summaryrefslogtreecommitdiffstats
path: root/drivers/mtd/nand/gpmi-nand/gpmi-lib.c
blob: ec4db2a359e5de9d7e6a57fb855698115e9722f1 (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
/*
 * Freescale GPMI NAND Flash Driver
 *
 * Copyright (C) 2008-2011 Freescale Semiconductor, Inc.
 * Copyright (C) 2008 Embedded Alley Solutions, Inc.
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * 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 02110-1301 USA.
 */
#include <linux/delay.h>
#include <linux/clk.h>
#include <linux/slab.h>

#include "gpmi-nand.h"
#include "gpmi-regs.h"
#include "bch-regs.h"

static struct timing_threshod timing_default_threshold = {
	.max_data_setup_cycles       = (BM_GPMI_TIMING0_DATA_SETUP >>
						BP_GPMI_TIMING0_DATA_SETUP),
	.internal_data_setup_in_ns   = 0,
	.max_sample_delay_factor     = (BM_GPMI_CTRL1_RDN_DELAY >>
						BP_GPMI_CTRL1_RDN_DELAY),
	.max_dll_clock_period_in_ns  = 32,
	.max_dll_delay_in_ns         = 16,
};

#define MXS_SET_ADDR		0x4
#define MXS_CLR_ADDR		0x8
/*
 * Clear the bit and poll it cleared.  This is usually called with
 * a reset address and mask being either SFTRST(bit 31) or CLKGATE
 * (bit 30).
 */
static int clear_poll_bit(void __iomem *addr, u32 mask)
{
	int timeout = 0x400;

	/* clear the bit */
	writel(mask, addr + MXS_CLR_ADDR);

	/*
	 * SFTRST needs 3 GPMI clocks to settle, the reference manual
	 * recommends to wait 1us.
	 */
	udelay(1);

	/* poll the bit becoming clear */
	while ((readl(addr) & mask) && --timeout)
		/* nothing */;

	return !timeout;
}

#define MODULE_CLKGATE		(1 << 30)
#define MODULE_SFTRST		(1 << 31)
/*
 * The current mxs_reset_block() will do two things:
 *  [1] enable the module.
 *  [2] reset the module.
 *
 * In most of the cases, it's ok.
 * But in MX23, there is a hardware bug in the BCH block (see erratum #2847).
 * If you try to soft reset the BCH block, it becomes unusable until
 * the next hard reset. This case occurs in the NAND boot mode. When the board
 * boots by NAND, the ROM of the chip will initialize the BCH blocks itself.
 * So If the driver tries to reset the BCH again, the BCH will not work anymore.
 * You will see a DMA timeout in this case. The bug has been fixed
 * in the following chips, such as MX28.
 *
 * To avoid this bug, just add a new parameter `just_enable` for
 * the mxs_reset_block(), and rewrite it here.
 */
static int gpmi_reset_block(void __iomem *reset_addr, bool just_enable)
{
	int ret;
	int timeout = 0x400;

	/* clear and poll SFTRST */
	ret = clear_poll_bit(reset_addr, MODULE_SFTRST);
	if (unlikely(ret))
		goto error;

	/* clear CLKGATE */
	writel(MODULE_CLKGATE, reset_addr + MXS_CLR_ADDR);

	if (!just_enable) {
		/* set SFTRST to reset the block */
		writel(MODULE_SFTRST, reset_addr + MXS_SET_ADDR);
		udelay(1);

		/* poll CLKGATE becoming set */
		while ((!(readl(reset_addr) & MODULE_CLKGATE)) && --timeout)
			/* nothing */;
		if (unlikely(!timeout))
			goto error;
	}

	/* clear and poll SFTRST */
	ret = clear_poll_bit(reset_addr, MODULE_SFTRST);
	if (unlikely(ret))
		goto error;

	/* clear and poll CLKGATE */
	ret = clear_poll_bit(reset_addr, MODULE_CLKGATE);
	if (unlikely(ret))
		goto error;

	return 0;

error:
	pr_err("%s(%p): module reset timeout\n", __func__, reset_addr);
	return -ETIMEDOUT;
}

static int __gpmi_enable_clk(struct gpmi_nand_data *this, bool v)
{
	struct clk *clk;
	int ret;
	int i;

	for (i = 0; i < GPMI_CLK_MAX; i++) {
		clk = this->resources.clock[i];
		if (!clk)
			break;

		if (v) {
			ret = clk_prepare_enable(clk);
			if (ret)
				goto err_clk;
		} else {
			clk_disable_unprepare(clk);
		}
	}
	return 0;

err_clk:
	for (; i > 0; i--)
		clk_disable_unprepare(this->resources.clock[i - 1]);
	return ret;
}

#define gpmi_enable_clk(x) __gpmi_enable_clk(x, true)
#define gpmi_disable_clk(x) __gpmi_enable_clk(x, false)

int gpmi_init(struct gpmi_nand_data *this)
{
	struct resources *r = &this->resources;
	int ret;

	ret = gpmi_enable_clk(this);
	if (ret)
		goto err_out;
	ret = gpmi_reset_block(r->gpmi_regs, false);
	if (ret)
		goto err_out;

	/*
	 * Reset BCH here, too. We got failures otherwise :(
	 * See later BCH reset for explanation of MX23 handling
	 */
	ret = gpmi_reset_block(r->bch_regs, GPMI_IS_MX23(this));
	if (ret)
		goto err_out;


	/* Choose NAND mode. */
	writel(BM_GPMI_CTRL1_GPMI_MODE, r->gpmi_regs + HW_GPMI_CTRL1_CLR);

	/* Set the IRQ polarity. */
	writel(BM_GPMI_CTRL1_ATA_IRQRDY_POLARITY,
				r->gpmi_regs + HW_GPMI_CTRL1_SET);

	/* Disable Write-Protection. */
	writel(BM_GPMI_CTRL1_DEV_RESET, r->gpmi_regs + HW_GPMI_CTRL1_SET);

	/* Select BCH ECC. */
	writel(BM_GPMI_CTRL1_BCH_MODE, r->gpmi_regs + HW_GPMI_CTRL1_SET);

	/*
	 * Decouple the chip select from dma channel. We use dma0 for all
	 * the chips.
	 */
	writel(BM_GPMI_CTRL1_DECOUPLE_CS, r->gpmi_regs + HW_GPMI_CTRL1_SET);

	gpmi_disable_clk(this);
	return 0;
err_out:
	return ret;
}

/* This function is very useful. It is called only when the bug occur. */
void gpmi_dump_info(struct gpmi_nand_data *this)
{
	struct resources *r = &this->resources;
	struct bch_geometry *geo = &this->bch_geometry;
	u32 reg;
	int i;

	dev_err(this->dev, "Show GPMI registers :\n");
	for (i = 0; i <= HW_GPMI_DEBUG / 0x10 + 1; i++) {
		reg = readl(r->gpmi_regs + i * 0x10);
		dev_err(this->dev, "offset 0x%.3x : 0x%.8x\n", i * 0x10, reg);
	}

	/* start to print out the BCH info */
	dev_err(this->dev, "Show BCH registers :\n");
	for (i = 0; i <= HW_BCH_VERSION / 0x10 + 1; i++) {
		reg = readl(r->bch_regs + i * 0x10);
		dev_err(this->dev, "offset 0x%.3x : 0x%.8x\n", i * 0x10, reg);
	}
	dev_err(this->dev, "BCH Geometry :\n"
		"GF length              : %u\n"
		"ECC Strength           : %u\n"
		"Page Size in Bytes     : %u\n"
		"Metadata Size in Bytes : %u\n"
		"ECC Chunk Size in Bytes: %u\n"
		"ECC Chunk Count        : %u\n"
		"Payload Size in Bytes  : %u\n"
		"Auxiliary Size in Bytes: %u\n"
		"Auxiliary Status Offset: %u\n"
		"Block Mark Byte Offset : %u\n"
		"Block Mark Bit Offset  : %u\n",
		geo->gf_len,
		geo->ecc_strength,
		geo->page_size,
		geo->metadata_size,
		geo->ecc_chunk_size,
		geo->ecc_chunk_count,
		geo->payload_size,
		geo->auxiliary_size,
		geo->auxiliary_status_offset,
		geo->block_mark_byte_offset,
		geo->block_mark_bit_offset);
}

/* Configures the geometry for BCH.  */
int bch_set_geometry(struct gpmi_nand_data *this)
{
	struct resources *r = &this->resources;
	struct bch_geometry *bch_geo = &this->bch_geometry;
	unsigned int block_count;
	unsigned int block_size;
	unsigned int metadata_size;
	unsigned int ecc_strength;
	unsigned int page_size;
	unsigned int gf_len;
	int ret;

	if (common_nfc_set_geometry(this))
		return !0;

	block_count   = bch_geo->ecc_chunk_count - 1;
	block_size    = bch_geo->ecc_chunk_size;
	metadata_size = bch_geo->metadata_size;
	ecc_strength  = bch_geo->ecc_strength >> 1;
	page_size     = bch_geo->page_size;
	gf_len        = bch_geo->gf_len;

	ret = gpmi_enable_clk(this);
	if (ret)
		goto err_out;

	/*
	* Due to erratum #2847 of the MX23, the BCH cannot be soft reset on this
	* chip, otherwise it will lock up. So we skip resetting BCH on the MX23.
	* On the other hand, the MX28 needs the reset, because one case has been
	* seen where the BCH produced ECC errors constantly after 10000
	* consecutive reboots. The latter case has not been seen on the MX23
	* yet, still we don't know if it could happen there as well.
	*/
	ret = gpmi_reset_block(r->bch_regs, GPMI_IS_MX23(this));
	if (ret)
		goto err_out;

	/* Configure layout 0. */
	writel(BF_BCH_FLASH0LAYOUT0_NBLOCKS(block_count)
			| BF_BCH_FLASH0LAYOUT0_META_SIZE(metadata_size)
			| BF_BCH_FLASH0LAYOUT0_ECC0(ecc_strength, this)
			| BF_BCH_FLASH0LAYOUT0_GF(gf_len, this)
			| BF_BCH_FLASH0LAYOUT0_DATA0_SIZE(block_size, this),
			r->bch_regs + HW_BCH_FLASH0LAYOUT0);

	writel(BF_BCH_FLASH0LAYOUT1_PAGE_SIZE(page_size)
			| BF_BCH_FLASH0LAYOUT1_ECCN(ecc_strength, this)
			| BF_BCH_FLASH0LAYOUT1_GF(gf_len, this)
			| BF_BCH_FLASH0LAYOUT1_DATAN_SIZE(block_size, this),
			r->bch_regs + HW_BCH_FLASH0LAYOUT1);

	/* Set *all* chip selects to use layout 0. */
	writel(0, r->bch_regs + HW_BCH_LAYOUTSELECT);

	/* Enable interrupts. */
	writel(BM_BCH_CTRL_COMPLETE_IRQ_EN,
				r->bch_regs + HW_BCH_CTRL_SET);

	gpmi_disable_clk(this);
	return 0;
err_out:
	return ret;
}

/* Converts time in nanoseconds to cycles. */
static unsigned int ns_to_cycles(unsigned int time,
			unsigned int period, unsigned int min)
{
	unsigned int k;

	k = (time + period - 1) / period;
	return max(k, min);
}

#define DEF_MIN_PROP_DELAY	5
#define DEF_MAX_PROP_DELAY	9
/* Apply timing to current hardware conditions. */
static int gpmi_nfc_compute_hardware_timing(struct gpmi_nand_data *this,
					struct gpmi_nfc_hardware_timing *hw)
{
	struct timing_threshod *nfc = &timing_default_threshold;
	struct resources *r = &this->resources;
	struct nand_chip *nand = &this->nand;
	struct nand_timing target = this->timing;
	bool improved_timing_is_available;
	unsigned long clock_frequency_in_hz;
	unsigned int clock_period_in_ns;
	bool dll_use_half_periods;
	unsigned int dll_delay_shift;
	unsigned int max_sample_delay_in_ns;
	unsigned int address_setup_in_cycles;
	unsigned int data_setup_in_ns;
	unsigned int data_setup_in_cycles;
	unsigned int data_hold_in_cycles;
	int ideal_sample_delay_in_ns;
	unsigned int sample_delay_factor;
	int tEYE;
	unsigned int min_prop_delay_in_ns = DEF_MIN_PROP_DELAY;
	unsigned int max_prop_delay_in_ns = DEF_MAX_PROP_DELAY;

	/*
	 * If there are multiple chips, we need to relax the timings to allow
	 * for signal distortion due to higher capacitance.
	 */
	if (nand->numchips > 2) {
		target.data_setup_in_ns    += 10;
		target.data_hold_in_ns     += 10;
		target.address_setup_in_ns += 10;
	} else if (nand->numchips > 1) {
		target.data_setup_in_ns    += 5;
		target.data_hold_in_ns     += 5;
		target.address_setup_in_ns += 5;
	}

	/* Check if improved timing information is available. */
	improved_timing_is_available =
		(target.tREA_in_ns  >= 0) &&
		(target.tRLOH_in_ns >= 0) &&
		(target.tRHOH_in_ns >= 0);

	/* Inspect the clock. */
	nfc->clock_frequency_in_hz = clk_get_rate(r->clock[0]);
	clock_frequency_in_hz = nfc->clock_frequency_in_hz;
	clock_period_in_ns    = NSEC_PER_SEC / clock_frequency_in_hz;

	/*
	 * The NFC quantizes setup and hold parameters in terms of clock cycles.
	 * Here, we quantize the setup and hold timing parameters to the
	 * next-highest clock period to make sure we apply at least the
	 * specified times.
	 *
	 * For data setup and data hold, the hardware interprets a value of zero
	 * as the largest possible delay. This is not what's intended by a zero
	 * in the input parameter, so we impose a minimum of one cycle.
	 */
	data_setup_in_cycles    = ns_to_cycles(target.data_setup_in_ns,
							clock_period_in_ns, 1);
	data_hold_in_cycles     = ns_to_cycles(target.data_hold_in_ns,
							clock_period_in_ns, 1);
	address_setup_in_cycles = ns_to_cycles(target.address_setup_in_ns,
							clock_period_in_ns, 0);

	/*
	 * The clock's period affects the sample delay in a number of ways:
	 *
	 * (1) The NFC HAL tells us the maximum clock period the sample delay
	 *     DLL can tolerate. If the clock period is greater than half that
	 *     maximum, we must configure the DLL to be driven by half periods.
	 *
	 * (2) We need to convert from an ideal sample delay, in ns, to a
	 *     "sample delay factor," which the NFC uses. This factor depends on
	 *     whether we're driving the DLL with full or half periods.
	 *     Paraphrasing the reference manual:
	 *
	 *         AD = SDF x 0.125 x RP
	 *
	 * where:
	 *
	 *     AD   is the applied delay, in ns.
	 *     SDF  is the sample delay factor, which is dimensionless.
	 *     RP   is the reference period, in ns, which is a full clock period
	 *          if the DLL is being driven by full periods, or half that if
	 *          the DLL is being driven by half periods.
	 *
	 * Let's re-arrange this in a way that's more useful to us:
	 *
	 *                        8
	 *         SDF  =  AD x ----
	 *                       RP
	 *
	 * The reference period is either the clock period or half that, so this
	 * is:
	 *
	 *                        8       AD x DDF
	 *         SDF  =  AD x -----  =  --------
	 *                      f x P        P
	 *
	 * where:
	 *
	 *       f  is 1 or 1/2, depending on how we're driving the DLL.
	 *       P  is the clock period.
	 *     DDF  is the DLL Delay Factor, a dimensionless value that
	 *          incorporates all the constants in the conversion.
	 *
	 * DDF will be either 8 or 16, both of which are powers of two. We can
	 * reduce the cost of this conversion by using bit shifts instead of
	 * multiplication or division. Thus:
	 *
	 *                 AD << DDS
	 *         SDF  =  ---------
	 *                     P
	 *
	 *     or
	 *
	 *         AD  =  (SDF >> DDS) x P
	 *
	 * where:
	 *
	 *     DDS  is the DLL Delay Shift, the logarithm to base 2 of the DDF.
	 */
	if (clock_period_in_ns > (nfc->max_dll_clock_period_in_ns >> 1)) {
		dll_use_half_periods = true;
		dll_delay_shift      = 3 + 1;
	} else {
		dll_use_half_periods = false;
		dll_delay_shift      = 3;
	}

	/*
	 * Compute the maximum sample delay the NFC allows, under current
	 * conditions. If the clock is running too slowly, no sample delay is
	 * possible.
	 */
	if (clock_period_in_ns > nfc->max_dll_clock_period_in_ns)
		max_sample_delay_in_ns = 0;
	else {
		/*
		 * Compute the delay implied by the largest sample delay factor
		 * the NFC allows.
		 */
		max_sample_delay_in_ns =
			(nfc->max_sample_delay_factor * clock_period_in_ns) >>
								dll_delay_shift;

		/*
		 * Check if the implied sample delay larger than the NFC
		 * actually allows.
		 */
		if (max_sample_delay_in_ns > nfc->max_dll_delay_in_ns)
			max_sample_delay_in_ns = nfc->max_dll_delay_in_ns;
	}

	/*
	 * Check if improved timing information is available. If not, we have to
	 * use a less-sophisticated algorithm.
	 */
	if (!improved_timing_is_available) {
		/*
		 * Fold the read setup time required by the NFC into the ideal
		 * sample delay.
		 */
		ideal_sample_delay_in_ns = target.gpmi_sample_delay_in_ns +
						nfc->internal_data_setup_in_ns;

		/*
		 * The ideal sample delay may be greater than the maximum
		 * allowed by the NFC. If so, we can trade off sample delay time
		 * for more data setup time.
		 *
		 * In each iteration of the following loop, we add a cycle to
		 * the data setup time and subtract a corresponding amount from
		 * the sample delay until we've satisified the constraints or
		 * can't do any better.
		 */
		while ((ideal_sample_delay_in_ns > max_sample_delay_in_ns) &&
			(data_setup_in_cycles < nfc->max_data_setup_cycles)) {

			data_setup_in_cycles++;
			ideal_sample_delay_in_ns -= clock_period_in_ns;

			if (ideal_sample_delay_in_ns < 0)
				ideal_sample_delay_in_ns = 0;

		}

		/*
		 * Compute the sample delay factor that corresponds most closely
		 * to the ideal sample delay. If the result is too large for the
		 * NFC, use the maximum value.
		 *
		 * Notice that we use the ns_to_cycles function to compute the
		 * sample delay factor. We do this because the form of the
		 * computation is the same as that for calculating cycles.
		 */
		sample_delay_factor =
			ns_to_cycles(
				ideal_sample_delay_in_ns << dll_delay_shift,
							clock_period_in_ns, 0);

		if (sample_delay_factor > nfc->max_sample_delay_factor)
			sample_delay_factor = nfc->max_sample_delay_factor;

		/* Skip to the part where we return our results. */
		goto return_results;
	}

	/*
	 * If control arrives here, we have more detailed timing information,
	 * so we can use a better algorithm.
	 */

	/*
	 * Fold the read setup time required by the NFC into the maximum
	 * propagation delay.
	 */
	max_prop_delay_in_ns += nfc->internal_data_setup_in_ns;

	/*
	 * Earlier, we computed the number of clock cycles required to satisfy
	 * the data setup time. Now, we need to know the actual nanoseconds.
	 */
	data_setup_in_ns = clock_period_in_ns * data_setup_in_cycles;

	/*
	 * Compute tEYE, the width of the data eye when reading from the NAND
	 * Flash. The eye width is fundamentally determined by the data setup
	 * time, perturbed by propagation delays and some characteristics of the
	 * NAND Flash device.
	 *
	 * start of the eye = max_prop_delay + tREA
	 * end of the eye   = min_prop_delay + tRHOH + data_setup
	 */
	tEYE = (int)min_prop_delay_in_ns + (int)target.tRHOH_in_ns +
							(int)data_setup_in_ns;

	tEYE -= (int)max_prop_delay_in_ns + (int)target.tREA_in_ns;

	/*
	 * The eye must be open. If it's not, we can try to open it by
	 * increasing its main forcer, the data setup time.
	 *
	 * In each iteration of the following loop, we increase the data setup
	 * time by a single clock cycle. We do this until either the eye is
	 * open or we run into NFC limits.
	 */
	while ((tEYE <= 0) &&
			(data_setup_in_cycles < nfc->max_data_setup_cycles)) {
		/* Give a cycle to data setup. */
		data_setup_in_cycles++;
		/* Synchronize the data setup time with the cycles. */
		data_setup_in_ns += clock_period_in_ns;
		/* Adjust tEYE accordingly. */
		tEYE += clock_period_in_ns;
	}

	/*
	 * When control arrives here, the eye is open. The ideal time to sample
	 * the data is in the center of the eye:
	 *
	 *     end of the eye + start of the eye
	 *     ---------------------------------  -  data_setup
	 *                    2
	 *
	 * After some algebra, this simplifies to the code immediately below.
	 */
	ideal_sample_delay_in_ns =
		((int)max_prop_delay_in_ns +
			(int)target.tREA_in_ns +
				(int)min_prop_delay_in_ns +
					(int)target.tRHOH_in_ns -
						(int)data_setup_in_ns) >> 1;

	/*
	 * The following figure illustrates some aspects of a NAND Flash read:
	 *
	 *
	 *           __                   _____________________________________
	 * RDN         \_________________/
	 *
	 *                                         <---- tEYE ----->
	 *                                        /-----------------\
	 * Read Data ----------------------------<                   >---------
	 *                                        \-----------------/
	 *             ^                 ^                 ^              ^
	 *             |                 |                 |              |
	 *             |<--Data Setup -->|<--Delay Time -->|              |
	 *             |                 |                 |              |
	 *             |                 |                                |
	 *             |                 |<--   Quantized Delay Time   -->|
	 *             |                 |                                |
	 *
	 *
	 * We have some issues we must now address:
	 *
	 * (1) The *ideal* sample delay time must not be negative. If it is, we
	 *     jam it to zero.
	 *
	 * (2) The *ideal* sample delay time must not be greater than that
	 *     allowed by the NFC. If it is, we can increase the data setup
	 *     time, which will reduce the delay between the end of the data
	 *     setup and the center of the eye. It will also make the eye
	 *     larger, which might help with the next issue...
	 *
	 * (3) The *quantized* sample delay time must not fall either before the
	 *     eye opens or after it closes (the latter is the problem
	 *     illustrated in the above figure).
	 */

	/* Jam a negative ideal sample delay to zero. */
	if (ideal_sample_delay_in_ns < 0)
		ideal_sample_delay_in_ns = 0;

	/*
	 * Extend the data setup as needed to reduce the ideal sample delay
	 * below the maximum permitted by the NFC.
	 */
	while ((ideal_sample_delay_in_ns > max_sample_delay_in_ns) &&
			(data_setup_in_cycles < nfc->max_data_setup_cycles)) {

		/* Give a cycle to data setup. */
		data_setup_in_cycles++;
		/* Synchronize the data setup time with the cycles. */
		data_setup_in_ns += clock_period_in_ns;
		/* Adjust tEYE accordingly. */
		tEYE += clock_period_in_ns;

		/*
		 * Decrease the ideal sample delay by one half cycle, to keep it
		 * in the middle of the eye.
		 */
		ideal_sample_delay_in_ns -= (clock_period_in_ns >> 1);

		/* Jam a negative ideal sample delay to zero. */
		if (ideal_sample_delay_in_ns < 0)
			ideal_sample_delay_in_ns = 0;
	}

	/*
	 * Compute the sample delay factor that corresponds to the ideal sample
	 * delay. If the result is too large, then use the maximum allowed
	 * value.
	 *
	 * Notice that we use the ns_to_cycles function to compute the sample
	 * delay factor. We do this because the form of the computation is the
	 * same as that for calculating cycles.
	 */
	sample_delay_factor =
		ns_to_cycles(ideal_sample_delay_in_ns << dll_delay_shift,
							clock_period_in_ns, 0);

	if (sample_delay_factor > nfc->max_sample_delay_factor)
		sample_delay_factor = nfc->max_sample_delay_factor;

	/*
	 * These macros conveniently encapsulate a computation we'll use to
	 * continuously evaluate whether or not the data sample delay is inside
	 * the eye.
	 */
	#define IDEAL_DELAY  ((int) ideal_sample_delay_in_ns)

	#define QUANTIZED_DELAY  \
		((int) ((sample_delay_factor * clock_period_in_ns) >> \
							dll_delay_shift))

	#define DELAY_ERROR  (abs(QUANTIZED_DELAY - IDEAL_DELAY))

	#define SAMPLE_IS_NOT_WITHIN_THE_EYE  (DELAY_ERROR > (tEYE >> 1))

	/*
	 * While the quantized sample time falls outside the eye, reduce the
	 * sample delay or extend the data setup to move the sampling point back
	 * toward the eye. Do not allow the number of data setup cycles to
	 * exceed the maximum allowed by the NFC.
	 */
	while (SAMPLE_IS_NOT_WITHIN_THE_EYE &&
			(data_setup_in_cycles < nfc->max_data_setup_cycles)) {
		/*
		 * If control arrives here, the quantized sample delay falls
		 * outside the eye. Check if it's before the eye opens, or after
		 * the eye closes.
		 */
		if (QUANTIZED_DELAY > IDEAL_DELAY) {
			/*
			 * If control arrives here, the quantized sample delay
			 * falls after the eye closes. Decrease the quantized
			 * delay time and then go back to re-evaluate.
			 */
			if (sample_delay_factor != 0)
				sample_delay_factor--;
			continue;
		}

		/*
		 * If control arrives here, the quantized sample delay falls
		 * before the eye opens. Shift the sample point by increasing
		 * data setup time. This will also make the eye larger.
		 */

		/* Give a cycle to data setup. */
		data_setup_in_cycles++;
		/* Synchronize the data setup time with the cycles. */
		data_setup_in_ns += clock_period_in_ns;
		/* Adjust tEYE accordingly. */
		tEYE += clock_period_in_ns;

		/*
		 * Decrease the ideal sample delay by one half cycle, to keep it
		 * in the middle of the eye.
		 */
		ideal_sample_delay_in_ns -= (clock_period_in_ns >> 1);

		/* ...and one less period for the delay time. */
		ideal_sample_delay_in_ns -= clock_period_in_ns;

		/* Jam a negative ideal sample delay to zero. */
		if (ideal_sample_delay_in_ns < 0)
			ideal_sample_delay_in_ns = 0;

		/*
		 * We have a new ideal sample delay, so re-compute the quantized
		 * delay.
		 */
		sample_delay_factor =
			ns_to_cycles(
				ideal_sample_delay_in_ns << dll_delay_shift,
							clock_period_in_ns, 0);

		if (sample_delay_factor > nfc->max_sample_delay_factor)
			sample_delay_factor = nfc->max_sample_delay_factor;
	}

	/* Control arrives here when we're ready to return our results. */
return_results:
	hw->data_setup_in_cycles    = data_setup_in_cycles;
	hw->data_hold_in_cycles     = data_hold_in_cycles;
	hw->address_setup_in_cycles = address_setup_in_cycles;
	hw->use_half_periods        = dll_use_half_periods;
	hw->sample_delay_factor     = sample_delay_factor;
	hw->device_busy_timeout     = GPMI_DEFAULT_BUSY_TIMEOUT;
	hw->wrn_dly_sel             = BV_GPMI_CTRL1_WRN_DLY_SEL_4_TO_8NS;

	/* Return success. */
	return 0;
}

/*
 * <1> Firstly, we should know what's the GPMI-clock means.
 *     The GPMI-clock is the internal clock in the gpmi nand controller.
 *     If you set 100MHz to gpmi nand controller, the GPMI-clock's period
 *     is 10ns. Mark the GPMI-clock's period as GPMI-clock-period.
 *
 * <2> Secondly, we should know what's the frequency on the nand chip pins.
 *     The frequency on the nand chip pins is derived from the GPMI-clock.
 *     We can get it from the following equation:
 *
 *         F = G / (DS + DH)
 *
 *         F  : the frequency on the nand chip pins.
 *         G  : the GPMI clock, such as 100MHz.
 *         DS : GPMI_HW_GPMI_TIMING0:DATA_SETUP
 *         DH : GPMI_HW_GPMI_TIMING0:DATA_HOLD
 *
 * <3> Thirdly, when the frequency on the nand chip pins is above 33MHz,
 *     the nand EDO(extended Data Out) timing could be applied.
 *     The GPMI implements a feedback read strobe to sample the read data.
 *     The feedback read strobe can be delayed to support the nand EDO timing
 *     where the read strobe may deasserts before the read data is valid, and
 *     read data is valid for some time after read strobe.
 *
 *     The following figure illustrates some aspects of a NAND Flash read:
 *
 *                   |<---tREA---->|
 *                   |             |
 *                   |         |   |
 *                   |<--tRP-->|   |
 *                   |         |   |
 *                  __          ___|__________________________________
 *     RDN            \________/   |
 *                                 |
 *                                 /---------\
 *     Read Data    --------------<           >---------
 *                                 \---------/
 *                                |     |
 *                                |<-D->|
 *     FeedbackRDN  ________             ____________
 *                          \___________/
 *
 *          D stands for delay, set in the HW_GPMI_CTRL1:RDN_DELAY.
 *
 *
 * <4> Now, we begin to describe how to compute the right RDN_DELAY.
 *
 *  4.1) From the aspect of the nand chip pins:
 *        Delay = (tREA + C - tRP)               {1}
 *
 *        tREA : the maximum read access time. From the ONFI nand standards,
 *               we know that tREA is 16ns in mode 5, tREA is 20ns is mode 4.
 *               Please check it in : www.onfi.org
 *        C    : a constant for adjust the delay. default is 4.
 *        tRP  : the read pulse width.
 *               Specified by the HW_GPMI_TIMING0:DATA_SETUP:
 *                    tRP = (GPMI-clock-period) * DATA_SETUP
 *
 *  4.2) From the aspect of the GPMI nand controller:
 *         Delay = RDN_DELAY * 0.125 * RP        {2}
 *
 *         RP   : the DLL reference period.
 *            if (GPMI-clock-period > DLL_THRETHOLD)
 *                   RP = GPMI-clock-period / 2;
 *            else
 *                   RP = GPMI-clock-period;
 *
 *            Set the HW_GPMI_CTRL1:HALF_PERIOD if GPMI-clock-period
 *            is greater DLL_THRETHOLD. In other SOCs, the DLL_THRETHOLD
 *            is 16ns, but in mx6q, we use 12ns.
 *
 *  4.3) since {1} equals {2}, we get:
 *
 *                    (tREA + 4 - tRP) * 8
 *         RDN_DELAY = ---------------------     {3}
 *                           RP
 *
 *  4.4) We only support the fastest asynchronous mode of ONFI nand.
 *       For some ONFI nand, the mode 4 is the fastest mode;
 *       while for some ONFI nand, the mode 5 is the fastest mode.
 *       So we only support the mode 4 and mode 5. It is no need to
 *       support other modes.
 */
static void gpmi_compute_edo_timing(struct gpmi_nand_data *this,
			struct gpmi_nfc_hardware_timing *hw)
{
	struct resources *r = &this->resources;
	unsigned long rate = clk_get_rate(r->clock[0]);
	int mode = this->timing_mode;
	int dll_threshold = this->devdata->max_chain_delay;
	unsigned long delay;
	unsigned long clk_period;
	int t_rea;
	int c = 4;
	int t_rp;
	int rp;

	/*
	 * [1] for GPMI_HW_GPMI_TIMING0:
	 *     The async mode requires 40MHz for mode 4, 50MHz for mode 5.
	 *     The GPMI can support 100MHz at most. So if we want to
	 *     get the 40MHz or 50MHz, we have to set DS=1, DH=1.
	 *     Set the ADDRESS_SETUP to 0 in mode 4.
	 */
	hw->data_setup_in_cycles = 1;
	hw->data_hold_in_cycles = 1;
	hw->address_setup_in_cycles = ((mode == 5) ? 1 : 0);

	/* [2] for GPMI_HW_GPMI_TIMING1 */
	hw->device_busy_timeout = 0x9000;

	/* [3] for GPMI_HW_GPMI_CTRL1 */
	hw->wrn_dly_sel = BV_GPMI_CTRL1_WRN_DLY_SEL_NO_DELAY;

	/*
	 * Enlarge 10 times for the numerator and denominator in {3}.
	 * This make us to get more accurate result.
	 */
	clk_period = NSEC_PER_SEC / (rate / 10);
	dll_threshold *= 10;
	t_rea = ((mode == 5) ? 16 : 20) * 10;
	c *= 10;

	t_rp = clk_period * 1; /* DATA_SETUP is 1 */

	if (clk_period > dll_threshold) {
		hw->use_half_periods = 1;
		rp = clk_period / 2;
	} else {
		hw->use_half_periods = 0;
		rp = clk_period;
	}

	/*
	 * Multiply the numerator with 10, we could do a round off:
	 *      7.8 round up to 8; 7.4 round down to 7.
	 */
	delay  = (((t_rea + c - t_rp) * 8) * 10) / rp;
	delay = (delay + 5) / 10;

	hw->sample_delay_factor = delay;
}

static int enable_edo_mode(struct gpmi_nand_data *this, int mode)
{
	struct resources  *r = &this->resources;
	struct nand_chip *nand = &this->nand;
	struct mtd_info	 *mtd = &this->mtd;
	uint8_t *feature;
	unsigned long rate;
	int ret;

	feature = kzalloc(ONFI_SUBFEATURE_PARAM_LEN, GFP_KERNEL);
	if (!feature)
		return -ENOMEM;

	nand->select_chip(mtd, 0);

	/* [1] send SET FEATURE commond to NAND */
	feature[0] = mode;
	ret = nand->onfi_set_features(mtd, nand,
				ONFI_FEATURE_ADDR_TIMING_MODE, feature);
	if (ret)
		goto err_out;

	/* [2] send GET FEATURE command to double-check the timing mode */
	memset(feature, 0, ONFI_SUBFEATURE_PARAM_LEN);
	ret = nand->onfi_get_features(mtd, nand,
				ONFI_FEATURE_ADDR_TIMING_MODE, feature);
	if (ret || feature[0] != mode)
		goto err_out;

	nand->select_chip(mtd, -1);

	/* [3] set the main IO clock, 100MHz for mode 5, 80MHz for mode 4. */
	rate = (mode == 5) ? 100000000 : 80000000;
	clk_set_rate(r->clock[0], rate);

	/* Let the gpmi_begin() re-compute the timing again. */
	this->flags &= ~GPMI_TIMING_INIT_OK;

	this->flags |= GPMI_ASYNC_EDO_ENABLED;
	this->timing_mode = mode;
	kfree(feature);
	dev_info(this->dev, "enable the asynchronous EDO mode %d\n", mode);
	return 0;

err_out:
	nand->select_chip(mtd, -1);
	kfree(feature);
	dev_err(this->dev, "mode:%d ,failed in set feature.\n", mode);
	return -EINVAL;
}

int gpmi_extra_init(struct gpmi_nand_data *this)
{
	struct nand_chip *chip = &this->nand;

	/* Enable the asynchronous EDO feature. */
	if (GPMI_IS_MX6Q(this) && chip->onfi_version) {
		int mode = onfi_get_async_timing_mode(chip);

		/* We only support the timing mode 4 and mode 5. */
		if (mode & ONFI_TIMING_MODE_5)
			mode = 5;
		else if (mode & ONFI_TIMING_MODE_4)
			mode = 4;
		else
			return 0;

		return enable_edo_mode(this, mode);
	}
	return 0;
}

/* Begin the I/O */
void gpmi_begin(struct gpmi_nand_data *this)
{
	struct resources *r = &this->resources;
	void __iomem *gpmi_regs = r->gpmi_regs;
	unsigned int   clock_period_in_ns;
	uint32_t       reg;
	unsigned int   dll_wait_time_in_us;
	struct gpmi_nfc_hardware_timing  hw;
	int ret;

	/* Enable the clock. */
	ret = gpmi_enable_clk(this);
	if (ret) {
		dev_err(this->dev, "We failed in enable the clk\n");
		goto err_out;
	}

	/* Only initialize the timing once */
	if (this->flags & GPMI_TIMING_INIT_OK)
		return;
	this->flags |= GPMI_TIMING_INIT_OK;

	if (this->flags & GPMI_ASYNC_EDO_ENABLED)
		gpmi_compute_edo_timing(this, &hw);
	else
		gpmi_nfc_compute_hardware_timing(this, &hw);

	/* [1] Set HW_GPMI_TIMING0 */
	reg = BF_GPMI_TIMING0_ADDRESS_SETUP(hw.address_setup_in_cycles) |
		BF_GPMI_TIMING0_DATA_HOLD(hw.data_hold_in_cycles)         |
		BF_GPMI_TIMING0_DATA_SETUP(hw.data_setup_in_cycles);

	writel(reg, gpmi_regs + HW_GPMI_TIMING0);

	/* [2] Set HW_GPMI_TIMING1 */
	writel(BF_GPMI_TIMING1_BUSY_TIMEOUT(hw.device_busy_timeout),
		gpmi_regs + HW_GPMI_TIMING1);

	/* [3] The following code is to set the HW_GPMI_CTRL1. */

	/* Set the WRN_DLY_SEL */
	writel(BM_GPMI_CTRL1_WRN_DLY_SEL, gpmi_regs + HW_GPMI_CTRL1_CLR);
	writel(BF_GPMI_CTRL1_WRN_DLY_SEL(hw.wrn_dly_sel),
					gpmi_regs + HW_GPMI_CTRL1_SET);

	/* DLL_ENABLE must be set to 0 when setting RDN_DELAY or HALF_PERIOD. */
	writel(BM_GPMI_CTRL1_DLL_ENABLE, gpmi_regs + HW_GPMI_CTRL1_CLR);

	/* Clear out the DLL control fields. */
	reg = BM_GPMI_CTRL1_RDN_DELAY | BM_GPMI_CTRL1_HALF_PERIOD;
	writel(reg, gpmi_regs + HW_GPMI_CTRL1_CLR);

	/* If no sample delay is called for, return immediately. */
	if (!hw.sample_delay_factor)
		return;

	/* Set RDN_DELAY or HALF_PERIOD. */
	reg = ((hw.use_half_periods) ? BM_GPMI_CTRL1_HALF_PERIOD : 0)
		| BF_GPMI_CTRL1_RDN_DELAY(hw.sample_delay_factor);

	writel(reg, gpmi_regs + HW_GPMI_CTRL1_SET);

	/* At last, we enable the DLL. */
	writel(BM_GPMI_CTRL1_DLL_ENABLE, gpmi_regs + HW_GPMI_CTRL1_SET);

	/*
	 * After we enable the GPMI DLL, we have to wait 64 clock cycles before
	 * we can use the GPMI. Calculate the amount of time we need to wait,
	 * in microseconds.
	 */
	clock_period_in_ns = NSEC_PER_SEC / clk_get_rate(r->clock[0]);
	dll_wait_time_in_us = (clock_period_in_ns * 64) / 1000;

	if (!dll_wait_time_in_us)
		dll_wait_time_in_us = 1;

	/* Wait for the DLL to settle. */
	udelay(dll_wait_time_in_us);

err_out:
	return;
}

void gpmi_end(struct gpmi_nand_data *this)
{
	gpmi_disable_clk(this);
}

/* Clears a BCH interrupt. */
void gpmi_clear_bch(struct gpmi_nand_data *this)
{
	struct resources *r = &this->resources;
	writel(BM_BCH_CTRL_COMPLETE_IRQ, r->bch_regs + HW_BCH_CTRL_CLR);
}

/* Returns the Ready/Busy status of the given chip. */
int gpmi_is_ready(struct gpmi_nand_data *this, unsigned chip)
{
	struct resources *r = &this->resources;
	uint32_t mask = 0;
	uint32_t reg = 0;

	if (GPMI_IS_MX23(this)) {
		mask = MX23_BM_GPMI_DEBUG_READY0 << chip;
		reg = readl(r->gpmi_regs + HW_GPMI_DEBUG);
	} else if (GPMI_IS_MX28(this) || GPMI_IS_MX6Q(this)) {
		/*
		 * In the imx6, all the ready/busy pins are bound
		 * together. So we only need to check chip 0.
		 */
		if (GPMI_IS_MX6Q(this))
			chip = 0;

		/* MX28 shares the same R/B register as MX6Q. */
		mask = MX28_BF_GPMI_STAT_READY_BUSY(1 << chip);
		reg = readl(r->gpmi_regs + HW_GPMI_STAT);
	} else
		dev_err(this->dev, "unknow arch.\n");
	return reg & mask;
}

static inline void set_dma_type(struct gpmi_nand_data *this,
					enum dma_ops_type type)
{
	this->last_dma_type = this->dma_type;
	this->dma_type = type;
}

int gpmi_send_command(struct gpmi_nand_data *this)
{
	struct dma_chan *channel = get_dma_chan(this);
	struct dma_async_tx_descriptor *desc;
	struct scatterlist *sgl;
	int chip = this->current_chip;
	u32 pio[3];

	/* [1] send out the PIO words */
	pio[0] = BF_GPMI_CTRL0_COMMAND_MODE(BV_GPMI_CTRL0_COMMAND_MODE__WRITE)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(BV_GPMI_CTRL0_ADDRESS__NAND_CLE)
		| BM_GPMI_CTRL0_ADDRESS_INCREMENT
		| BF_GPMI_CTRL0_XFER_COUNT(this->command_length);
	pio[1] = pio[2] = 0;
	desc = dmaengine_prep_slave_sg(channel,
					(struct scatterlist *)pio,
					ARRAY_SIZE(pio), DMA_TRANS_NONE, 0);
	if (!desc)
		return -EINVAL;

	/* [2] send out the COMMAND + ADDRESS string stored in @buffer */
	sgl = &this->cmd_sgl;

	sg_init_one(sgl, this->cmd_buffer, this->command_length);
	dma_map_sg(this->dev, sgl, 1, DMA_TO_DEVICE);
	desc = dmaengine_prep_slave_sg(channel,
				sgl, 1, DMA_MEM_TO_DEV,
				DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	/* [3] submit the DMA */
	set_dma_type(this, DMA_FOR_COMMAND);
	return start_dma_without_bch_irq(this, desc);
}

int gpmi_send_data(struct gpmi_nand_data *this)
{
	struct dma_async_tx_descriptor *desc;
	struct dma_chan *channel = get_dma_chan(this);
	int chip = this->current_chip;
	uint32_t command_mode;
	uint32_t address;
	u32 pio[2];

	/* [1] PIO */
	command_mode = BV_GPMI_CTRL0_COMMAND_MODE__WRITE;
	address      = BV_GPMI_CTRL0_ADDRESS__NAND_DATA;

	pio[0] = BF_GPMI_CTRL0_COMMAND_MODE(command_mode)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(address)
		| BF_GPMI_CTRL0_XFER_COUNT(this->upper_len);
	pio[1] = 0;
	desc = dmaengine_prep_slave_sg(channel, (struct scatterlist *)pio,
					ARRAY_SIZE(pio), DMA_TRANS_NONE, 0);
	if (!desc)
		return -EINVAL;

	/* [2] send DMA request */
	prepare_data_dma(this, DMA_TO_DEVICE);
	desc = dmaengine_prep_slave_sg(channel, &this->data_sgl,
					1, DMA_MEM_TO_DEV,
					DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	/* [3] submit the DMA */
	set_dma_type(this, DMA_FOR_WRITE_DATA);
	return start_dma_without_bch_irq(this, desc);
}

int gpmi_read_data(struct gpmi_nand_data *this)
{
	struct dma_async_tx_descriptor *desc;
	struct dma_chan *channel = get_dma_chan(this);
	int chip = this->current_chip;
	u32 pio[2];

	/* [1] : send PIO */
	pio[0] = BF_GPMI_CTRL0_COMMAND_MODE(BV_GPMI_CTRL0_COMMAND_MODE__READ)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(BV_GPMI_CTRL0_ADDRESS__NAND_DATA)
		| BF_GPMI_CTRL0_XFER_COUNT(this->upper_len);
	pio[1] = 0;
	desc = dmaengine_prep_slave_sg(channel,
					(struct scatterlist *)pio,
					ARRAY_SIZE(pio), DMA_TRANS_NONE, 0);
	if (!desc)
		return -EINVAL;

	/* [2] : send DMA request */
	prepare_data_dma(this, DMA_FROM_DEVICE);
	desc = dmaengine_prep_slave_sg(channel, &this->data_sgl,
					1, DMA_DEV_TO_MEM,
					DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	/* [3] : submit the DMA */
	set_dma_type(this, DMA_FOR_READ_DATA);
	return start_dma_without_bch_irq(this, desc);
}

int gpmi_send_page(struct gpmi_nand_data *this,
			dma_addr_t payload, dma_addr_t auxiliary)
{
	struct bch_geometry *geo = &this->bch_geometry;
	uint32_t command_mode;
	uint32_t address;
	uint32_t ecc_command;
	uint32_t buffer_mask;
	struct dma_async_tx_descriptor *desc;
	struct dma_chan *channel = get_dma_chan(this);
	int chip = this->current_chip;
	u32 pio[6];

	/* A DMA descriptor that does an ECC page read. */
	command_mode = BV_GPMI_CTRL0_COMMAND_MODE__WRITE;
	address      = BV_GPMI_CTRL0_ADDRESS__NAND_DATA;
	ecc_command  = BV_GPMI_ECCCTRL_ECC_CMD__BCH_ENCODE;
	buffer_mask  = BV_GPMI_ECCCTRL_BUFFER_MASK__BCH_PAGE |
				BV_GPMI_ECCCTRL_BUFFER_MASK__BCH_AUXONLY;

	pio[0] = BF_GPMI_CTRL0_COMMAND_MODE(command_mode)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(address)
		| BF_GPMI_CTRL0_XFER_COUNT(0);
	pio[1] = 0;
	pio[2] = BM_GPMI_ECCCTRL_ENABLE_ECC
		| BF_GPMI_ECCCTRL_ECC_CMD(ecc_command)
		| BF_GPMI_ECCCTRL_BUFFER_MASK(buffer_mask);
	pio[3] = geo->page_size;
	pio[4] = payload;
	pio[5] = auxiliary;

	desc = dmaengine_prep_slave_sg(channel,
					(struct scatterlist *)pio,
					ARRAY_SIZE(pio), DMA_TRANS_NONE,
					DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	set_dma_type(this, DMA_FOR_WRITE_ECC_PAGE);
	return start_dma_with_bch_irq(this, desc);
}

int gpmi_read_page(struct gpmi_nand_data *this,
				dma_addr_t payload, dma_addr_t auxiliary)
{
	struct bch_geometry *geo = &this->bch_geometry;
	uint32_t command_mode;
	uint32_t address;
	uint32_t ecc_command;
	uint32_t buffer_mask;
	struct dma_async_tx_descriptor *desc;
	struct dma_chan *channel = get_dma_chan(this);
	int chip = this->current_chip;
	u32 pio[6];

	/* [1] Wait for the chip to report ready. */
	command_mode = BV_GPMI_CTRL0_COMMAND_MODE__WAIT_FOR_READY;
	address      = BV_GPMI_CTRL0_ADDRESS__NAND_DATA;

	pio[0] =  BF_GPMI_CTRL0_COMMAND_MODE(command_mode)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(address)
		| BF_GPMI_CTRL0_XFER_COUNT(0);
	pio[1] = 0;
	desc = dmaengine_prep_slave_sg(channel,
				(struct scatterlist *)pio, 2,
				DMA_TRANS_NONE, 0);
	if (!desc)
		return -EINVAL;

	/* [2] Enable the BCH block and read. */
	command_mode = BV_GPMI_CTRL0_COMMAND_MODE__READ;
	address      = BV_GPMI_CTRL0_ADDRESS__NAND_DATA;
	ecc_command  = BV_GPMI_ECCCTRL_ECC_CMD__BCH_DECODE;
	buffer_mask  = BV_GPMI_ECCCTRL_BUFFER_MASK__BCH_PAGE
			| BV_GPMI_ECCCTRL_BUFFER_MASK__BCH_AUXONLY;

	pio[0] =  BF_GPMI_CTRL0_COMMAND_MODE(command_mode)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(address)
		| BF_GPMI_CTRL0_XFER_COUNT(geo->page_size);

	pio[1] = 0;
	pio[2] =  BM_GPMI_ECCCTRL_ENABLE_ECC
		| BF_GPMI_ECCCTRL_ECC_CMD(ecc_command)
		| BF_GPMI_ECCCTRL_BUFFER_MASK(buffer_mask);
	pio[3] = geo->page_size;
	pio[4] = payload;
	pio[5] = auxiliary;
	desc = dmaengine_prep_slave_sg(channel,
					(struct scatterlist *)pio,
					ARRAY_SIZE(pio), DMA_TRANS_NONE,
					DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	/* [3] Disable the BCH block */
	command_mode = BV_GPMI_CTRL0_COMMAND_MODE__WAIT_FOR_READY;
	address      = BV_GPMI_CTRL0_ADDRESS__NAND_DATA;

	pio[0] = BF_GPMI_CTRL0_COMMAND_MODE(command_mode)
		| BM_GPMI_CTRL0_WORD_LENGTH
		| BF_GPMI_CTRL0_CS(chip, this)
		| BF_GPMI_CTRL0_LOCK_CS(LOCK_CS_ENABLE, this)
		| BF_GPMI_CTRL0_ADDRESS(address)
		| BF_GPMI_CTRL0_XFER_COUNT(geo->page_size);
	pio[1] = 0;
	pio[2] = 0; /* clear GPMI_HW_GPMI_ECCCTRL, disable the BCH. */
	desc = dmaengine_prep_slave_sg(channel,
				(struct scatterlist *)pio, 3,
				DMA_TRANS_NONE,
				DMA_PREP_INTERRUPT | DMA_CTRL_ACK);
	if (!desc)
		return -EINVAL;

	/* [4] submit the DMA */
	set_dma_type(this, DMA_FOR_READ_ECC_PAGE);
	return start_dma_with_bch_irq(this, desc);
}