summaryrefslogtreecommitdiff
path: root/drivers/net/wireless/intel/iwlwifi/pcie/tx.c
blob: de50418adae5013173facd2bcfaf05c9bb5bf90b (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
1315
1316
1317
1318
1319
1320
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
1342
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421
1422
1423
1424
1425
1426
1427
1428
1429
1430
1431
1432
1433
1434
1435
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
1514
1515
1516
1517
1518
1519
1520
1521
1522
1523
1524
1525
1526
1527
1528
1529
1530
1531
1532
1533
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564
1565
1566
1567
1568
1569
1570
1571
1572
1573
1574
1575
1576
1577
1578
1579
1580
1581
1582
1583
1584
1585
1586
1587
1588
1589
1590
1591
1592
1593
1594
1595
1596
1597
1598
1599
1600
1601
1602
1603
1604
1605
1606
1607
1608
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620
1621
1622
1623
1624
1625
1626
1627
1628
1629
1630
1631
1632
1633
1634
1635
1636
1637
1638
1639
1640
1641
1642
1643
1644
1645
1646
1647
1648
1649
1650
1651
1652
1653
1654
1655
1656
1657
1658
1659
1660
1661
1662
1663
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
1691
1692
1693
1694
1695
1696
1697
1698
1699
1700
1701
1702
1703
1704
1705
1706
1707
1708
1709
1710
1711
1712
1713
1714
1715
1716
1717
1718
1719
1720
1721
1722
1723
1724
1725
1726
1727
1728
1729
1730
1731
1732
1733
1734
1735
1736
1737
1738
1739
1740
1741
1742
1743
1744
1745
1746
1747
1748
1749
1750
1751
1752
1753
1754
1755
1756
1757
1758
1759
1760
1761
1762
1763
1764
1765
1766
1767
1768
1769
1770
1771
1772
1773
1774
1775
1776
1777
1778
1779
1780
1781
1782
1783
1784
1785
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
1800
1801
1802
1803
1804
1805
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
1816
1817
1818
1819
1820
1821
1822
1823
1824
1825
1826
1827
1828
1829
1830
1831
1832
1833
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
1856
1857
1858
1859
1860
1861
1862
1863
1864
1865
1866
1867
1868
1869
1870
1871
1872
1873
1874
1875
1876
1877
1878
1879
1880
1881
1882
1883
1884
1885
1886
1887
1888
1889
1890
1891
1892
1893
1894
1895
1896
1897
1898
1899
1900
1901
1902
1903
1904
1905
1906
1907
1908
1909
1910
1911
1912
1913
1914
1915
1916
1917
1918
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
1930
1931
1932
1933
1934
1935
1936
1937
1938
1939
1940
1941
1942
1943
1944
1945
1946
1947
1948
1949
1950
1951
1952
1953
1954
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
2031
2032
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
2067
2068
2069
2070
2071
2072
2073
2074
2075
2076
2077
2078
2079
2080
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
2136
2137
2138
2139
2140
2141
2142
2143
2144
2145
2146
2147
2148
2149
2150
2151
2152
2153
2154
2155
2156
2157
2158
2159
2160
2161
2162
2163
2164
2165
2166
2167
2168
2169
2170
2171
2172
2173
2174
2175
2176
2177
2178
2179
2180
2181
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
2198
2199
2200
2201
2202
2203
2204
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
2226
2227
2228
2229
2230
2231
2232
2233
2234
2235
2236
2237
2238
2239
2240
2241
2242
2243
2244
2245
2246
2247
2248
2249
2250
2251
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
2286
2287
2288
2289
2290
2291
2292
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
2352
2353
2354
2355
2356
2357
2358
2359
2360
2361
2362
2363
2364
2365
2366
2367
2368
2369
2370
2371
2372
2373
2374
2375
2376
2377
2378
2379
2380
2381
2382
2383
2384
2385
2386
2387
2388
2389
2390
2391
2392
2393
2394
2395
2396
2397
2398
2399
2400
2401
2402
2403
2404
2405
2406
2407
2408
2409
2410
2411
2412
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
/******************************************************************************
 *
 * Copyright(c) 2003 - 2014 Intel Corporation. All rights reserved.
 * Copyright(c) 2013 - 2015 Intel Mobile Communications GmbH
 * Copyright(c) 2016 - 2017 Intel Deutschland GmbH
 *
 * Portions of this file are derived from the ipw3945 project, as well
 * as portions of the ieee80211 subsystem header files.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms of version 2 of the GNU General Public License as
 * published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License along with
 * this program; if not, write to the Free Software Foundation, Inc.,
 * 51 Franklin Street, Fifth Floor, Boston, MA 02110, USA
 *
 * The full GNU General Public License is included in this distribution in the
 * file called LICENSE.
 *
 * Contact Information:
 *  Intel Linux Wireless <linuxwifi@intel.com>
 * Intel Corporation, 5200 N.E. Elam Young Parkway, Hillsboro, OR 97124-6497
 *
 *****************************************************************************/
#include <linux/etherdevice.h>
#include <linux/ieee80211.h>
#include <linux/slab.h>
#include <linux/sched.h>
#include <linux/pm_runtime.h>
#include <net/ip6_checksum.h>
#include <net/tso.h>

#include "iwl-debug.h"
#include "iwl-csr.h"
#include "iwl-prph.h"
#include "iwl-io.h"
#include "iwl-scd.h"
#include "iwl-op-mode.h"
#include "internal.h"
/* FIXME: need to abstract out TX command (once we know what it looks like) */
#include "dvm/commands.h"

#define IWL_TX_CRC_SIZE 4
#define IWL_TX_DELIMITER_SIZE 4

/*************** DMA-QUEUE-GENERAL-FUNCTIONS  *****
 * DMA services
 *
 * Theory of operation
 *
 * A Tx or Rx queue resides in host DRAM, and is comprised of a circular buffer
 * of buffer descriptors, each of which points to one or more data buffers for
 * the device to read from or fill.  Driver and device exchange status of each
 * queue via "read" and "write" pointers.  Driver keeps minimum of 2 empty
 * entries in each circular buffer, to protect against confusing empty and full
 * queue states.
 *
 * The device reads or writes the data in the queues via the device's several
 * DMA/FIFO channels.  Each queue is mapped to a single DMA channel.
 *
 * For Tx queue, there are low mark and high mark limits. If, after queuing
 * the packet for Tx, free space become < low mark, Tx queue stopped. When
 * reclaiming packets (on 'tx done IRQ), if free space become > high mark,
 * Tx queue resumed.
 *
 ***************************************************/

int iwl_queue_space(const struct iwl_txq *q)
{
	unsigned int max;
	unsigned int used;

	/*
	 * To avoid ambiguity between empty and completely full queues, there
	 * should always be less than TFD_QUEUE_SIZE_MAX elements in the queue.
	 * If q->n_window is smaller than TFD_QUEUE_SIZE_MAX, there is no need
	 * to reserve any queue entries for this purpose.
	 */
	if (q->n_window < TFD_QUEUE_SIZE_MAX)
		max = q->n_window;
	else
		max = TFD_QUEUE_SIZE_MAX - 1;

	/*
	 * TFD_QUEUE_SIZE_MAX is a power of 2, so the following is equivalent to
	 * modulo by TFD_QUEUE_SIZE_MAX and is well defined.
	 */
	used = (q->write_ptr - q->read_ptr) & (TFD_QUEUE_SIZE_MAX - 1);

	if (WARN_ON(used > max))
		return 0;

	return max - used;
}

/*
 * iwl_queue_init - Initialize queue's high/low-water and read/write indexes
 */
static int iwl_queue_init(struct iwl_txq *q, int slots_num)
{
	q->n_window = slots_num;

	/* slots_num must be power-of-two size, otherwise
	 * get_cmd_index is broken. */
	if (WARN_ON(!is_power_of_2(slots_num)))
		return -EINVAL;

	q->low_mark = q->n_window / 4;
	if (q->low_mark < 4)
		q->low_mark = 4;

	q->high_mark = q->n_window / 8;
	if (q->high_mark < 2)
		q->high_mark = 2;

	q->write_ptr = 0;
	q->read_ptr = 0;

	return 0;
}

int iwl_pcie_alloc_dma_ptr(struct iwl_trans *trans,
			   struct iwl_dma_ptr *ptr, size_t size)
{
	if (WARN_ON(ptr->addr))
		return -EINVAL;

	ptr->addr = dma_alloc_coherent(trans->dev, size,
				       &ptr->dma, GFP_KERNEL);
	if (!ptr->addr)
		return -ENOMEM;
	ptr->size = size;
	return 0;
}

void iwl_pcie_free_dma_ptr(struct iwl_trans *trans, struct iwl_dma_ptr *ptr)
{
	if (unlikely(!ptr->addr))
		return;

	dma_free_coherent(trans->dev, ptr->size, ptr->addr, ptr->dma);
	memset(ptr, 0, sizeof(*ptr));
}

static void iwl_pcie_txq_stuck_timer(unsigned long data)
{
	struct iwl_txq *txq = (void *)data;
	struct iwl_trans_pcie *trans_pcie = txq->trans_pcie;
	struct iwl_trans *trans = iwl_trans_pcie_get_trans(trans_pcie);

	spin_lock(&txq->lock);
	/* check if triggered erroneously */
	if (txq->read_ptr == txq->write_ptr) {
		spin_unlock(&txq->lock);
		return;
	}
	spin_unlock(&txq->lock);

	iwl_trans_pcie_log_scd_error(trans, txq);

	iwl_force_nmi(trans);
}

/*
 * iwl_pcie_txq_update_byte_cnt_tbl - Set up entry in Tx byte-count array
 */
static void iwl_pcie_txq_update_byte_cnt_tbl(struct iwl_trans *trans,
					     struct iwl_txq *txq, u16 byte_cnt,
					     int num_tbs)
{
	struct iwlagn_scd_bc_tbl *scd_bc_tbl;
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int write_ptr = txq->write_ptr;
	int txq_id = txq->id;
	u8 sec_ctl = 0;
	u16 len = byte_cnt + IWL_TX_CRC_SIZE + IWL_TX_DELIMITER_SIZE;
	__le16 bc_ent;
	struct iwl_tx_cmd *tx_cmd =
		(void *)txq->entries[txq->write_ptr].cmd->payload;
	u8 sta_id = tx_cmd->sta_id;

	scd_bc_tbl = trans_pcie->scd_bc_tbls.addr;

	sec_ctl = tx_cmd->sec_ctl;

	switch (sec_ctl & TX_CMD_SEC_MSK) {
	case TX_CMD_SEC_CCM:
		len += IEEE80211_CCMP_MIC_LEN;
		break;
	case TX_CMD_SEC_TKIP:
		len += IEEE80211_TKIP_ICV_LEN;
		break;
	case TX_CMD_SEC_WEP:
		len += IEEE80211_WEP_IV_LEN + IEEE80211_WEP_ICV_LEN;
		break;
	}
	if (trans_pcie->bc_table_dword)
		len = DIV_ROUND_UP(len, 4);

	if (WARN_ON(len > 0xFFF || write_ptr >= TFD_QUEUE_SIZE_MAX))
		return;

	bc_ent = cpu_to_le16(len | (sta_id << 12));

	scd_bc_tbl[txq_id].tfd_offset[write_ptr] = bc_ent;

	if (write_ptr < TFD_QUEUE_SIZE_BC_DUP)
		scd_bc_tbl[txq_id].
			tfd_offset[TFD_QUEUE_SIZE_MAX + write_ptr] = bc_ent;
}

static void iwl_pcie_txq_inval_byte_cnt_tbl(struct iwl_trans *trans,
					    struct iwl_txq *txq)
{
	struct iwl_trans_pcie *trans_pcie =
		IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwlagn_scd_bc_tbl *scd_bc_tbl = trans_pcie->scd_bc_tbls.addr;
	int txq_id = txq->id;
	int read_ptr = txq->read_ptr;
	u8 sta_id = 0;
	__le16 bc_ent;
	struct iwl_tx_cmd *tx_cmd =
		(void *)txq->entries[read_ptr].cmd->payload;

	WARN_ON(read_ptr >= TFD_QUEUE_SIZE_MAX);

	if (txq_id != trans_pcie->cmd_queue)
		sta_id = tx_cmd->sta_id;

	bc_ent = cpu_to_le16(1 | (sta_id << 12));

	scd_bc_tbl[txq_id].tfd_offset[read_ptr] = bc_ent;

	if (read_ptr < TFD_QUEUE_SIZE_BC_DUP)
		scd_bc_tbl[txq_id].
			tfd_offset[TFD_QUEUE_SIZE_MAX + read_ptr] = bc_ent;
}

/*
 * iwl_pcie_txq_inc_wr_ptr - Send new write index to hardware
 */
static void iwl_pcie_txq_inc_wr_ptr(struct iwl_trans *trans,
				    struct iwl_txq *txq)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	u32 reg = 0;
	int txq_id = txq->id;

	lockdep_assert_held(&txq->lock);

	/*
	 * explicitly wake up the NIC if:
	 * 1. shadow registers aren't enabled
	 * 2. NIC is woken up for CMD regardless of shadow outside this function
	 * 3. there is a chance that the NIC is asleep
	 */
	if (!trans->cfg->base_params->shadow_reg_enable &&
	    txq_id != trans_pcie->cmd_queue &&
	    test_bit(STATUS_TPOWER_PMI, &trans->status)) {
		/*
		 * wake up nic if it's powered down ...
		 * uCode will wake up, and interrupt us again, so next
		 * time we'll skip this part.
		 */
		reg = iwl_read32(trans, CSR_UCODE_DRV_GP1);

		if (reg & CSR_UCODE_DRV_GP1_BIT_MAC_SLEEP) {
			IWL_DEBUG_INFO(trans, "Tx queue %d requesting wakeup, GP1 = 0x%x\n",
				       txq_id, reg);
			iwl_set_bit(trans, CSR_GP_CNTRL,
				    CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ);
			txq->need_update = true;
			return;
		}
	}

	/*
	 * if not in power-save mode, uCode will never sleep when we're
	 * trying to tx (during RFKILL, we're not trying to tx).
	 */
	IWL_DEBUG_TX(trans, "Q:%d WR: 0x%x\n", txq_id, txq->write_ptr);
	if (!txq->block)
		iwl_write32(trans, HBUS_TARG_WRPTR,
			    txq->write_ptr | (txq_id << 8));
}

void iwl_pcie_txq_check_wrptrs(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int i;

	for (i = 0; i < trans->cfg->base_params->num_of_queues; i++) {
		struct iwl_txq *txq = trans_pcie->txq[i];

		spin_lock_bh(&txq->lock);
		if (txq->need_update) {
			iwl_pcie_txq_inc_wr_ptr(trans, txq);
			txq->need_update = false;
		}
		spin_unlock_bh(&txq->lock);
	}
}

static inline dma_addr_t iwl_pcie_tfd_tb_get_addr(struct iwl_trans *trans,
						  void *_tfd, u8 idx)
{

	if (trans->cfg->use_tfh) {
		struct iwl_tfh_tfd *tfd = _tfd;
		struct iwl_tfh_tb *tb = &tfd->tbs[idx];

		return (dma_addr_t)(le64_to_cpu(tb->addr));
	} else {
		struct iwl_tfd *tfd = _tfd;
		struct iwl_tfd_tb *tb = &tfd->tbs[idx];
		dma_addr_t addr = get_unaligned_le32(&tb->lo);
		dma_addr_t hi_len;

		if (sizeof(dma_addr_t) <= sizeof(u32))
			return addr;

		hi_len = le16_to_cpu(tb->hi_n_len) & 0xF;

		/*
		 * shift by 16 twice to avoid warnings on 32-bit
		 * (where this code never runs anyway due to the
		 * if statement above)
		 */
		return addr | ((hi_len << 16) << 16);
	}
}

static inline void iwl_pcie_tfd_set_tb(struct iwl_trans *trans, void *tfd,
				       u8 idx, dma_addr_t addr, u16 len)
{
	struct iwl_tfd *tfd_fh = (void *)tfd;
	struct iwl_tfd_tb *tb = &tfd_fh->tbs[idx];

	u16 hi_n_len = len << 4;

	put_unaligned_le32(addr, &tb->lo);
	hi_n_len |= iwl_get_dma_hi_addr(addr);

	tb->hi_n_len = cpu_to_le16(hi_n_len);

	tfd_fh->num_tbs = idx + 1;
}

static inline u8 iwl_pcie_tfd_get_num_tbs(struct iwl_trans *trans, void *_tfd)
{
	if (trans->cfg->use_tfh) {
		struct iwl_tfh_tfd *tfd = _tfd;

		return le16_to_cpu(tfd->num_tbs) & 0x1f;
	} else {
		struct iwl_tfd *tfd = _tfd;

		return tfd->num_tbs & 0x1f;
	}
}

static void iwl_pcie_tfd_unmap(struct iwl_trans *trans,
			       struct iwl_cmd_meta *meta,
			       struct iwl_txq *txq, int index)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int i, num_tbs;
	void *tfd = iwl_pcie_get_tfd(trans_pcie, txq, index);

	/* Sanity check on number of chunks */
	num_tbs = iwl_pcie_tfd_get_num_tbs(trans, tfd);

	if (num_tbs >= trans_pcie->max_tbs) {
		IWL_ERR(trans, "Too many chunks: %i\n", num_tbs);
		/* @todo issue fatal error, it is quite serious situation */
		return;
	}

	/* first TB is never freed - it's the bidirectional DMA data */

	for (i = 1; i < num_tbs; i++) {
		if (meta->tbs & BIT(i))
			dma_unmap_page(trans->dev,
				       iwl_pcie_tfd_tb_get_addr(trans, tfd, i),
				       iwl_pcie_tfd_tb_get_len(trans, tfd, i),
				       DMA_TO_DEVICE);
		else
			dma_unmap_single(trans->dev,
					 iwl_pcie_tfd_tb_get_addr(trans, tfd,
								  i),
					 iwl_pcie_tfd_tb_get_len(trans, tfd,
								 i),
					 DMA_TO_DEVICE);
	}

	if (trans->cfg->use_tfh) {
		struct iwl_tfh_tfd *tfd_fh = (void *)tfd;

		tfd_fh->num_tbs = 0;
	} else {
		struct iwl_tfd *tfd_fh = (void *)tfd;

		tfd_fh->num_tbs = 0;
	}

}

/*
 * iwl_pcie_txq_free_tfd - Free all chunks referenced by TFD [txq->q.read_ptr]
 * @trans - transport private data
 * @txq - tx queue
 * @dma_dir - the direction of the DMA mapping
 *
 * Does NOT advance any TFD circular buffer read/write indexes
 * Does NOT free the TFD itself (which is within circular buffer)
 */
void iwl_pcie_txq_free_tfd(struct iwl_trans *trans, struct iwl_txq *txq)
{
	/* rd_ptr is bounded by TFD_QUEUE_SIZE_MAX and
	 * idx is bounded by n_window
	 */
	int rd_ptr = txq->read_ptr;
	int idx = get_cmd_index(txq, rd_ptr);

	lockdep_assert_held(&txq->lock);

	/* We have only q->n_window txq->entries, but we use
	 * TFD_QUEUE_SIZE_MAX tfds
	 */
	iwl_pcie_tfd_unmap(trans, &txq->entries[idx].meta, txq, rd_ptr);

	/* free SKB */
	if (txq->entries) {
		struct sk_buff *skb;

		skb = txq->entries[idx].skb;

		/* Can be called from irqs-disabled context
		 * If skb is not NULL, it means that the whole queue is being
		 * freed and that the queue is not empty - free the skb
		 */
		if (skb) {
			iwl_op_mode_free_skb(trans->op_mode, skb);
			txq->entries[idx].skb = NULL;
		}
	}
}

static int iwl_pcie_txq_build_tfd(struct iwl_trans *trans, struct iwl_txq *txq,
				  dma_addr_t addr, u16 len, bool reset)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	void *tfd;
	u32 num_tbs;

	tfd = txq->tfds + trans_pcie->tfd_size * txq->write_ptr;

	if (reset)
		memset(tfd, 0, trans_pcie->tfd_size);

	num_tbs = iwl_pcie_tfd_get_num_tbs(trans, tfd);

	/* Each TFD can point to a maximum max_tbs Tx buffers */
	if (num_tbs >= trans_pcie->max_tbs) {
		IWL_ERR(trans, "Error can not send more than %d chunks\n",
			trans_pcie->max_tbs);
		return -EINVAL;
	}

	if (WARN(addr & ~IWL_TX_DMA_MASK,
		 "Unaligned address = %llx\n", (unsigned long long)addr))
		return -EINVAL;

	iwl_pcie_tfd_set_tb(trans, tfd, num_tbs, addr, len);

	return num_tbs;
}

int iwl_pcie_txq_alloc(struct iwl_trans *trans, struct iwl_txq *txq,
		       int slots_num, bool cmd_queue)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	size_t tfd_sz = trans_pcie->tfd_size * TFD_QUEUE_SIZE_MAX;
	size_t tb0_buf_sz;
	int i;

	if (WARN_ON(txq->entries || txq->tfds))
		return -EINVAL;

	setup_timer(&txq->stuck_timer, iwl_pcie_txq_stuck_timer,
		    (unsigned long)txq);
	txq->trans_pcie = trans_pcie;

	txq->n_window = slots_num;

	txq->entries = kcalloc(slots_num,
			       sizeof(struct iwl_pcie_txq_entry),
			       GFP_KERNEL);

	if (!txq->entries)
		goto error;

	if (cmd_queue)
		for (i = 0; i < slots_num; i++) {
			txq->entries[i].cmd =
				kmalloc(sizeof(struct iwl_device_cmd),
					GFP_KERNEL);
			if (!txq->entries[i].cmd)
				goto error;
		}

	/* Circular buffer of transmit frame descriptors (TFDs),
	 * shared with device */
	txq->tfds = dma_alloc_coherent(trans->dev, tfd_sz,
				       &txq->dma_addr, GFP_KERNEL);
	if (!txq->tfds)
		goto error;

	BUILD_BUG_ON(IWL_FIRST_TB_SIZE_ALIGN != sizeof(*txq->first_tb_bufs));

	tb0_buf_sz = sizeof(*txq->first_tb_bufs) * slots_num;

	txq->first_tb_bufs = dma_alloc_coherent(trans->dev, tb0_buf_sz,
					      &txq->first_tb_dma,
					      GFP_KERNEL);
	if (!txq->first_tb_bufs)
		goto err_free_tfds;

	return 0;
err_free_tfds:
	dma_free_coherent(trans->dev, tfd_sz, txq->tfds, txq->dma_addr);
error:
	if (txq->entries && cmd_queue)
		for (i = 0; i < slots_num; i++)
			kfree(txq->entries[i].cmd);
	kfree(txq->entries);
	txq->entries = NULL;

	return -ENOMEM;

}

int iwl_pcie_txq_init(struct iwl_trans *trans, struct iwl_txq *txq,
		      int slots_num, bool cmd_queue)
{
	int ret;

	txq->need_update = false;

	/* TFD_QUEUE_SIZE_MAX must be power-of-two size, otherwise
	 * iwl_queue_inc_wrap and iwl_queue_dec_wrap are broken. */
	BUILD_BUG_ON(TFD_QUEUE_SIZE_MAX & (TFD_QUEUE_SIZE_MAX - 1));

	/* Initialize queue's high/low-water marks, and head/tail indexes */
	ret = iwl_queue_init(txq, slots_num);
	if (ret)
		return ret;

	spin_lock_init(&txq->lock);

	if (cmd_queue) {
		static struct lock_class_key iwl_pcie_cmd_queue_lock_class;

		lockdep_set_class(&txq->lock, &iwl_pcie_cmd_queue_lock_class);
	}

	__skb_queue_head_init(&txq->overflow_q);

	return 0;
}

static void iwl_pcie_free_tso_page(struct iwl_trans_pcie *trans_pcie,
				   struct sk_buff *skb)
{
	struct page **page_ptr;

	page_ptr = (void *)((u8 *)skb->cb + trans_pcie->page_offs);

	if (*page_ptr) {
		__free_page(*page_ptr);
		*page_ptr = NULL;
	}
}

static void iwl_pcie_clear_cmd_in_flight(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);

	lockdep_assert_held(&trans_pcie->reg_lock);

	if (trans_pcie->ref_cmd_in_flight) {
		trans_pcie->ref_cmd_in_flight = false;
		IWL_DEBUG_RPM(trans, "clear ref_cmd_in_flight - unref\n");
		iwl_trans_unref(trans);
	}

	if (!trans->cfg->base_params->apmg_wake_up_wa)
		return;
	if (WARN_ON(!trans_pcie->cmd_hold_nic_awake))
		return;

	trans_pcie->cmd_hold_nic_awake = false;
	__iwl_trans_pcie_clear_bit(trans, CSR_GP_CNTRL,
				   CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ);
}

/*
 * iwl_pcie_txq_unmap -  Unmap any remaining DMA mappings and free skb's
 */
static void iwl_pcie_txq_unmap(struct iwl_trans *trans, int txq_id)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];

	spin_lock_bh(&txq->lock);
	while (txq->write_ptr != txq->read_ptr) {
		IWL_DEBUG_TX_REPLY(trans, "Q %d Free %d\n",
				   txq_id, txq->read_ptr);

		if (txq_id != trans_pcie->cmd_queue) {
			struct sk_buff *skb = txq->entries[txq->read_ptr].skb;

			if (WARN_ON_ONCE(!skb))
				continue;

			iwl_pcie_free_tso_page(trans_pcie, skb);
		}
		iwl_pcie_txq_free_tfd(trans, txq);
		txq->read_ptr = iwl_queue_inc_wrap(txq->read_ptr);

		if (txq->read_ptr == txq->write_ptr) {
			unsigned long flags;

			spin_lock_irqsave(&trans_pcie->reg_lock, flags);
			if (txq_id != trans_pcie->cmd_queue) {
				IWL_DEBUG_RPM(trans, "Q %d - last tx freed\n",
					      txq->id);
				iwl_trans_unref(trans);
			} else {
				iwl_pcie_clear_cmd_in_flight(trans);
			}
			spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
		}
	}

	while (!skb_queue_empty(&txq->overflow_q)) {
		struct sk_buff *skb = __skb_dequeue(&txq->overflow_q);

		iwl_op_mode_free_skb(trans->op_mode, skb);
	}

	spin_unlock_bh(&txq->lock);

	/* just in case - this queue may have been stopped */
	iwl_wake_queue(trans, txq);
}

/*
 * iwl_pcie_txq_free - Deallocate DMA queue.
 * @txq: Transmit queue to deallocate.
 *
 * Empty queue by removing and destroying all BD's.
 * Free all buffers.
 * 0-fill, but do not free "txq" descriptor structure.
 */
static void iwl_pcie_txq_free(struct iwl_trans *trans, int txq_id)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];
	struct device *dev = trans->dev;
	int i;

	if (WARN_ON(!txq))
		return;

	iwl_pcie_txq_unmap(trans, txq_id);

	/* De-alloc array of command/tx buffers */
	if (txq_id == trans_pcie->cmd_queue)
		for (i = 0; i < txq->n_window; i++) {
			kzfree(txq->entries[i].cmd);
			kzfree(txq->entries[i].free_buf);
		}

	/* De-alloc circular buffer of TFDs */
	if (txq->tfds) {
		dma_free_coherent(dev,
				  trans_pcie->tfd_size * TFD_QUEUE_SIZE_MAX,
				  txq->tfds, txq->dma_addr);
		txq->dma_addr = 0;
		txq->tfds = NULL;

		dma_free_coherent(dev,
				  sizeof(*txq->first_tb_bufs) * txq->n_window,
				  txq->first_tb_bufs, txq->first_tb_dma);
	}

	kfree(txq->entries);
	txq->entries = NULL;

	del_timer_sync(&txq->stuck_timer);

	/* 0-fill queue descriptor structure */
	memset(txq, 0, sizeof(*txq));
}

void iwl_pcie_tx_start(struct iwl_trans *trans, u32 scd_base_addr)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int nq = trans->cfg->base_params->num_of_queues;
	int chan;
	u32 reg_val;
	int clear_dwords = (SCD_TRANS_TBL_OFFSET_QUEUE(nq) -
				SCD_CONTEXT_MEM_LOWER_BOUND) / sizeof(u32);

	/* make sure all queue are not stopped/used */
	memset(trans_pcie->queue_stopped, 0, sizeof(trans_pcie->queue_stopped));
	memset(trans_pcie->queue_used, 0, sizeof(trans_pcie->queue_used));

	trans_pcie->scd_base_addr =
		iwl_read_prph(trans, SCD_SRAM_BASE_ADDR);

	WARN_ON(scd_base_addr != 0 &&
		scd_base_addr != trans_pcie->scd_base_addr);

	/* reset context data, TX status and translation data */
	iwl_trans_write_mem(trans, trans_pcie->scd_base_addr +
				   SCD_CONTEXT_MEM_LOWER_BOUND,
			    NULL, clear_dwords);

	iwl_write_prph(trans, SCD_DRAM_BASE_ADDR,
		       trans_pcie->scd_bc_tbls.dma >> 10);

	/* The chain extension of the SCD doesn't work well. This feature is
	 * enabled by default by the HW, so we need to disable it manually.
	 */
	if (trans->cfg->base_params->scd_chain_ext_wa)
		iwl_write_prph(trans, SCD_CHAINEXT_EN, 0);

	iwl_trans_ac_txq_enable(trans, trans_pcie->cmd_queue,
				trans_pcie->cmd_fifo,
				trans_pcie->cmd_q_wdg_timeout);

	/* Activate all Tx DMA/FIFO channels */
	iwl_scd_activate_fifos(trans);

	/* Enable DMA channel */
	for (chan = 0; chan < FH_TCSR_CHNL_NUM; chan++)
		iwl_write_direct32(trans, FH_TCSR_CHNL_TX_CONFIG_REG(chan),
				   FH_TCSR_TX_CONFIG_REG_VAL_DMA_CHNL_ENABLE |
				   FH_TCSR_TX_CONFIG_REG_VAL_DMA_CREDIT_ENABLE);

	/* Update FH chicken bits */
	reg_val = iwl_read_direct32(trans, FH_TX_CHICKEN_BITS_REG);
	iwl_write_direct32(trans, FH_TX_CHICKEN_BITS_REG,
			   reg_val | FH_TX_CHICKEN_BITS_SCD_AUTO_RETRY_EN);

	/* Enable L1-Active */
	if (trans->cfg->device_family < IWL_DEVICE_FAMILY_8000)
		iwl_clear_bits_prph(trans, APMG_PCIDEV_STT_REG,
				    APMG_PCIDEV_STT_VAL_L1_ACT_DIS);
}

void iwl_trans_pcie_tx_reset(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int txq_id;

	/*
	 * we should never get here in gen2 trans mode return early to avoid
	 * having invalid accesses
	 */
	if (WARN_ON_ONCE(trans->cfg->gen2))
		return;

	for (txq_id = 0; txq_id < trans->cfg->base_params->num_of_queues;
	     txq_id++) {
		struct iwl_txq *txq = trans_pcie->txq[txq_id];
		if (trans->cfg->use_tfh)
			iwl_write_direct64(trans,
					   FH_MEM_CBBC_QUEUE(trans, txq_id),
					   txq->dma_addr);
		else
			iwl_write_direct32(trans,
					   FH_MEM_CBBC_QUEUE(trans, txq_id),
					   txq->dma_addr >> 8);
		iwl_pcie_txq_unmap(trans, txq_id);
		txq->read_ptr = 0;
		txq->write_ptr = 0;
	}

	/* Tell NIC where to find the "keep warm" buffer */
	iwl_write_direct32(trans, FH_KW_MEM_ADDR_REG,
			   trans_pcie->kw.dma >> 4);

	/*
	 * Send 0 as the scd_base_addr since the device may have be reset
	 * while we were in WoWLAN in which case SCD_SRAM_BASE_ADDR will
	 * contain garbage.
	 */
	iwl_pcie_tx_start(trans, 0);
}

static void iwl_pcie_tx_stop_fh(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	unsigned long flags;
	int ch, ret;
	u32 mask = 0;

	spin_lock(&trans_pcie->irq_lock);

	if (!iwl_trans_grab_nic_access(trans, &flags))
		goto out;

	/* Stop each Tx DMA channel */
	for (ch = 0; ch < FH_TCSR_CHNL_NUM; ch++) {
		iwl_write32(trans, FH_TCSR_CHNL_TX_CONFIG_REG(ch), 0x0);
		mask |= FH_TSSR_TX_STATUS_REG_MSK_CHNL_IDLE(ch);
	}

	/* Wait for DMA channels to be idle */
	ret = iwl_poll_bit(trans, FH_TSSR_TX_STATUS_REG, mask, mask, 5000);
	if (ret < 0)
		IWL_ERR(trans,
			"Failing on timeout while stopping DMA channel %d [0x%08x]\n",
			ch, iwl_read32(trans, FH_TSSR_TX_STATUS_REG));

	iwl_trans_release_nic_access(trans, &flags);

out:
	spin_unlock(&trans_pcie->irq_lock);
}

/*
 * iwl_pcie_tx_stop - Stop all Tx DMA channels
 */
int iwl_pcie_tx_stop(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int txq_id;

	/* Turn off all Tx DMA fifos */
	iwl_scd_deactivate_fifos(trans);

	/* Turn off all Tx DMA channels */
	iwl_pcie_tx_stop_fh(trans);

	/*
	 * This function can be called before the op_mode disabled the
	 * queues. This happens when we have an rfkill interrupt.
	 * Since we stop Tx altogether - mark the queues as stopped.
	 */
	memset(trans_pcie->queue_stopped, 0, sizeof(trans_pcie->queue_stopped));
	memset(trans_pcie->queue_used, 0, sizeof(trans_pcie->queue_used));

	/* This can happen: start_hw, stop_device */
	if (!trans_pcie->txq_memory)
		return 0;

	/* Unmap DMA from host system and free skb's */
	for (txq_id = 0; txq_id < trans->cfg->base_params->num_of_queues;
	     txq_id++)
		iwl_pcie_txq_unmap(trans, txq_id);

	return 0;
}

/*
 * iwl_trans_tx_free - Free TXQ Context
 *
 * Destroy all TX DMA queues and structures
 */
void iwl_pcie_tx_free(struct iwl_trans *trans)
{
	int txq_id;
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);

	memset(trans_pcie->queue_used, 0, sizeof(trans_pcie->queue_used));

	/* Tx queues */
	if (trans_pcie->txq_memory) {
		for (txq_id = 0;
		     txq_id < trans->cfg->base_params->num_of_queues;
		     txq_id++) {
			iwl_pcie_txq_free(trans, txq_id);
			trans_pcie->txq[txq_id] = NULL;
		}
	}

	kfree(trans_pcie->txq_memory);
	trans_pcie->txq_memory = NULL;

	iwl_pcie_free_dma_ptr(trans, &trans_pcie->kw);

	iwl_pcie_free_dma_ptr(trans, &trans_pcie->scd_bc_tbls);
}

/*
 * iwl_pcie_tx_alloc - allocate TX context
 * Allocate all Tx DMA structures and initialize them
 */
static int iwl_pcie_tx_alloc(struct iwl_trans *trans)
{
	int ret;
	int txq_id, slots_num;
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);

	u16 scd_bc_tbls_size = trans->cfg->base_params->num_of_queues *
			sizeof(struct iwlagn_scd_bc_tbl);

	/*It is not allowed to alloc twice, so warn when this happens.
	 * We cannot rely on the previous allocation, so free and fail */
	if (WARN_ON(trans_pcie->txq_memory)) {
		ret = -EINVAL;
		goto error;
	}

	ret = iwl_pcie_alloc_dma_ptr(trans, &trans_pcie->scd_bc_tbls,
				   scd_bc_tbls_size);
	if (ret) {
		IWL_ERR(trans, "Scheduler BC Table allocation failed\n");
		goto error;
	}

	/* Alloc keep-warm buffer */
	ret = iwl_pcie_alloc_dma_ptr(trans, &trans_pcie->kw, IWL_KW_SIZE);
	if (ret) {
		IWL_ERR(trans, "Keep Warm allocation failed\n");
		goto error;
	}

	trans_pcie->txq_memory = kcalloc(trans->cfg->base_params->num_of_queues,
					 sizeof(struct iwl_txq), GFP_KERNEL);
	if (!trans_pcie->txq_memory) {
		IWL_ERR(trans, "Not enough memory for txq\n");
		ret = -ENOMEM;
		goto error;
	}

	/* Alloc and init all Tx queues, including the command queue (#4/#9) */
	for (txq_id = 0; txq_id < trans->cfg->base_params->num_of_queues;
	     txq_id++) {
		bool cmd_queue = (txq_id == trans_pcie->cmd_queue);

		slots_num = cmd_queue ? TFD_CMD_SLOTS : TFD_TX_CMD_SLOTS;
		trans_pcie->txq[txq_id] = &trans_pcie->txq_memory[txq_id];
		ret = iwl_pcie_txq_alloc(trans, trans_pcie->txq[txq_id],
					 slots_num, cmd_queue);
		if (ret) {
			IWL_ERR(trans, "Tx %d queue alloc failed\n", txq_id);
			goto error;
		}
		trans_pcie->txq[txq_id]->id = txq_id;
	}

	return 0;

error:
	iwl_pcie_tx_free(trans);

	return ret;
}

int iwl_pcie_tx_init(struct iwl_trans *trans)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int ret;
	int txq_id, slots_num;
	bool alloc = false;

	if (!trans_pcie->txq_memory) {
		ret = iwl_pcie_tx_alloc(trans);
		if (ret)
			goto error;
		alloc = true;
	}

	spin_lock(&trans_pcie->irq_lock);

	/* Turn off all Tx DMA fifos */
	iwl_scd_deactivate_fifos(trans);

	/* Tell NIC where to find the "keep warm" buffer */
	iwl_write_direct32(trans, FH_KW_MEM_ADDR_REG,
			   trans_pcie->kw.dma >> 4);

	spin_unlock(&trans_pcie->irq_lock);

	/* Alloc and init all Tx queues, including the command queue (#4/#9) */
	for (txq_id = 0; txq_id < trans->cfg->base_params->num_of_queues;
	     txq_id++) {
		bool cmd_queue = (txq_id == trans_pcie->cmd_queue);

		slots_num = cmd_queue ? TFD_CMD_SLOTS : TFD_TX_CMD_SLOTS;
		ret = iwl_pcie_txq_init(trans, trans_pcie->txq[txq_id],
					slots_num, cmd_queue);
		if (ret) {
			IWL_ERR(trans, "Tx %d queue init failed\n", txq_id);
			goto error;
		}

		/*
		 * Tell nic where to find circular buffer of TFDs for a
		 * given Tx queue, and enable the DMA channel used for that
		 * queue.
		 * Circular buffer (TFD queue in DRAM) physical base address
		 */
		iwl_write_direct32(trans, FH_MEM_CBBC_QUEUE(trans, txq_id),
				   trans_pcie->txq[txq_id]->dma_addr >> 8);
	}

	iwl_set_bits_prph(trans, SCD_GP_CTRL, SCD_GP_CTRL_AUTO_ACTIVE_MODE);
	if (trans->cfg->base_params->num_of_queues > 20)
		iwl_set_bits_prph(trans, SCD_GP_CTRL,
				  SCD_GP_CTRL_ENABLE_31_QUEUES);

	return 0;
error:
	/*Upon error, free only if we allocated something */
	if (alloc)
		iwl_pcie_tx_free(trans);
	return ret;
}

static inline void iwl_pcie_txq_progress(struct iwl_txq *txq)
{
	lockdep_assert_held(&txq->lock);

	if (!txq->wd_timeout)
		return;

	/*
	 * station is asleep and we send data - that must
	 * be uAPSD or PS-Poll. Don't rearm the timer.
	 */
	if (txq->frozen)
		return;

	/*
	 * if empty delete timer, otherwise move timer forward
	 * since we're making progress on this queue
	 */
	if (txq->read_ptr == txq->write_ptr)
		del_timer(&txq->stuck_timer);
	else
		mod_timer(&txq->stuck_timer, jiffies + txq->wd_timeout);
}

/* Frees buffers until index _not_ inclusive */
void iwl_trans_pcie_reclaim(struct iwl_trans *trans, int txq_id, int ssn,
			    struct sk_buff_head *skbs)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];
	int tfd_num = ssn & (TFD_QUEUE_SIZE_MAX - 1);
	int last_to_free;

	/* This function is not meant to release cmd queue*/
	if (WARN_ON(txq_id == trans_pcie->cmd_queue))
		return;

	spin_lock_bh(&txq->lock);

	if (!test_bit(txq_id, trans_pcie->queue_used)) {
		IWL_DEBUG_TX_QUEUES(trans, "Q %d inactive - ignoring idx %d\n",
				    txq_id, ssn);
		goto out;
	}

	if (txq->read_ptr == tfd_num)
		goto out;

	IWL_DEBUG_TX_REPLY(trans, "[Q %d] %d -> %d (%d)\n",
			   txq_id, txq->read_ptr, tfd_num, ssn);

	/*Since we free until index _not_ inclusive, the one before index is
	 * the last we will free. This one must be used */
	last_to_free = iwl_queue_dec_wrap(tfd_num);

	if (!iwl_queue_used(txq, last_to_free)) {
		IWL_ERR(trans,
			"%s: Read index for DMA queue txq id (%d), last_to_free %d is out of range [0-%d] %d %d.\n",
			__func__, txq_id, last_to_free, TFD_QUEUE_SIZE_MAX,
			txq->write_ptr, txq->read_ptr);
		goto out;
	}

	if (WARN_ON(!skb_queue_empty(skbs)))
		goto out;

	for (;
	     txq->read_ptr != tfd_num;
	     txq->read_ptr = iwl_queue_inc_wrap(txq->read_ptr)) {
		struct sk_buff *skb = txq->entries[txq->read_ptr].skb;

		if (WARN_ON_ONCE(!skb))
			continue;

		iwl_pcie_free_tso_page(trans_pcie, skb);

		__skb_queue_tail(skbs, skb);

		txq->entries[txq->read_ptr].skb = NULL;

		if (!trans->cfg->use_tfh)
			iwl_pcie_txq_inval_byte_cnt_tbl(trans, txq);

		iwl_pcie_txq_free_tfd(trans, txq);
	}

	iwl_pcie_txq_progress(txq);

	if (iwl_queue_space(txq) > txq->low_mark &&
	    test_bit(txq_id, trans_pcie->queue_stopped)) {
		struct sk_buff_head overflow_skbs;

		__skb_queue_head_init(&overflow_skbs);
		skb_queue_splice_init(&txq->overflow_q, &overflow_skbs);

		/*
		 * This is tricky: we are in reclaim path which is non
		 * re-entrant, so noone will try to take the access the
		 * txq data from that path. We stopped tx, so we can't
		 * have tx as well. Bottom line, we can unlock and re-lock
		 * later.
		 */
		spin_unlock_bh(&txq->lock);

		while (!skb_queue_empty(&overflow_skbs)) {
			struct sk_buff *skb = __skb_dequeue(&overflow_skbs);
			struct iwl_device_cmd *dev_cmd_ptr;

			dev_cmd_ptr = *(void **)((u8 *)skb->cb +
						 trans_pcie->dev_cmd_offs);

			/*
			 * Note that we can very well be overflowing again.
			 * In that case, iwl_queue_space will be small again
			 * and we won't wake mac80211's queue.
			 */
			iwl_trans_pcie_tx(trans, skb, dev_cmd_ptr, txq_id);
		}
		spin_lock_bh(&txq->lock);

		if (iwl_queue_space(txq) > txq->low_mark)
			iwl_wake_queue(trans, txq);
	}

	if (txq->read_ptr == txq->write_ptr) {
		IWL_DEBUG_RPM(trans, "Q %d - last tx reclaimed\n", txq->id);
		iwl_trans_unref(trans);
	}

out:
	spin_unlock_bh(&txq->lock);
}

static int iwl_pcie_set_cmd_in_flight(struct iwl_trans *trans,
				      const struct iwl_host_cmd *cmd)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	int ret;

	lockdep_assert_held(&trans_pcie->reg_lock);

	if (!(cmd->flags & CMD_SEND_IN_IDLE) &&
	    !trans_pcie->ref_cmd_in_flight) {
		trans_pcie->ref_cmd_in_flight = true;
		IWL_DEBUG_RPM(trans, "set ref_cmd_in_flight - ref\n");
		iwl_trans_ref(trans);
	}

	/*
	 * wake up the NIC to make sure that the firmware will see the host
	 * command - we will let the NIC sleep once all the host commands
	 * returned. This needs to be done only on NICs that have
	 * apmg_wake_up_wa set.
	 */
	if (trans->cfg->base_params->apmg_wake_up_wa &&
	    !trans_pcie->cmd_hold_nic_awake) {
		__iwl_trans_pcie_set_bit(trans, CSR_GP_CNTRL,
					 CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ);

		ret = iwl_poll_bit(trans, CSR_GP_CNTRL,
				   CSR_GP_CNTRL_REG_VAL_MAC_ACCESS_EN,
				   (CSR_GP_CNTRL_REG_FLAG_MAC_CLOCK_READY |
				    CSR_GP_CNTRL_REG_FLAG_GOING_TO_SLEEP),
				   15000);
		if (ret < 0) {
			__iwl_trans_pcie_clear_bit(trans, CSR_GP_CNTRL,
					CSR_GP_CNTRL_REG_FLAG_MAC_ACCESS_REQ);
			IWL_ERR(trans, "Failed to wake NIC for hcmd\n");
			return -EIO;
		}
		trans_pcie->cmd_hold_nic_awake = true;
	}

	return 0;
}

/*
 * iwl_pcie_cmdq_reclaim - Reclaim TX command queue entries already Tx'd
 *
 * When FW advances 'R' index, all entries between old and new 'R' index
 * need to be reclaimed. As result, some free space forms.  If there is
 * enough free space (> low mark), wake the stack that feeds us.
 */
static void iwl_pcie_cmdq_reclaim(struct iwl_trans *trans, int txq_id, int idx)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];
	unsigned long flags;
	int nfreed = 0;

	lockdep_assert_held(&txq->lock);

	if ((idx >= TFD_QUEUE_SIZE_MAX) || (!iwl_queue_used(txq, idx))) {
		IWL_ERR(trans,
			"%s: Read index for DMA queue txq id (%d), index %d is out of range [0-%d] %d %d.\n",
			__func__, txq_id, idx, TFD_QUEUE_SIZE_MAX,
			txq->write_ptr, txq->read_ptr);
		return;
	}

	for (idx = iwl_queue_inc_wrap(idx); txq->read_ptr != idx;
	     txq->read_ptr = iwl_queue_inc_wrap(txq->read_ptr)) {

		if (nfreed++ > 0) {
			IWL_ERR(trans, "HCMD skipped: index (%d) %d %d\n",
				idx, txq->write_ptr, txq->read_ptr);
			iwl_force_nmi(trans);
		}
	}

	if (txq->read_ptr == txq->write_ptr) {
		spin_lock_irqsave(&trans_pcie->reg_lock, flags);
		iwl_pcie_clear_cmd_in_flight(trans);
		spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
	}

	iwl_pcie_txq_progress(txq);
}

static int iwl_pcie_txq_set_ratid_map(struct iwl_trans *trans, u16 ra_tid,
				 u16 txq_id)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	u32 tbl_dw_addr;
	u32 tbl_dw;
	u16 scd_q2ratid;

	scd_q2ratid = ra_tid & SCD_QUEUE_RA_TID_MAP_RATID_MSK;

	tbl_dw_addr = trans_pcie->scd_base_addr +
			SCD_TRANS_TBL_OFFSET_QUEUE(txq_id);

	tbl_dw = iwl_trans_read_mem32(trans, tbl_dw_addr);

	if (txq_id & 0x1)
		tbl_dw = (scd_q2ratid << 16) | (tbl_dw & 0x0000FFFF);
	else
		tbl_dw = scd_q2ratid | (tbl_dw & 0xFFFF0000);

	iwl_trans_write_mem32(trans, tbl_dw_addr, tbl_dw);

	return 0;
}

/* Receiver address (actually, Rx station's index into station table),
 * combined with Traffic ID (QOS priority), in format used by Tx Scheduler */
#define BUILD_RAxTID(sta_id, tid)	(((sta_id) << 4) + (tid))

bool iwl_trans_pcie_txq_enable(struct iwl_trans *trans, int txq_id, u16 ssn,
			       const struct iwl_trans_txq_scd_cfg *cfg,
			       unsigned int wdg_timeout)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];
	int fifo = -1;
	bool scd_bug = false;

	if (test_and_set_bit(txq_id, trans_pcie->queue_used))
		WARN_ONCE(1, "queue %d already used - expect issues", txq_id);

	txq->wd_timeout = msecs_to_jiffies(wdg_timeout);

	if (cfg) {
		fifo = cfg->fifo;

		/* Disable the scheduler prior configuring the cmd queue */
		if (txq_id == trans_pcie->cmd_queue &&
		    trans_pcie->scd_set_active)
			iwl_scd_enable_set_active(trans, 0);

		/* Stop this Tx queue before configuring it */
		iwl_scd_txq_set_inactive(trans, txq_id);

		/* Set this queue as a chain-building queue unless it is CMD */
		if (txq_id != trans_pcie->cmd_queue)
			iwl_scd_txq_set_chain(trans, txq_id);

		if (cfg->aggregate) {
			u16 ra_tid = BUILD_RAxTID(cfg->sta_id, cfg->tid);

			/* Map receiver-address / traffic-ID to this queue */
			iwl_pcie_txq_set_ratid_map(trans, ra_tid, txq_id);

			/* enable aggregations for the queue */
			iwl_scd_txq_enable_agg(trans, txq_id);
			txq->ampdu = true;
		} else {
			/*
			 * disable aggregations for the queue, this will also
			 * make the ra_tid mapping configuration irrelevant
			 * since it is now a non-AGG queue.
			 */
			iwl_scd_txq_disable_agg(trans, txq_id);

			ssn = txq->read_ptr;
		}
	} else {
		/*
		 * If we need to move the SCD write pointer by steps of
		 * 0x40, 0x80 or 0xc0, it gets stuck. Avoids this and let
		 * the op_mode know by returning true later.
		 * Do this only in case cfg is NULL since this trick can
		 * be done only if we have DQA enabled which is true for mvm
		 * only. And mvm never sets a cfg pointer.
		 * This is really ugly, but this is the easiest way out for
		 * this sad hardware issue.
		 * This bug has been fixed on devices 9000 and up.
		 */
		scd_bug = !trans->cfg->mq_rx_supported &&
			!((ssn - txq->write_ptr) & 0x3f) &&
			(ssn != txq->write_ptr);
		if (scd_bug)
			ssn++;
	}

	/* Place first TFD at index corresponding to start sequence number.
	 * Assumes that ssn_idx is valid (!= 0xFFF) */
	txq->read_ptr = (ssn & 0xff);
	txq->write_ptr = (ssn & 0xff);
	iwl_write_direct32(trans, HBUS_TARG_WRPTR,
			   (ssn & 0xff) | (txq_id << 8));

	if (cfg) {
		u8 frame_limit = cfg->frame_limit;

		iwl_write_prph(trans, SCD_QUEUE_RDPTR(txq_id), ssn);

		/* Set up Tx window size and frame limit for this queue */
		iwl_trans_write_mem32(trans, trans_pcie->scd_base_addr +
				SCD_CONTEXT_QUEUE_OFFSET(txq_id), 0);
		iwl_trans_write_mem32(trans,
			trans_pcie->scd_base_addr +
			SCD_CONTEXT_QUEUE_OFFSET(txq_id) + sizeof(u32),
			SCD_QUEUE_CTX_REG2_VAL(WIN_SIZE, frame_limit) |
			SCD_QUEUE_CTX_REG2_VAL(FRAME_LIMIT, frame_limit));

		/* Set up status area in SRAM, map to Tx DMA/FIFO, activate */
		iwl_write_prph(trans, SCD_QUEUE_STATUS_BITS(txq_id),
			       (1 << SCD_QUEUE_STTS_REG_POS_ACTIVE) |
			       (cfg->fifo << SCD_QUEUE_STTS_REG_POS_TXF) |
			       (1 << SCD_QUEUE_STTS_REG_POS_WSL) |
			       SCD_QUEUE_STTS_REG_MSK);

		/* enable the scheduler for this queue (only) */
		if (txq_id == trans_pcie->cmd_queue &&
		    trans_pcie->scd_set_active)
			iwl_scd_enable_set_active(trans, BIT(txq_id));

		IWL_DEBUG_TX_QUEUES(trans,
				    "Activate queue %d on FIFO %d WrPtr: %d\n",
				    txq_id, fifo, ssn & 0xff);
	} else {
		IWL_DEBUG_TX_QUEUES(trans,
				    "Activate queue %d WrPtr: %d\n",
				    txq_id, ssn & 0xff);
	}

	return scd_bug;
}

void iwl_trans_pcie_txq_set_shared_mode(struct iwl_trans *trans, u32 txq_id,
					bool shared_mode)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[txq_id];

	txq->ampdu = !shared_mode;
}

void iwl_trans_pcie_txq_disable(struct iwl_trans *trans, int txq_id,
				bool configure_scd)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	u32 stts_addr = trans_pcie->scd_base_addr +
			SCD_TX_STTS_QUEUE_OFFSET(txq_id);
	static const u32 zero_val[4] = {};

	trans_pcie->txq[txq_id]->frozen_expiry_remainder = 0;
	trans_pcie->txq[txq_id]->frozen = false;

	/*
	 * Upon HW Rfkill - we stop the device, and then stop the queues
	 * in the op_mode. Just for the sake of the simplicity of the op_mode,
	 * allow the op_mode to call txq_disable after it already called
	 * stop_device.
	 */
	if (!test_and_clear_bit(txq_id, trans_pcie->queue_used)) {
		WARN_ONCE(test_bit(STATUS_DEVICE_ENABLED, &trans->status),
			  "queue %d not used", txq_id);
		return;
	}

	if (configure_scd) {
		iwl_scd_txq_set_inactive(trans, txq_id);

		iwl_trans_write_mem(trans, stts_addr, (void *)zero_val,
				    ARRAY_SIZE(zero_val));
	}

	iwl_pcie_txq_unmap(trans, txq_id);
	trans_pcie->txq[txq_id]->ampdu = false;

	IWL_DEBUG_TX_QUEUES(trans, "Deactivate queue %d\n", txq_id);
}

/*************** HOST COMMAND QUEUE FUNCTIONS   *****/

/*
 * iwl_pcie_enqueue_hcmd - enqueue a uCode command
 * @priv: device private data point
 * @cmd: a pointer to the ucode command structure
 *
 * The function returns < 0 values to indicate the operation
 * failed. On success, it returns the index (>= 0) of command in the
 * command queue.
 */
static int iwl_pcie_enqueue_hcmd(struct iwl_trans *trans,
				 struct iwl_host_cmd *cmd)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[trans_pcie->cmd_queue];
	struct iwl_device_cmd *out_cmd;
	struct iwl_cmd_meta *out_meta;
	unsigned long flags;
	void *dup_buf = NULL;
	dma_addr_t phys_addr;
	int idx;
	u16 copy_size, cmd_size, tb0_size;
	bool had_nocopy = false;
	u8 group_id = iwl_cmd_groupid(cmd->id);
	int i, ret;
	u32 cmd_pos;
	const u8 *cmddata[IWL_MAX_CMD_TBS_PER_TFD];
	u16 cmdlen[IWL_MAX_CMD_TBS_PER_TFD];

	if (WARN(!trans->wide_cmd_header &&
		 group_id > IWL_ALWAYS_LONG_GROUP,
		 "unsupported wide command %#x\n", cmd->id))
		return -EINVAL;

	if (group_id != 0) {
		copy_size = sizeof(struct iwl_cmd_header_wide);
		cmd_size = sizeof(struct iwl_cmd_header_wide);
	} else {
		copy_size = sizeof(struct iwl_cmd_header);
		cmd_size = sizeof(struct iwl_cmd_header);
	}

	/* need one for the header if the first is NOCOPY */
	BUILD_BUG_ON(IWL_MAX_CMD_TBS_PER_TFD > IWL_NUM_OF_TBS - 1);

	for (i = 0; i < IWL_MAX_CMD_TBS_PER_TFD; i++) {
		cmddata[i] = cmd->data[i];
		cmdlen[i] = cmd->len[i];

		if (!cmd->len[i])
			continue;

		/* need at least IWL_FIRST_TB_SIZE copied */
		if (copy_size < IWL_FIRST_TB_SIZE) {
			int copy = IWL_FIRST_TB_SIZE - copy_size;

			if (copy > cmdlen[i])
				copy = cmdlen[i];
			cmdlen[i] -= copy;
			cmddata[i] += copy;
			copy_size += copy;
		}

		if (cmd->dataflags[i] & IWL_HCMD_DFL_NOCOPY) {
			had_nocopy = true;
			if (WARN_ON(cmd->dataflags[i] & IWL_HCMD_DFL_DUP)) {
				idx = -EINVAL;
				goto free_dup_buf;
			}
		} else if (cmd->dataflags[i] & IWL_HCMD_DFL_DUP) {
			/*
			 * This is also a chunk that isn't copied
			 * to the static buffer so set had_nocopy.
			 */
			had_nocopy = true;

			/* only allowed once */
			if (WARN_ON(dup_buf)) {
				idx = -EINVAL;
				goto free_dup_buf;
			}

			dup_buf = kmemdup(cmddata[i], cmdlen[i],
					  GFP_ATOMIC);
			if (!dup_buf)
				return -ENOMEM;
		} else {
			/* NOCOPY must not be followed by normal! */
			if (WARN_ON(had_nocopy)) {
				idx = -EINVAL;
				goto free_dup_buf;
			}
			copy_size += cmdlen[i];
		}
		cmd_size += cmd->len[i];
	}

	/*
	 * If any of the command structures end up being larger than
	 * the TFD_MAX_PAYLOAD_SIZE and they aren't dynamically
	 * allocated into separate TFDs, then we will need to
	 * increase the size of the buffers.
	 */
	if (WARN(copy_size > TFD_MAX_PAYLOAD_SIZE,
		 "Command %s (%#x) is too large (%d bytes)\n",
		 iwl_get_cmd_string(trans, cmd->id),
		 cmd->id, copy_size)) {
		idx = -EINVAL;
		goto free_dup_buf;
	}

	spin_lock_bh(&txq->lock);

	if (iwl_queue_space(txq) < ((cmd->flags & CMD_ASYNC) ? 2 : 1)) {
		spin_unlock_bh(&txq->lock);

		IWL_ERR(trans, "No space in command queue\n");
		iwl_op_mode_cmd_queue_full(trans->op_mode);
		idx = -ENOSPC;
		goto free_dup_buf;
	}

	idx = get_cmd_index(txq, txq->write_ptr);
	out_cmd = txq->entries[idx].cmd;
	out_meta = &txq->entries[idx].meta;

	memset(out_meta, 0, sizeof(*out_meta));	/* re-initialize to NULL */
	if (cmd->flags & CMD_WANT_SKB)
		out_meta->source = cmd;

	/* set up the header */
	if (group_id != 0) {
		out_cmd->hdr_wide.cmd = iwl_cmd_opcode(cmd->id);
		out_cmd->hdr_wide.group_id = group_id;
		out_cmd->hdr_wide.version = iwl_cmd_version(cmd->id);
		out_cmd->hdr_wide.length =
			cpu_to_le16(cmd_size -
				    sizeof(struct iwl_cmd_header_wide));
		out_cmd->hdr_wide.reserved = 0;
		out_cmd->hdr_wide.sequence =
			cpu_to_le16(QUEUE_TO_SEQ(trans_pcie->cmd_queue) |
						 INDEX_TO_SEQ(txq->write_ptr));

		cmd_pos = sizeof(struct iwl_cmd_header_wide);
		copy_size = sizeof(struct iwl_cmd_header_wide);
	} else {
		out_cmd->hdr.cmd = iwl_cmd_opcode(cmd->id);
		out_cmd->hdr.sequence =
			cpu_to_le16(QUEUE_TO_SEQ(trans_pcie->cmd_queue) |
						 INDEX_TO_SEQ(txq->write_ptr));
		out_cmd->hdr.group_id = 0;

		cmd_pos = sizeof(struct iwl_cmd_header);
		copy_size = sizeof(struct iwl_cmd_header);
	}

	/* and copy the data that needs to be copied */
	for (i = 0; i < IWL_MAX_CMD_TBS_PER_TFD; i++) {
		int copy;

		if (!cmd->len[i])
			continue;

		/* copy everything if not nocopy/dup */
		if (!(cmd->dataflags[i] & (IWL_HCMD_DFL_NOCOPY |
					   IWL_HCMD_DFL_DUP))) {
			copy = cmd->len[i];

			memcpy((u8 *)out_cmd + cmd_pos, cmd->data[i], copy);
			cmd_pos += copy;
			copy_size += copy;
			continue;
		}

		/*
		 * Otherwise we need at least IWL_FIRST_TB_SIZE copied
		 * in total (for bi-directional DMA), but copy up to what
		 * we can fit into the payload for debug dump purposes.
		 */
		copy = min_t(int, TFD_MAX_PAYLOAD_SIZE - cmd_pos, cmd->len[i]);

		memcpy((u8 *)out_cmd + cmd_pos, cmd->data[i], copy);
		cmd_pos += copy;

		/* However, treat copy_size the proper way, we need it below */
		if (copy_size < IWL_FIRST_TB_SIZE) {
			copy = IWL_FIRST_TB_SIZE - copy_size;

			if (copy > cmd->len[i])
				copy = cmd->len[i];
			copy_size += copy;
		}
	}

	IWL_DEBUG_HC(trans,
		     "Sending command %s (%.2x.%.2x), seq: 0x%04X, %d bytes at %d[%d]:%d\n",
		     iwl_get_cmd_string(trans, cmd->id),
		     group_id, out_cmd->hdr.cmd,
		     le16_to_cpu(out_cmd->hdr.sequence),
		     cmd_size, txq->write_ptr, idx, trans_pcie->cmd_queue);

	/* start the TFD with the minimum copy bytes */
	tb0_size = min_t(int, copy_size, IWL_FIRST_TB_SIZE);
	memcpy(&txq->first_tb_bufs[idx], &out_cmd->hdr, tb0_size);
	iwl_pcie_txq_build_tfd(trans, txq,
			       iwl_pcie_get_first_tb_dma(txq, idx),
			       tb0_size, true);

	/* map first command fragment, if any remains */
	if (copy_size > tb0_size) {
		phys_addr = dma_map_single(trans->dev,
					   ((u8 *)&out_cmd->hdr) + tb0_size,
					   copy_size - tb0_size,
					   DMA_TO_DEVICE);
		if (dma_mapping_error(trans->dev, phys_addr)) {
			iwl_pcie_tfd_unmap(trans, out_meta, txq,
					   txq->write_ptr);
			idx = -ENOMEM;
			goto out;
		}

		iwl_pcie_txq_build_tfd(trans, txq, phys_addr,
				       copy_size - tb0_size, false);
	}

	/* map the remaining (adjusted) nocopy/dup fragments */
	for (i = 0; i < IWL_MAX_CMD_TBS_PER_TFD; i++) {
		const void *data = cmddata[i];

		if (!cmdlen[i])
			continue;
		if (!(cmd->dataflags[i] & (IWL_HCMD_DFL_NOCOPY |
					   IWL_HCMD_DFL_DUP)))
			continue;
		if (cmd->dataflags[i] & IWL_HCMD_DFL_DUP)
			data = dup_buf;
		phys_addr = dma_map_single(trans->dev, (void *)data,
					   cmdlen[i], DMA_TO_DEVICE);
		if (dma_mapping_error(trans->dev, phys_addr)) {
			iwl_pcie_tfd_unmap(trans, out_meta, txq,
					   txq->write_ptr);
			idx = -ENOMEM;
			goto out;
		}

		iwl_pcie_txq_build_tfd(trans, txq, phys_addr, cmdlen[i], false);
	}

	BUILD_BUG_ON(IWL_TFH_NUM_TBS > sizeof(out_meta->tbs) * BITS_PER_BYTE);
	out_meta->flags = cmd->flags;
	if (WARN_ON_ONCE(txq->entries[idx].free_buf))
		kzfree(txq->entries[idx].free_buf);
	txq->entries[idx].free_buf = dup_buf;

	trace_iwlwifi_dev_hcmd(trans->dev, cmd, cmd_size, &out_cmd->hdr_wide);

	/* start timer if queue currently empty */
	if (txq->read_ptr == txq->write_ptr && txq->wd_timeout)
		mod_timer(&txq->stuck_timer, jiffies + txq->wd_timeout);

	spin_lock_irqsave(&trans_pcie->reg_lock, flags);
	ret = iwl_pcie_set_cmd_in_flight(trans, cmd);
	if (ret < 0) {
		idx = ret;
		spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);
		goto out;
	}

	/* Increment and update queue's write index */
	txq->write_ptr = iwl_queue_inc_wrap(txq->write_ptr);
	iwl_pcie_txq_inc_wr_ptr(trans, txq);

	spin_unlock_irqrestore(&trans_pcie->reg_lock, flags);

 out:
	spin_unlock_bh(&txq->lock);
 free_dup_buf:
	if (idx < 0)
		kfree(dup_buf);
	return idx;
}

/*
 * iwl_pcie_hcmd_complete - Pull unused buffers off the queue and reclaim them
 * @rxb: Rx buffer to reclaim
 */
void iwl_pcie_hcmd_complete(struct iwl_trans *trans,
			    struct iwl_rx_cmd_buffer *rxb)
{
	struct iwl_rx_packet *pkt = rxb_addr(rxb);
	u16 sequence = le16_to_cpu(pkt->hdr.sequence);
	u8 group_id;
	u32 cmd_id;
	int txq_id = SEQ_TO_QUEUE(sequence);
	int index = SEQ_TO_INDEX(sequence);
	int cmd_index;
	struct iwl_device_cmd *cmd;
	struct iwl_cmd_meta *meta;
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[trans_pcie->cmd_queue];

	/* If a Tx command is being handled and it isn't in the actual
	 * command queue then there a command routing bug has been introduced
	 * in the queue management code. */
	if (WARN(txq_id != trans_pcie->cmd_queue,
		 "wrong command queue %d (should be %d), sequence 0x%X readp=%d writep=%d\n",
		 txq_id, trans_pcie->cmd_queue, sequence, txq->read_ptr,
		 txq->write_ptr)) {
		iwl_print_hex_error(trans, pkt, 32);
		return;
	}

	spin_lock_bh(&txq->lock);

	cmd_index = get_cmd_index(txq, index);
	cmd = txq->entries[cmd_index].cmd;
	meta = &txq->entries[cmd_index].meta;
	group_id = cmd->hdr.group_id;
	cmd_id = iwl_cmd_id(cmd->hdr.cmd, group_id, 0);

	iwl_pcie_tfd_unmap(trans, meta, txq, index);

	/* Input error checking is done when commands are added to queue. */
	if (meta->flags & CMD_WANT_SKB) {
		struct page *p = rxb_steal_page(rxb);

		meta->source->resp_pkt = pkt;
		meta->source->_rx_page_addr = (unsigned long)page_address(p);
		meta->source->_rx_page_order = trans_pcie->rx_page_order;
	}

	if (meta->flags & CMD_WANT_ASYNC_CALLBACK)
		iwl_op_mode_async_cb(trans->op_mode, cmd);

	iwl_pcie_cmdq_reclaim(trans, txq_id, index);

	if (!(meta->flags & CMD_ASYNC)) {
		if (!test_bit(STATUS_SYNC_HCMD_ACTIVE, &trans->status)) {
			IWL_WARN(trans,
				 "HCMD_ACTIVE already clear for command %s\n",
				 iwl_get_cmd_string(trans, cmd_id));
		}
		clear_bit(STATUS_SYNC_HCMD_ACTIVE, &trans->status);
		IWL_DEBUG_INFO(trans, "Clearing HCMD_ACTIVE for command %s\n",
			       iwl_get_cmd_string(trans, cmd_id));
		wake_up(&trans_pcie->wait_command_queue);
	}

	if (meta->flags & CMD_MAKE_TRANS_IDLE) {
		IWL_DEBUG_INFO(trans, "complete %s - mark trans as idle\n",
			       iwl_get_cmd_string(trans, cmd->hdr.cmd));
		set_bit(STATUS_TRANS_IDLE, &trans->status);
		wake_up(&trans_pcie->d0i3_waitq);
	}

	if (meta->flags & CMD_WAKE_UP_TRANS) {
		IWL_DEBUG_INFO(trans, "complete %s - clear trans idle flag\n",
			       iwl_get_cmd_string(trans, cmd->hdr.cmd));
		clear_bit(STATUS_TRANS_IDLE, &trans->status);
		wake_up(&trans_pcie->d0i3_waitq);
	}

	meta->flags = 0;

	spin_unlock_bh(&txq->lock);
}

#define HOST_COMPLETE_TIMEOUT	(2 * HZ)

static int iwl_pcie_send_hcmd_async(struct iwl_trans *trans,
				    struct iwl_host_cmd *cmd)
{
	int ret;

	/* An asynchronous command can not expect an SKB to be set. */
	if (WARN_ON(cmd->flags & CMD_WANT_SKB))
		return -EINVAL;

	ret = iwl_pcie_enqueue_hcmd(trans, cmd);
	if (ret < 0) {
		IWL_ERR(trans,
			"Error sending %s: enqueue_hcmd failed: %d\n",
			iwl_get_cmd_string(trans, cmd->id), ret);
		return ret;
	}
	return 0;
}

static int iwl_pcie_send_hcmd_sync(struct iwl_trans *trans,
				   struct iwl_host_cmd *cmd)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_txq *txq = trans_pcie->txq[trans_pcie->cmd_queue];
	int cmd_idx;
	int ret;

	IWL_DEBUG_INFO(trans, "Attempting to send sync command %s\n",
		       iwl_get_cmd_string(trans, cmd->id));

	if (WARN(test_and_set_bit(STATUS_SYNC_HCMD_ACTIVE,
				  &trans->status),
		 "Command %s: a command is already active!\n",
		 iwl_get_cmd_string(trans, cmd->id)))
		return -EIO;

	IWL_DEBUG_INFO(trans, "Setting HCMD_ACTIVE for command %s\n",
		       iwl_get_cmd_string(trans, cmd->id));

	if (pm_runtime_suspended(&trans_pcie->pci_dev->dev)) {
		ret = wait_event_timeout(trans_pcie->d0i3_waitq,
				 pm_runtime_active(&trans_pcie->pci_dev->dev),
				 msecs_to_jiffies(IWL_TRANS_IDLE_TIMEOUT));
		if (!ret) {
			IWL_ERR(trans, "Timeout exiting D0i3 before hcmd\n");
			return -ETIMEDOUT;
		}
	}

	cmd_idx = iwl_pcie_enqueue_hcmd(trans, cmd);
	if (cmd_idx < 0) {
		ret = cmd_idx;
		clear_bit(STATUS_SYNC_HCMD_ACTIVE, &trans->status);
		IWL_ERR(trans,
			"Error sending %s: enqueue_hcmd failed: %d\n",
			iwl_get_cmd_string(trans, cmd->id), ret);
		return ret;
	}

	ret = wait_event_timeout(trans_pcie->wait_command_queue,
				 !test_bit(STATUS_SYNC_HCMD_ACTIVE,
					   &trans->status),
				 HOST_COMPLETE_TIMEOUT);
	if (!ret) {
		IWL_ERR(trans, "Error sending %s: time out after %dms.\n",
			iwl_get_cmd_string(trans, cmd->id),
			jiffies_to_msecs(HOST_COMPLETE_TIMEOUT));

		IWL_ERR(trans, "Current CMD queue read_ptr %d write_ptr %d\n",
			txq->read_ptr, txq->write_ptr);

		clear_bit(STATUS_SYNC_HCMD_ACTIVE, &trans->status);
		IWL_DEBUG_INFO(trans, "Clearing HCMD_ACTIVE for command %s\n",
			       iwl_get_cmd_string(trans, cmd->id));
		ret = -ETIMEDOUT;

		iwl_force_nmi(trans);
		iwl_trans_fw_error(trans);

		goto cancel;
	}

	if (test_bit(STATUS_FW_ERROR, &trans->status)) {
		IWL_ERR(trans, "FW error in SYNC CMD %s\n",
			iwl_get_cmd_string(trans, cmd->id));
		dump_stack();
		ret = -EIO;
		goto cancel;
	}

	if (!(cmd->flags & CMD_SEND_IN_RFKILL) &&
	    test_bit(STATUS_RFKILL_OPMODE, &trans->status)) {
		IWL_DEBUG_RF_KILL(trans, "RFKILL in SYNC CMD... no rsp\n");
		ret = -ERFKILL;
		goto cancel;
	}

	if ((cmd->flags & CMD_WANT_SKB) && !cmd->resp_pkt) {
		IWL_ERR(trans, "Error: Response NULL in '%s'\n",
			iwl_get_cmd_string(trans, cmd->id));
		ret = -EIO;
		goto cancel;
	}

	return 0;

cancel:
	if (cmd->flags & CMD_WANT_SKB) {
		/*
		 * Cancel the CMD_WANT_SKB flag for the cmd in the
		 * TX cmd queue. Otherwise in case the cmd comes
		 * in later, it will possibly set an invalid
		 * address (cmd->meta.source).
		 */
		txq->entries[cmd_idx].meta.flags &= ~CMD_WANT_SKB;
	}

	if (cmd->resp_pkt) {
		iwl_free_resp(cmd);
		cmd->resp_pkt = NULL;
	}

	return ret;
}

int iwl_trans_pcie_send_hcmd(struct iwl_trans *trans, struct iwl_host_cmd *cmd)
{
	if (!(cmd->flags & CMD_SEND_IN_RFKILL) &&
	    test_bit(STATUS_RFKILL_OPMODE, &trans->status)) {
		IWL_DEBUG_RF_KILL(trans, "Dropping CMD 0x%x: RF KILL\n",
				  cmd->id);
		return -ERFKILL;
	}

	if (cmd->flags & CMD_ASYNC)
		return iwl_pcie_send_hcmd_async(trans, cmd);

	/* We still can fail on RFKILL that can be asserted while we wait */
	return iwl_pcie_send_hcmd_sync(trans, cmd);
}

static int iwl_fill_data_tbs(struct iwl_trans *trans, struct sk_buff *skb,
			     struct iwl_txq *txq, u8 hdr_len,
			     struct iwl_cmd_meta *out_meta,
			     struct iwl_device_cmd *dev_cmd, u16 tb1_len)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	u16 tb2_len;
	int i;

	/*
	 * Set up TFD's third entry to point directly to remainder
	 * of skb's head, if any
	 */
	tb2_len = skb_headlen(skb) - hdr_len;

	if (tb2_len > 0) {
		dma_addr_t tb2_phys = dma_map_single(trans->dev,
						     skb->data + hdr_len,
						     tb2_len, DMA_TO_DEVICE);
		if (unlikely(dma_mapping_error(trans->dev, tb2_phys))) {
			iwl_pcie_tfd_unmap(trans, out_meta, txq,
					   txq->write_ptr);
			return -EINVAL;
		}
		iwl_pcie_txq_build_tfd(trans, txq, tb2_phys, tb2_len, false);
	}

	/* set up the remaining entries to point to the data */
	for (i = 0; i < skb_shinfo(skb)->nr_frags; i++) {
		const skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
		dma_addr_t tb_phys;
		int tb_idx;

		if (!skb_frag_size(frag))
			continue;

		tb_phys = skb_frag_dma_map(trans->dev, frag, 0,
					   skb_frag_size(frag), DMA_TO_DEVICE);

		if (unlikely(dma_mapping_error(trans->dev, tb_phys))) {
			iwl_pcie_tfd_unmap(trans, out_meta, txq,
					   txq->write_ptr);
			return -EINVAL;
		}
		tb_idx = iwl_pcie_txq_build_tfd(trans, txq, tb_phys,
						skb_frag_size(frag), false);

		out_meta->tbs |= BIT(tb_idx);
	}

	trace_iwlwifi_dev_tx(trans->dev, skb,
			     iwl_pcie_get_tfd(trans_pcie, txq, txq->write_ptr),
			     trans_pcie->tfd_size,
			     &dev_cmd->hdr, IWL_FIRST_TB_SIZE + tb1_len,
			     hdr_len);
	trace_iwlwifi_dev_tx_data(trans->dev, skb, hdr_len);
	return 0;
}

#ifdef CONFIG_INET
struct iwl_tso_hdr_page *get_page_hdr(struct iwl_trans *trans, size_t len)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct iwl_tso_hdr_page *p = this_cpu_ptr(trans_pcie->tso_hdr_page);

	if (!p->page)
		goto alloc;

	/* enough room on this page */
	if (p->pos + len < (u8 *)page_address(p->page) + PAGE_SIZE)
		return p;

	/* We don't have enough room on this page, get a new one. */
	__free_page(p->page);

alloc:
	p->page = alloc_page(GFP_ATOMIC);
	if (!p->page)
		return NULL;
	p->pos = page_address(p->page);
	return p;
}

static void iwl_compute_pseudo_hdr_csum(void *iph, struct tcphdr *tcph,
					bool ipv6, unsigned int len)
{
	if (ipv6) {
		struct ipv6hdr *iphv6 = iph;

		tcph->check = ~csum_ipv6_magic(&iphv6->saddr, &iphv6->daddr,
					       len + tcph->doff * 4,
					       IPPROTO_TCP, 0);
	} else {
		struct iphdr *iphv4 = iph;

		ip_send_check(iphv4);
		tcph->check = ~csum_tcpudp_magic(iphv4->saddr, iphv4->daddr,
						 len + tcph->doff * 4,
						 IPPROTO_TCP, 0);
	}
}

static int iwl_fill_data_tbs_amsdu(struct iwl_trans *trans, struct sk_buff *skb,
				   struct iwl_txq *txq, u8 hdr_len,
				   struct iwl_cmd_meta *out_meta,
				   struct iwl_device_cmd *dev_cmd, u16 tb1_len)
{
	struct iwl_tx_cmd *tx_cmd = (void *)dev_cmd->payload;
	struct iwl_trans_pcie *trans_pcie = txq->trans_pcie;
	struct ieee80211_hdr *hdr = (void *)skb->data;
	unsigned int snap_ip_tcp_hdrlen, ip_hdrlen, total_len, hdr_room;
	unsigned int mss = skb_shinfo(skb)->gso_size;
	u16 length, iv_len, amsdu_pad;
	u8 *start_hdr;
	struct iwl_tso_hdr_page *hdr_page;
	struct page **page_ptr;
	int ret;
	struct tso_t tso;

	/* if the packet is protected, then it must be CCMP or GCMP */
	BUILD_BUG_ON(IEEE80211_CCMP_HDR_LEN != IEEE80211_GCMP_HDR_LEN);
	iv_len = ieee80211_has_protected(hdr->frame_control) ?
		IEEE80211_CCMP_HDR_LEN : 0;

	trace_iwlwifi_dev_tx(trans->dev, skb,
			     iwl_pcie_get_tfd(trans_pcie, txq, txq->write_ptr),
			     trans_pcie->tfd_size,
			     &dev_cmd->hdr, IWL_FIRST_TB_SIZE + tb1_len, 0);

	ip_hdrlen = skb_transport_header(skb) - skb_network_header(skb);
	snap_ip_tcp_hdrlen = 8 + ip_hdrlen + tcp_hdrlen(skb);
	total_len = skb->len - snap_ip_tcp_hdrlen - hdr_len - iv_len;
	amsdu_pad = 0;

	/* total amount of header we may need for this A-MSDU */
	hdr_room = DIV_ROUND_UP(total_len, mss) *
		(3 + snap_ip_tcp_hdrlen + sizeof(struct ethhdr)) + iv_len;

	/* Our device supports 9 segments at most, it will fit in 1 page */
	hdr_page = get_page_hdr(trans, hdr_room);
	if (!hdr_page)
		return -ENOMEM;

	get_page(hdr_page->page);
	start_hdr = hdr_page->pos;
	page_ptr = (void *)((u8 *)skb->cb + trans_pcie->page_offs);
	*page_ptr = hdr_page->page;
	memcpy(hdr_page->pos, skb->data + hdr_len, iv_len);
	hdr_page->pos += iv_len;

	/*
	 * Pull the ieee80211 header + IV to be able to use TSO core,
	 * we will restore it for the tx_status flow.
	 */
	skb_pull(skb, hdr_len + iv_len);

	/*
	 * Remove the length of all the headers that we don't actually
	 * have in the MPDU by themselves, but that we duplicate into
	 * all the different MSDUs inside the A-MSDU.
	 */
	le16_add_cpu(&tx_cmd->len, -snap_ip_tcp_hdrlen);

	tso_start(skb, &tso);

	while (total_len) {
		/* this is the data left for this subframe */
		unsigned int data_left =
			min_t(unsigned int, mss, total_len);
		struct sk_buff *csum_skb = NULL;
		unsigned int hdr_tb_len;
		dma_addr_t hdr_tb_phys;
		struct tcphdr *tcph;
		u8 *iph, *subf_hdrs_start = hdr_page->pos;

		total_len -= data_left;

		memset(hdr_page->pos, 0, amsdu_pad);
		hdr_page->pos += amsdu_pad;
		amsdu_pad = (4 - (sizeof(struct ethhdr) + snap_ip_tcp_hdrlen +
				  data_left)) & 0x3;
		ether_addr_copy(hdr_page->pos, ieee80211_get_DA(hdr));
		hdr_page->pos += ETH_ALEN;
		ether_addr_copy(hdr_page->pos, ieee80211_get_SA(hdr));
		hdr_page->pos += ETH_ALEN;

		length = snap_ip_tcp_hdrlen + data_left;
		*((__be16 *)hdr_page->pos) = cpu_to_be16(length);
		hdr_page->pos += sizeof(length);

		/*
		 * This will copy the SNAP as well which will be considered
		 * as MAC header.
		 */
		tso_build_hdr(skb, hdr_page->pos, &tso, data_left, !total_len);
		iph = hdr_page->pos + 8;
		tcph = (void *)(iph + ip_hdrlen);

		/* For testing on current hardware only */
		if (trans_pcie->sw_csum_tx) {
			csum_skb = alloc_skb(data_left + tcp_hdrlen(skb),
					     GFP_ATOMIC);
			if (!csum_skb) {
				ret = -ENOMEM;
				goto out_unmap;
			}

			iwl_compute_pseudo_hdr_csum(iph, tcph,
						    skb->protocol ==
							htons(ETH_P_IPV6),
						    data_left);

			skb_put_data(csum_skb, tcph, tcp_hdrlen(skb));
			skb_reset_transport_header(csum_skb);
			csum_skb->csum_start =
				(unsigned char *)tcp_hdr(csum_skb) -
						 csum_skb->head;
		}

		hdr_page->pos += snap_ip_tcp_hdrlen;

		hdr_tb_len = hdr_page->pos - start_hdr;
		hdr_tb_phys = dma_map_single(trans->dev, start_hdr,
					     hdr_tb_len, DMA_TO_DEVICE);
		if (unlikely(dma_mapping_error(trans->dev, hdr_tb_phys))) {
			dev_kfree_skb(csum_skb);
			ret = -EINVAL;
			goto out_unmap;
		}
		iwl_pcie_txq_build_tfd(trans, txq, hdr_tb_phys,
				       hdr_tb_len, false);
		trace_iwlwifi_dev_tx_tso_chunk(trans->dev, start_hdr,
					       hdr_tb_len);
		/* add this subframe's headers' length to the tx_cmd */
		le16_add_cpu(&tx_cmd->len, hdr_page->pos - subf_hdrs_start);

		/* prepare the start_hdr for the next subframe */
		start_hdr = hdr_page->pos;

		/* put the payload */
		while (data_left) {
			unsigned int size = min_t(unsigned int, tso.size,
						  data_left);
			dma_addr_t tb_phys;

			if (trans_pcie->sw_csum_tx)
				skb_put_data(csum_skb, tso.data, size);

			tb_phys = dma_map_single(trans->dev, tso.data,
						 size, DMA_TO_DEVICE);
			if (unlikely(dma_mapping_error(trans->dev, tb_phys))) {
				dev_kfree_skb(csum_skb);
				ret = -EINVAL;
				goto out_unmap;
			}

			iwl_pcie_txq_build_tfd(trans, txq, tb_phys,
					       size, false);
			trace_iwlwifi_dev_tx_tso_chunk(trans->dev, tso.data,
						       size);

			data_left -= size;
			tso_build_data(skb, &tso, size);
		}

		/* For testing on early hardware only */
		if (trans_pcie->sw_csum_tx) {
			__wsum csum;

			csum = skb_checksum(csum_skb,
					    skb_checksum_start_offset(csum_skb),
					    csum_skb->len -
					    skb_checksum_start_offset(csum_skb),
					    0);
			dev_kfree_skb(csum_skb);
			dma_sync_single_for_cpu(trans->dev, hdr_tb_phys,
						hdr_tb_len, DMA_TO_DEVICE);
			tcph->check = csum_fold(csum);
			dma_sync_single_for_device(trans->dev, hdr_tb_phys,
						   hdr_tb_len, DMA_TO_DEVICE);
		}
	}

	/* re -add the WiFi header and IV */
	skb_push(skb, hdr_len + iv_len);

	return 0;

out_unmap:
	iwl_pcie_tfd_unmap(trans, out_meta, txq, txq->write_ptr);
	return ret;
}
#else /* CONFIG_INET */
static int iwl_fill_data_tbs_amsdu(struct iwl_trans *trans, struct sk_buff *skb,
				   struct iwl_txq *txq, u8 hdr_len,
				   struct iwl_cmd_meta *out_meta,
				   struct iwl_device_cmd *dev_cmd, u16 tb1_len)
{
	/* No A-MSDU without CONFIG_INET */
	WARN_ON(1);

	return -1;
}
#endif /* CONFIG_INET */

int iwl_trans_pcie_tx(struct iwl_trans *trans, struct sk_buff *skb,
		      struct iwl_device_cmd *dev_cmd, int txq_id)
{
	struct iwl_trans_pcie *trans_pcie = IWL_TRANS_GET_PCIE_TRANS(trans);
	struct ieee80211_hdr *hdr;
	struct iwl_tx_cmd *tx_cmd = (struct iwl_tx_cmd *)dev_cmd->payload;
	struct iwl_cmd_meta *out_meta;
	struct iwl_txq *txq;
	dma_addr_t tb0_phys, tb1_phys, scratch_phys;
	void *tb1_addr;
	void *tfd;
	u16 len, tb1_len;
	bool wait_write_ptr;
	__le16 fc;
	u8 hdr_len;
	u16 wifi_seq;
	bool amsdu;

	txq = trans_pcie->txq[txq_id];

	if (WARN_ONCE(!test_bit(txq_id, trans_pcie->queue_used),
		      "TX on unused queue %d\n", txq_id))
		return -EINVAL;

	if (unlikely(trans_pcie->sw_csum_tx &&
		     skb->ip_summed == CHECKSUM_PARTIAL)) {
		int offs = skb_checksum_start_offset(skb);
		int csum_offs = offs + skb->csum_offset;
		__wsum csum;

		if (skb_ensure_writable(skb, csum_offs + sizeof(__sum16)))
			return -1;

		csum = skb_checksum(skb, offs, skb->len - offs, 0);
		*(__sum16 *)(skb->data + csum_offs) = csum_fold(csum);

		skb->ip_summed = CHECKSUM_UNNECESSARY;
	}

	if (skb_is_nonlinear(skb) &&
	    skb_shinfo(skb)->nr_frags > IWL_PCIE_MAX_FRAGS(trans_pcie) &&
	    __skb_linearize(skb))
		return -ENOMEM;

	/* mac80211 always puts the full header into the SKB's head,
	 * so there's no need to check if it's readable there
	 */
	hdr = (struct ieee80211_hdr *)skb->data;
	fc = hdr->frame_control;
	hdr_len = ieee80211_hdrlen(fc);

	spin_lock(&txq->lock);

	if (iwl_queue_space(txq) < txq->high_mark) {
		iwl_stop_queue(trans, txq);

		/* don't put the packet on the ring, if there is no room */
		if (unlikely(iwl_queue_space(txq) < 3)) {
			struct iwl_device_cmd **dev_cmd_ptr;

			dev_cmd_ptr = (void *)((u8 *)skb->cb +
					       trans_pcie->dev_cmd_offs);

			*dev_cmd_ptr = dev_cmd;
			__skb_queue_tail(&txq->overflow_q, skb);

			spin_unlock(&txq->lock);
			return 0;
		}
	}

	/* In AGG mode, the index in the ring must correspond to the WiFi
	 * sequence number. This is a HW requirements to help the SCD to parse
	 * the BA.
	 * Check here that the packets are in the right place on the ring.
	 */
	wifi_seq = IEEE80211_SEQ_TO_SN(le16_to_cpu(hdr->seq_ctrl));
	WARN_ONCE(txq->ampdu &&
		  (wifi_seq & 0xff) != txq->write_ptr,
		  "Q: %d WiFi Seq %d tfdNum %d",
		  txq_id, wifi_seq, txq->write_ptr);

	/* Set up driver data for this TFD */
	txq->entries[txq->write_ptr].skb = skb;
	txq->entries[txq->write_ptr].cmd = dev_cmd;

	dev_cmd->hdr.sequence =
		cpu_to_le16((u16)(QUEUE_TO_SEQ(txq_id) |
			    INDEX_TO_SEQ(txq->write_ptr)));

	tb0_phys = iwl_pcie_get_first_tb_dma(txq, txq->write_ptr);
	scratch_phys = tb0_phys + sizeof(struct iwl_cmd_header) +
		       offsetof(struct iwl_tx_cmd, scratch);

	tx_cmd->dram_lsb_ptr = cpu_to_le32(scratch_phys);
	tx_cmd->dram_msb_ptr = iwl_get_dma_hi_addr(scratch_phys);

	/* Set up first empty entry in queue's array of Tx/cmd buffers */
	out_meta = &txq->entries[txq->write_ptr].meta;
	out_meta->flags = 0;

	/*
	 * The second TB (tb1) points to the remainder of the TX command
	 * and the 802.11 header - dword aligned size
	 * (This calculation modifies the TX command, so do it before the
	 * setup of the first TB)
	 */
	len = sizeof(struct iwl_tx_cmd) + sizeof(struct iwl_cmd_header) +
	      hdr_len - IWL_FIRST_TB_SIZE;
	/* do not align A-MSDU to dword as the subframe header aligns it */
	amsdu = ieee80211_is_data_qos(fc) &&
		(*ieee80211_get_qos_ctl(hdr) &
		 IEEE80211_QOS_CTL_A_MSDU_PRESENT);
	if (trans_pcie->sw_csum_tx || !amsdu) {
		tb1_len = ALIGN(len, 4);
		/* Tell NIC about any 2-byte padding after MAC header */
		if (tb1_len != len)
			tx_cmd->tx_flags |= TX_CMD_FLG_MH_PAD_MSK;
	} else {
		tb1_len = len;
	}

	/*
	 * The first TB points to bi-directional DMA data, we'll
	 * memcpy the data into it later.
	 */
	iwl_pcie_txq_build_tfd(trans, txq, tb0_phys,
			       IWL_FIRST_TB_SIZE, true);

	/* there must be data left over for TB1 or this code must be changed */
	BUILD_BUG_ON(sizeof(struct iwl_tx_cmd) < IWL_FIRST_TB_SIZE);

	/* map the data for TB1 */
	tb1_addr = ((u8 *)&dev_cmd->hdr) + IWL_FIRST_TB_SIZE;
	tb1_phys = dma_map_single(trans->dev, tb1_addr, tb1_len, DMA_TO_DEVICE);
	if (unlikely(dma_mapping_error(trans->dev, tb1_phys)))
		goto out_err;
	iwl_pcie_txq_build_tfd(trans, txq, tb1_phys, tb1_len, false);

	if (amsdu) {
		if (unlikely(iwl_fill_data_tbs_amsdu(trans, skb, txq, hdr_len,
						     out_meta, dev_cmd,
						     tb1_len)))
			goto out_err;
	} else if (unlikely(iwl_fill_data_tbs(trans, skb, txq, hdr_len,
				       out_meta, dev_cmd, tb1_len))) {
		goto out_err;
	}

	/* building the A-MSDU might have changed this data, so memcpy it now */
	memcpy(&txq->first_tb_bufs[txq->write_ptr], &dev_cmd->hdr,
	       IWL_FIRST_TB_SIZE);

	tfd = iwl_pcie_get_tfd(trans_pcie, txq, txq->write_ptr);
	/* Set up entry for this TFD in Tx byte-count array */
	iwl_pcie_txq_update_byte_cnt_tbl(trans, txq, le16_to_cpu(tx_cmd->len),
					 iwl_pcie_tfd_get_num_tbs(trans, tfd));

	wait_write_ptr = ieee80211_has_morefrags(fc);

	/* start timer if queue currently empty */
	if (txq->read_ptr == txq->write_ptr) {
		if (txq->wd_timeout) {
			/*
			 * If the TXQ is active, then set the timer, if not,
			 * set the timer in remainder so that the timer will
			 * be armed with the right value when the station will
			 * wake up.
			 */
			if (!txq->frozen)
				mod_timer(&txq->stuck_timer,
					  jiffies + txq->wd_timeout);
			else
				txq->frozen_expiry_remainder = txq->wd_timeout;
		}
		IWL_DEBUG_RPM(trans, "Q: %d first tx - take ref\n", txq->id);
		iwl_trans_ref(trans);
	}

	/* Tell device the write index *just past* this latest filled TFD */
	txq->write_ptr = iwl_queue_inc_wrap(txq->write_ptr);
	if (!wait_write_ptr)
		iwl_pcie_txq_inc_wr_ptr(trans, txq);

	/*
	 * At this point the frame is "transmitted" successfully
	 * and we will get a TX status notification eventually.
	 */
	spin_unlock(&txq->lock);
	return 0;
out_err:
	spin_unlock(&txq->lock);
	return -1;
}