aboutsummaryrefslogtreecommitdiffstats
path: root/uart-loader/lib/ecc_512.c
blob: 29647c6fe2ba9c29522e32c0508b351a8b76de04 (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
/*
 * (C) Copyright 2000 Texas Instruments
 *
 * This file os based on the following u-boot file:
 *	common/cmd_nand.c
 *
 * See file CREDITS for list of people who contributed to this
 * project.
 *
 * 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., 59 Temple Place, Suite 330, Boston,
 * MA 02111-1307 USA
 */
#include <common.h>

#ifdef CFG_SW_ECC_512

/*
 * invparity is a 256 byte table that contains the odd parity
 * for each byte. So if the number of bits in a byte is even,
 * the array element is 1, and when the number of bits is odd
 * the array eleemnt is 0.
 */
static const char invparity[256] = {
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	0, 1, 1, 0, 1, 0, 0, 1, 1, 0, 0, 1, 0, 1, 1, 0,
	1, 0, 0, 1, 0, 1, 1, 0, 0, 1, 1, 0, 1, 0, 0, 1
};

/*
 * bitsperbyte contains the number of bits per byte
 * this is only used for testing and repairing parity
 * (a precalculated value slightly improves performance)
 */
static const char bitsperbyte[256] = {
	0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4,
	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
	1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
	2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6,
	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
	3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7,
	4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8,
};

/*
 * addressbits is a lookup table to filter out the bits from the xor-ed
 * ecc data that identify the faulty location.
 * this is only used for repairing parity
 * see the comments in nand_correct_data for more details
 */
static const char addressbits[256] = {
	0x00, 0x00, 0x01, 0x01, 0x00, 0x00, 0x01, 0x01,
	0x02, 0x02, 0x03, 0x03, 0x02, 0x02, 0x03, 0x03,
	0x00, 0x00, 0x01, 0x01, 0x00, 0x00, 0x01, 0x01,
	0x02, 0x02, 0x03, 0x03, 0x02, 0x02, 0x03, 0x03,
	0x04, 0x04, 0x05, 0x05, 0x04, 0x04, 0x05, 0x05,
	0x06, 0x06, 0x07, 0x07, 0x06, 0x06, 0x07, 0x07,
	0x04, 0x04, 0x05, 0x05, 0x04, 0x04, 0x05, 0x05,
	0x06, 0x06, 0x07, 0x07, 0x06, 0x06, 0x07, 0x07,
	0x00, 0x00, 0x01, 0x01, 0x00, 0x00, 0x01, 0x01,
	0x02, 0x02, 0x03, 0x03, 0x02, 0x02, 0x03, 0x03,
	0x00, 0x00, 0x01, 0x01, 0x00, 0x00, 0x01, 0x01,
	0x02, 0x02, 0x03, 0x03, 0x02, 0x02, 0x03, 0x03,
	0x04, 0x04, 0x05, 0x05, 0x04, 0x04, 0x05, 0x05,
	0x06, 0x06, 0x07, 0x07, 0x06, 0x06, 0x07, 0x07,
	0x04, 0x04, 0x05, 0x05, 0x04, 0x04, 0x05, 0x05,
	0x06, 0x06, 0x07, 0x07, 0x06, 0x06, 0x07, 0x07,
	0x08, 0x08, 0x09, 0x09, 0x08, 0x08, 0x09, 0x09,
	0x0a, 0x0a, 0x0b, 0x0b, 0x0a, 0x0a, 0x0b, 0x0b,
	0x08, 0x08, 0x09, 0x09, 0x08, 0x08, 0x09, 0x09,
	0x0a, 0x0a, 0x0b, 0x0b, 0x0a, 0x0a, 0x0b, 0x0b,
	0x0c, 0x0c, 0x0d, 0x0d, 0x0c, 0x0c, 0x0d, 0x0d,
	0x0e, 0x0e, 0x0f, 0x0f, 0x0e, 0x0e, 0x0f, 0x0f,
	0x0c, 0x0c, 0x0d, 0x0d, 0x0c, 0x0c, 0x0d, 0x0d,
	0x0e, 0x0e, 0x0f, 0x0f, 0x0e, 0x0e, 0x0f, 0x0f,
	0x08, 0x08, 0x09, 0x09, 0x08, 0x08, 0x09, 0x09,
	0x0a, 0x0a, 0x0b, 0x0b, 0x0a, 0x0a, 0x0b, 0x0b,
	0x08, 0x08, 0x09, 0x09, 0x08, 0x08, 0x09, 0x09,
	0x0a, 0x0a, 0x0b, 0x0b, 0x0a, 0x0a, 0x0b, 0x0b,
	0x0c, 0x0c, 0x0d, 0x0d, 0x0c, 0x0c, 0x0d, 0x0d,
	0x0e, 0x0e, 0x0f, 0x0f, 0x0e, 0x0e, 0x0f, 0x0f,
	0x0c, 0x0c, 0x0d, 0x0d, 0x0c, 0x0c, 0x0d, 0x0d,
	0x0e, 0x0e, 0x0f, 0x0f, 0x0e, 0x0e, 0x0f, 0x0f
};

/*
 * nand_calculate_ecc - [NAND Interface] Calculate 3-byte ECC for 256/512-byte
 *			 block
 * @mtd:	MTD block structure
 * @buf:	input buffer with raw data
 * @code:	output buffer with ECC
 */
void nand_calculate_ecc(const u_char *buf, u_char *code)
{
	int i;
	const uint32_t *bp = (uint32_t *)buf;
	/* 256 or 512 bytes/ecc  */
	int eccsize = 512;
	const uint32_t eccsize_mult = eccsize >> 8;
	uint32_t cur;		/* current value in buffer */
	/* rp0..rp15..rp17 are the various accumulated parities (per byte) */
	uint32_t rp0, rp1, rp2, rp3, rp4, rp5, rp6, rp7;
	uint32_t rp8, rp9, rp10, rp11, rp12, rp13, rp14, rp15, rp16;
	uint32_t rp17 = 0;
	uint32_t par;		/* the cumulative parity for all data */
	uint32_t tmppar;	/* the cumulative parity for this iteration;
				   for rp12, rp14 and rp16 at the end of the
				   loop */
	par = 0;
	rp4 = 0;
	rp6 = 0;
	rp8 = 0;
	rp10 = 0;
	rp12 = 0;
	rp14 = 0;
	rp16 = 0;

	/*
	 * The loop is unrolled a number of times;
	 * This avoids if statements to decide on which rp value to update
	 * Also we process the data by longwords.
	 * Note: passing unaligned data might give a performance penalty.
	 * It is assumed that the buffers are aligned.
	 * tmppar is the cumulative sum of this iteration.
	 * needed for calculating rp12, rp14, rp16 and par
	 * also used as a performance improvement for rp6, rp8 and rp10
	 */
	for (i = 0; i < eccsize_mult << 2; i++) {
		cur = *bp++;
		tmppar = cur;
		rp4 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp6 ^= tmppar;
		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp8 ^= tmppar;

		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		rp6 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp6 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp10 ^= tmppar;

		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		rp6 ^= cur;
		rp8 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp6 ^= cur;
		rp8 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		rp8 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp8 ^= cur;

		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		rp6 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp6 ^= cur;
		cur = *bp++;
		tmppar ^= cur;
		rp4 ^= cur;
		cur = *bp++;
		tmppar ^= cur;

		par ^= tmppar;
		if ((i & 0x1) == 0)
			rp12 ^= tmppar;
		if ((i & 0x2) == 0)
			rp14 ^= tmppar;
		if (eccsize_mult == 2 && (i & 0x4) == 0)
			rp16 ^= tmppar;
	}

	/*
	 * handle the fact that we use longword operations
	 * we'll bring rp4..rp14..rp16 back to single byte entities by
	 * shifting and xoring first fold the upper and lower 16 bits,
	 * then the upper and lower 8 bits.
	 */
	rp4 ^= (rp4 >> 16);
	rp4 ^= (rp4 >> 8);
	rp4 &= 0xff;
	rp6 ^= (rp6 >> 16);
	rp6 ^= (rp6 >> 8);
	rp6 &= 0xff;
	rp8 ^= (rp8 >> 16);
	rp8 ^= (rp8 >> 8);
	rp8 &= 0xff;
	rp10 ^= (rp10 >> 16);
	rp10 ^= (rp10 >> 8);
	rp10 &= 0xff;
	rp12 ^= (rp12 >> 16);
	rp12 ^= (rp12 >> 8);
	rp12 &= 0xff;
	rp14 ^= (rp14 >> 16);
	rp14 ^= (rp14 >> 8);
	rp14 &= 0xff;
	if (eccsize_mult == 2) {
		rp16 ^= (rp16 >> 16);
		rp16 ^= (rp16 >> 8);
		rp16 &= 0xff;
	}

	/*
	 * we also need to calculate the row parity for rp0..rp3
	 * This is present in par, because par is now
	 * rp3 rp3 rp2 rp2 in little endian and
	 * rp2 rp2 rp3 rp3 in big endian
	 * as well as
	 * rp1 rp0 rp1 rp0 in little endian and
	 * rp0 rp1 rp0 rp1 in big endian
	 * First calculate rp2 and rp3
	 */
	rp3 = (par >> 16);
	rp3 ^= (rp3 >> 8);
	rp3 &= 0xff;
	rp2 = par & 0xffff;
	rp2 ^= (rp2 >> 8);
	rp2 &= 0xff;

	/* reduce par to 16 bits then calculate rp1 and rp0 */
	par ^= (par >> 16);
	rp1 = (par >> 8) & 0xff;
	rp0 = (par & 0xff);

	/* finally reduce par to 8 bits */
	par ^= (par >> 8);
	par &= 0xff;

	/*
	 * and calculate rp5..rp15..rp17
	 * note that par = rp4 ^ rp5 and due to the commutative property
	 * of the ^ operator we can say:
	 * rp5 = (par ^ rp4);
	 * The & 0xff seems superfluous, but benchmarking learned that
	 * leaving it out gives slightly worse results. No idea why, probably
	 * it has to do with the way the pipeline in pentium is organized.
	 */
	rp5 = (par ^ rp4) & 0xff;
	rp7 = (par ^ rp6) & 0xff;
	rp9 = (par ^ rp8) & 0xff;
	rp11 = (par ^ rp10) & 0xff;
	rp13 = (par ^ rp12) & 0xff;
	rp15 = (par ^ rp14) & 0xff;
	if (eccsize_mult == 2)
		rp17 = (par ^ rp16) & 0xff;

	/*
	 * Finally calculate the ecc bits.
	 * Again here it might seem that there are performance optimisations
	 * possible, but benchmarks showed that on the system this is developed
	 * the code below is the fastest
	 */
	code[1] =
	    (invparity[rp7] << 7) |
	    (invparity[rp6] << 6) |
	    (invparity[rp5] << 5) |
	    (invparity[rp4] << 4) |
	    (invparity[rp3] << 3) |
	    (invparity[rp2] << 2) |
	    (invparity[rp1] << 1) |
	    (invparity[rp0]);
	code[0] =
	    (invparity[rp15] << 7) |
	    (invparity[rp14] << 6) |
	    (invparity[rp13] << 5) |
	    (invparity[rp12] << 4) |
	    (invparity[rp11] << 3) |
	    (invparity[rp10] << 2) |
	    (invparity[rp9] << 1)  |
	    (invparity[rp8]);
	if (eccsize_mult == 1)
		code[2] =
		    (invparity[par & 0xf0] << 7) |
		    (invparity[par & 0x0f] << 6) |
		    (invparity[par & 0xcc] << 5) |
		    (invparity[par & 0x33] << 4) |
		    (invparity[par & 0xaa] << 3) |
		    (invparity[par & 0x55] << 2) |
		    3;
	else
		code[2] =
		    (invparity[par & 0xf0] << 7) |
		    (invparity[par & 0x0f] << 6) |
		    (invparity[par & 0xcc] << 5) |
		    (invparity[par & 0x33] << 4) |
		    (invparity[par & 0xaa] << 3) |
		    (invparity[par & 0x55] << 2) |
		    (invparity[rp17] << 1) |
		    (invparity[rp16] << 0);
}

/**
 * nand_correct_data - [NAND Interface] Detect and correct bit error(s)
 * @mtd:	MTD block structure
 * @buf:	raw data read from the chip
 * @read_ecc:	ECC from the chip
 * @calc_ecc:	the ECC calculated from raw data
 *
 * Detect and correct a 1 bit error for 256/512 byte block
 */
int nand_correct_data(unsigned char *buf,
		      unsigned char *read_ecc, unsigned char *calc_ecc)
{
	unsigned char b0, b1, b2;
	uint32_t byte_addr;
	unsigned char bit_addr;
	/* 256 or 512 bytes/ecc  */
	int eccsize = 512;
	const uint32_t eccsize_mult = eccsize >> 8;
	/*
	 * b0 to b2 indicate which bit is faulty (if any)
	 * we might need the xor result  more than once,
	 * so keep them in a local var
	*/
	b0 = read_ecc[1] ^ calc_ecc[1];
	b1 = read_ecc[0] ^ calc_ecc[0];
	b2 = read_ecc[2] ^ calc_ecc[2];

	/* check if there are any bitfaults */

	/* repeated if statements are slightly more efficient than switch ... */
	/* ordered in order of likelihood */

	if ((b0 | b1 | b2) == 0)
		return 0;	/* no error */

	if ((((b0 ^ (b0 >> 1)) & 0x55) == 0x55) &&
	    (((b1 ^ (b1 >> 1)) & 0x55) == 0x55) &&
	    ((eccsize_mult == 1 && ((b2 ^ (b2 >> 1)) & 0x54) == 0x54) ||
	     (eccsize_mult == 2 && ((b2 ^ (b2 >> 1)) & 0x55) == 0x55))) {
	/* single bit error */
		/*
		 * rp17/rp15/13/11/9/7/5/3/1 indicate which byte is the faulty
		 * byte, cp 5/3/1 indicate the faulty bit.
		 * A lookup table (called addressbits) is used to filter
		 * the bits from the byte they are in.
		 * A marginal optimisation is possible by having three
		 * different lookup tables.
		 * One as we have now (for b0), one for b2
		 * (that would avoid the >> 1), and one for b1 (with all values
		 * << 4). However it was felt that introducing two more tables
		 * hardly justify the gain.
		 *
		 * The b2 shift is there to get rid of the lowest two bits.
		 * We could also do addressbits[b2] >> 1 but for the
		 * performace it does not make any difference
		 */
		if (eccsize_mult == 1)
			byte_addr = (addressbits[b1] << 4) + addressbits[b0];
		else
			byte_addr = (addressbits[b2 & 0x3] << 8) +
				    (addressbits[b1] << 4) + addressbits[b0];
		bit_addr = addressbits[b2 >> 2];
		/* flip the bit */
		buf[byte_addr] ^= (1 << bit_addr);
		return 1;

	}
	/* count nr of bits; use table lookup, faster than calculating it */
	if ((bitsperbyte[b0] + bitsperbyte[b1] + bitsperbyte[b2]) == 1)
		return 1;	/* error in ecc data; no action needed */

	return -1;
}
#endif /* CFG_SW_ECC_512 */