1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Generic binary BCH encoding/decoding library
4  *
5  * Copyright © 2011 Parrot S.A.
6  *
7  * Author: Ivan Djelic <ivan.djelic@parrot.com>
8  *
9  * Description:
10  *
11  * This library provides runtime configurable encoding/decoding of binary
12  * Bose-Chaudhuri-Hocquenghem (BCH) codes.
13  *
14  * Call init_bch to get a pointer to a newly allocated bch_control structure for
15  * the given m (Galois field order), t (error correction capability) and
16  * (optional) primitive polynomial parameters.
17  *
18  * Call encode_bch to compute and store ecc parity bytes to a given buffer.
19  * Call decode_bch to detect and locate errors in received data.
20  *
21  * On systems supporting hw BCH features, intermediate results may be provided
22  * to decode_bch in order to skip certain steps. See decode_bch() documentation
23  * for details.
24  *
25  * Option CONFIG_BCH_CONST_PARAMS can be used to force fixed values of
26  * parameters m and t; thus allowing extra compiler optimizations and providing
27  * better (up to 2x) encoding performance. Using this option makes sense when
28  * (m,t) are fixed and known in advance, e.g. when using BCH error correction
29  * on a particular NAND flash device.
30  *
31  * Algorithmic details:
32  *
33  * Encoding is performed by processing 32 input bits in parallel, using 4
34  * remainder lookup tables.
35  *
36  * The final stage of decoding involves the following internal steps:
37  * a. Syndrome computation
38  * b. Error locator polynomial computation using Berlekamp-Massey algorithm
39  * c. Error locator root finding (by far the most expensive step)
40  *
41  * In this implementation, step c is not performed using the usual Chien search.
42  * Instead, an alternative approach described in [1] is used. It consists in
43  * factoring the error locator polynomial using the Berlekamp Trace algorithm
44  * (BTA) down to a certain degree (4), after which ad hoc low-degree polynomial
45  * solving techniques [2] are used. The resulting algorithm, called BTZ, yields
46  * much better performance than Chien search for usual (m,t) values (typically
47  * m >= 13, t < 32, see [1]).
48  *
49  * [1] B. Biswas, V. Herbert. Efficient root finding of polynomials over fields
50  * of characteristic 2, in: Western European Workshop on Research in Cryptology
51  * - WEWoRC 2009, Graz, Austria, LNCS, Springer, July 2009, to appear.
52  * [2] [Zin96] V.A. Zinoviev. On the solution of equations of degree 10 over
53  * finite fields GF(2^q). In Rapport de recherche INRIA no 2829, 1996.
54  */
55 
56 #ifndef USE_HOSTCC
57 #include <common.h>
58 #include <log.h>
59 #include <malloc.h>
60 #include <ubi_uboot.h>
61 #include <dm/devres.h>
62 
63 #include <linux/bitops.h>
64 #else
65 #include <errno.h>
66 #if defined(__FreeBSD__) || defined(__DragonFly__)
67 #include <sys/endian.h>
68 #elif defined(__APPLE__)
69 #include <machine/endian.h>
70 #include <libkern/OSByteOrder.h>
71 #else
72 #include <endian.h>
73 #endif
74 #include <stdint.h>
75 #include <stdlib.h>
76 #include <string.h>
77 
78 #undef cpu_to_be32
79 #if defined(__APPLE__)
80 #define cpu_to_be32 OSSwapHostToBigInt32
81 #else
82 #define cpu_to_be32 htobe32
83 #endif
84 #define DIV_ROUND_UP(n,d) (((n) + (d) - 1) / (d))
85 #define kmalloc(size, flags)	malloc(size)
86 #define kzalloc(size, flags)	calloc(1, size)
87 #define kfree free
88 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]))
89 #endif
90 
91 #include <asm/byteorder.h>
92 #include <linux/bch.h>
93 
94 #if defined(CONFIG_BCH_CONST_PARAMS)
95 #define GF_M(_p)               (CONFIG_BCH_CONST_M)
96 #define GF_T(_p)               (CONFIG_BCH_CONST_T)
97 #define GF_N(_p)               ((1 << (CONFIG_BCH_CONST_M))-1)
98 #else
99 #define GF_M(_p)               ((_p)->m)
100 #define GF_T(_p)               ((_p)->t)
101 #define GF_N(_p)               ((_p)->n)
102 #endif
103 
104 #define BCH_ECC_WORDS(_p)      DIV_ROUND_UP(GF_M(_p)*GF_T(_p), 32)
105 #define BCH_ECC_BYTES(_p)      DIV_ROUND_UP(GF_M(_p)*GF_T(_p), 8)
106 
107 #ifndef dbg
108 #define dbg(_fmt, args...)     do {} while (0)
109 #endif
110 
111 /*
112  * represent a polynomial over GF(2^m)
113  */
114 struct gf_poly {
115 	unsigned int deg;    /* polynomial degree */
116 	unsigned int c[0];   /* polynomial terms */
117 };
118 
119 /* given its degree, compute a polynomial size in bytes */
120 #define GF_POLY_SZ(_d) (sizeof(struct gf_poly)+((_d)+1)*sizeof(unsigned int))
121 
122 /* polynomial of degree 1 */
123 struct gf_poly_deg1 {
124 	struct gf_poly poly;
125 	unsigned int   c[2];
126 };
127 
128 #ifdef USE_HOSTCC
129 #if !defined(__DragonFly__) && !defined(__FreeBSD__) && !defined(__APPLE__)
fls(int x)130 static int fls(int x)
131 {
132 	int r = 32;
133 
134 	if (!x)
135 		return 0;
136 	if (!(x & 0xffff0000u)) {
137 		x <<= 16;
138 		r -= 16;
139 	}
140 	if (!(x & 0xff000000u)) {
141 		x <<= 8;
142 		r -= 8;
143 	}
144 	if (!(x & 0xf0000000u)) {
145 		x <<= 4;
146 		r -= 4;
147 	}
148 	if (!(x & 0xc0000000u)) {
149 		x <<= 2;
150 		r -= 2;
151 	}
152 	if (!(x & 0x80000000u)) {
153 		x <<= 1;
154 		r -= 1;
155 	}
156 	return r;
157 }
158 #endif
159 #endif
160 
161 /*
162  * same as encode_bch(), but process input data one byte at a time
163  */
encode_bch_unaligned(struct bch_control * bch,const unsigned char * data,unsigned int len,uint32_t * ecc)164 static void encode_bch_unaligned(struct bch_control *bch,
165 				 const unsigned char *data, unsigned int len,
166 				 uint32_t *ecc)
167 {
168 	int i;
169 	const uint32_t *p;
170 	const int l = BCH_ECC_WORDS(bch)-1;
171 
172 	while (len--) {
173 		p = bch->mod8_tab + (l+1)*(((ecc[0] >> 24)^(*data++)) & 0xff);
174 
175 		for (i = 0; i < l; i++)
176 			ecc[i] = ((ecc[i] << 8)|(ecc[i+1] >> 24))^(*p++);
177 
178 		ecc[l] = (ecc[l] << 8)^(*p);
179 	}
180 }
181 
182 /*
183  * convert ecc bytes to aligned, zero-padded 32-bit ecc words
184  */
load_ecc8(struct bch_control * bch,uint32_t * dst,const uint8_t * src)185 static void load_ecc8(struct bch_control *bch, uint32_t *dst,
186 		      const uint8_t *src)
187 {
188 	uint8_t pad[4] = {0, 0, 0, 0};
189 	unsigned int i, nwords = BCH_ECC_WORDS(bch)-1;
190 
191 	for (i = 0; i < nwords; i++, src += 4)
192 		dst[i] = (src[0] << 24)|(src[1] << 16)|(src[2] << 8)|src[3];
193 
194 	memcpy(pad, src, BCH_ECC_BYTES(bch)-4*nwords);
195 	dst[nwords] = (pad[0] << 24)|(pad[1] << 16)|(pad[2] << 8)|pad[3];
196 }
197 
198 /*
199  * convert 32-bit ecc words to ecc bytes
200  */
store_ecc8(struct bch_control * bch,uint8_t * dst,const uint32_t * src)201 static void store_ecc8(struct bch_control *bch, uint8_t *dst,
202 		       const uint32_t *src)
203 {
204 	uint8_t pad[4];
205 	unsigned int i, nwords = BCH_ECC_WORDS(bch)-1;
206 
207 	for (i = 0; i < nwords; i++) {
208 		*dst++ = (src[i] >> 24);
209 		*dst++ = (src[i] >> 16) & 0xff;
210 		*dst++ = (src[i] >>  8) & 0xff;
211 		*dst++ = (src[i] >>  0) & 0xff;
212 	}
213 	pad[0] = (src[nwords] >> 24);
214 	pad[1] = (src[nwords] >> 16) & 0xff;
215 	pad[2] = (src[nwords] >>  8) & 0xff;
216 	pad[3] = (src[nwords] >>  0) & 0xff;
217 	memcpy(dst, pad, BCH_ECC_BYTES(bch)-4*nwords);
218 }
219 
220 /**
221  * encode_bch - calculate BCH ecc parity of data
222  * @bch:   BCH control structure
223  * @data:  data to encode
224  * @len:   data length in bytes
225  * @ecc:   ecc parity data, must be initialized by caller
226  *
227  * The @ecc parity array is used both as input and output parameter, in order to
228  * allow incremental computations. It should be of the size indicated by member
229  * @ecc_bytes of @bch, and should be initialized to 0 before the first call.
230  *
231  * The exact number of computed ecc parity bits is given by member @ecc_bits of
232  * @bch; it may be less than m*t for large values of t.
233  */
encode_bch(struct bch_control * bch,const uint8_t * data,unsigned int len,uint8_t * ecc)234 void encode_bch(struct bch_control *bch, const uint8_t *data,
235 		unsigned int len, uint8_t *ecc)
236 {
237 	const unsigned int l = BCH_ECC_WORDS(bch)-1;
238 	unsigned int i, mlen;
239 	unsigned long m;
240 	uint32_t w, r[l+1];
241 	const uint32_t * const tab0 = bch->mod8_tab;
242 	const uint32_t * const tab1 = tab0 + 256*(l+1);
243 	const uint32_t * const tab2 = tab1 + 256*(l+1);
244 	const uint32_t * const tab3 = tab2 + 256*(l+1);
245 	const uint32_t *pdata, *p0, *p1, *p2, *p3;
246 
247 	if (ecc) {
248 		/* load ecc parity bytes into internal 32-bit buffer */
249 		load_ecc8(bch, bch->ecc_buf, ecc);
250 	} else {
251 		memset(bch->ecc_buf, 0, sizeof(r));
252 	}
253 
254 	/* process first unaligned data bytes */
255 	m = ((unsigned long)data) & 3;
256 	if (m) {
257 		mlen = (len < (4-m)) ? len : 4-m;
258 		encode_bch_unaligned(bch, data, mlen, bch->ecc_buf);
259 		data += mlen;
260 		len  -= mlen;
261 	}
262 
263 	/* process 32-bit aligned data words */
264 	pdata = (uint32_t *)data;
265 	mlen  = len/4;
266 	data += 4*mlen;
267 	len  -= 4*mlen;
268 	memcpy(r, bch->ecc_buf, sizeof(r));
269 
270 	/*
271 	 * split each 32-bit word into 4 polynomials of weight 8 as follows:
272 	 *
273 	 * 31 ...24  23 ...16  15 ... 8  7 ... 0
274 	 * xxxxxxxx  yyyyyyyy  zzzzzzzz  tttttttt
275 	 *                               tttttttt  mod g = r0 (precomputed)
276 	 *                     zzzzzzzz  00000000  mod g = r1 (precomputed)
277 	 *           yyyyyyyy  00000000  00000000  mod g = r2 (precomputed)
278 	 * xxxxxxxx  00000000  00000000  00000000  mod g = r3 (precomputed)
279 	 * xxxxxxxx  yyyyyyyy  zzzzzzzz  tttttttt  mod g = r0^r1^r2^r3
280 	 */
281 	while (mlen--) {
282 		/* input data is read in big-endian format */
283 		w = r[0]^cpu_to_be32(*pdata++);
284 		p0 = tab0 + (l+1)*((w >>  0) & 0xff);
285 		p1 = tab1 + (l+1)*((w >>  8) & 0xff);
286 		p2 = tab2 + (l+1)*((w >> 16) & 0xff);
287 		p3 = tab3 + (l+1)*((w >> 24) & 0xff);
288 
289 		for (i = 0; i < l; i++)
290 			r[i] = r[i+1]^p0[i]^p1[i]^p2[i]^p3[i];
291 
292 		r[l] = p0[l]^p1[l]^p2[l]^p3[l];
293 	}
294 	memcpy(bch->ecc_buf, r, sizeof(r));
295 
296 	/* process last unaligned bytes */
297 	if (len)
298 		encode_bch_unaligned(bch, data, len, bch->ecc_buf);
299 
300 	/* store ecc parity bytes into original parity buffer */
301 	if (ecc)
302 		store_ecc8(bch, ecc, bch->ecc_buf);
303 }
304 
modulo(struct bch_control * bch,unsigned int v)305 static inline int modulo(struct bch_control *bch, unsigned int v)
306 {
307 	const unsigned int n = GF_N(bch);
308 	while (v >= n) {
309 		v -= n;
310 		v = (v & n) + (v >> GF_M(bch));
311 	}
312 	return v;
313 }
314 
315 /*
316  * shorter and faster modulo function, only works when v < 2N.
317  */
mod_s(struct bch_control * bch,unsigned int v)318 static inline int mod_s(struct bch_control *bch, unsigned int v)
319 {
320 	const unsigned int n = GF_N(bch);
321 	return (v < n) ? v : v-n;
322 }
323 
deg(unsigned int poly)324 static inline int deg(unsigned int poly)
325 {
326 	/* polynomial degree is the most-significant bit index */
327 	return fls(poly)-1;
328 }
329 
parity(unsigned int x)330 static inline int parity(unsigned int x)
331 {
332 	/*
333 	 * public domain code snippet, lifted from
334 	 * http://www-graphics.stanford.edu/~seander/bithacks.html
335 	 */
336 	x ^= x >> 1;
337 	x ^= x >> 2;
338 	x = (x & 0x11111111U) * 0x11111111U;
339 	return (x >> 28) & 1;
340 }
341 
342 /* Galois field basic operations: multiply, divide, inverse, etc. */
343 
gf_mul(struct bch_control * bch,unsigned int a,unsigned int b)344 static inline unsigned int gf_mul(struct bch_control *bch, unsigned int a,
345 				  unsigned int b)
346 {
347 	return (a && b) ? bch->a_pow_tab[mod_s(bch, bch->a_log_tab[a]+
348 					       bch->a_log_tab[b])] : 0;
349 }
350 
gf_sqr(struct bch_control * bch,unsigned int a)351 static inline unsigned int gf_sqr(struct bch_control *bch, unsigned int a)
352 {
353 	return a ? bch->a_pow_tab[mod_s(bch, 2*bch->a_log_tab[a])] : 0;
354 }
355 
gf_div(struct bch_control * bch,unsigned int a,unsigned int b)356 static inline unsigned int gf_div(struct bch_control *bch, unsigned int a,
357 				  unsigned int b)
358 {
359 	return a ? bch->a_pow_tab[mod_s(bch, bch->a_log_tab[a]+
360 					GF_N(bch)-bch->a_log_tab[b])] : 0;
361 }
362 
gf_inv(struct bch_control * bch,unsigned int a)363 static inline unsigned int gf_inv(struct bch_control *bch, unsigned int a)
364 {
365 	return bch->a_pow_tab[GF_N(bch)-bch->a_log_tab[a]];
366 }
367 
a_pow(struct bch_control * bch,int i)368 static inline unsigned int a_pow(struct bch_control *bch, int i)
369 {
370 	return bch->a_pow_tab[modulo(bch, i)];
371 }
372 
a_log(struct bch_control * bch,unsigned int x)373 static inline int a_log(struct bch_control *bch, unsigned int x)
374 {
375 	return bch->a_log_tab[x];
376 }
377 
a_ilog(struct bch_control * bch,unsigned int x)378 static inline int a_ilog(struct bch_control *bch, unsigned int x)
379 {
380 	return mod_s(bch, GF_N(bch)-bch->a_log_tab[x]);
381 }
382 
383 /*
384  * compute 2t syndromes of ecc polynomial, i.e. ecc(a^j) for j=1..2t
385  */
compute_syndromes(struct bch_control * bch,uint32_t * ecc,unsigned int * syn)386 static void compute_syndromes(struct bch_control *bch, uint32_t *ecc,
387 			      unsigned int *syn)
388 {
389 	int i, j, s;
390 	unsigned int m;
391 	uint32_t poly;
392 	const int t = GF_T(bch);
393 
394 	s = bch->ecc_bits;
395 
396 	/* make sure extra bits in last ecc word are cleared */
397 	m = ((unsigned int)s) & 31;
398 	if (m)
399 		ecc[s/32] &= ~((1u << (32-m))-1);
400 	memset(syn, 0, 2*t*sizeof(*syn));
401 
402 	/* compute v(a^j) for j=1 .. 2t-1 */
403 	do {
404 		poly = *ecc++;
405 		s -= 32;
406 		while (poly) {
407 			i = deg(poly);
408 			for (j = 0; j < 2*t; j += 2)
409 				syn[j] ^= a_pow(bch, (j+1)*(i+s));
410 
411 			poly ^= (1 << i);
412 		}
413 	} while (s > 0);
414 
415 	/* v(a^(2j)) = v(a^j)^2 */
416 	for (j = 0; j < t; j++)
417 		syn[2*j+1] = gf_sqr(bch, syn[j]);
418 }
419 
gf_poly_copy(struct gf_poly * dst,struct gf_poly * src)420 static void gf_poly_copy(struct gf_poly *dst, struct gf_poly *src)
421 {
422 	memcpy(dst, src, GF_POLY_SZ(src->deg));
423 }
424 
compute_error_locator_polynomial(struct bch_control * bch,const unsigned int * syn)425 static int compute_error_locator_polynomial(struct bch_control *bch,
426 					    const unsigned int *syn)
427 {
428 	const unsigned int t = GF_T(bch);
429 	const unsigned int n = GF_N(bch);
430 	unsigned int i, j, tmp, l, pd = 1, d = syn[0];
431 	struct gf_poly *elp = bch->elp;
432 	struct gf_poly *pelp = bch->poly_2t[0];
433 	struct gf_poly *elp_copy = bch->poly_2t[1];
434 	int k, pp = -1;
435 
436 	memset(pelp, 0, GF_POLY_SZ(2*t));
437 	memset(elp, 0, GF_POLY_SZ(2*t));
438 
439 	pelp->deg = 0;
440 	pelp->c[0] = 1;
441 	elp->deg = 0;
442 	elp->c[0] = 1;
443 
444 	/* use simplified binary Berlekamp-Massey algorithm */
445 	for (i = 0; (i < t) && (elp->deg <= t); i++) {
446 		if (d) {
447 			k = 2*i-pp;
448 			gf_poly_copy(elp_copy, elp);
449 			/* e[i+1](X) = e[i](X)+di*dp^-1*X^2(i-p)*e[p](X) */
450 			tmp = a_log(bch, d)+n-a_log(bch, pd);
451 			for (j = 0; j <= pelp->deg; j++) {
452 				if (pelp->c[j]) {
453 					l = a_log(bch, pelp->c[j]);
454 					elp->c[j+k] ^= a_pow(bch, tmp+l);
455 				}
456 			}
457 			/* compute l[i+1] = max(l[i]->c[l[p]+2*(i-p]) */
458 			tmp = pelp->deg+k;
459 			if (tmp > elp->deg) {
460 				elp->deg = tmp;
461 				gf_poly_copy(pelp, elp_copy);
462 				pd = d;
463 				pp = 2*i;
464 			}
465 		}
466 		/* di+1 = S(2i+3)+elp[i+1].1*S(2i+2)+...+elp[i+1].lS(2i+3-l) */
467 		if (i < t-1) {
468 			d = syn[2*i+2];
469 			for (j = 1; j <= elp->deg; j++)
470 				d ^= gf_mul(bch, elp->c[j], syn[2*i+2-j]);
471 		}
472 	}
473 	dbg("elp=%s\n", gf_poly_str(elp));
474 	return (elp->deg > t) ? -1 : (int)elp->deg;
475 }
476 
477 /*
478  * solve a m x m linear system in GF(2) with an expected number of solutions,
479  * and return the number of found solutions
480  */
solve_linear_system(struct bch_control * bch,unsigned int * rows,unsigned int * sol,int nsol)481 static int solve_linear_system(struct bch_control *bch, unsigned int *rows,
482 			       unsigned int *sol, int nsol)
483 {
484 	const int m = GF_M(bch);
485 	unsigned int tmp, mask;
486 	int rem, c, r, p, k, param[m];
487 
488 	k = 0;
489 	mask = 1 << m;
490 
491 	/* Gaussian elimination */
492 	for (c = 0; c < m; c++) {
493 		rem = 0;
494 		p = c-k;
495 		/* find suitable row for elimination */
496 		for (r = p; r < m; r++) {
497 			if (rows[r] & mask) {
498 				if (r != p) {
499 					tmp = rows[r];
500 					rows[r] = rows[p];
501 					rows[p] = tmp;
502 				}
503 				rem = r+1;
504 				break;
505 			}
506 		}
507 		if (rem) {
508 			/* perform elimination on remaining rows */
509 			tmp = rows[p];
510 			for (r = rem; r < m; r++) {
511 				if (rows[r] & mask)
512 					rows[r] ^= tmp;
513 			}
514 		} else {
515 			/* elimination not needed, store defective row index */
516 			param[k++] = c;
517 		}
518 		mask >>= 1;
519 	}
520 	/* rewrite system, inserting fake parameter rows */
521 	if (k > 0) {
522 		p = k;
523 		for (r = m-1; r >= 0; r--) {
524 			if ((r > m-1-k) && rows[r])
525 				/* system has no solution */
526 				return 0;
527 
528 			rows[r] = (p && (r == param[p-1])) ?
529 				p--, 1u << (m-r) : rows[r-p];
530 		}
531 	}
532 
533 	if (nsol != (1 << k))
534 		/* unexpected number of solutions */
535 		return 0;
536 
537 	for (p = 0; p < nsol; p++) {
538 		/* set parameters for p-th solution */
539 		for (c = 0; c < k; c++)
540 			rows[param[c]] = (rows[param[c]] & ~1)|((p >> c) & 1);
541 
542 		/* compute unique solution */
543 		tmp = 0;
544 		for (r = m-1; r >= 0; r--) {
545 			mask = rows[r] & (tmp|1);
546 			tmp |= parity(mask) << (m-r);
547 		}
548 		sol[p] = tmp >> 1;
549 	}
550 	return nsol;
551 }
552 
553 /*
554  * this function builds and solves a linear system for finding roots of a degree
555  * 4 affine monic polynomial X^4+aX^2+bX+c over GF(2^m).
556  */
find_affine4_roots(struct bch_control * bch,unsigned int a,unsigned int b,unsigned int c,unsigned int * roots)557 static int find_affine4_roots(struct bch_control *bch, unsigned int a,
558 			      unsigned int b, unsigned int c,
559 			      unsigned int *roots)
560 {
561 	int i, j, k;
562 	const int m = GF_M(bch);
563 	unsigned int mask = 0xff, t, rows[16] = {0,};
564 
565 	j = a_log(bch, b);
566 	k = a_log(bch, a);
567 	rows[0] = c;
568 
569 	/* buid linear system to solve X^4+aX^2+bX+c = 0 */
570 	for (i = 0; i < m; i++) {
571 		rows[i+1] = bch->a_pow_tab[4*i]^
572 			(a ? bch->a_pow_tab[mod_s(bch, k)] : 0)^
573 			(b ? bch->a_pow_tab[mod_s(bch, j)] : 0);
574 		j++;
575 		k += 2;
576 	}
577 	/*
578 	 * transpose 16x16 matrix before passing it to linear solver
579 	 * warning: this code assumes m < 16
580 	 */
581 	for (j = 8; j != 0; j >>= 1, mask ^= (mask << j)) {
582 		for (k = 0; k < 16; k = (k+j+1) & ~j) {
583 			t = ((rows[k] >> j)^rows[k+j]) & mask;
584 			rows[k] ^= (t << j);
585 			rows[k+j] ^= t;
586 		}
587 	}
588 	return solve_linear_system(bch, rows, roots, 4);
589 }
590 
591 /*
592  * compute root r of a degree 1 polynomial over GF(2^m) (returned as log(1/r))
593  */
find_poly_deg1_roots(struct bch_control * bch,struct gf_poly * poly,unsigned int * roots)594 static int find_poly_deg1_roots(struct bch_control *bch, struct gf_poly *poly,
595 				unsigned int *roots)
596 {
597 	int n = 0;
598 
599 	if (poly->c[0])
600 		/* poly[X] = bX+c with c!=0, root=c/b */
601 		roots[n++] = mod_s(bch, GF_N(bch)-bch->a_log_tab[poly->c[0]]+
602 				   bch->a_log_tab[poly->c[1]]);
603 	return n;
604 }
605 
606 /*
607  * compute roots of a degree 2 polynomial over GF(2^m)
608  */
find_poly_deg2_roots(struct bch_control * bch,struct gf_poly * poly,unsigned int * roots)609 static int find_poly_deg2_roots(struct bch_control *bch, struct gf_poly *poly,
610 				unsigned int *roots)
611 {
612 	int n = 0, i, l0, l1, l2;
613 	unsigned int u, v, r;
614 
615 	if (poly->c[0] && poly->c[1]) {
616 
617 		l0 = bch->a_log_tab[poly->c[0]];
618 		l1 = bch->a_log_tab[poly->c[1]];
619 		l2 = bch->a_log_tab[poly->c[2]];
620 
621 		/* using z=a/bX, transform aX^2+bX+c into z^2+z+u (u=ac/b^2) */
622 		u = a_pow(bch, l0+l2+2*(GF_N(bch)-l1));
623 		/*
624 		 * let u = sum(li.a^i) i=0..m-1; then compute r = sum(li.xi):
625 		 * r^2+r = sum(li.(xi^2+xi)) = sum(li.(a^i+Tr(a^i).a^k)) =
626 		 * u + sum(li.Tr(a^i).a^k) = u+a^k.Tr(sum(li.a^i)) = u+a^k.Tr(u)
627 		 * i.e. r and r+1 are roots iff Tr(u)=0
628 		 */
629 		r = 0;
630 		v = u;
631 		while (v) {
632 			i = deg(v);
633 			r ^= bch->xi_tab[i];
634 			v ^= (1 << i);
635 		}
636 		/* verify root */
637 		if ((gf_sqr(bch, r)^r) == u) {
638 			/* reverse z=a/bX transformation and compute log(1/r) */
639 			roots[n++] = modulo(bch, 2*GF_N(bch)-l1-
640 					    bch->a_log_tab[r]+l2);
641 			roots[n++] = modulo(bch, 2*GF_N(bch)-l1-
642 					    bch->a_log_tab[r^1]+l2);
643 		}
644 	}
645 	return n;
646 }
647 
648 /*
649  * compute roots of a degree 3 polynomial over GF(2^m)
650  */
find_poly_deg3_roots(struct bch_control * bch,struct gf_poly * poly,unsigned int * roots)651 static int find_poly_deg3_roots(struct bch_control *bch, struct gf_poly *poly,
652 				unsigned int *roots)
653 {
654 	int i, n = 0;
655 	unsigned int a, b, c, a2, b2, c2, e3, tmp[4];
656 
657 	if (poly->c[0]) {
658 		/* transform polynomial into monic X^3 + a2X^2 + b2X + c2 */
659 		e3 = poly->c[3];
660 		c2 = gf_div(bch, poly->c[0], e3);
661 		b2 = gf_div(bch, poly->c[1], e3);
662 		a2 = gf_div(bch, poly->c[2], e3);
663 
664 		/* (X+a2)(X^3+a2X^2+b2X+c2) = X^4+aX^2+bX+c (affine) */
665 		c = gf_mul(bch, a2, c2);           /* c = a2c2      */
666 		b = gf_mul(bch, a2, b2)^c2;        /* b = a2b2 + c2 */
667 		a = gf_sqr(bch, a2)^b2;            /* a = a2^2 + b2 */
668 
669 		/* find the 4 roots of this affine polynomial */
670 		if (find_affine4_roots(bch, a, b, c, tmp) == 4) {
671 			/* remove a2 from final list of roots */
672 			for (i = 0; i < 4; i++) {
673 				if (tmp[i] != a2)
674 					roots[n++] = a_ilog(bch, tmp[i]);
675 			}
676 		}
677 	}
678 	return n;
679 }
680 
681 /*
682  * compute roots of a degree 4 polynomial over GF(2^m)
683  */
find_poly_deg4_roots(struct bch_control * bch,struct gf_poly * poly,unsigned int * roots)684 static int find_poly_deg4_roots(struct bch_control *bch, struct gf_poly *poly,
685 				unsigned int *roots)
686 {
687 	int i, l, n = 0;
688 	unsigned int a, b, c, d, e = 0, f, a2, b2, c2, e4;
689 
690 	if (poly->c[0] == 0)
691 		return 0;
692 
693 	/* transform polynomial into monic X^4 + aX^3 + bX^2 + cX + d */
694 	e4 = poly->c[4];
695 	d = gf_div(bch, poly->c[0], e4);
696 	c = gf_div(bch, poly->c[1], e4);
697 	b = gf_div(bch, poly->c[2], e4);
698 	a = gf_div(bch, poly->c[3], e4);
699 
700 	/* use Y=1/X transformation to get an affine polynomial */
701 	if (a) {
702 		/* first, eliminate cX by using z=X+e with ae^2+c=0 */
703 		if (c) {
704 			/* compute e such that e^2 = c/a */
705 			f = gf_div(bch, c, a);
706 			l = a_log(bch, f);
707 			l += (l & 1) ? GF_N(bch) : 0;
708 			e = a_pow(bch, l/2);
709 			/*
710 			 * use transformation z=X+e:
711 			 * z^4+e^4 + a(z^3+ez^2+e^2z+e^3) + b(z^2+e^2) +cz+ce+d
712 			 * z^4 + az^3 + (ae+b)z^2 + (ae^2+c)z+e^4+be^2+ae^3+ce+d
713 			 * z^4 + az^3 + (ae+b)z^2 + e^4+be^2+d
714 			 * z^4 + az^3 +     b'z^2 + d'
715 			 */
716 			d = a_pow(bch, 2*l)^gf_mul(bch, b, f)^d;
717 			b = gf_mul(bch, a, e)^b;
718 		}
719 		/* now, use Y=1/X to get Y^4 + b/dY^2 + a/dY + 1/d */
720 		if (d == 0)
721 			/* assume all roots have multiplicity 1 */
722 			return 0;
723 
724 		c2 = gf_inv(bch, d);
725 		b2 = gf_div(bch, a, d);
726 		a2 = gf_div(bch, b, d);
727 	} else {
728 		/* polynomial is already affine */
729 		c2 = d;
730 		b2 = c;
731 		a2 = b;
732 	}
733 	/* find the 4 roots of this affine polynomial */
734 	if (find_affine4_roots(bch, a2, b2, c2, roots) == 4) {
735 		for (i = 0; i < 4; i++) {
736 			/* post-process roots (reverse transformations) */
737 			f = a ? gf_inv(bch, roots[i]) : roots[i];
738 			roots[i] = a_ilog(bch, f^e);
739 		}
740 		n = 4;
741 	}
742 	return n;
743 }
744 
745 /*
746  * build monic, log-based representation of a polynomial
747  */
gf_poly_logrep(struct bch_control * bch,const struct gf_poly * a,int * rep)748 static void gf_poly_logrep(struct bch_control *bch,
749 			   const struct gf_poly *a, int *rep)
750 {
751 	int i, d = a->deg, l = GF_N(bch)-a_log(bch, a->c[a->deg]);
752 
753 	/* represent 0 values with -1; warning, rep[d] is not set to 1 */
754 	for (i = 0; i < d; i++)
755 		rep[i] = a->c[i] ? mod_s(bch, a_log(bch, a->c[i])+l) : -1;
756 }
757 
758 /*
759  * compute polynomial Euclidean division remainder in GF(2^m)[X]
760  */
gf_poly_mod(struct bch_control * bch,struct gf_poly * a,const struct gf_poly * b,int * rep)761 static void gf_poly_mod(struct bch_control *bch, struct gf_poly *a,
762 			const struct gf_poly *b, int *rep)
763 {
764 	int la, p, m;
765 	unsigned int i, j, *c = a->c;
766 	const unsigned int d = b->deg;
767 
768 	if (a->deg < d)
769 		return;
770 
771 	/* reuse or compute log representation of denominator */
772 	if (!rep) {
773 		rep = bch->cache;
774 		gf_poly_logrep(bch, b, rep);
775 	}
776 
777 	for (j = a->deg; j >= d; j--) {
778 		if (c[j]) {
779 			la = a_log(bch, c[j]);
780 			p = j-d;
781 			for (i = 0; i < d; i++, p++) {
782 				m = rep[i];
783 				if (m >= 0)
784 					c[p] ^= bch->a_pow_tab[mod_s(bch,
785 								     m+la)];
786 			}
787 		}
788 	}
789 	a->deg = d-1;
790 	while (!c[a->deg] && a->deg)
791 		a->deg--;
792 }
793 
794 /*
795  * compute polynomial Euclidean division quotient in GF(2^m)[X]
796  */
gf_poly_div(struct bch_control * bch,struct gf_poly * a,const struct gf_poly * b,struct gf_poly * q)797 static void gf_poly_div(struct bch_control *bch, struct gf_poly *a,
798 			const struct gf_poly *b, struct gf_poly *q)
799 {
800 	if (a->deg >= b->deg) {
801 		q->deg = a->deg-b->deg;
802 		/* compute a mod b (modifies a) */
803 		gf_poly_mod(bch, a, b, NULL);
804 		/* quotient is stored in upper part of polynomial a */
805 		memcpy(q->c, &a->c[b->deg], (1+q->deg)*sizeof(unsigned int));
806 	} else {
807 		q->deg = 0;
808 		q->c[0] = 0;
809 	}
810 }
811 
812 /*
813  * compute polynomial GCD (Greatest Common Divisor) in GF(2^m)[X]
814  */
gf_poly_gcd(struct bch_control * bch,struct gf_poly * a,struct gf_poly * b)815 static struct gf_poly *gf_poly_gcd(struct bch_control *bch, struct gf_poly *a,
816 				   struct gf_poly *b)
817 {
818 	struct gf_poly *tmp;
819 
820 	dbg("gcd(%s,%s)=", gf_poly_str(a), gf_poly_str(b));
821 
822 	if (a->deg < b->deg) {
823 		tmp = b;
824 		b = a;
825 		a = tmp;
826 	}
827 
828 	while (b->deg > 0) {
829 		gf_poly_mod(bch, a, b, NULL);
830 		tmp = b;
831 		b = a;
832 		a = tmp;
833 	}
834 
835 	dbg("%s\n", gf_poly_str(a));
836 
837 	return a;
838 }
839 
840 /*
841  * Given a polynomial f and an integer k, compute Tr(a^kX) mod f
842  * This is used in Berlekamp Trace algorithm for splitting polynomials
843  */
compute_trace_bk_mod(struct bch_control * bch,int k,const struct gf_poly * f,struct gf_poly * z,struct gf_poly * out)844 static void compute_trace_bk_mod(struct bch_control *bch, int k,
845 				 const struct gf_poly *f, struct gf_poly *z,
846 				 struct gf_poly *out)
847 {
848 	const int m = GF_M(bch);
849 	int i, j;
850 
851 	/* z contains z^2j mod f */
852 	z->deg = 1;
853 	z->c[0] = 0;
854 	z->c[1] = bch->a_pow_tab[k];
855 
856 	out->deg = 0;
857 	memset(out, 0, GF_POLY_SZ(f->deg));
858 
859 	/* compute f log representation only once */
860 	gf_poly_logrep(bch, f, bch->cache);
861 
862 	for (i = 0; i < m; i++) {
863 		/* add a^(k*2^i)(z^(2^i) mod f) and compute (z^(2^i) mod f)^2 */
864 		for (j = z->deg; j >= 0; j--) {
865 			out->c[j] ^= z->c[j];
866 			z->c[2*j] = gf_sqr(bch, z->c[j]);
867 			z->c[2*j+1] = 0;
868 		}
869 		if (z->deg > out->deg)
870 			out->deg = z->deg;
871 
872 		if (i < m-1) {
873 			z->deg *= 2;
874 			/* z^(2(i+1)) mod f = (z^(2^i) mod f)^2 mod f */
875 			gf_poly_mod(bch, z, f, bch->cache);
876 		}
877 	}
878 	while (!out->c[out->deg] && out->deg)
879 		out->deg--;
880 
881 	dbg("Tr(a^%d.X) mod f = %s\n", k, gf_poly_str(out));
882 }
883 
884 /*
885  * factor a polynomial using Berlekamp Trace algorithm (BTA)
886  */
factor_polynomial(struct bch_control * bch,int k,struct gf_poly * f,struct gf_poly ** g,struct gf_poly ** h)887 static void factor_polynomial(struct bch_control *bch, int k, struct gf_poly *f,
888 			      struct gf_poly **g, struct gf_poly **h)
889 {
890 	struct gf_poly *f2 = bch->poly_2t[0];
891 	struct gf_poly *q  = bch->poly_2t[1];
892 	struct gf_poly *tk = bch->poly_2t[2];
893 	struct gf_poly *z  = bch->poly_2t[3];
894 	struct gf_poly *gcd;
895 
896 	dbg("factoring %s...\n", gf_poly_str(f));
897 
898 	*g = f;
899 	*h = NULL;
900 
901 	/* tk = Tr(a^k.X) mod f */
902 	compute_trace_bk_mod(bch, k, f, z, tk);
903 
904 	if (tk->deg > 0) {
905 		/* compute g = gcd(f, tk) (destructive operation) */
906 		gf_poly_copy(f2, f);
907 		gcd = gf_poly_gcd(bch, f2, tk);
908 		if (gcd->deg < f->deg) {
909 			/* compute h=f/gcd(f,tk); this will modify f and q */
910 			gf_poly_div(bch, f, gcd, q);
911 			/* store g and h in-place (clobbering f) */
912 			*h = &((struct gf_poly_deg1 *)f)[gcd->deg].poly;
913 			gf_poly_copy(*g, gcd);
914 			gf_poly_copy(*h, q);
915 		}
916 	}
917 }
918 
919 /*
920  * find roots of a polynomial, using BTZ algorithm; see the beginning of this
921  * file for details
922  */
find_poly_roots(struct bch_control * bch,unsigned int k,struct gf_poly * poly,unsigned int * roots)923 static int find_poly_roots(struct bch_control *bch, unsigned int k,
924 			   struct gf_poly *poly, unsigned int *roots)
925 {
926 	int cnt;
927 	struct gf_poly *f1, *f2;
928 
929 	switch (poly->deg) {
930 		/* handle low degree polynomials with ad hoc techniques */
931 	case 1:
932 		cnt = find_poly_deg1_roots(bch, poly, roots);
933 		break;
934 	case 2:
935 		cnt = find_poly_deg2_roots(bch, poly, roots);
936 		break;
937 	case 3:
938 		cnt = find_poly_deg3_roots(bch, poly, roots);
939 		break;
940 	case 4:
941 		cnt = find_poly_deg4_roots(bch, poly, roots);
942 		break;
943 	default:
944 		/* factor polynomial using Berlekamp Trace Algorithm (BTA) */
945 		cnt = 0;
946 		if (poly->deg && (k <= GF_M(bch))) {
947 			factor_polynomial(bch, k, poly, &f1, &f2);
948 			if (f1)
949 				cnt += find_poly_roots(bch, k+1, f1, roots);
950 			if (f2)
951 				cnt += find_poly_roots(bch, k+1, f2, roots+cnt);
952 		}
953 		break;
954 	}
955 	return cnt;
956 }
957 
958 #if defined(USE_CHIEN_SEARCH)
959 /*
960  * exhaustive root search (Chien) implementation - not used, included only for
961  * reference/comparison tests
962  */
chien_search(struct bch_control * bch,unsigned int len,struct gf_poly * p,unsigned int * roots)963 static int chien_search(struct bch_control *bch, unsigned int len,
964 			struct gf_poly *p, unsigned int *roots)
965 {
966 	int m;
967 	unsigned int i, j, syn, syn0, count = 0;
968 	const unsigned int k = 8*len+bch->ecc_bits;
969 
970 	/* use a log-based representation of polynomial */
971 	gf_poly_logrep(bch, p, bch->cache);
972 	bch->cache[p->deg] = 0;
973 	syn0 = gf_div(bch, p->c[0], p->c[p->deg]);
974 
975 	for (i = GF_N(bch)-k+1; i <= GF_N(bch); i++) {
976 		/* compute elp(a^i) */
977 		for (j = 1, syn = syn0; j <= p->deg; j++) {
978 			m = bch->cache[j];
979 			if (m >= 0)
980 				syn ^= a_pow(bch, m+j*i);
981 		}
982 		if (syn == 0) {
983 			roots[count++] = GF_N(bch)-i;
984 			if (count == p->deg)
985 				break;
986 		}
987 	}
988 	return (count == p->deg) ? count : 0;
989 }
990 #define find_poly_roots(_p, _k, _elp, _loc) chien_search(_p, len, _elp, _loc)
991 #endif /* USE_CHIEN_SEARCH */
992 
993 /**
994  * decode_bch - decode received codeword and find bit error locations
995  * @bch:      BCH control structure
996  * @data:     received data, ignored if @calc_ecc is provided
997  * @len:      data length in bytes, must always be provided
998  * @recv_ecc: received ecc, if NULL then assume it was XORed in @calc_ecc
999  * @calc_ecc: calculated ecc, if NULL then calc_ecc is computed from @data
1000  * @syn:      hw computed syndrome data (if NULL, syndrome is calculated)
1001  * @errloc:   output array of error locations
1002  *
1003  * Returns:
1004  *  The number of errors found, or -EBADMSG if decoding failed, or -EINVAL if
1005  *  invalid parameters were provided
1006  *
1007  * Depending on the available hw BCH support and the need to compute @calc_ecc
1008  * separately (using encode_bch()), this function should be called with one of
1009  * the following parameter configurations -
1010  *
1011  * by providing @data and @recv_ecc only:
1012  *   decode_bch(@bch, @data, @len, @recv_ecc, NULL, NULL, @errloc)
1013  *
1014  * by providing @recv_ecc and @calc_ecc:
1015  *   decode_bch(@bch, NULL, @len, @recv_ecc, @calc_ecc, NULL, @errloc)
1016  *
1017  * by providing ecc = recv_ecc XOR calc_ecc:
1018  *   decode_bch(@bch, NULL, @len, NULL, ecc, NULL, @errloc)
1019  *
1020  * by providing syndrome results @syn:
1021  *   decode_bch(@bch, NULL, @len, NULL, NULL, @syn, @errloc)
1022  *
1023  * Once decode_bch() has successfully returned with a positive value, error
1024  * locations returned in array @errloc should be interpreted as follows -
1025  *
1026  * if (errloc[n] >= 8*len), then n-th error is located in ecc (no need for
1027  * data correction)
1028  *
1029  * if (errloc[n] < 8*len), then n-th error is located in data and can be
1030  * corrected with statement data[errloc[n]/8] ^= 1 << (errloc[n] % 8);
1031  *
1032  * Note that this function does not perform any data correction by itself, it
1033  * merely indicates error locations.
1034  */
decode_bch(struct bch_control * bch,const uint8_t * data,unsigned int len,const uint8_t * recv_ecc,const uint8_t * calc_ecc,const unsigned int * syn,unsigned int * errloc)1035 int decode_bch(struct bch_control *bch, const uint8_t *data, unsigned int len,
1036 	       const uint8_t *recv_ecc, const uint8_t *calc_ecc,
1037 	       const unsigned int *syn, unsigned int *errloc)
1038 {
1039 	const unsigned int ecc_words = BCH_ECC_WORDS(bch);
1040 	unsigned int nbits;
1041 	int i, err, nroots;
1042 	uint32_t sum;
1043 
1044 	/* sanity check: make sure data length can be handled */
1045 	if (8*len > (bch->n-bch->ecc_bits))
1046 		return -EINVAL;
1047 
1048 	/* if caller does not provide syndromes, compute them */
1049 	if (!syn) {
1050 		if (!calc_ecc) {
1051 			/* compute received data ecc into an internal buffer */
1052 			if (!data || !recv_ecc)
1053 				return -EINVAL;
1054 			encode_bch(bch, data, len, NULL);
1055 		} else {
1056 			/* load provided calculated ecc */
1057 			load_ecc8(bch, bch->ecc_buf, calc_ecc);
1058 		}
1059 		/* load received ecc or assume it was XORed in calc_ecc */
1060 		if (recv_ecc) {
1061 			load_ecc8(bch, bch->ecc_buf2, recv_ecc);
1062 			/* XOR received and calculated ecc */
1063 			for (i = 0, sum = 0; i < (int)ecc_words; i++) {
1064 				bch->ecc_buf[i] ^= bch->ecc_buf2[i];
1065 				sum |= bch->ecc_buf[i];
1066 			}
1067 			if (!sum)
1068 				/* no error found */
1069 				return 0;
1070 		}
1071 		compute_syndromes(bch, bch->ecc_buf, bch->syn);
1072 		syn = bch->syn;
1073 	}
1074 
1075 	err = compute_error_locator_polynomial(bch, syn);
1076 	if (err > 0) {
1077 		nroots = find_poly_roots(bch, 1, bch->elp, errloc);
1078 		if (err != nroots)
1079 			err = -1;
1080 	}
1081 	if (err > 0) {
1082 		/* post-process raw error locations for easier correction */
1083 		nbits = (len*8)+bch->ecc_bits;
1084 		for (i = 0; i < err; i++) {
1085 			if (errloc[i] >= nbits) {
1086 				err = -1;
1087 				break;
1088 			}
1089 			errloc[i] = nbits-1-errloc[i];
1090 			errloc[i] = (errloc[i] & ~7)|(7-(errloc[i] & 7));
1091 		}
1092 	}
1093 	return (err >= 0) ? err : -EBADMSG;
1094 }
1095 
1096 /*
1097  * generate Galois field lookup tables
1098  */
build_gf_tables(struct bch_control * bch,unsigned int poly)1099 static int build_gf_tables(struct bch_control *bch, unsigned int poly)
1100 {
1101 	unsigned int i, x = 1;
1102 	const unsigned int k = 1 << deg(poly);
1103 
1104 	/* primitive polynomial must be of degree m */
1105 	if (k != (1u << GF_M(bch)))
1106 		return -1;
1107 
1108 	for (i = 0; i < GF_N(bch); i++) {
1109 		bch->a_pow_tab[i] = x;
1110 		bch->a_log_tab[x] = i;
1111 		if (i && (x == 1))
1112 			/* polynomial is not primitive (a^i=1 with 0<i<2^m-1) */
1113 			return -1;
1114 		x <<= 1;
1115 		if (x & k)
1116 			x ^= poly;
1117 	}
1118 	bch->a_pow_tab[GF_N(bch)] = 1;
1119 	bch->a_log_tab[0] = 0;
1120 
1121 	return 0;
1122 }
1123 
1124 /*
1125  * compute generator polynomial remainder tables for fast encoding
1126  */
build_mod8_tables(struct bch_control * bch,const uint32_t * g)1127 static void build_mod8_tables(struct bch_control *bch, const uint32_t *g)
1128 {
1129 	int i, j, b, d;
1130 	uint32_t data, hi, lo, *tab;
1131 	const int l = BCH_ECC_WORDS(bch);
1132 	const int plen = DIV_ROUND_UP(bch->ecc_bits+1, 32);
1133 	const int ecclen = DIV_ROUND_UP(bch->ecc_bits, 32);
1134 
1135 	memset(bch->mod8_tab, 0, 4*256*l*sizeof(*bch->mod8_tab));
1136 
1137 	for (i = 0; i < 256; i++) {
1138 		/* p(X)=i is a small polynomial of weight <= 8 */
1139 		for (b = 0; b < 4; b++) {
1140 			/* we want to compute (p(X).X^(8*b+deg(g))) mod g(X) */
1141 			tab = bch->mod8_tab + (b*256+i)*l;
1142 			data = i << (8*b);
1143 			while (data) {
1144 				d = deg(data);
1145 				/* subtract X^d.g(X) from p(X).X^(8*b+deg(g)) */
1146 				data ^= g[0] >> (31-d);
1147 				for (j = 0; j < ecclen; j++) {
1148 					hi = (d < 31) ? g[j] << (d+1) : 0;
1149 					lo = (j+1 < plen) ?
1150 						g[j+1] >> (31-d) : 0;
1151 					tab[j] ^= hi|lo;
1152 				}
1153 			}
1154 		}
1155 	}
1156 }
1157 
1158 /*
1159  * build a base for factoring degree 2 polynomials
1160  */
build_deg2_base(struct bch_control * bch)1161 static int build_deg2_base(struct bch_control *bch)
1162 {
1163 	const int m = GF_M(bch);
1164 	int i, j, r;
1165 	unsigned int sum, x, y, remaining, ak = 0, xi[m];
1166 
1167 	/* find k s.t. Tr(a^k) = 1 and 0 <= k < m */
1168 	for (i = 0; i < m; i++) {
1169 		for (j = 0, sum = 0; j < m; j++)
1170 			sum ^= a_pow(bch, i*(1 << j));
1171 
1172 		if (sum) {
1173 			ak = bch->a_pow_tab[i];
1174 			break;
1175 		}
1176 	}
1177 	/* find xi, i=0..m-1 such that xi^2+xi = a^i+Tr(a^i).a^k */
1178 	remaining = m;
1179 	memset(xi, 0, sizeof(xi));
1180 
1181 	for (x = 0; (x <= GF_N(bch)) && remaining; x++) {
1182 		y = gf_sqr(bch, x)^x;
1183 		for (i = 0; i < 2; i++) {
1184 			r = a_log(bch, y);
1185 			if (y && (r < m) && !xi[r]) {
1186 				bch->xi_tab[r] = x;
1187 				xi[r] = 1;
1188 				remaining--;
1189 				dbg("x%d = %x\n", r, x);
1190 				break;
1191 			}
1192 			y ^= ak;
1193 		}
1194 	}
1195 	/* should not happen but check anyway */
1196 	return remaining ? -1 : 0;
1197 }
1198 
bch_alloc(size_t size,int * err)1199 static void *bch_alloc(size_t size, int *err)
1200 {
1201 	void *ptr;
1202 
1203 	ptr = kmalloc(size, GFP_KERNEL);
1204 	if (ptr == NULL)
1205 		*err = 1;
1206 	return ptr;
1207 }
1208 
1209 /*
1210  * compute generator polynomial for given (m,t) parameters.
1211  */
compute_generator_polynomial(struct bch_control * bch)1212 static uint32_t *compute_generator_polynomial(struct bch_control *bch)
1213 {
1214 	const unsigned int m = GF_M(bch);
1215 	const unsigned int t = GF_T(bch);
1216 	int n, err = 0;
1217 	unsigned int i, j, nbits, r, word, *roots;
1218 	struct gf_poly *g;
1219 	uint32_t *genpoly;
1220 
1221 	g = bch_alloc(GF_POLY_SZ(m*t), &err);
1222 	roots = bch_alloc((bch->n+1)*sizeof(*roots), &err);
1223 	genpoly = bch_alloc(DIV_ROUND_UP(m*t+1, 32)*sizeof(*genpoly), &err);
1224 
1225 	if (err) {
1226 		kfree(genpoly);
1227 		genpoly = NULL;
1228 		goto finish;
1229 	}
1230 
1231 	/* enumerate all roots of g(X) */
1232 	memset(roots , 0, (bch->n+1)*sizeof(*roots));
1233 	for (i = 0; i < t; i++) {
1234 		for (j = 0, r = 2*i+1; j < m; j++) {
1235 			roots[r] = 1;
1236 			r = mod_s(bch, 2*r);
1237 		}
1238 	}
1239 	/* build generator polynomial g(X) */
1240 	g->deg = 0;
1241 	g->c[0] = 1;
1242 	for (i = 0; i < GF_N(bch); i++) {
1243 		if (roots[i]) {
1244 			/* multiply g(X) by (X+root) */
1245 			r = bch->a_pow_tab[i];
1246 			g->c[g->deg+1] = 1;
1247 			for (j = g->deg; j > 0; j--)
1248 				g->c[j] = gf_mul(bch, g->c[j], r)^g->c[j-1];
1249 
1250 			g->c[0] = gf_mul(bch, g->c[0], r);
1251 			g->deg++;
1252 		}
1253 	}
1254 	/* store left-justified binary representation of g(X) */
1255 	n = g->deg+1;
1256 	i = 0;
1257 
1258 	while (n > 0) {
1259 		nbits = (n > 32) ? 32 : n;
1260 		for (j = 0, word = 0; j < nbits; j++) {
1261 			if (g->c[n-1-j])
1262 				word |= 1u << (31-j);
1263 		}
1264 		genpoly[i++] = word;
1265 		n -= nbits;
1266 	}
1267 	bch->ecc_bits = g->deg;
1268 
1269 finish:
1270 	kfree(g);
1271 	kfree(roots);
1272 
1273 	return genpoly;
1274 }
1275 
1276 /**
1277  * init_bch - initialize a BCH encoder/decoder
1278  * @m:          Galois field order, should be in the range 5-15
1279  * @t:          maximum error correction capability, in bits
1280  * @prim_poly:  user-provided primitive polynomial (or 0 to use default)
1281  *
1282  * Returns:
1283  *  a newly allocated BCH control structure if successful, NULL otherwise
1284  *
1285  * This initialization can take some time, as lookup tables are built for fast
1286  * encoding/decoding; make sure not to call this function from a time critical
1287  * path. Usually, init_bch() should be called on module/driver init and
1288  * free_bch() should be called to release memory on exit.
1289  *
1290  * You may provide your own primitive polynomial of degree @m in argument
1291  * @prim_poly, or let init_bch() use its default polynomial.
1292  *
1293  * Once init_bch() has successfully returned a pointer to a newly allocated
1294  * BCH control structure, ecc length in bytes is given by member @ecc_bytes of
1295  * the structure.
1296  */
init_bch(int m,int t,unsigned int prim_poly)1297 struct bch_control *init_bch(int m, int t, unsigned int prim_poly)
1298 {
1299 	int err = 0;
1300 	unsigned int i, words;
1301 	uint32_t *genpoly;
1302 	struct bch_control *bch = NULL;
1303 
1304 	const int min_m = 5;
1305 	const int max_m = 15;
1306 
1307 	/* default primitive polynomials */
1308 	static const unsigned int prim_poly_tab[] = {
1309 		0x25, 0x43, 0x83, 0x11d, 0x211, 0x409, 0x805, 0x1053, 0x201b,
1310 		0x402b, 0x8003,
1311 	};
1312 
1313 #if defined(CONFIG_BCH_CONST_PARAMS)
1314 	if ((m != (CONFIG_BCH_CONST_M)) || (t != (CONFIG_BCH_CONST_T))) {
1315 		printk(KERN_ERR "bch encoder/decoder was configured to support "
1316 		       "parameters m=%d, t=%d only!\n",
1317 		       CONFIG_BCH_CONST_M, CONFIG_BCH_CONST_T);
1318 		goto fail;
1319 	}
1320 #endif
1321 	if ((m < min_m) || (m > max_m))
1322 		/*
1323 		 * values of m greater than 15 are not currently supported;
1324 		 * supporting m > 15 would require changing table base type
1325 		 * (uint16_t) and a small patch in matrix transposition
1326 		 */
1327 		goto fail;
1328 
1329 	/* sanity checks */
1330 	if ((t < 1) || (m*t >= ((1 << m)-1)))
1331 		/* invalid t value */
1332 		goto fail;
1333 
1334 	/* select a primitive polynomial for generating GF(2^m) */
1335 	if (prim_poly == 0)
1336 		prim_poly = prim_poly_tab[m-min_m];
1337 
1338 	bch = kzalloc(sizeof(*bch), GFP_KERNEL);
1339 	if (bch == NULL)
1340 		goto fail;
1341 
1342 	bch->m = m;
1343 	bch->t = t;
1344 	bch->n = (1 << m)-1;
1345 	words  = DIV_ROUND_UP(m*t, 32);
1346 	bch->ecc_bytes = DIV_ROUND_UP(m*t, 8);
1347 	bch->a_pow_tab = bch_alloc((1+bch->n)*sizeof(*bch->a_pow_tab), &err);
1348 	bch->a_log_tab = bch_alloc((1+bch->n)*sizeof(*bch->a_log_tab), &err);
1349 	bch->mod8_tab  = bch_alloc(words*1024*sizeof(*bch->mod8_tab), &err);
1350 	bch->ecc_buf   = bch_alloc(words*sizeof(*bch->ecc_buf), &err);
1351 	bch->ecc_buf2  = bch_alloc(words*sizeof(*bch->ecc_buf2), &err);
1352 	bch->xi_tab    = bch_alloc(m*sizeof(*bch->xi_tab), &err);
1353 	bch->syn       = bch_alloc(2*t*sizeof(*bch->syn), &err);
1354 	bch->cache     = bch_alloc(2*t*sizeof(*bch->cache), &err);
1355 	bch->elp       = bch_alloc((t+1)*sizeof(struct gf_poly_deg1), &err);
1356 
1357 	for (i = 0; i < ARRAY_SIZE(bch->poly_2t); i++)
1358 		bch->poly_2t[i] = bch_alloc(GF_POLY_SZ(2*t), &err);
1359 
1360 	if (err)
1361 		goto fail;
1362 
1363 	err = build_gf_tables(bch, prim_poly);
1364 	if (err)
1365 		goto fail;
1366 
1367 	/* use generator polynomial for computing encoding tables */
1368 	genpoly = compute_generator_polynomial(bch);
1369 	if (genpoly == NULL)
1370 		goto fail;
1371 
1372 	build_mod8_tables(bch, genpoly);
1373 	kfree(genpoly);
1374 
1375 	err = build_deg2_base(bch);
1376 	if (err)
1377 		goto fail;
1378 
1379 	return bch;
1380 
1381 fail:
1382 	free_bch(bch);
1383 	return NULL;
1384 }
1385 
1386 /**
1387  *  free_bch - free the BCH control structure
1388  *  @bch:    BCH control structure to release
1389  */
free_bch(struct bch_control * bch)1390 void free_bch(struct bch_control *bch)
1391 {
1392 	unsigned int i;
1393 
1394 	if (bch) {
1395 		kfree(bch->a_pow_tab);
1396 		kfree(bch->a_log_tab);
1397 		kfree(bch->mod8_tab);
1398 		kfree(bch->ecc_buf);
1399 		kfree(bch->ecc_buf2);
1400 		kfree(bch->xi_tab);
1401 		kfree(bch->syn);
1402 		kfree(bch->cache);
1403 		kfree(bch->elp);
1404 
1405 		for (i = 0; i < ARRAY_SIZE(bch->poly_2t); i++)
1406 			kfree(bch->poly_2t[i]);
1407 
1408 		kfree(bch);
1409 	}
1410 }
1411