64, pa, rxout, &flags,
			     di->rxbufsize);
		rxout = nextrxd(di, rxout);
	}

	di->rxout = rxout;

	/* update the chip lastdscr pointer */
	W_REG(&di->d64rxregs->ptr,
	      di->rcvptrbase + I2B(rxout, struct dma64desc));

	return ring_empty;
}

void dma_rxreclaim(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	struct sk_buff *p;

	DMA_TRACE(("%s: dma_rxreclaim\n", di->name));

	while ((p = _dma_getnextrxp(di, true)))
		brcmu_pkt_buf_free_skb(p);
}

void dma_counterreset(struct dma_pub *pub)
{
	/* reset all software counters */
	pub->rxgiants = 0;
	pub->rxnobuf = 0;
	pub->txnobuf = 0;
}

/* get the address of the var in order to change later */
unsigned long dma_getvar(struct dma_pub *pub, const char *name)
{
	struct dma_info *di = (struct dma_info *)pub;

	if (!strcmp(name, "&txavail"))
		return (unsigned long)&(di->dma.txavail);
	return 0;
}

/* 64-bit DMA functions */

void dma_txinit(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	u32 control = D64_XC_XE;

	DMA_TRACE(("%s: dma_txinit\n", di->name));

	if (di->ntxd == 0)
		return;

	di->txin = di->txout = 0;
	di->dma.txavail = di->ntxd - 1;

	/* clear tx descriptor ring */
	memset(di->txd64, '\0', (di->ntxd * sizeof(struct dma64desc)));

	/* DMA engine with out alignment requirement requires table to be inited
	 * before enabling the engine
	 */
	if (!di->aligndesc_4k)
		_dma_ddtable_init(di, DMA_TX, di->txdpa);

	if ((di->dma.dmactrlflags & DMA_CTRL_PEN) == 0)
		control |= D64_XC_PD;
	OR_REG(&di->d64txregs->control, control);

	/* DMA engine with alignment requirement requires table to be inited
	 * before enabling the engine
	 */
	if (di->aligndesc_4k)
		_dma_ddtable_init(di, DMA_TX, di->txdpa);
}

void dma_txsuspend(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;

	DMA_TRACE(("%s: dma_txsuspend\n", di->name));

	if (di->ntxd == 0)
		return;

	OR_REG(&di->d64txregs->control, D64_XC_SE);
}

void dma_txresume(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;

	DMA_TRACE(("%s: dma_txresume\n", di->name));

	if (di->ntxd == 0)
		return;

	AND_REG(&di->d64txregs->control, ~D64_XC_SE);
}

bool dma_txsuspended(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;

	return (di->ntxd == 0) ||
	    ((R_REG(&di->d64txregs->control) & D64_XC_SE) ==
	     D64_XC_SE);
}

void dma_txreclaim(struct dma_pub *pub, enum txd_range range)
{
	struct dma_info *di = (struct dma_info *)pub;
	struct sk_buff *p;

	DMA_TRACE(("%s: dma_txreclaim %s\n", di->name,
		   (range == DMA_RANGE_ALL) ? "all" :
		   ((range ==
		     DMA_RANGE_TRANSMITTED) ? "transmitted" :
		    "transferred")));

	if (di->txin == di->txout)
		return;

	while ((p = dma_getnexttxp(pub, range))) {
		/* For unframed data, we don't have any packets to free */
		if (!(di->dma.dmactrlflags & DMA_CTRL_UNFRAMED))
			brcmu_pkt_buf_free_skb(p);
	}
}

bool dma_txreset(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	u32 status;

	if (di->ntxd == 0)
		return true;

	/* suspend tx DMA first */
	W_REG(&di->d64txregs->control, D64_XC_SE);
	SPINWAIT(((status =
		   (R_REG(&di->d64txregs->status0) & D64_XS0_XS_MASK))
		  != D64_XS0_XS_DISABLED) && (status != D64_XS0_XS_IDLE)
		 && (status != D64_XS0_XS_STOPPED), 10000);

	W_REG(&di->d64txregs->control, 0);
	SPINWAIT(((status =
		   (R_REG(&di->d64txregs->status0) & D64_XS0_XS_MASK))
		  != D64_XS0_XS_DISABLED), 10000);

	/* wait for the last transaction to complete */
	udelay(300);

	return status == D64_XS0_XS_DISABLED;
}

bool dma_rxreset(struct dma_pub *pub)
{
	struct dma_info *di = (struct dma_info *)pub;
	u32 status;

	if (di->nrxd == 0)
		return true;

	W_REG(&di->d64rxregs->control, 0);
	SPINWAIT(((status =
		   (R_REG(&di->d64rxregs->status0) & D64_RS0_RS_MASK))
		  != D64_RS0_RS_DISABLED), 10000);

	return status == D64_RS0_RS_DISABLED;
}

/*
 * !! tx entry routine
 * WARNING: call must check the return value for error.
 *   the error(toss frames) could be fatal and cause many subsequent hard
 *   to debug problems
 */
int dma_txfast(struct dma_pub *pub, struct sk_buff *p0, bool commit)
{
	struct dma_info *di = (struct dma_info *)pub;
	struct sk_buff *p, *next;
	unsigned char *data;
	uint len;
	u16 txout;
	u32 flags = 0;
	dma_addr_t pa;

	DMA_TRACE(("%s: dma_txfast\n", di->name));

	txout = di->txout;

	/*
	 * Walk the chain of packet buffers
	 * allocating and initializing transmit descriptor entries.
	 */
	for (p = p0; p; p = next) {
		data = p->data;
		len = p->len;
		next = p->next;

		/* return nonzero if out of tx descriptors */
		if (nexttxd(di, txout) == di->txin)
			goto outoftxd;

		if (len == 0)
			continue;

		/* get physical address of buffer start */
		pa = pci_map_single(di->pbus, data, len, PCI_DMA_TODEVICE);

		flags = 0;
		if (p == p0)
			flags |= D64_CTRL1_SOF;

		/* With a DMA segment list, Descriptor table is filled
		 * using the segment list instead of looping over
		 * buffers in multi-chain DMA. Therefore, EOF for SGLIST
		 * is when end of segment list is reached.
		 */
		if (next == NULL)
			flags |= (D64_CTRL1_IOC | D64_CTRL1_EOF);
		if (txout == (di->ntxd - 1))
			flags |= D64_CTRL1_EOT;

		dma64_dd_upd(di, di->txd64, pa, txout, &flags, len);

		txout = nexttxd(di, txout);
	}

	/* if last txd eof not set, fix it */
	if (!(flags & D64_CTRL1_EOF))
		di->txd64[prevtxd(di, txout)].ctrl1 =
		     cpu_to_le32(flags | D64_CTRL1_IOC | D64_CTRL1_EOF);

	/* save the packet */
	di->txp[prevtxd(di, txout)] = p0;

	/* bump the tx descriptor index */
	di->txout = txout;

	/* kick the chip */
	if (commit)
		W_REG(&di->d64txregs->ptr,
		      di->xmtptrbase + I2B(txout, struct dma64desc));

	/* tx flow control */
	di->dma.txavail = di->ntxd - ntxdactive(di, di->txin, di->txout) - 1;

	return 0;

 outoftxd:
	DMA_ERROR(("%s: dma_txfast: out of txds !!!\n", di->name));
	brcmu_pkt_buf_free_skb(p0);
	di->dma.txavail = 0;
	di->dma.txnobuf++;
	return -1;
}

/*
 * Reclaim next completed txd (txds if using chained buffers) in the range
 * specified and return associated packet.
 * If range is DMA_RANGE_TRANSMITTED, reclaim descriptors that have be
 * transmitted as noted by the hardware "CurrDescr" pointer.
 * If range is DMA_RANGE_TRANSFERED, reclaim descriptors that have be
 * transferred by the DMA as noted by the hardware "ActiveDescr" pointer.
 * If range is DMA_RANGE_ALL, reclaim all txd(s) posted to the ring and
 * return associated packet regardless of the value of hardware pointers.
 */
struct sk_buff *dma_getnexttxp(struct dma_pub *pub, enum txd_range range)
{
	struct dma_info *di = (struct dma_info *)pub;
	u16 start, end, i;
	u16 active_desc;
	struct sk_buff *txp;

	DMA_TRACE(("%s: dma_getnexttxp %s\n", di->name,
		   (range == DMA_RANGE_ALL) ? "all" :
		   ((range ==
		     DMA_RANGE_TRANSMITTED) ? "transmitted" :
		    "transferred")));

	if (di->ntxd == 0)
		return NULL;

	txp = NULL;

	start = di->txin;
	if (range == DMA_RANGE_ALL)
		end = di->txout;
	else {
		struct dma64regs __iomem *dregs = di->d64txregs;

		end = (u16) (B2I(((R_REG(&dregs->status0) &
				 D64_XS0_CD_MASK) -
				 di->xmtptrbase) & D64_XS0_CD_MASK,
				 struct dma64desc));

		if (range == DMA_RANGE_TRANSFERED) {
			active_desc =
			    (u16) (R_REG(&dregs->status1) &
				      D64_XS1_AD_MASK);
			active_desc =
			    (active_desc - di->xmtptrbase) & D64_XS0_CD_MASK;
			active_desc = B2I(active_desc, struct dma64desc);
			if (end != active_desc)
				end = prevtxd(di, active_desc);
		}
	}

	if ((start == 0) && (end > di->txout))
		goto bogus;

	for (i = start; i != end && !txp; i = nexttxd(di, i)) {
		dma_addr_t pa;
		uint size;

		pa = le32_to_cpu(di->txd64[i].addrlow) - di->dataoffsetlow;

		size =
		    (le32_to_cpu(di->txd64[i].ctrl2) &
		     D64_CTRL2_BC_MASK);

		di->txd64[i].addrlow = cpu_to_le32(0xdeadbeef);
		di->txd64[i].addrhigh = cpu_to_le32(0xdeadbeef);

		txp = di->txp[i];
		di->txp[i] = NULL;

		pci_unmap_single(di->pbus, pa, size, PCI_DMA_TODEVICE);
	}

	di->txin = i;

	/* tx flow control */
	di->dma.txavail = di->ntxd - ntxdactive(di, di->txin, di->txout) - 1;

	return txp;

 bogus:
	DMA_NONE(("dma_getnexttxp: bogus curr: start %d end %d txout %d "
		  "force %d\n", start, end, di->txout, forceall));
	return NULL;
}

/*
 * Mac80211 initiated actions sometimes require packets in the DMA queue to be
 * modified. The modified portion of the packet is not under control of the DMA
 * engine. This function calls a caller-supplied function for each packet in
 * the caller specified dma chain.
 */
void dma_walk_packets(struct dma_pub *dmah, void (*callback_fnc)
		      (void *pkt, void *arg_a), void *arg_a)
{
	struct dma_info *di = (struct dma_info *) dmah;
	uint i =   di->txin;
	uint end = di->txout;
	struct sk_buff *skb;
	struct ieee80211_tx_info *tx_info;

	while (i != end) {
		skb = (struct sk_buff *)di->txp[i];
		if (skb != NULL) {
			tx_info = (struct ieee80211_tx_info *)skb->cb;
			(callback_fnc)(tx_info, arg_a);
		}
		i = nexttxd(di, i);
	}
}
OpenPOWER on IntegriCloud