summaryrefslogtreecommitdiffstats
path: root/drivers/usb/musb/musb_host.c
blob: d227a71d85e19cffef3a11662bb9a23affa18fb3 (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
2447
2448
2449
2450
2451
2452
2453
2454
2455
2456
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
2479
2480
2481
2482
2483
2484
2485
2486
2487
2488
2489
2490
2491
2492
2493
2494
2495
2496
2497
2498
2499
2500
2501
2502
2503
2504
2505
2506
2507
2508
2509
2510
2511
2512
2513
2514
2515
2516
2517
2518
2519
2520
2521
2522
2523
2524
2525
2526
2527
2528
2529
2530
2531
2532
2533
2534
2535
2536
2537
2538
2539
2540
2541
2542
2543
2544
2545
2546
2547
2548
2549
2550
2551
2552
2553
2554
2555
2556
2557
2558
2559
2560
2561
2562
2563
2564
2565
2566
2567
2568
2569
2570
2571
2572
2573
2574
2575
2576
2577
2578
2579
2580
2581
2582
2583
2584
2585
2586
2587
2588
2589
2590
2591
2592
2593
2594
2595
2596
2597
2598
2599
2600
2601
2602
2603
2604
2605
2606
2607
2608
2609
2610
2611
2612
2613
2614
2615
2616
2617
2618
2619
2620
2621
2622
2623
2624
2625
2626
2627
2628
2629
2630
2631
2632
2633
2634
2635
2636
2637
2638
2639
2640
2641
2642
2643
2644
2645
2646
2647
2648
2649
2650
2651
2652
2653
2654
2655
2656
2657
2658
2659
2660
2661
2662
2663
2664
2665
2666
2667
2668
2669
2670
2671
2672
2673
2674
2675
2676
2677
2678
2679
2680
2681
2682
2683
2684
2685
2686
2687
2688
2689
2690
2691
2692
2693
2694
2695
2696
2697
2698
2699
2700
2701
2702
2703
2704
2705
2706
2707
2708
2709
2710
2711
2712
2713
2714
2715
2716
2717
2718
2719
2720
2721
2722
2723
2724
2725
2726
2727
2728
2729
2730
2731
2732
2733
2734
2735
2736
2737
2738
2739
2740
2741
2742
2743
2744
2745
2746
2747
2748
2749
2750
2751
2752
2753
2754
2755
2756
2757
2758
2759
2760
2761
2762
2763
2764
2765
2766
2767
2768
2769
2770
2771
2772
2773
2774
2775
2776
2777
2778
2779
2780
2781
2782
2783
2784
2785
2786
2787
2788
2789
2790
2791
2792
2793
2794
2795
2796
2797
2798
2799
2800
2801
2802
2803
2804
2805
2806
2807
2808
2809
2810
2811
2812
2813
2814
2815
2816
2817
2818
2819
2820
2821
2822
2823
2824
2825
2826
2827
2828
2829
2830
2831
2832
2833
2834
2835
/*
 * MUSB OTG driver host support
 *
 * Copyright 2005 Mentor Graphics Corporation
 * Copyright (C) 2005-2006 by Texas Instruments
 * Copyright (C) 2006-2007 Nokia Corporation
 * Copyright (C) 2008-2009 MontaVista Software, Inc. <source@mvista.com>
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License
 * version 2 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 St, Fifth Floor, Boston, MA
 * 02110-1301 USA
 *
 * THIS SOFTWARE IS PROVIDED "AS IS" AND ANY EXPRESS OR IMPLIED
 * WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN
 * NO EVENT SHALL THE AUTHORS BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF
 * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
 * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 *
 */

#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/delay.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/errno.h>
#include <linux/list.h>
#include <linux/dma-mapping.h>

#include "musb_core.h"
#include "musb_host.h"

/* MUSB HOST status 22-mar-2006
 *
 * - There's still lots of partial code duplication for fault paths, so
 *   they aren't handled as consistently as they need to be.
 *
 * - PIO mostly behaved when last tested.
 *     + including ep0, with all usbtest cases 9, 10
 *     + usbtest 14 (ep0out) doesn't seem to run at all
 *     + double buffered OUT/TX endpoints saw stalls(!) with certain usbtest
 *       configurations, but otherwise double buffering passes basic tests.
 *     + for 2.6.N, for N > ~10, needs API changes for hcd framework.
 *
 * - DMA (CPPI) ... partially behaves, not currently recommended
 *     + about 1/15 the speed of typical EHCI implementations (PCI)
 *     + RX, all too often reqpkt seems to misbehave after tx
 *     + TX, no known issues (other than evident silicon issue)
 *
 * - DMA (Mentor/OMAP) ...has at least toggle update problems
 *
 * - [23-feb-2009] minimal traffic scheduling to avoid bulk RX packet
 *   starvation ... nothing yet for TX, interrupt, or bulk.
 *
 * - Not tested with HNP, but some SRP paths seem to behave.
 *
 * NOTE 24-August-2006:
 *
 * - Bulk traffic finally uses both sides of hardware ep1, freeing up an
 *   extra endpoint for periodic use enabling hub + keybd + mouse.  That
 *   mostly works, except that with "usbnet" it's easy to trigger cases
 *   with "ping" where RX loses.  (a) ping to davinci, even "ping -f",
 *   fine; but (b) ping _from_ davinci, even "ping -c 1", ICMP RX loses
 *   although ARP RX wins.  (That test was done with a full speed link.)
 */


/*
 * NOTE on endpoint usage:
 *
 * CONTROL transfers all go through ep0.  BULK ones go through dedicated IN
 * and OUT endpoints ... hardware is dedicated for those "async" queue(s).
 * (Yes, bulk _could_ use more of the endpoints than that, and would even
 * benefit from it.)
 *
 * INTERUPPT and ISOCHRONOUS transfers are scheduled to the other endpoints.
 * So far that scheduling is both dumb and optimistic:  the endpoint will be
 * "claimed" until its software queue is no longer refilled.  No multiplexing
 * of transfers between endpoints, or anything clever.
 */

struct musb *hcd_to_musb(struct usb_hcd *hcd)
{
	return *(struct musb **) hcd->hcd_priv;
}


static void musb_ep_program(struct musb *musb, u8 epnum,
			struct urb *urb, int is_out,
			u8 *buf, u32 offset, u32 len);

/*
 * Clear TX fifo. Needed to avoid BABBLE errors.
 */
static void musb_h_tx_flush_fifo(struct musb_hw_ep *ep)
{
	struct musb	*musb = ep->musb;
	void __iomem	*epio = ep->regs;
	u16		csr;
	int		retries = 1000;

	csr = musb_readw(epio, MUSB_TXCSR);
	while (csr & MUSB_TXCSR_FIFONOTEMPTY) {
		csr |= MUSB_TXCSR_FLUSHFIFO | MUSB_TXCSR_TXPKTRDY;
		musb_writew(epio, MUSB_TXCSR, csr);
		csr = musb_readw(epio, MUSB_TXCSR);

		/*
		 * FIXME: sometimes the tx fifo flush failed, it has been
		 * observed during device disconnect on AM335x.
		 *
		 * To reproduce the issue, ensure tx urb(s) are queued when
		 * unplug the usb device which is connected to AM335x usb
		 * host port.
		 *
		 * I found using a usb-ethernet device and running iperf
		 * (client on AM335x) has very high chance to trigger it.
		 *
		 * Better to turn on dev_dbg() in musb_cleanup_urb() with
		 * CPPI enabled to see the issue when aborting the tx channel.
		 */
		if (dev_WARN_ONCE(musb->controller, retries-- < 1,
				"Could not flush host TX%d fifo: csr: %04x\n",
				ep->epnum, csr))
			return;
	}
}

static void musb_h_ep0_flush_fifo(struct musb_hw_ep *ep)
{
	void __iomem	*epio = ep->regs;
	u16		csr;
	int		retries = 5;

	/* scrub any data left in the fifo */
	do {
		csr = musb_readw(epio, MUSB_TXCSR);
		if (!(csr & (MUSB_CSR0_TXPKTRDY | MUSB_CSR0_RXPKTRDY)))
			break;
		musb_writew(epio, MUSB_TXCSR, MUSB_CSR0_FLUSHFIFO);
		csr = musb_readw(epio, MUSB_TXCSR);
		udelay(10);
	} while (--retries);

	WARN(!retries, "Could not flush host TX%d fifo: csr: %04x\n",
			ep->epnum, csr);

	/* and reset for the next transfer */
	musb_writew(epio, MUSB_TXCSR, 0);
}

/*
 * Start transmit. Caller is responsible for locking shared resources.
 * musb must be locked.
 */
static inline void musb_h_tx_start(struct musb_hw_ep *ep)
{
	u16	txcsr;

	/* NOTE: no locks here; caller should lock and select EP */
	if (ep->epnum) {
		txcsr = musb_readw(ep->regs, MUSB_TXCSR);
		txcsr |= MUSB_TXCSR_TXPKTRDY | MUSB_TXCSR_H_WZC_BITS;
		musb_writew(ep->regs, MUSB_TXCSR, txcsr);
	} else {
		txcsr = MUSB_CSR0_H_SETUPPKT | MUSB_CSR0_TXPKTRDY;
		musb_writew(ep->regs, MUSB_CSR0, txcsr);
	}

}

static inline void musb_h_tx_dma_start(struct musb_hw_ep *ep)
{
	u16	txcsr;

	/* NOTE: no locks here; caller should lock and select EP */
	txcsr = musb_readw(ep->regs, MUSB_TXCSR);
	txcsr |= MUSB_TXCSR_DMAENAB | MUSB_TXCSR_H_WZC_BITS;
	if (is_cppi_enabled(ep->musb))
		txcsr |= MUSB_TXCSR_DMAMODE;
	musb_writew(ep->regs, MUSB_TXCSR, txcsr);
}

static void musb_ep_set_qh(struct musb_hw_ep *ep, int is_in, struct musb_qh *qh)
{
	if (is_in != 0 || ep->is_shared_fifo)
		ep->in_qh  = qh;
	if (is_in == 0 || ep->is_shared_fifo)
		ep->out_qh = qh;
}

static struct musb_qh *musb_ep_get_qh(struct musb_hw_ep *ep, int is_in)
{
	return is_in ? ep->in_qh : ep->out_qh;
}

/*
 * Start the URB at the front of an endpoint's queue
 * end must be claimed from the caller.
 *
 * Context: controller locked, irqs blocked
 */
static void
musb_start_urb(struct musb *musb, int is_in, struct musb_qh *qh)
{
	u16			frame;
	u32			len;
	void __iomem		*mbase =  musb->mregs;
	struct urb		*urb = next_urb(qh);
	void			*buf = urb->transfer_buffer;
	u32			offset = 0;
	struct musb_hw_ep	*hw_ep = qh->hw_ep;
	unsigned		pipe = urb->pipe;
	u8			address = usb_pipedevice(pipe);
	int			epnum = hw_ep->epnum;

	/* initialize software qh state */
	qh->offset = 0;
	qh->segsize = 0;

	/* gather right source of data */
	switch (qh->type) {
	case USB_ENDPOINT_XFER_CONTROL:
		/* control transfers always start with SETUP */
		is_in = 0;
		musb->ep0_stage = MUSB_EP0_START;
		buf = urb->setup_packet;
		len = 8;
		break;
	case USB_ENDPOINT_XFER_ISOC:
		qh->iso_idx = 0;
		qh->frame = 0;
		offset = urb->iso_frame_desc[0].offset;
		len = urb->iso_frame_desc[0].length;
		break;
	default:		/* bulk, interrupt */
		/* actual_length may be nonzero on retry paths */
		buf = urb->transfer_buffer + urb->actual_length;
		len = urb->transfer_buffer_length - urb->actual_length;
	}

	dev_dbg(musb->controller, "qh %p urb %p dev%d ep%d%s%s, hw_ep %d, %p/%d\n",
			qh, urb, address, qh->epnum,
			is_in ? "in" : "out",
			({char *s; switch (qh->type) {
			case USB_ENDPOINT_XFER_CONTROL:	s = ""; break;
			case USB_ENDPOINT_XFER_BULK:	s = "-bulk"; break;
			case USB_ENDPOINT_XFER_ISOC:	s = "-iso"; break;
			default:			s = "-intr"; break;
			} s; }),
			epnum, buf + offset, len);

	/* Configure endpoint */
	musb_ep_set_qh(hw_ep, is_in, qh);
	musb_ep_program(musb, epnum, urb, !is_in, buf, offset, len);

	/* transmit may have more work: start it when it is time */
	if (is_in)
		return;

	/* determine if the time is right for a periodic transfer */
	switch (qh->type) {
	case USB_ENDPOINT_XFER_ISOC:
	case USB_ENDPOINT_XFER_INT:
		dev_dbg(musb->controller, "check whether there's still time for periodic Tx\n");
		frame = musb_readw(mbase, MUSB_FRAME);
		/* FIXME this doesn't implement that scheduling policy ...
		 * or handle framecounter wrapping
		 */
		if (1) {	/* Always assume URB_ISO_ASAP */
			/* REVISIT the SOF irq handler shouldn't duplicate
			 * this code; and we don't init urb->start_frame...
			 */
			qh->frame = 0;
			goto start;
		} else {
			qh->frame = urb->start_frame;
			/* enable SOF interrupt so we can count down */
			dev_dbg(musb->controller, "SOF for %d\n", epnum);
#if 1 /* ifndef	CONFIG_ARCH_DAVINCI */
			musb_writeb(mbase, MUSB_INTRUSBE, 0xff);
#endif
		}
		break;
	default:
start:
		dev_dbg(musb->controller, "Start TX%d %s\n", epnum,
			hw_ep->tx_channel ? "dma" : "pio");

		if (!hw_ep->tx_channel)
			musb_h_tx_start(hw_ep);
		else if (is_cppi_enabled(musb) || tusb_dma_omap(musb))
			musb_h_tx_dma_start(hw_ep);
	}
}

/* Context: caller owns controller lock, IRQs are blocked */
static void musb_giveback(struct musb *musb, struct urb *urb, int status)
__releases(musb->lock)
__acquires(musb->lock)
{
	dev_dbg(musb->controller,
			"complete %p %pF (%d), dev%d ep%d%s, %d/%d\n",
			urb, urb->complete, status,
			usb_pipedevice(urb->pipe),
			usb_pipeendpoint(urb->pipe),
			usb_pipein(urb->pipe) ? "in" : "out",
			urb->actual_length, urb->transfer_buffer_length
			);

	usb_hcd_unlink_urb_from_ep(musb->hcd, urb);
	spin_unlock(&musb->lock);
	usb_hcd_giveback_urb(musb->hcd, urb, status);
	spin_lock(&musb->lock);
}

/* For bulk/interrupt endpoints only */
static inline void musb_save_toggle(struct musb_qh *qh, int is_in,
				    struct urb *urb)
{
	void __iomem		*epio = qh->hw_ep->regs;
	u16			csr;

	/*
	 * FIXME: the current Mentor DMA code seems to have
	 * problems getting toggle correct.
	 */

	if (is_in)
		csr = musb_readw(epio, MUSB_RXCSR) & MUSB_RXCSR_H_DATATOGGLE;
	else
		csr = musb_readw(epio, MUSB_TXCSR) & MUSB_TXCSR_H_DATATOGGLE;

	usb_settoggle(urb->dev, qh->epnum, !is_in, csr ? 1 : 0);
}

/*
 * Advance this hardware endpoint's queue, completing the specified URB and
 * advancing to either the next URB queued to that qh, or else invalidating
 * that qh and advancing to the next qh scheduled after the current one.
 *
 * Context: caller owns controller lock, IRQs are blocked
 */
static void musb_advance_schedule(struct musb *musb, struct urb *urb,
				  struct musb_hw_ep *hw_ep, int is_in)
{
	struct musb_qh		*qh = musb_ep_get_qh(hw_ep, is_in);
	struct musb_hw_ep	*ep = qh->hw_ep;
	int			ready = qh->is_ready;
	int			status;

	status = (urb->status == -EINPROGRESS) ? 0 : urb->status;

	/* save toggle eagerly, for paranoia */
	switch (qh->type) {
	case USB_ENDPOINT_XFER_BULK:
	case USB_ENDPOINT_XFER_INT:
		musb_save_toggle(qh, is_in, urb);
		break;
	case USB_ENDPOINT_XFER_ISOC:
		if (status == 0 && urb->error_count)
			status = -EXDEV;
		break;
	}

	qh->is_ready = 0;
	musb_giveback(musb, urb, status);
	qh->is_ready = ready;

	/* reclaim resources (and bandwidth) ASAP; deschedule it, and
	 * invalidate qh as soon as list_empty(&hep->urb_list)
	 */
	if (list_empty(&qh->hep->urb_list)) {
		struct list_head	*head;
		struct dma_controller	*dma = musb->dma_controller;

		if (is_in) {
			ep->rx_reinit = 1;
			if (ep->rx_channel) {
				dma->channel_release(ep->rx_channel);
				ep->rx_channel = NULL;
			}
		} else {
			ep->tx_reinit = 1;
			if (ep->tx_channel) {
				dma->channel_release(ep->tx_channel);
				ep->tx_channel = NULL;
			}
		}

		/* Clobber old pointers to this qh */
		musb_ep_set_qh(ep, is_in, NULL);
		qh->hep->hcpriv = NULL;

		switch (qh->type) {

		case USB_ENDPOINT_XFER_CONTROL:
		case USB_ENDPOINT_XFER_BULK:
			/* fifo policy for these lists, except that NAKing
			 * should rotate a qh to the end (for fairness).
			 */
			if (qh->mux == 1) {
				head = qh->ring.prev;
				list_del(&qh->ring);
				kfree(qh);
				qh = first_qh(head);
				break;
			}

		case USB_ENDPOINT_XFER_ISOC:
		case USB_ENDPOINT_XFER_INT:
			/* this is where periodic bandwidth should be
			 * de-allocated if it's tracked and allocated;
			 * and where we'd update the schedule tree...
			 */
			kfree(qh);
			qh = NULL;
			break;
		}
	}

	/*
	 * The pipe must be broken if current urb->status is set, so don't
	 * start next urb.
	 * TODO: to minimize the risk of regression, only check urb->status
	 * for RX, until we have a test case to understand the behavior of TX.
	 */
	if ((!status || !is_in) && qh && qh->is_ready) {
		dev_dbg(musb->controller, "... next ep%d %cX urb %p\n",
		    hw_ep->epnum, is_in ? 'R' : 'T', next_urb(qh));
		musb_start_urb(musb, is_in, qh);
	}
}

static u16 musb_h_flush_rxfifo(struct musb_hw_ep *hw_ep, u16 csr)
{
	/* we don't want fifo to fill itself again;
	 * ignore dma (various models),
	 * leave toggle alone (may not have been saved yet)
	 */
	csr |= MUSB_RXCSR_FLUSHFIFO | MUSB_RXCSR_RXPKTRDY;
	csr &= ~(MUSB_RXCSR_H_REQPKT
		| MUSB_RXCSR_H_AUTOREQ
		| MUSB_RXCSR_AUTOCLEAR);

	/* write 2x to allow double buffering */
	musb_writew(hw_ep->regs, MUSB_RXCSR, csr);
	musb_writew(hw_ep->regs, MUSB_RXCSR, csr);

	/* flush writebuffer */
	return musb_readw(hw_ep->regs, MUSB_RXCSR);
}

/*
 * PIO RX for a packet (or part of it).
 */
static bool
musb_host_packet_rx(struct musb *musb, struct urb *urb, u8 epnum, u8 iso_err)
{
	u16			rx_count;
	u8			*buf;
	u16			csr;
	bool			done = false;
	u32			length;
	int			do_flush = 0;
	struct musb_hw_ep	*hw_ep = musb->endpoints + epnum;
	void __iomem		*epio = hw_ep->regs;
	struct musb_qh		*qh = hw_ep->in_qh;
	int			pipe = urb->pipe;
	void			*buffer = urb->transfer_buffer;

	/* musb_ep_select(mbase, epnum); */
	rx_count = musb_readw(epio, MUSB_RXCOUNT);
	dev_dbg(musb->controller, "RX%d count %d, buffer %p len %d/%d\n", epnum, rx_count,
			urb->transfer_buffer, qh->offset,
			urb->transfer_buffer_length);

	/* unload FIFO */
	if (usb_pipeisoc(pipe)) {
		int					status = 0;
		struct usb_iso_packet_descriptor	*d;

		if (iso_err) {
			status = -EILSEQ;
			urb->error_count++;
		}

		d = urb->iso_frame_desc + qh->iso_idx;
		buf = buffer + d->offset;
		length = d->length;
		if (rx_count > length) {
			if (status == 0) {
				status = -EOVERFLOW;
				urb->error_count++;
			}
			dev_dbg(musb->controller, "** OVERFLOW %d into %d\n", rx_count, length);
			do_flush = 1;
		} else
			length = rx_count;
		urb->actual_length += length;
		d->actual_length = length;

		d->status = status;

		/* see if we are done */
		done = (++qh->iso_idx >= urb->number_of_packets);
	} else {
		/* non-isoch */
		buf = buffer + qh->offset;
		length = urb->transfer_buffer_length - qh->offset;
		if (rx_count > length) {
			if (urb->status == -EINPROGRESS)
				urb->status = -EOVERFLOW;
			dev_dbg(musb->controller, "** OVERFLOW %d into %d\n", rx_count, length);
			do_flush = 1;
		} else
			length = rx_count;
		urb->actual_length += length;
		qh->offset += length;

		/* see if we are done */
		done = (urb->actual_length == urb->transfer_buffer_length)
			|| (rx_count < qh->maxpacket)
			|| (urb->status != -EINPROGRESS);
		if (done
				&& (urb->status == -EINPROGRESS)
				&& (urb->transfer_flags & URB_SHORT_NOT_OK)
				&& (urb->actual_length
					< urb->transfer_buffer_length))
			urb->status = -EREMOTEIO;
	}

	musb_read_fifo(hw_ep, length, buf);

	csr = musb_readw(epio, MUSB_RXCSR);
	csr |= MUSB_RXCSR_H_WZC_BITS;
	if (unlikely(do_flush))
		musb_h_flush_rxfifo(hw_ep, csr);
	else {
		/* REVISIT this assumes AUTOCLEAR is never set */
		csr &= ~(MUSB_RXCSR_RXPKTRDY | MUSB_RXCSR_H_REQPKT);
		if (!done)
			csr |= MUSB_RXCSR_H_REQPKT;
		musb_writew(epio, MUSB_RXCSR, csr);
	}

	return done;
}

/* we don't always need to reinit a given side of an endpoint...
 * when we do, use tx/rx reinit routine and then construct a new CSR
 * to address data toggle, NYET, and DMA or PIO.
 *
 * it's possible that driver bugs (especially for DMA) or aborting a
 * transfer might have left the endpoint busier than it should be.
 * the busy/not-empty tests are basically paranoia.
 */
static void
musb_rx_reinit(struct musb *musb, struct musb_qh *qh, u8 epnum)
{
	struct musb_hw_ep *ep = musb->endpoints + epnum;
	u16	csr;

	/* NOTE:  we know the "rx" fifo reinit never triggers for ep0.
	 * That always uses tx_reinit since ep0 repurposes TX register
	 * offsets; the initial SETUP packet is also a kind of OUT.
	 */

	/* if programmed for Tx, put it in RX mode */
	if (ep->is_shared_fifo) {
		csr = musb_readw(ep->regs, MUSB_TXCSR);
		if (csr & MUSB_TXCSR_MODE) {
			musb_h_tx_flush_fifo(ep);
			csr = musb_readw(ep->regs, MUSB_TXCSR);
			musb_writew(ep->regs, MUSB_TXCSR,
				    csr | MUSB_TXCSR_FRCDATATOG);
		}

		/*
		 * Clear the MODE bit (and everything else) to enable Rx.
		 * NOTE: we mustn't clear the DMAMODE bit before DMAENAB.
		 */
		if (csr & MUSB_TXCSR_DMAMODE)
			musb_writew(ep->regs, MUSB_TXCSR, MUSB_TXCSR_DMAMODE);
		musb_writew(ep->regs, MUSB_TXCSR, 0);

	/* scrub all previous state, clearing toggle */
	}
	csr = musb_readw(ep->regs, MUSB_RXCSR);
	if (csr & MUSB_RXCSR_RXPKTRDY)
		WARNING("rx%d, packet/%d ready?\n", ep->epnum,
			musb_readw(ep->regs, MUSB_RXCOUNT));

	musb_h_flush_rxfifo(ep, MUSB_RXCSR_CLRDATATOG);

	/* target addr and (for multipoint) hub addr/port */
	if (musb->is_multipoint) {
		musb_write_rxfunaddr(musb, epnum, qh->addr_reg);
		musb_write_rxhubaddr(musb, epnum, qh->h_addr_reg);
		musb_write_rxhubport(musb, epnum, qh->h_port_reg);
	} else
		musb_writeb(musb->mregs, MUSB_FADDR, qh->addr_reg);

	/* protocol/endpoint, interval/NAKlimit, i/o size */
	musb_writeb(ep->regs, MUSB_RXTYPE, qh->type_reg);
	musb_writeb(ep->regs, MUSB_RXINTERVAL, qh->intv_reg);
	/* NOTE: bulk combining rewrites high bits of maxpacket */
	/* Set RXMAXP with the FIFO size of the endpoint
	 * to disable double buffer mode.
	 */
	if (musb->double_buffer_not_ok)
		musb_writew(ep->regs, MUSB_RXMAXP, ep->max_packet_sz_rx);
	else
		musb_writew(ep->regs, MUSB_RXMAXP,
				qh->maxpacket | ((qh->hb_mult - 1) << 11));

	ep->rx_reinit = 0;
}

static void musb_tx_dma_set_mode_mentor(struct dma_controller *dma,
		struct musb_hw_ep *hw_ep, struct musb_qh *qh,
		struct urb *urb, u32 offset,
		u32 *length, u8 *mode)
{
	struct dma_channel	*channel = hw_ep->tx_channel;
	void __iomem		*epio = hw_ep->regs;
	u16			pkt_size = qh->maxpacket;
	u16			csr;

	if (*length > channel->max_len)
		*length = channel->max_len;

	csr = musb_readw(epio, MUSB_TXCSR);
	if (*length > pkt_size) {
		*mode = 1;
		csr |= MUSB_TXCSR_DMAMODE | MUSB_TXCSR_DMAENAB;
		/* autoset shouldn't be set in high bandwidth */
		/*
		 * Enable Autoset according to table
		 * below
		 * bulk_split hb_mult	Autoset_Enable
		 *	0	1	Yes(Normal)
		 *	0	>1	No(High BW ISO)
		 *	1	1	Yes(HS bulk)
		 *	1	>1	Yes(FS bulk)
		 */
		if (qh->hb_mult == 1 || (qh->hb_mult > 1 &&
					can_bulk_split(hw_ep->musb, qh->type)))
			csr |= MUSB_TXCSR_AUTOSET;
	} else {
		*mode = 0;
		csr &= ~(MUSB_TXCSR_AUTOSET | MUSB_TXCSR_DMAMODE);
		csr |= MUSB_TXCSR_DMAENAB; /* against programmer's guide */
	}
	channel->desired_mode = *mode;
	musb_writew(epio, MUSB_TXCSR, csr);
}

static void musb_tx_dma_set_mode_cppi_tusb(struct dma_controller *dma,
					   struct musb_hw_ep *hw_ep,
					   struct musb_qh *qh,
					   struct urb *urb,
					   u32 offset,
					   u32 *length,
					   u8 *mode)
{
	struct dma_channel *channel = hw_ep->tx_channel;

	channel->actual_len = 0;

	/*
	 * TX uses "RNDIS" mode automatically but needs help
	 * to identify the zero-length-final-packet case.
	 */
	*mode = (urb->transfer_flags & URB_ZERO_PACKET) ? 1 : 0;
}

static bool musb_tx_dma_program(struct dma_controller *dma,
		struct musb_hw_ep *hw_ep, struct musb_qh *qh,
		struct urb *urb, u32 offset, u32 length)
{
	struct dma_channel	*channel = hw_ep->tx_channel;
	u16			pkt_size = qh->maxpacket;
	u8			mode;

	if (musb_dma_inventra(hw_ep->musb) || musb_dma_ux500(hw_ep->musb))
		musb_tx_dma_set_mode_mentor(dma, hw_ep, qh, urb, offset,
					    &length, &mode);
	else if (is_cppi_enabled(hw_ep->musb) || tusb_dma_omap(hw_ep->musb))
		musb_tx_dma_set_mode_cppi_tusb(dma, hw_ep, qh, urb, offset,
					       &length, &mode);
	else
		return false;

	qh->segsize = length;

	/*
	 * Ensure the data reaches to main memory before starting
	 * DMA transfer
	 */
	wmb();

	if (!dma->channel_program(channel, pkt_size, mode,
			urb->transfer_dma + offset, length)) {
		void __iomem *epio = hw_ep->regs;
		u16 csr;

		dma->channel_release(channel);
		hw_ep->tx_channel = NULL;

		csr = musb_readw(epio, MUSB_TXCSR);
		csr &= ~(MUSB_TXCSR_AUTOSET | MUSB_TXCSR_DMAENAB);
		musb_writew(epio, MUSB_TXCSR, csr | MUSB_TXCSR_H_WZC_BITS);
		return false;
	}
	return true;
}

/*
 * Program an HDRC endpoint as per the given URB
 * Context: irqs blocked, controller lock held
 */
static void musb_ep_program(struct musb *musb, u8 epnum,
			struct urb *urb, int is_out,
			u8 *buf, u32 offset, u32 len)
{
	struct dma_controller	*dma_controller;
	struct dma_channel	*dma_channel;
	u8			dma_ok;
	void __iomem		*mbase = musb->mregs;
	struct musb_hw_ep	*hw_ep = musb->endpoints + epnum;
	void __iomem		*epio = hw_ep->regs;
	struct musb_qh		*qh = musb_ep_get_qh(hw_ep, !is_out);
	u16			packet_sz = qh->maxpacket;
	u8			use_dma = 1;
	u16			csr;

	dev_dbg(musb->controller, "%s hw%d urb %p spd%d dev%d ep%d%s "
				"h_addr%02x h_port%02x bytes %d\n",
			is_out ? "-->" : "<--",
			epnum, urb, urb->dev->speed,
			qh->addr_reg, qh->epnum, is_out ? "out" : "in",
			qh->h_addr_reg, qh->h_port_reg,
			len);

	musb_ep_select(mbase, epnum);

	if (is_out && !len) {
		use_dma = 0;
		csr = musb_readw(epio, MUSB_TXCSR);
		csr &= ~MUSB_TXCSR_DMAENAB;
		musb_writew(epio, MUSB_TXCSR, csr);
		hw_ep->tx_channel = NULL;
	}

	/* candidate for DMA? */
	dma_controller = musb->dma_controller;
	if (use_dma && is_dma_capable() && epnum && dma_controller) {
		dma_channel = is_out ? hw_ep->tx_channel : hw_ep->rx_channel;
		if (!dma_channel) {
			dma_channel = dma_controller->channel_alloc(
					dma_controller, hw_ep, is_out);
			if (is_out)
				hw_ep->tx_channel = dma_channel;
			else
				hw_ep->rx_channel = dma_channel;
		}
	} else
		dma_channel = NULL;

	/* make sure we clear DMAEnab, autoSet bits from previous run */

	/* OUT/transmit/EP0 or IN/receive? */
	if (is_out) {
		u16	csr;
		u16	int_txe;
		u16	load_count;

		csr = musb_readw(epio, MUSB_TXCSR);

		/* disable interrupt in case we flush */
		int_txe = musb->intrtxe;
		musb_writew(mbase, MUSB_INTRTXE, int_txe & ~(1 << epnum));

		/* general endpoint setup */
		if (epnum) {
			/* flush all old state, set default */
			/*
			 * We could be flushing valid
			 * packets in double buffering
			 * case
			 */
			if (!hw_ep->tx_double_buffered)
				musb_h_tx_flush_fifo(hw_ep);

			/*
			 * We must not clear the DMAMODE bit before or in
			 * the same cycle with the DMAENAB bit, so we clear
			 * the latter first...
			 */
			csr &= ~(MUSB_TXCSR_H_NAKTIMEOUT
					| MUSB_TXCSR_AUTOSET
					| MUSB_TXCSR_DMAENAB
					| MUSB_TXCSR_FRCDATATOG
					| MUSB_TXCSR_H_RXSTALL
					| MUSB_TXCSR_H_ERROR
					| MUSB_TXCSR_TXPKTRDY
					);
			csr |= MUSB_TXCSR_MODE;

			if (!hw_ep->tx_double_buffered) {
				if (usb_gettoggle(urb->dev, qh->epnum, 1))
					csr |= MUSB_TXCSR_H_WR_DATATOGGLE
						| MUSB_TXCSR_H_DATATOGGLE;
				else
					csr |= MUSB_TXCSR_CLRDATATOG;
			}

			musb_writew(epio, MUSB_TXCSR, csr);
			/* REVISIT may need to clear FLUSHFIFO ... */
			csr &= ~MUSB_TXCSR_DMAMODE;
			musb_writew(epio, MUSB_TXCSR, csr);
			csr = musb_readw(epio, MUSB_TXCSR);
		} else {
			/* endpoint 0: just flush */
			musb_h_ep0_flush_fifo(hw_ep);
		}

		/* target addr and (for multipoint) hub addr/port */
		if (musb->is_multipoint) {
			musb_write_txfunaddr(musb, epnum, qh->addr_reg);
			musb_write_txhubaddr(musb, epnum, qh->h_addr_reg);
			musb_write_txhubport(musb, epnum, qh->h_port_reg);
/* FIXME if !epnum, do the same for RX ... */
		} else
			musb_writeb(mbase, MUSB_FADDR, qh->addr_reg);

		/* protocol/endpoint/interval/NAKlimit */
		if (epnum) {
			musb_writeb(epio, MUSB_TXTYPE, qh->type_reg);
			if (musb->double_buffer_not_ok) {
				musb_writew(epio, MUSB_TXMAXP,
						hw_ep->max_packet_sz_tx);
			} else if (can_bulk_split(musb, qh->type)) {
				qh->hb_mult = hw_ep->max_packet_sz_tx
						/ packet_sz;
				musb_writew(epio, MUSB_TXMAXP, packet_sz
					| ((qh->hb_mult) - 1) << 11);
			} else {
				musb_writew(epio, MUSB_TXMAXP,
						qh->maxpacket |
						((qh->hb_mult - 1) << 11));
			}
			musb_writeb(epio, MUSB_TXINTERVAL, qh->intv_reg);
		} else {
			musb_writeb(epio, MUSB_NAKLIMIT0, qh->intv_reg);
			if (musb->is_multipoint)
				musb_writeb(epio, MUSB_TYPE0,
						qh->type_reg);
		}

		if (can_bulk_split(musb, qh->type))
			load_count = min((u32) hw_ep->max_packet_sz_tx,
						len);
		else
			load_count = min((u32) packet_sz, len);

		if (dma_channel && musb_tx_dma_program(dma_controller,
					hw_ep, qh, urb, offset, len))
			load_count = 0;

		if (load_count) {
			/* PIO to load FIFO */
			qh->segsize = load_count;
			if (!buf) {
				sg_miter_start(&qh->sg_miter, urb->sg, 1,
						SG_MITER_ATOMIC
						| SG_MITER_FROM_SG);
				if (!sg_miter_next(&qh->sg_miter)) {
					dev_err(musb->controller,
							"error: sg"
							"list empty\n");
					sg_miter_stop(&qh->sg_miter);
					goto finish;
				}
				buf = qh->sg_miter.addr + urb->sg->offset +
					urb->actual_length;
				load_count = min_t(u32, load_count,
						qh->sg_miter.length);
				musb_write_fifo(hw_ep, load_count, buf);
				qh->sg_miter.consumed = load_count;
				sg_miter_stop(&qh->sg_miter);
			} else
				musb_write_fifo(hw_ep, load_count, buf);
		}
finish:
		/* re-enable interrupt */
		musb_writew(mbase, MUSB_INTRTXE, int_txe);

	/* IN/receive */
	} else {
		u16	csr;

		if (hw_ep->rx_reinit) {
			musb_rx_reinit(musb, qh, epnum);

			/* init new state: toggle and NYET, maybe DMA later */
			if (usb_gettoggle(urb->dev, qh->epnum, 0))
				csr = MUSB_RXCSR_H_WR_DATATOGGLE
					| MUSB_RXCSR_H_DATATOGGLE;
			else
				csr = 0;
			if (qh->type == USB_ENDPOINT_XFER_INT)
				csr |= MUSB_RXCSR_DISNYET;

		} else {
			csr = musb_readw(hw_ep->regs, MUSB_RXCSR);

			if (csr & (MUSB_RXCSR_RXPKTRDY
					| MUSB_RXCSR_DMAENAB
					| MUSB_RXCSR_H_REQPKT))
				ERR("broken !rx_reinit, ep%d csr %04x\n",
						hw_ep->epnum, csr);

			/* scrub any stale state, leaving toggle alone */
			csr &= MUSB_RXCSR_DISNYET;
		}

		/* kick things off */

		if ((is_cppi_enabled(musb) || tusb_dma_omap(musb)) && dma_channel) {
			/* Candidate for DMA */
			dma_channel->actual_len = 0L;
			qh->segsize = len;

			/* AUTOREQ is in a DMA register */
			musb_writew(hw_ep->regs, MUSB_RXCSR, csr);
			csr = musb_readw(hw_ep->regs, MUSB_RXCSR);

			/*
			 * Unless caller treats short RX transfers as
			 * errors, we dare not queue multiple transfers.
			 */
			dma_ok = dma_controller->channel_program(dma_channel,
					packet_sz, !(urb->transfer_flags &
						     URB_SHORT_NOT_OK),
					urb->transfer_dma + offset,
					qh->segsize);
			if (!dma_ok) {
				dma_controller->channel_release(dma_channel);
				hw_ep->rx_channel = dma_channel = NULL;
			} else
				csr |= MUSB_RXCSR_DMAENAB;
		}

		csr |= MUSB_RXCSR_H_REQPKT;
		dev_dbg(musb->controller, "RXCSR%d := %04x\n", epnum, csr);
		musb_writew(hw_ep->regs, MUSB_RXCSR, csr);
		csr = musb_readw(hw_ep->regs, MUSB_RXCSR);
	}
}

/* Schedule next QH from musb->in_bulk/out_bulk and move the current qh to
 * the end; avoids starvation for other endpoints.
 */
static void musb_bulk_nak_timeout(struct musb *musb, struct musb_hw_ep *ep,
	int is_in)
{
	struct dma_channel	*dma;
	struct urb		*urb;
	void __iomem		*mbase = musb->mregs;
	void __iomem		*epio = ep->regs;
	struct musb_qh		*cur_qh, *next_qh;
	u16			rx_csr, tx_csr;

	musb_ep_select(mbase, ep->epnum);
	if (is_in) {
		dma = is_dma_capable() ? ep->rx_channel : NULL;

		/*
		 * Need to stop the transaction by clearing REQPKT first
		 * then the NAK Timeout bit ref MUSBMHDRC USB 2.0 HIGH-SPEED
		 * DUAL-ROLE CONTROLLER Programmer's Guide, section 9.2.2
		 */
		rx_csr = musb_readw(epio, MUSB_RXCSR);
		rx_csr |= MUSB_RXCSR_H_WZC_BITS;
		rx_csr &= ~MUSB_RXCSR_H_REQPKT;
		musb_writew(epio, MUSB_RXCSR, rx_csr);
		rx_csr &= ~MUSB_RXCSR_DATAERROR;
		musb_writew(epio, MUSB_RXCSR, rx_csr);

		cur_qh = first_qh(&musb->in_bulk);
	} else {
		dma = is_dma_capable() ? ep->tx_channel : NULL;

		/* clear nak timeout bit */
		tx_csr = musb_readw(epio, MUSB_TXCSR);
		tx_csr |= MUSB_TXCSR_H_WZC_BITS;
		tx_csr &= ~MUSB_TXCSR_H_NAKTIMEOUT;
		musb_writew(epio, MUSB_TXCSR, tx_csr);

		cur_qh = first_qh(&musb->out_bulk);
	}
	if (cur_qh) {
		urb = next_urb(cur_qh);
		if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
			dma->status = MUSB_DMA_STATUS_CORE_ABORT;
			musb->dma_controller->channel_abort(dma);
			urb->actual_length += dma->actual_len;
			dma->actual_len = 0L;
		}
		musb_save_toggle(cur_qh, is_in, urb);

		if (is_in) {
			/* move cur_qh to end of queue */
			list_move_tail(&cur_qh->ring, &musb->in_bulk);

			/* get the next qh from musb->in_bulk */
			next_qh = first_qh(&musb->in_bulk);

			/* set rx_reinit and schedule the next qh */
			ep->rx_reinit = 1;
		} else {
			/* move cur_qh to end of queue */
			list_move_tail(&cur_qh->ring, &musb->out_bulk);

			/* get the next qh from musb->out_bulk */
			next_qh = first_qh(&musb->out_bulk);

			/* set tx_reinit and schedule the next qh */
			ep->tx_reinit = 1;
		}
		musb_start_urb(musb, is_in, next_qh);
	}
}

/*
 * Service the default endpoint (ep0) as host.
 * Return true until it's time to start the status stage.
 */
static bool musb_h_ep0_continue(struct musb *musb, u16 len, struct urb *urb)
{
	bool			 more = false;
	u8			*fifo_dest = NULL;
	u16			fifo_count = 0;
	struct musb_hw_ep	*hw_ep = musb->control_ep;
	struct musb_qh		*qh = hw_ep->in_qh;
	struct usb_ctrlrequest	*request;

	switch (musb->ep0_stage) {
	case MUSB_EP0_IN:
		fifo_dest = urb->transfer_buffer + urb->actual_length;
		fifo_count = min_t(size_t, len, urb->transfer_buffer_length -
				   urb->actual_length);
		if (fifo_count < len)
			urb->status = -EOVERFLOW;

		musb_read_fifo(hw_ep, fifo_count, fifo_dest);

		urb->actual_length += fifo_count;
		if (len < qh->maxpacket) {
			/* always terminate on short read; it's
			 * rarely reported as an error.
			 */
		} else if (urb->actual_length <
				urb->transfer_buffer_length)
			more = true;
		break;
	case MUSB_EP0_START:
		request = (struct usb_ctrlrequest *) urb->setup_packet;

		if (!request->wLength) {
			dev_dbg(musb->controller, "start no-DATA\n");
			break;
		} else if (request->bRequestType & USB_DIR_IN) {
			dev_dbg(musb->controller, "start IN-DATA\n");
			musb->ep0_stage = MUSB_EP0_IN;
			more = true;
			break;
		} else {
			dev_dbg(musb->controller, "start OUT-DATA\n");
			musb->ep0_stage = MUSB_EP0_OUT;
			more = true;
		}
		/* FALLTHROUGH */
	case MUSB_EP0_OUT:
		fifo_count = min_t(size_t, qh->maxpacket,
				   urb->transfer_buffer_length -
				   urb->actual_length);
		if (fifo_count) {
			fifo_dest = (u8 *) (urb->transfer_buffer
					+ urb->actual_length);
			dev_dbg(musb->controller, "Sending %d byte%s to ep0 fifo %p\n",
					fifo_count,
					(fifo_count == 1) ? "" : "s",
					fifo_dest);
			musb_write_fifo(hw_ep, fifo_count, fifo_dest);

			urb->actual_length += fifo_count;
			more = true;
		}
		break;
	default:
		ERR("bogus ep0 stage %d\n", musb->ep0_stage);
		break;
	}

	return more;
}

/*
 * Handle default endpoint interrupt as host. Only called in IRQ time
 * from musb_interrupt().
 *
 * called with controller irqlocked
 */
irqreturn_t musb_h_ep0_irq(struct musb *musb)
{
	struct urb		*urb;
	u16			csr, len;
	int			status = 0;
	void __iomem		*mbase = musb->mregs;
	struct musb_hw_ep	*hw_ep = musb->control_ep;
	void __iomem		*epio = hw_ep->regs;
	struct musb_qh		*qh = hw_ep->in_qh;
	bool			complete = false;
	irqreturn_t		retval = IRQ_NONE;

	/* ep0 only has one queue, "in" */
	urb = next_urb(qh);

	musb_ep_select(mbase, 0);
	csr = musb_readw(epio, MUSB_CSR0);
	len = (csr & MUSB_CSR0_RXPKTRDY)
			? musb_readb(epio, MUSB_COUNT0)
			: 0;

	dev_dbg(musb->controller, "<== csr0 %04x, qh %p, count %d, urb %p, stage %d\n",
		csr, qh, len, urb, musb->ep0_stage);

	/* if we just did status stage, we are done */
	if (MUSB_EP0_STATUS == musb->ep0_stage) {
		retval = IRQ_HANDLED;
		complete = true;
	}

	/* prepare status */
	if (csr & MUSB_CSR0_H_RXSTALL) {
		dev_dbg(musb->controller, "STALLING ENDPOINT\n");
		status = -EPIPE;

	} else if (csr & MUSB_CSR0_H_ERROR) {
		dev_dbg(musb->controller, "no response, csr0 %04x\n", csr);
		status = -EPROTO;

	} else if (csr & MUSB_CSR0_H_NAKTIMEOUT) {
		dev_dbg(musb->controller, "control NAK timeout\n");

		/* NOTE:  this code path would be a good place to PAUSE a
		 * control transfer, if another one is queued, so that
		 * ep0 is more likely to stay busy.  That's already done
		 * for bulk RX transfers.
		 *
		 * if (qh->ring.next != &musb->control), then
		 * we have a candidate... NAKing is *NOT* an error
		 */
		musb_writew(epio, MUSB_CSR0, 0);
		retval = IRQ_HANDLED;
	}

	if (status) {
		dev_dbg(musb->controller, "aborting\n");
		retval = IRQ_HANDLED;
		if (urb)
			urb->status = status;
		complete = true;

		/* use the proper sequence to abort the transfer */
		if (csr & MUSB_CSR0_H_REQPKT) {
			csr &= ~MUSB_CSR0_H_REQPKT;
			musb_writew(epio, MUSB_CSR0, csr);
			csr &= ~MUSB_CSR0_H_NAKTIMEOUT;
			musb_writew(epio, MUSB_CSR0, csr);
		} else {
			musb_h_ep0_flush_fifo(hw_ep);
		}

		musb_writeb(epio, MUSB_NAKLIMIT0, 0);

		/* clear it */
		musb_writew(epio, MUSB_CSR0, 0);
	}

	if (unlikely(!urb)) {
		/* stop endpoint since we have no place for its data, this
		 * SHOULD NEVER HAPPEN! */
		ERR("no URB for end 0\n");

		musb_h_ep0_flush_fifo(hw_ep);
		goto done;
	}

	if (!complete) {
		/* call common logic and prepare response */
		if (musb_h_ep0_continue(musb, len, urb)) {
			/* more packets required */
			csr = (MUSB_EP0_IN == musb->ep0_stage)
				?  MUSB_CSR0_H_REQPKT : MUSB_CSR0_TXPKTRDY;
		} else {
			/* data transfer complete; perform status phase */
			if (usb_pipeout(urb->pipe)
					|| !urb->transfer_buffer_length)
				csr = MUSB_CSR0_H_STATUSPKT
					| MUSB_CSR0_H_REQPKT;
			else
				csr = MUSB_CSR0_H_STATUSPKT
					| MUSB_CSR0_TXPKTRDY;

			/* disable ping token in status phase */
			csr |= MUSB_CSR0_H_DIS_PING;

			/* flag status stage */
			musb->ep0_stage = MUSB_EP0_STATUS;

			dev_dbg(musb->controller, "ep0 STATUS, csr %04x\n", csr);

		}
		musb_writew(epio, MUSB_CSR0, csr);
		retval = IRQ_HANDLED;
	} else
		musb->ep0_stage = MUSB_EP0_IDLE;

	/* call completion handler if done */
	if (complete)
		musb_advance_schedule(musb, urb, hw_ep, 1);
done:
	return retval;
}


#ifdef CONFIG_USB_INVENTRA_DMA

/* Host side TX (OUT) using Mentor DMA works as follows:
	submit_urb ->
		- if queue was empty, Program Endpoint
		- ... which starts DMA to fifo in mode 1 or 0

	DMA Isr (transfer complete) -> TxAvail()
		- Stop DMA (~DmaEnab)	(<--- Alert ... currently happens
					only in musb_cleanup_urb)
		- TxPktRdy has to be set in mode 0 or for
			short packets in mode 1.
*/

#endif

/* Service a Tx-Available or dma completion irq for the endpoint */
void musb_host_tx(struct musb *musb, u8 epnum)
{
	int			pipe;
	bool			done = false;
	u16			tx_csr;
	size_t			length = 0;
	size_t			offset = 0;
	struct musb_hw_ep	*hw_ep = musb->endpoints + epnum;
	void __iomem		*epio = hw_ep->regs;
	struct musb_qh		*qh = hw_ep->out_qh;
	struct urb		*urb = next_urb(qh);
	u32			status = 0;
	void __iomem		*mbase = musb->mregs;
	struct dma_channel	*dma;
	bool			transfer_pending = false;

	musb_ep_select(mbase, epnum);
	tx_csr = musb_readw(epio, MUSB_TXCSR);

	/* with CPPI, DMA sometimes triggers "extra" irqs */
	if (!urb) {
		dev_dbg(musb->controller, "extra TX%d ready, csr %04x\n", epnum, tx_csr);
		return;
	}

	pipe = urb->pipe;
	dma = is_dma_capable() ? hw_ep->tx_channel : NULL;
	dev_dbg(musb->controller, "OUT/TX%d end, csr %04x%s\n", epnum, tx_csr,
			dma ? ", dma" : "");

	/* check for errors */
	if (tx_csr & MUSB_TXCSR_H_RXSTALL) {
		/* dma was disabled, fifo flushed */
		dev_dbg(musb->controller, "TX end %d stall\n", epnum);

		/* stall; record URB status */
		status = -EPIPE;

	} else if (tx_csr & MUSB_TXCSR_H_ERROR) {
		/* (NON-ISO) dma was disabled, fifo flushed */
		dev_dbg(musb->controller, "TX 3strikes on ep=%d\n", epnum);

		status = -ETIMEDOUT;

	} else if (tx_csr & MUSB_TXCSR_H_NAKTIMEOUT) {
		if (USB_ENDPOINT_XFER_BULK == qh->type && qh->mux == 1
				&& !list_is_singular(&musb->out_bulk)) {
			dev_dbg(musb->controller,
				"NAK timeout on TX%d ep\n", epnum);
			musb_bulk_nak_timeout(musb, hw_ep, 0);
		} else {
			dev_dbg(musb->controller,
				"TX end=%d device not responding\n", epnum);
			/* NOTE:  this code path would be a good place to PAUSE a
			 * transfer, if there's some other (nonperiodic) tx urb
			 * that could use this fifo.  (dma complicates it...)
			 * That's already done for bulk RX transfers.
			 *
			 * if (bulk && qh->ring.next != &musb->out_bulk), then
			 * we have a candidate... NAKing is *NOT* an error
			 */
			musb_ep_select(mbase, epnum);
			musb_writew(epio, MUSB_TXCSR,
					MUSB_TXCSR_H_WZC_BITS
					| MUSB_TXCSR_TXPKTRDY);
		}
			return;
	}

done:
	if (status) {
		if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
			dma->status = MUSB_DMA_STATUS_CORE_ABORT;
			musb->dma_controller->channel_abort(dma);
		}

		/* do the proper sequence to abort the transfer in the
		 * usb core; the dma engine should already be stopped.
		 */
		musb_h_tx_flush_fifo(hw_ep);
		tx_csr &= ~(MUSB_TXCSR_AUTOSET
				| MUSB_TXCSR_DMAENAB
				| MUSB_TXCSR_H_ERROR
				| MUSB_TXCSR_H_RXSTALL
				| MUSB_TXCSR_H_NAKTIMEOUT
				);

		musb_ep_select(mbase, epnum);
		musb_writew(epio, MUSB_TXCSR, tx_csr);
		/* REVISIT may need to clear FLUSHFIFO ... */
		musb_writew(epio, MUSB_TXCSR, tx_csr);
		musb_writeb(epio, MUSB_TXINTERVAL, 0);

		done = true;
	}

	/* second cppi case */
	if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
		dev_dbg(musb->controller, "extra TX%d ready, csr %04x\n", epnum, tx_csr);
		return;
	}

	if (is_dma_capable() && dma && !status) {
		/*
		 * DMA has completed.  But if we're using DMA mode 1 (multi
		 * packet DMA), we need a terminal TXPKTRDY interrupt before
		 * we can consider this transfer completed, lest we trash
		 * its last packet when writing the next URB's data.  So we
		 * switch back to mode 0 to get that interrupt; we'll come
		 * back here once it happens.
		 */
		if (tx_csr & MUSB_TXCSR_DMAMODE) {
			/*
			 * We shouldn't clear DMAMODE with DMAENAB set; so
			 * clear them in a safe order.  That should be OK
			 * once TXPKTRDY has been set (and I've never seen
			 * it being 0 at this moment -- DMA interrupt latency
			 * is significant) but if it hasn't been then we have
			 * no choice but to stop being polite and ignore the
			 * programmer's guide... :-)
			 *
			 * Note that we must write TXCSR with TXPKTRDY cleared
			 * in order not to re-trigger the packet send (this bit
			 * can't be cleared by CPU), and there's another caveat:
			 * TXPKTRDY may be set shortly and then cleared in the
			 * double-buffered FIFO mode, so we do an extra TXCSR
			 * read for debouncing...
			 */
			tx_csr &= musb_readw(epio, MUSB_TXCSR);
			if (tx_csr & MUSB_TXCSR_TXPKTRDY) {
				tx_csr &= ~(MUSB_TXCSR_DMAENAB |
					    MUSB_TXCSR_TXPKTRDY);
				musb_writew(epio, MUSB_TXCSR,
					    tx_csr | MUSB_TXCSR_H_WZC_BITS);
			}
			tx_csr &= ~(MUSB_TXCSR_DMAMODE |
				    MUSB_TXCSR_TXPKTRDY);
			musb_writew(epio, MUSB_TXCSR,
				    tx_csr | MUSB_TXCSR_H_WZC_BITS);

			/*
			 * There is no guarantee that we'll get an interrupt
			 * after clearing DMAMODE as we might have done this
			 * too late (after TXPKTRDY was cleared by controller).
			 * Re-read TXCSR as we have spoiled its previous value.
			 */
			tx_csr = musb_readw(epio, MUSB_TXCSR);
		}

		/*
		 * We may get here from a DMA completion or TXPKTRDY interrupt.
		 * In any case, we must check the FIFO status here and bail out
		 * only if the FIFO still has data -- that should prevent the
		 * "missed" TXPKTRDY interrupts and deal with double-buffered
		 * FIFO mode too...
		 */
		if (tx_csr & (MUSB_TXCSR_FIFONOTEMPTY | MUSB_TXCSR_TXPKTRDY)) {
			dev_dbg(musb->controller, "DMA complete but packet still in FIFO, "
			    "CSR %04x\n", tx_csr);
			return;
		}
	}

	if (!status || dma || usb_pipeisoc(pipe)) {
		if (dma)
			length = dma->actual_len;
		else
			length = qh->segsize;
		qh->offset += length;

		if (usb_pipeisoc(pipe)) {
			struct usb_iso_packet_descriptor	*d;

			d = urb->iso_frame_desc + qh->iso_idx;
			d->actual_length = length;
			d->status = status;
			if (++qh->iso_idx >= urb->number_of_packets) {
				done = true;
			} else {
				d++;
				offset = d->offset;
				length = d->length;
			}
		} else if (dma && urb->transfer_buffer_length == qh->offset) {
			done = true;
		} else {
			/* see if we need to send more data, or ZLP */
			if (qh->segsize < qh->maxpacket)
				done = true;
			else if (qh->offset == urb->transfer_buffer_length
					&& !(urb->transfer_flags
						& URB_ZERO_PACKET))
				done = true;
			if (!done) {
				offset = qh->offset;
				length = urb->transfer_buffer_length - offset;
				transfer_pending = true;
			}
		}
	}

	/* urb->status != -EINPROGRESS means request has been faulted,
	 * so we must abort this transfer after cleanup
	 */
	if (urb->status != -EINPROGRESS) {
		done = true;
		if (status == 0)
			status = urb->status;
	}

	if (done) {
		/* set status */
		urb->status = status;
		urb->actual_length = qh->offset;
		musb_advance_schedule(musb, urb, hw_ep, USB_DIR_OUT);
		return;
	} else if ((usb_pipeisoc(pipe) || transfer_pending) && dma) {
		if (musb_tx_dma_program(musb->dma_controller, hw_ep, qh, urb,
				offset, length)) {
			if (is_cppi_enabled(musb) || tusb_dma_omap(musb))
				musb_h_tx_dma_start(hw_ep);
			return;
		}
	} else	if (tx_csr & MUSB_TXCSR_DMAENAB) {
		dev_dbg(musb->controller, "not complete, but DMA enabled?\n");
		return;
	}

	/*
	 * PIO: start next packet in this URB.
	 *
	 * REVISIT: some docs say that when hw_ep->tx_double_buffered,
	 * (and presumably, FIFO is not half-full) we should write *two*
	 * packets before updating TXCSR; other docs disagree...
	 */
	if (length > qh->maxpacket)
		length = qh->maxpacket;
	/* Unmap the buffer so that CPU can use it */
	usb_hcd_unmap_urb_for_dma(musb->hcd, urb);

	/*
	 * We need to map sg if the transfer_buffer is
	 * NULL.
	 */
	if (!urb->transfer_buffer)
		qh->use_sg = true;

	if (qh->use_sg) {
		/* sg_miter_start is already done in musb_ep_program */
		if (!sg_miter_next(&qh->sg_miter)) {
			dev_err(musb->controller, "error: sg list empty\n");
			sg_miter_stop(&qh->sg_miter);
			status = -EINVAL;
			goto done;
		}
		urb->transfer_buffer = qh->sg_miter.addr;
		length = min_t(u32, length, qh->sg_miter.length);
		musb_write_fifo(hw_ep, length, urb->transfer_buffer);
		qh->sg_miter.consumed = length;
		sg_miter_stop(&qh->sg_miter);
	} else {
		musb_write_fifo(hw_ep, length, urb->transfer_buffer + offset);
	}

	qh->segsize = length;

	if (qh->use_sg) {
		if (offset + length >= urb->transfer_buffer_length)
			qh->use_sg = false;
	}

	musb_ep_select(mbase, epnum);
	musb_writew(epio, MUSB_TXCSR,
			MUSB_TXCSR_H_WZC_BITS | MUSB_TXCSR_TXPKTRDY);
}

#ifdef CONFIG_USB_TI_CPPI41_DMA
/* Seems to set up ISO for cppi41 and not advance len. See commit c57c41d */
static int musb_rx_dma_iso_cppi41(struct dma_controller *dma,
				  struct musb_hw_ep *hw_ep,
				  struct musb_qh *qh,
				  struct urb *urb,
				  size_t len)
{
	struct dma_channel *channel = hw_ep->rx_channel;
	void __iomem *epio = hw_ep->regs;
	dma_addr_t *buf;
	u32 length, res;
	u16 val;

	buf = (void *)urb->iso_frame_desc[qh->iso_idx].offset +
		(u32)urb->transfer_dma;

	length = urb->iso_frame_desc[qh->iso_idx].length;

	val = musb_readw(epio, MUSB_RXCSR);
	val |= MUSB_RXCSR_DMAENAB;
	musb_writew(hw_ep->regs, MUSB_RXCSR, val);

	res = dma->channel_program(channel, qh->maxpacket, 0,
				   (u32)buf, length);

	return res;
}
#else
static inline int musb_rx_dma_iso_cppi41(struct dma_controller *dma,
					 struct musb_hw_ep *hw_ep,
					 struct musb_qh *qh,
					 struct urb *urb,
					 size_t len)
{
	return false;
}
#endif

#if defined(CONFIG_USB_INVENTRA_DMA) || defined(CONFIG_USB_UX500_DMA) || \
	defined(CONFIG_USB_TI_CPPI41_DMA)
/* Host side RX (IN) using Mentor DMA works as follows:
	submit_urb ->
		- if queue was empty, ProgramEndpoint
		- first IN token is sent out (by setting ReqPkt)
	LinuxIsr -> RxReady()
	/\	=> first packet is received
	|	- Set in mode 0 (DmaEnab, ~ReqPkt)
	|		-> DMA Isr (transfer complete) -> RxReady()
	|		    - Ack receive (~RxPktRdy), turn off DMA (~DmaEnab)
	|		    - if urb not complete, send next IN token (ReqPkt)
	|			   |		else complete urb.
	|			   |
	---------------------------
 *
 * Nuances of mode 1:
 *	For short packets, no ack (+RxPktRdy) is sent automatically
 *	(even if AutoClear is ON)
 *	For full packets, ack (~RxPktRdy) and next IN token (+ReqPkt) is sent
 *	automatically => major problem, as collecting the next packet becomes
 *	difficult. Hence mode 1 is not used.
 *
 * REVISIT
 *	All we care about at this driver level is that
 *       (a) all URBs terminate with REQPKT cleared and fifo(s) empty;
 *       (b) termination conditions are: short RX, or buffer full;
 *       (c) fault modes include
 *           - iff URB_SHORT_NOT_OK, short RX status is -EREMOTEIO.
 *             (and that endpoint's dma queue stops immediately)
 *           - overflow (full, PLUS more bytes in the terminal packet)
 *
 *	So for example, usb-storage sets URB_SHORT_NOT_OK, and would
 *	thus be a great candidate for using mode 1 ... for all but the
 *	last packet of one URB's transfer.
 */
static int musb_rx_dma_inventra_cppi41(struct dma_controller *dma,
				       struct musb_hw_ep *hw_ep,
				       struct musb_qh *qh,
				       struct urb *urb,
				       size_t len)
{
	struct dma_channel *channel = hw_ep->rx_channel;
	void __iomem *epio = hw_ep->regs;
	u16 val;
	int pipe;
	bool done;

	pipe = urb->pipe;

	if (usb_pipeisoc(pipe)) {
		struct usb_iso_packet_descriptor *d;

		d = urb->iso_frame_desc + qh->iso_idx;
		d->actual_length = len;

		/* even if there was an error, we did the dma
		 * for iso_frame_desc->length
		 */
		if (d->status != -EILSEQ && d->status != -EOVERFLOW)
			d->status = 0;

		if (++qh->iso_idx >= urb->number_of_packets) {
			done = true;
		} else {
			/* REVISIT: Why ignore return value here? */
			if (musb_dma_cppi41(hw_ep->musb))
				done = musb_rx_dma_iso_cppi41(dma, hw_ep, qh,
							      urb, len);
			done = false;
		}

	} else  {
		/* done if urb buffer is full or short packet is recd */
		done = (urb->actual_length + len >=
			urb->transfer_buffer_length
			|| channel->actual_len < qh->maxpacket
			|| channel->rx_packet_done);
	}

	/* send IN token for next packet, without AUTOREQ */
	if (!done) {
		val = musb_readw(epio, MUSB_RXCSR);
		val |= MUSB_RXCSR_H_REQPKT;
		musb_writew(epio, MUSB_RXCSR, MUSB_RXCSR_H_WZC_BITS | val);
	}

	return done;
}

/* Disadvantage of using mode 1:
 *	It's basically usable only for mass storage class; essentially all
 *	other protocols also terminate transfers on short packets.
 *
 * Details:
 *	An extra IN token is sent at the end of the transfer (due to AUTOREQ)
 *	If you try to use mode 1 for (transfer_buffer_length - 512), and try
 *	to use the extra IN token to grab the last packet using mode 0, then
 *	the problem is that you cannot be sure when the device will send the
 *	last packet and RxPktRdy set. Sometimes the packet is recd too soon
 *	such that it gets lost when RxCSR is re-set at the end of the mode 1
 *	transfer, while sometimes it is recd just a little late so that if you
 *	try to configure for mode 0 soon after the mode 1 transfer is
 *	completed, you will find rxcount 0. Okay, so you might think why not
 *	wait for an interrupt when the pkt is recd. Well, you won't get any!
 */
static int musb_rx_dma_in_inventra_cppi41(struct dma_controller *dma,
					  struct musb_hw_ep *hw_ep,
					  struct musb_qh *qh,
					  struct urb *urb,
					  size_t len,
					  u8 iso_err)
{
	struct musb *musb = hw_ep->musb;
	void __iomem *epio = hw_ep->regs;
	struct dma_channel *channel = hw_ep->rx_channel;
	u16 rx_count, val;
	int length, pipe, done;
	dma_addr_t buf;

	rx_count = musb_readw(epio, MUSB_RXCOUNT);
	pipe = urb->pipe;

	if (usb_pipeisoc(pipe)) {
		int d_status = 0;
		struct usb_iso_packet_descriptor *d;

		d = urb->iso_frame_desc + qh->iso_idx;

		if (iso_err) {
			d_status = -EILSEQ;
			urb->error_count++;
		}
		if (rx_count > d->length) {
			if (d_status == 0) {
				d_status = -EOVERFLOW;
				urb->error_count++;
			}
			dev_dbg(musb->controller, "** OVERFLOW %d into %d\n",
				rx_count, d->length);

			length = d->length;
		} else
			length = rx_count;
		d->status = d_status;
		buf = urb->transfer_dma + d->offset;
	} else {
		length = rx_count;
		buf = urb->transfer_dma + urb->actual_length;
	}

	channel->desired_mode = 0;
#ifdef USE_MODE1
	/* because of the issue below, mode 1 will
	 * only rarely behave with correct semantics.
	 */
	if ((urb->transfer_flags & URB_SHORT_NOT_OK)
	    && (urb->transfer_buffer_length - urb->actual_length)
	    > qh->maxpacket)
		channel->desired_mode = 1;
	if (rx_count < hw_ep->max_packet_sz_rx) {
		length = rx_count;
		channel->desired_mode = 0;
	} else {
		length = urb->transfer_buffer_length;
	}
#endif

	/* See comments above on disadvantages of using mode 1 */
	val = musb_readw(epio, MUSB_RXCSR);
	val &= ~MUSB_RXCSR_H_REQPKT;

	if (channel->desired_mode == 0)
		val &= ~MUSB_RXCSR_H_AUTOREQ;
	else
		val |= MUSB_RXCSR_H_AUTOREQ;
	val |= MUSB_RXCSR_DMAENAB;

	/* autoclear shouldn't be set in high bandwidth */
	if (qh->hb_mult == 1)
		val |= MUSB_RXCSR_AUTOCLEAR;

	musb_writew(epio, MUSB_RXCSR, MUSB_RXCSR_H_WZC_BITS | val);

	/* REVISIT if when actual_length != 0,
	 * transfer_buffer_length needs to be
	 * adjusted first...
	 */
	done = dma->channel_program(channel, qh->maxpacket,
				   channel->desired_mode,
				   buf, length);

	if (!done) {
		dma->channel_release(channel);
		hw_ep->rx_channel = NULL;
		channel = NULL;
		val = musb_readw(epio, MUSB_RXCSR);
		val &= ~(MUSB_RXCSR_DMAENAB
			 | MUSB_RXCSR_H_AUTOREQ
			 | MUSB_RXCSR_AUTOCLEAR);
		musb_writew(epio, MUSB_RXCSR, val);
	}

	return done;
}
#else
static inline int musb_rx_dma_inventra_cppi41(struct dma_controller *dma,
					      struct musb_hw_ep *hw_ep,
					      struct musb_qh *qh,
					      struct urb *urb,
					      size_t len)
{
	return false;
}

static inline int musb_rx_dma_in_inventra_cppi41(struct dma_controller *dma,
						 struct musb_hw_ep *hw_ep,
						 struct musb_qh *qh,
						 struct urb *urb,
						 size_t len,
						 u8 iso_err)
{
	return false;
}
#endif

/*
 * Service an RX interrupt for the given IN endpoint; docs cover bulk, iso,
 * and high-bandwidth IN transfer cases.
 */
void musb_host_rx(struct musb *musb, u8 epnum)
{
	struct urb		*urb;
	struct musb_hw_ep	*hw_ep = musb->endpoints + epnum;
	struct dma_controller	*c = musb->dma_controller;
	void __iomem		*epio = hw_ep->regs;
	struct musb_qh		*qh = hw_ep->in_qh;
	size_t			xfer_len;
	void __iomem		*mbase = musb->mregs;
	int			pipe;
	u16			rx_csr, val;
	bool			iso_err = false;
	bool			done = false;
	u32			status;
	struct dma_channel	*dma;
	unsigned int sg_flags = SG_MITER_ATOMIC | SG_MITER_TO_SG;

	musb_ep_select(mbase, epnum);

	urb = next_urb(qh);
	dma = is_dma_capable() ? hw_ep->rx_channel : NULL;
	status = 0;
	xfer_len = 0;

	rx_csr = musb_readw(epio, MUSB_RXCSR);
	val = rx_csr;

	if (unlikely(!urb)) {
		/* REVISIT -- THIS SHOULD NEVER HAPPEN ... but, at least
		 * usbtest #11 (unlinks) triggers it regularly, sometimes
		 * with fifo full.  (Only with DMA??)
		 */
		dev_dbg(musb->controller, "BOGUS RX%d ready, csr %04x, count %d\n", epnum, val,
			musb_readw(epio, MUSB_RXCOUNT));
		musb_h_flush_rxfifo(hw_ep, MUSB_RXCSR_CLRDATATOG);
		return;
	}

	pipe = urb->pipe;

	dev_dbg(musb->controller, "<== hw %d rxcsr %04x, urb actual %d (+dma %zu)\n",
		epnum, rx_csr, urb->actual_length,
		dma ? dma->actual_len : 0);

	/* check for errors, concurrent stall & unlink is not really
	 * handled yet! */
	if (rx_csr & MUSB_RXCSR_H_RXSTALL) {
		dev_dbg(musb->controller, "RX end %d STALL\n", epnum);

		/* stall; record URB status */
		status = -EPIPE;

	} else if (rx_csr & MUSB_RXCSR_H_ERROR) {
		dev_dbg(musb->controller, "end %d RX proto error\n", epnum);

		status = -EPROTO;
		musb_writeb(epio, MUSB_RXINTERVAL, 0);

		rx_csr &= ~MUSB_RXCSR_H_ERROR;
		musb_writew(epio, MUSB_RXCSR, rx_csr);

	} else if (rx_csr & MUSB_RXCSR_DATAERROR) {

		if (USB_ENDPOINT_XFER_ISOC != qh->type) {
			dev_dbg(musb->controller, "RX end %d NAK timeout\n", epnum);

			/* NOTE: NAKing is *NOT* an error, so we want to
			 * continue.  Except ... if there's a request for
			 * another QH, use that instead of starving it.
			 *
			 * Devices like Ethernet and serial adapters keep
			 * reads posted at all times, which will starve
			 * other devices without this logic.
			 */
			if (usb_pipebulk(urb->pipe)
					&& qh->mux == 1
					&& !list_is_singular(&musb->in_bulk)) {
				musb_bulk_nak_timeout(musb, hw_ep, 1);
				return;
			}
			musb_ep_select(mbase, epnum);
			rx_csr |= MUSB_RXCSR_H_WZC_BITS;
			rx_csr &= ~MUSB_RXCSR_DATAERROR;
			musb_writew(epio, MUSB_RXCSR, rx_csr);

			goto finish;
		} else {
			dev_dbg(musb->controller, "RX end %d ISO data error\n", epnum);
			/* packet error reported later */
			iso_err = true;
		}
	} else if (rx_csr & MUSB_RXCSR_INCOMPRX) {
		dev_dbg(musb->controller, "end %d high bandwidth incomplete ISO packet RX\n",
				epnum);
		status = -EPROTO;
	}

	/* faults abort the transfer */
	if (status) {
		/* clean up dma and collect transfer count */
		if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
			dma->status = MUSB_DMA_STATUS_CORE_ABORT;
			musb->dma_controller->channel_abort(dma);
			xfer_len = dma->actual_len;
		}
		musb_h_flush_rxfifo(hw_ep, MUSB_RXCSR_CLRDATATOG);
		musb_writeb(epio, MUSB_RXINTERVAL, 0);
		done = true;
		goto finish;
	}

	if (unlikely(dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY)) {
		/* SHOULD NEVER HAPPEN ... but at least DaVinci has done it */
		ERR("RX%d dma busy, csr %04x\n", epnum, rx_csr);
		goto finish;
	}

	/* thorough shutdown for now ... given more precise fault handling
	 * and better queueing support, we might keep a DMA pipeline going
	 * while processing this irq for earlier completions.
	 */

	/* FIXME this is _way_ too much in-line logic for Mentor DMA */
	if (!musb_dma_inventra(musb) && !musb_dma_ux500(musb) &&
	    (rx_csr & MUSB_RXCSR_H_REQPKT)) {
		/* REVISIT this happened for a while on some short reads...
		 * the cleanup still needs investigation... looks bad...
		 * and also duplicates dma cleanup code above ... plus,
		 * shouldn't this be the "half full" double buffer case?
		 */
		if (dma_channel_status(dma) == MUSB_DMA_STATUS_BUSY) {
			dma->status = MUSB_DMA_STATUS_CORE_ABORT;
			musb->dma_controller->channel_abort(dma);
			xfer_len = dma->actual_len;
			done = true;
		}

		dev_dbg(musb->controller, "RXCSR%d %04x, reqpkt, len %zu%s\n", epnum, rx_csr,
				xfer_len, dma ? ", dma" : "");
		rx_csr &= ~MUSB_RXCSR_H_REQPKT;

		musb_ep_select(mbase, epnum);
		musb_writew(epio, MUSB_RXCSR,
				MUSB_RXCSR_H_WZC_BITS | rx_csr);
	}

	if (dma && (rx_csr & MUSB_RXCSR_DMAENAB)) {
		xfer_len = dma->actual_len;

		val &= ~(MUSB_RXCSR_DMAENAB
			| MUSB_RXCSR_H_AUTOREQ
			| MUSB_RXCSR_AUTOCLEAR
			| MUSB_RXCSR_RXPKTRDY);
		musb_writew(hw_ep->regs, MUSB_RXCSR, val);

		if (musb_dma_inventra(musb) || musb_dma_ux500(musb) ||
		    musb_dma_cppi41(musb)) {
			    done = musb_rx_dma_inventra_cppi41(c, hw_ep, qh, urb, xfer_len);
			    dev_dbg(hw_ep->musb->controller,
				    "ep %d dma %s, rxcsr %04x, rxcount %d\n",
				    epnum, done ? "off" : "reset",
				    musb_readw(epio, MUSB_RXCSR),
				    musb_readw(epio, MUSB_RXCOUNT));
		} else {
			done = true;
		}

	} else if (urb->status == -EINPROGRESS) {
		/* if no errors, be sure a packet is ready for unloading */
		if (unlikely(!(rx_csr & MUSB_RXCSR_RXPKTRDY))) {
			status = -EPROTO;
			ERR("Rx interrupt with no errors or packet!\n");

			/* FIXME this is another "SHOULD NEVER HAPPEN" */

/* SCRUB (RX) */
			/* do the proper sequence to abort the transfer */
			musb_ep_select(mbase, epnum);
			val &= ~MUSB_RXCSR_H_REQPKT;
			musb_writew(epio, MUSB_RXCSR, val);
			goto finish;
		}

		/* we are expecting IN packets */
		if ((musb_dma_inventra(musb) || musb_dma_ux500(musb) ||
		    musb_dma_cppi41(musb)) && dma) {
			dev_dbg(hw_ep->musb->controller,
				"RX%d count %d, buffer 0x%llx len %d/%d\n",
				epnum, musb_readw(epio, MUSB_RXCOUNT),
				(unsigned long long) urb->transfer_dma
				+ urb->actual_length,
				qh->offset,
				urb->transfer_buffer_length);

			if (musb_rx_dma_in_inventra_cppi41(c, hw_ep, qh, urb,
							   xfer_len, iso_err))
				goto finish;
			else
				dev_err(musb->controller, "error: rx_dma failed\n");
		}

		if (!dma) {
			unsigned int received_len;

			/* Unmap the buffer so that CPU can use it */
			usb_hcd_unmap_urb_for_dma(musb->hcd, urb);

			/*
			 * We need to map sg if the transfer_buffer is
			 * NULL.
			 */
			if (!urb->transfer_buffer) {
				qh->use_sg = true;
				sg_miter_start(&qh->sg_miter, urb->sg, 1,
						sg_flags);
			}

			if (qh->use_sg) {
				if (!sg_miter_next(&qh->sg_miter)) {
					dev_err(musb->controller, "error: sg list empty\n");
					sg_miter_stop(&qh->sg_miter);
					status = -EINVAL;
					done = true;
					goto finish;
				}
				urb->transfer_buffer = qh->sg_miter.addr;
				received_len = urb->actual_length;
				qh->offset = 0x0;
				done = musb_host_packet_rx(musb, urb, epnum,
						iso_err);
				/* Calculate the number of bytes received */
				received_len = urb->actual_length -
					received_len;
				qh->sg_miter.consumed = received_len;
				sg_miter_stop(&qh->sg_miter);
			} else {
				done = musb_host_packet_rx(musb, urb,
						epnum, iso_err);
			}
			dev_dbg(musb->controller, "read %spacket\n", done ? "last " : "");
		}
	}

finish:
	urb->actual_length += xfer_len;
	qh->offset += xfer_len;
	if (done) {
		if (qh->use_sg)
			qh->use_sg = false;

		if (urb->status == -EINPROGRESS)
			urb->status = status;
		musb_advance_schedule(musb, urb, hw_ep, USB_DIR_IN);
	}
}

/* schedule nodes correspond to peripheral endpoints, like an OHCI QH.
 * the software schedule associates multiple such nodes with a given
 * host side hardware endpoint + direction; scheduling may activate
 * that hardware endpoint.
 */
static int musb_schedule(
	struct musb		*musb,
	struct musb_qh		*qh,
	int			is_in)
{
	int			idle = 0;
	int			best_diff;
	int			best_end, epnum;
	struct musb_hw_ep	*hw_ep = NULL;
	struct list_head	*head = NULL;
	u8			toggle;
	u8			txtype;
	struct urb		*urb = next_urb(qh);

	/* use fixed hardware for control and bulk */
	if (qh->type == USB_ENDPOINT_XFER_CONTROL) {
		head = &musb->control;
		hw_ep = musb->control_ep;
		goto success;
	}

	/* else, periodic transfers get muxed to other endpoints */

	/*
	 * We know this qh hasn't been scheduled, so all we need to do
	 * is choose which hardware endpoint to put it on ...
	 *
	 * REVISIT what we really want here is a regular schedule tree
	 * like e.g. OHCI uses.
	 */
	best_diff = 4096;
	best_end = -1;

	for (epnum = 1, hw_ep = musb->endpoints + 1;
			epnum < musb->nr_endpoints;
			epnum++, hw_ep++) {
		int	diff;

		if (musb_ep_get_qh(hw_ep, is_in) != NULL)
			continue;

		if (hw_ep == musb->bulk_ep)
			continue;

		if (is_in)
			diff = hw_ep->max_packet_sz_rx;
		else
			diff = hw_ep->max_packet_sz_tx;
		diff -= (qh->maxpacket * qh->hb_mult);

		if (diff >= 0 && best_diff > diff) {

			/*
			 * Mentor controller has a bug in that if we schedule
			 * a BULK Tx transfer on an endpoint that had earlier
			 * handled ISOC then the BULK transfer has to start on
			 * a zero toggle.  If the BULK transfer starts on a 1
			 * toggle then this transfer will fail as the mentor
			 * controller starts the Bulk transfer on a 0 toggle
			 * irrespective of the programming of the toggle bits
			 * in the TXCSR register.  Check for this condition
			 * while allocating the EP for a Tx Bulk transfer.  If
			 * so skip this EP.
			 */
			hw_ep = musb->endpoints + epnum;
			toggle = usb_gettoggle(urb->dev, qh->epnum, !is_in);
			txtype = (musb_readb(hw_ep->regs, MUSB_TXTYPE)
					>> 4) & 0x3;
			if (!is_in && (qh->type == USB_ENDPOINT_XFER_BULK) &&
				toggle && (txtype == USB_ENDPOINT_XFER_ISOC))
				continue;

			best_diff = diff;
			best_end = epnum;
		}
	}
	/* use bulk reserved ep1 if no other ep is free */
	if (best_end < 0 && qh->type == USB_ENDPOINT_XFER_BULK) {
		hw_ep = musb->bulk_ep;
		if (is_in)
			head = &musb->in_bulk;
		else
			head = &musb->out_bulk;

		/* Enable bulk RX/TX NAK timeout scheme when bulk requests are
		 * multiplexed. This scheme does not work in high speed to full
		 * speed scenario as NAK interrupts are not coming from a
		 * full speed device connected to a high speed device.
		 * NAK timeout interval is 8 (128 uframe or 16ms) for HS and
		 * 4 (8 frame or 8ms) for FS device.
		 */
		if (qh->dev)
			qh->intv_reg =
				(USB_SPEED_HIGH == qh->dev->speed) ? 8 : 4;
		goto success;
	} else if (best_end < 0) {
		return -ENOSPC;
	}

	idle = 1;
	qh->mux = 0;
	hw_ep = musb->endpoints + best_end;
	dev_dbg(musb->controller, "qh %p periodic slot %d\n", qh, best_end);
success:
	if (head) {
		idle = list_empty(head);
		list_add_tail(&qh->ring, head);
		qh->mux = 1;
	}
	qh->hw_ep = hw_ep;
	qh->hep->hcpriv = qh;
	if (idle)
		musb_start_urb(musb, is_in, qh);
	return 0;
}

static int musb_urb_enqueue(
	struct usb_hcd			*hcd,
	struct urb			*urb,
	gfp_t				mem_flags)
{
	unsigned long			flags;
	struct musb			*musb = hcd_to_musb(hcd);
	struct usb_host_endpoint	*hep = urb->ep;
	struct musb_qh			*qh;
	struct usb_endpoint_descriptor	*epd = &hep->desc;
	int				ret;
	unsigned			type_reg;
	unsigned			interval;

	/* host role must be active */
	if (!is_host_active(musb) || !musb->is_active)
		return -ENODEV;

	spin_lock_irqsave(&musb->lock, flags);
	ret = usb_hcd_link_urb_to_ep(hcd, urb);
	qh = ret ? NULL : hep->hcpriv;
	if (qh)
		urb->hcpriv = qh;
	spin_unlock_irqrestore(&musb->lock, flags);

	/* DMA mapping was already done, if needed, and this urb is on
	 * hep->urb_list now ... so we're done, unless hep wasn't yet
	 * scheduled onto a live qh.
	 *
	 * REVISIT best to keep hep->hcpriv valid until the endpoint gets
	 * disabled, testing for empty qh->ring and avoiding qh setup costs
	 * except for the first urb queued after a config change.
	 */
	if (qh || ret)
		return ret;

	/* Allocate and initialize qh, minimizing the work done each time
	 * hw_ep gets reprogrammed, or with irqs blocked.  Then schedule it.
	 *
	 * REVISIT consider a dedicated qh kmem_cache, so it's harder
	 * for bugs in other kernel code to break this driver...
	 */
	qh = kzalloc(sizeof *qh, mem_flags);
	if (!qh) {
		spin_lock_irqsave(&musb->lock, flags);
		usb_hcd_unlink_urb_from_ep(hcd, urb);
		spin_unlock_irqrestore(&musb->lock, flags);
		return -ENOMEM;
	}

	qh->hep = hep;
	qh->dev = urb->dev;
	INIT_LIST_HEAD(&qh->ring);
	qh->is_ready = 1;

	qh->maxpacket = usb_endpoint_maxp(epd);
	qh->type = usb_endpoint_type(epd);

	/* Bits 11 & 12 of wMaxPacketSize encode high bandwidth multiplier.
	 * Some musb cores don't support high bandwidth ISO transfers; and
	 * we don't (yet!) support high bandwidth interrupt transfers.
	 */
	qh->hb_mult = 1 + ((qh->maxpacket >> 11) & 0x03);
	if (qh->hb_mult > 1) {
		int ok = (qh->type == USB_ENDPOINT_XFER_ISOC);

		if (ok)
			ok = (usb_pipein(urb->pipe) && musb->hb_iso_rx)
				|| (usb_pipeout(urb->pipe) && musb->hb_iso_tx);
		if (!ok) {
			ret = -EMSGSIZE;
			goto done;
		}
		qh->maxpacket &= 0x7ff;
	}

	qh->epnum = usb_endpoint_num(epd);

	/* NOTE: urb->dev->devnum is wrong during SET_ADDRESS */
	qh->addr_reg = (u8) usb_pipedevice(urb->pipe);

	/* precompute rxtype/txtype/type0 register */
	type_reg = (qh->type << 4) | qh->epnum;
	switch (urb->dev->speed) {
	case USB_SPEED_LOW:
		type_reg |= 0xc0;
		break;
	case USB_SPEED_FULL:
		type_reg |= 0x80;
		break;
	default:
		type_reg |= 0x40;
	}
	qh->type_reg = type_reg;

	/* Precompute RXINTERVAL/TXINTERVAL register */
	switch (qh->type) {
	case USB_ENDPOINT_XFER_INT:
		/*
		 * Full/low speeds use the  linear encoding,
		 * high speed uses the logarithmic encoding.
		 */
		if (urb->dev->speed <= USB_SPEED_FULL) {
			interval = max_t(u8, epd->bInterval, 1);
			break;
		}
		/* FALLTHROUGH */
	case USB_ENDPOINT_XFER_ISOC:
		/* ISO always uses logarithmic encoding */
		interval = min_t(u8, epd->bInterval, 16);
		break;
	default:
		/* REVISIT we actually want to use NAK limits, hinting to the
		 * transfer scheduling logic to try some other qh, e.g. try
		 * for 2 msec first:
		 *
		 * interval = (USB_SPEED_HIGH == urb->dev->speed) ? 16 : 2;
		 *
		 * The downside of disabling this is that transfer scheduling
		 * gets VERY unfair for nonperiodic transfers; a misbehaving
		 * peripheral could make that hurt.  That's perfectly normal
		 * for reads from network or serial adapters ... so we have
		 * partial NAKlimit support for bulk RX.
		 *
		 * The upside of disabling it is simpler transfer scheduling.
		 */
		interval = 0;
	}
	qh->intv_reg = interval;

	/* precompute addressing for external hub/tt ports */
	if (musb->is_multipoint) {
		struct usb_device	*parent = urb->dev->parent;

		if (parent != hcd->self.root_hub) {
			qh->h_addr_reg = (u8) parent->devnum;

			/* set up tt info if needed */
			if (urb->dev->tt) {
				qh->h_port_reg = (u8) urb->dev->ttport;
				if (urb->dev->tt->hub)
					qh->h_addr_reg =
						(u8) urb->dev->tt->hub->devnum;
				if (urb->dev->tt->multi)
					qh->h_addr_reg |= 0x80;
			}
		}
	}

	/* invariant: hep->hcpriv is null OR the qh that's already scheduled.
	 * until we get real dma queues (with an entry for each urb/buffer),
	 * we only have work to do in the former case.
	 */
	spin_lock_irqsave(&musb->lock, flags);
	if (hep->hcpriv || !next_urb(qh)) {
		/* some concurrent activity submitted another urb to hep...
		 * odd, rare, error prone, but legal.
		 */
		kfree(qh);
		qh = NULL;
		ret = 0;
	} else
		ret = musb_schedule(musb, qh,
				epd->bEndpointAddress & USB_ENDPOINT_DIR_MASK);

	if (ret == 0) {
		urb->hcpriv = qh;
		/* FIXME set urb->start_frame for iso/intr, it's tested in
		 * musb_start_urb(), but otherwise only konicawc cares ...
		 */
	}
	spin_unlock_irqrestore(&musb->lock, flags);

done:
	if (ret != 0) {
		spin_lock_irqsave(&musb->lock, flags);
		usb_hcd_unlink_urb_from_ep(hcd, urb);
		spin_unlock_irqrestore(&musb->lock, flags);
		kfree(qh);
	}
	return ret;
}


/*
 * abort a transfer that's at the head of a hardware queue.
 * called with controller locked, irqs blocked
 * that hardware queue advances to the next transfer, unless prevented
 */
static int musb_cleanup_urb(struct urb *urb, struct musb_qh *qh)
{
	struct musb_hw_ep	*ep = qh->hw_ep;
	struct musb		*musb = ep->musb;
	void __iomem		*epio = ep->regs;
	unsigned		hw_end = ep->epnum;
	void __iomem		*regs = ep->musb->mregs;
	int			is_in = usb_pipein(urb->pipe);
	int			status = 0;
	u16			csr;

	musb_ep_select(regs, hw_end);

	if (is_dma_capable()) {
		struct dma_channel	*dma;

		dma = is_in ? ep->rx_channel : ep->tx_channel;
		if (dma) {
			status = ep->musb->dma_controller->channel_abort(dma);
			dev_dbg(musb->controller,
				"abort %cX%d DMA for urb %p --> %d\n",
				is_in ? 'R' : 'T', ep->epnum,
				urb, status);
			urb->actual_length += dma->actual_len;
		}
	}

	/* turn off DMA requests, discard state, stop polling ... */
	if (ep->epnum && is_in) {
		/* giveback saves bulk toggle */
		csr = musb_h_flush_rxfifo(ep, 0);

		/* REVISIT we still get an irq; should likely clear the
		 * endpoint's irq status here to avoid bogus irqs.
		 * clearing that status is platform-specific...
		 */
	} else if (ep->epnum) {
		musb_h_tx_flush_fifo(ep);
		csr = musb_readw(epio, MUSB_TXCSR);
		csr &= ~(MUSB_TXCSR_AUTOSET
			| MUSB_TXCSR_DMAENAB
			| MUSB_TXCSR_H_RXSTALL
			| MUSB_TXCSR_H_NAKTIMEOUT
			| MUSB_TXCSR_H_ERROR
			| MUSB_TXCSR_TXPKTRDY);
		musb_writew(epio, MUSB_TXCSR, csr);
		/* REVISIT may need to clear FLUSHFIFO ... */
		musb_writew(epio, MUSB_TXCSR, csr);
		/* flush cpu writebuffer */
		csr = musb_readw(epio, MUSB_TXCSR);
	} else  {
		musb_h_ep0_flush_fifo(ep);
	}
	if (status == 0)
		musb_advance_schedule(ep->musb, urb, ep, is_in);
	return status;
}

static int musb_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status)
{
	struct musb		*musb = hcd_to_musb(hcd);
	struct musb_qh		*qh;
	unsigned long		flags;
	int			is_in  = usb_pipein(urb->pipe);
	int			ret;

	dev_dbg(musb->controller, "urb=%p, dev%d ep%d%s\n", urb,
			usb_pipedevice(urb->pipe),
			usb_pipeendpoint(urb->pipe),
			is_in ? "in" : "out");

	spin_lock_irqsave(&musb->lock, flags);
	ret = usb_hcd_check_unlink_urb(hcd, urb, status);
	if (ret)
		goto done;

	qh = urb->hcpriv;
	if (!qh)
		goto done;

	/*
	 * Any URB not actively programmed into endpoint hardware can be
	 * immediately given back; that's any URB not at the head of an
	 * endpoint queue, unless someday we get real DMA queues.  And even
	 * if it's at the head, it might not be known to the hardware...
	 *
	 * Otherwise abort current transfer, pending DMA, etc.; urb->status
	 * has already been updated.  This is a synchronous abort; it'd be
	 * OK to hold off until after some IRQ, though.
	 *
	 * NOTE: qh is invalid unless !list_empty(&hep->urb_list)
	 */
	if (!qh->is_ready
			|| urb->urb_list.prev != &qh->hep->urb_list
			|| musb_ep_get_qh(qh->hw_ep, is_in) != qh) {
		int	ready = qh->is_ready;

		qh->is_ready = 0;
		musb_giveback(musb, urb, 0);
		qh->is_ready = ready;

		/* If nothing else (usually musb_giveback) is using it
		 * and its URB list has emptied, recycle this qh.
		 */
		if (ready && list_empty(&qh->hep->urb_list)) {
			qh->hep->hcpriv = NULL;
			list_del(&qh->ring);
			kfree(qh);
		}
	} else
		ret = musb_cleanup_urb(urb, qh);
done:
	spin_unlock_irqrestore(&musb->lock, flags);
	return ret;
}

/* disable an endpoint */
static void
musb_h_disable(struct usb_hcd *hcd, struct usb_host_endpoint *hep)
{
	u8			is_in = hep->desc.bEndpointAddress & USB_DIR_IN;
	unsigned long		flags;
	struct musb		*musb = hcd_to_musb(hcd);
	struct musb_qh		*qh;
	struct urb		*urb;

	spin_lock_irqsave(&musb->lock, flags);

	qh = hep->hcpriv;
	if (qh == NULL)
		goto exit;

	/* NOTE: qh is invalid unless !list_empty(&hep->urb_list) */

	/* Kick the first URB off the hardware, if needed */
	qh->is_ready = 0;
	if (musb_ep_get_qh(qh->hw_ep, is_in) == qh) {
		urb = next_urb(qh);

		/* make software (then hardware) stop ASAP */
		if (!urb->unlinked)
			urb->status = -ESHUTDOWN;

		/* cleanup */
		musb_cleanup_urb(urb, qh);

		/* Then nuke all the others ... and advance the
		 * queue on hw_ep (e.g. bulk ring) when we're done.
		 */
		while (!list_empty(&hep->urb_list)) {
			urb = next_urb(qh);
			urb->status = -ESHUTDOWN;
			musb_advance_schedule(musb, urb, qh->hw_ep, is_in);
		}
	} else {
		/* Just empty the queue; the hardware is busy with
		 * other transfers, and since !qh->is_ready nothing
		 * will activate any of these as it advances.
		 */
		while (!list_empty(&hep->urb_list))
			musb_giveback(musb, next_urb(qh), -ESHUTDOWN);

		hep->hcpriv = NULL;
		list_del(&qh->ring);
		kfree(qh);
	}
exit:
	spin_unlock_irqrestore(&musb->lock, flags);
}

static int musb_h_get_frame_number(struct usb_hcd *hcd)
{
	struct musb	*musb = hcd_to_musb(hcd);

	return musb_readw(musb->mregs, MUSB_FRAME);
}

static int musb_h_start(struct usb_hcd *hcd)
{
	struct musb	*musb = hcd_to_musb(hcd);

	/* NOTE: musb_start() is called when the hub driver turns
	 * on port power, or when (OTG) peripheral starts.
	 */
	hcd->state = HC_STATE_RUNNING;
	musb->port1_status = 0;
	return 0;
}

static void musb_h_stop(struct usb_hcd *hcd)
{
	musb_stop(hcd_to_musb(hcd));
	hcd->state = HC_STATE_HALT;
}

static int musb_bus_suspend(struct usb_hcd *hcd)
{
	struct musb	*musb = hcd_to_musb(hcd);
	u8		devctl;

	musb_port_suspend(musb, true);

	if (!is_host_active(musb))
		return 0;

	switch (musb->xceiv->otg->state) {
	case OTG_STATE_A_SUSPEND:
		return 0;
	case OTG_STATE_A_WAIT_VRISE:
		/* ID could be grounded even if there's no device
		 * on the other end of the cable.  NOTE that the
		 * A_WAIT_VRISE timers are messy with MUSB...
		 */
		devctl = musb_readb(musb->mregs, MUSB_DEVCTL);
		if ((devctl & MUSB_DEVCTL_VBUS) == MUSB_DEVCTL_VBUS)
			musb->xceiv->otg->state = OTG_STATE_A_WAIT_BCON;
		break;
	default:
		break;
	}

	if (musb->is_active) {
		WARNING("trying to suspend as %s while active\n",
				usb_otg_state_string(musb->xceiv->otg->state));
		return -EBUSY;
	} else
		return 0;
}

static int musb_bus_resume(struct usb_hcd *hcd)
{
	struct musb *musb = hcd_to_musb(hcd);

	if (musb->config &&
	    musb->config->host_port_deassert_reset_at_resume)
		musb_port_reset(musb, false);

	return 0;
}

#ifndef CONFIG_MUSB_PIO_ONLY

#define MUSB_USB_DMA_ALIGN 4

struct musb_temp_buffer {
	void *kmalloc_ptr;
	void *old_xfer_buffer;
	u8 data[0];
};

static void musb_free_temp_buffer(struct urb *urb)
{
	enum dma_data_direction dir;
	struct musb_temp_buffer *temp;
	size_t length;

	if (!(urb->transfer_flags & URB_ALIGNED_TEMP_BUFFER))
		return;

	dir = usb_urb_dir_in(urb) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;

	temp = container_of(urb->transfer_buffer, struct musb_temp_buffer,
			    data);

	if (dir == DMA_FROM_DEVICE) {
		if (usb_pipeisoc(urb->pipe))
			length = urb->transfer_buffer_length;
		else
			length = urb->actual_length;

		memcpy(temp->old_xfer_buffer, temp->data, length);
	}
	urb->transfer_buffer = temp->old_xfer_buffer;
	kfree(temp->kmalloc_ptr);

	urb->transfer_flags &= ~URB_ALIGNED_TEMP_BUFFER;
}

static int musb_alloc_temp_buffer(struct urb *urb, gfp_t mem_flags)
{
	enum dma_data_direction dir;
	struct musb_temp_buffer *temp;
	void *kmalloc_ptr;
	size_t kmalloc_size;

	if (urb->num_sgs || urb->sg ||
	    urb->transfer_buffer_length == 0 ||
	    !((uintptr_t)urb->transfer_buffer & (MUSB_USB_DMA_ALIGN - 1)))
		return 0;

	dir = usb_urb_dir_in(urb) ? DMA_FROM_DEVICE : DMA_TO_DEVICE;

	/* Allocate a buffer with enough padding for alignment */
	kmalloc_size = urb->transfer_buffer_length +
		sizeof(struct musb_temp_buffer) + MUSB_USB_DMA_ALIGN - 1;

	kmalloc_ptr = kmalloc(kmalloc_size, mem_flags);
	if (!kmalloc_ptr)
		return -ENOMEM;

	/* Position our struct temp_buffer such that data is aligned */
	temp = PTR_ALIGN(kmalloc_ptr, MUSB_USB_DMA_ALIGN);


	temp->kmalloc_ptr = kmalloc_ptr;
	temp->old_xfer_buffer = urb->transfer_buffer;
	if (dir == DMA_TO_DEVICE)
		memcpy(temp->data, urb->transfer_buffer,
		       urb->transfer_buffer_length);
	urb->transfer_buffer = temp->data;

	urb->transfer_flags |= URB_ALIGNED_TEMP_BUFFER;

	return 0;
}

static int musb_map_urb_for_dma(struct usb_hcd *hcd, struct urb *urb,
				      gfp_t mem_flags)
{
	struct musb	*musb = hcd_to_musb(hcd);
	int ret;

	/*
	 * The DMA engine in RTL1.8 and above cannot handle
	 * DMA addresses that are not aligned to a 4 byte boundary.
	 * For such engine implemented (un)map_urb_for_dma hooks.
	 * Do not use these hooks for RTL<1.8
	 */
	if (musb->hwvers < MUSB_HWVERS_1800)
		return usb_hcd_map_urb_for_dma(hcd, urb, mem_flags);

	ret = musb_alloc_temp_buffer(urb, mem_flags);
	if (ret)
		return ret;

	ret = usb_hcd_map_urb_for_dma(hcd, urb, mem_flags);
	if (ret)
		musb_free_temp_buffer(urb);

	return ret;
}

static void musb_unmap_urb_for_dma(struct usb_hcd *hcd, struct urb *urb)
{
	struct musb	*musb = hcd_to_musb(hcd);

	usb_hcd_unmap_urb_for_dma(hcd, urb);

	/* Do not use this hook for RTL<1.8 (see description above) */
	if (musb->hwvers < MUSB_HWVERS_1800)
		return;

	musb_free_temp_buffer(urb);
}
#endif /* !CONFIG_MUSB_PIO_ONLY */

static const struct hc_driver musb_hc_driver = {
	.description		= "musb-hcd",
	.product_desc		= "MUSB HDRC host driver",
	.hcd_priv_size		= sizeof(struct musb *),
	.flags			= HCD_USB2 | HCD_MEMORY,

	/* not using irq handler or reset hooks from usbcore, since
	 * those must be shared with peripheral code for OTG configs
	 */

	.start			= musb_h_start,
	.stop			= musb_h_stop,

	.get_frame_number	= musb_h_get_frame_number,

	.urb_enqueue		= musb_urb_enqueue,
	.urb_dequeue		= musb_urb_dequeue,
	.endpoint_disable	= musb_h_disable,

#ifndef CONFIG_MUSB_PIO_ONLY
	.map_urb_for_dma	= musb_map_urb_for_dma,
	.unmap_urb_for_dma	= musb_unmap_urb_for_dma,
#endif

	.hub_status_data	= musb_hub_status_data,
	.hub_control		= musb_hub_control,
	.bus_suspend		= musb_bus_suspend,
	.bus_resume		= musb_bus_resume,
	/* .start_port_reset	= NULL, */
	/* .hub_irq_enable	= NULL, */
};

int musb_host_alloc(struct musb *musb)
{
	struct device	*dev = musb->controller;

	/* usbcore sets dev->driver_data to hcd, and sometimes uses that... */
	musb->hcd = usb_create_hcd(&musb_hc_driver, dev, dev_name(dev));
	if (!musb->hcd)
		return -EINVAL;

	*musb->hcd->hcd_priv = (unsigned long) musb;
	musb->hcd->self.uses_pio_for_control = 1;
	musb->hcd->uses_new_polling = 1;
	musb->hcd->has_tt = 1;

	return 0;
}

void musb_host_cleanup(struct musb *musb)
{
	if (musb->port_mode == MUSB_PORT_MODE_GADGET)
		return;
	usb_remove_hcd(musb->hcd);
}

void musb_host_free(struct musb *musb)
{
	usb_put_hcd(musb->hcd);
}

int musb_host_setup(struct musb *musb, int power_budget)
{
	int ret;
	struct usb_hcd *hcd = musb->hcd;

	MUSB_HST_MODE(musb);
	musb->xceiv->otg->default_a = 1;
	musb->xceiv->otg->state = OTG_STATE_A_IDLE;

	otg_set_host(musb->xceiv->otg, &hcd->self);
	hcd->self.otg_port = 1;
	musb->xceiv->otg->host = &hcd->self;
	hcd->power_budget = 2 * (power_budget ? : 250);

	ret = usb_add_hcd(hcd, 0, 0);
	if (ret < 0)
		return ret;

	device_wakeup_enable(hcd->self.controller);
	return 0;
}

void musb_host_resume_root_hub(struct musb *musb)
{
	usb_hcd_resume_root_hub(musb->hcd);
}

void musb_host_poke_root_hub(struct musb *musb)
{
	MUSB_HST_MODE(musb);
	if (musb->hcd->status_urb)
		usb_hcd_poll_rh_status(musb->hcd);
	else
		usb_hcd_resume_root_hub(musb->hcd);
}