xref: /openbmc/u-boot/lib/bch.c (revision 7e8702a00fe88590c2ece93061da4a40adf3c242)
1  /*
2   * Generic binary BCH encoding/decoding library
3   *
4   * SPDX-License-Identifier:	GPL-2.0
5   *
6   * Copyright © 2011 Parrot S.A.
7   *
8   * Author: Ivan Djelic <ivan.djelic@parrot.com>
9   *
10   * Description:
11   *
12   * This library provides runtime configurable encoding/decoding of binary
13   * Bose-Chaudhuri-Hocquenghem (BCH) codes.
14   *
15   * Call init_bch to get a pointer to a newly allocated bch_control structure for
16   * the given m (Galois field order), t (error correction capability) and
17   * (optional) primitive polynomial parameters.
18   *
19   * Call encode_bch to compute and store ecc parity bytes to a given buffer.
20   * Call decode_bch to detect and locate errors in received data.
21   *
22   * On systems supporting hw BCH features, intermediate results may be provided
23   * to decode_bch in order to skip certain steps. See decode_bch() documentation
24   * for details.
25   *
26   * Option CONFIG_BCH_CONST_PARAMS can be used to force fixed values of
27   * parameters m and t; thus allowing extra compiler optimizations and providing
28   * better (up to 2x) encoding performance. Using this option makes sense when
29   * (m,t) are fixed and known in advance, e.g. when using BCH error correction
30   * on a particular NAND flash device.
31   *
32   * Algorithmic details:
33   *
34   * Encoding is performed by processing 32 input bits in parallel, using 4
35   * remainder lookup tables.
36   *
37   * The final stage of decoding involves the following internal steps:
38   * a. Syndrome computation
39   * b. Error locator polynomial computation using Berlekamp-Massey algorithm
40   * c. Error locator root finding (by far the most expensive step)
41   *
42   * In this implementation, step c is not performed using the usual Chien search.
43   * Instead, an alternative approach described in [1] is used. It consists in
44   * factoring the error locator polynomial using the Berlekamp Trace algorithm
45   * (BTA) down to a certain degree (4), after which ad hoc low-degree polynomial
46   * solving techniques [2] are used. The resulting algorithm, called BTZ, yields
47   * much better performance than Chien search for usual (m,t) values (typically
48   * m >= 13, t < 32, see [1]).
49   *
50   * [1] B. Biswas, V. Herbert. Efficient root finding of polynomials over fields
51   * of characteristic 2, in: Western European Workshop on Research in Cryptology
52   * - WEWoRC 2009, Graz, Austria, LNCS, Springer, July 2009, to appear.
53   * [2] [Zin96] V.A. Zinoviev. On the solution of equations of degree 10 over
54   * finite fields GF(2^q). In Rapport de recherche INRIA no 2829, 1996.
55   */
56  
57  #include <common.h>
58  #include <ubi_uboot.h>
59  
60  #include <linux/bitops.h>
61  #include <asm/byteorder.h>
62  #include <linux/bch.h>
63  
64  #if defined(CONFIG_BCH_CONST_PARAMS)
65  #define GF_M(_p)               (CONFIG_BCH_CONST_M)
66  #define GF_T(_p)               (CONFIG_BCH_CONST_T)
67  #define GF_N(_p)               ((1 << (CONFIG_BCH_CONST_M))-1)
68  #else
69  #define GF_M(_p)               ((_p)->m)
70  #define GF_T(_p)               ((_p)->t)
71  #define GF_N(_p)               ((_p)->n)
72  #endif
73  
74  #define BCH_ECC_WORDS(_p)      DIV_ROUND_UP(GF_M(_p)*GF_T(_p), 32)
75  #define BCH_ECC_BYTES(_p)      DIV_ROUND_UP(GF_M(_p)*GF_T(_p), 8)
76  
77  #ifndef dbg
78  #define dbg(_fmt, args...)     do {} while (0)
79  #endif
80  
81  /*
82   * represent a polynomial over GF(2^m)
83   */
84  struct gf_poly {
85  	unsigned int deg;    /* polynomial degree */
86  	unsigned int c[0];   /* polynomial terms */
87  };
88  
89  /* given its degree, compute a polynomial size in bytes */
90  #define GF_POLY_SZ(_d) (sizeof(struct gf_poly)+((_d)+1)*sizeof(unsigned int))
91  
92  /* polynomial of degree 1 */
93  struct gf_poly_deg1 {
94  	struct gf_poly poly;
95  	unsigned int   c[2];
96  };
97  
98  /*
99   * same as encode_bch(), but process input data one byte at a time
100   */
101  static void encode_bch_unaligned(struct bch_control *bch,
102  				 const unsigned char *data, unsigned int len,
103  				 uint32_t *ecc)
104  {
105  	int i;
106  	const uint32_t *p;
107  	const int l = BCH_ECC_WORDS(bch)-1;
108  
109  	while (len--) {
110  		p = bch->mod8_tab + (l+1)*(((ecc[0] >> 24)^(*data++)) & 0xff);
111  
112  		for (i = 0; i < l; i++)
113  			ecc[i] = ((ecc[i] << 8)|(ecc[i+1] >> 24))^(*p++);
114  
115  		ecc[l] = (ecc[l] << 8)^(*p);
116  	}
117  }
118  
119  /*
120   * convert ecc bytes to aligned, zero-padded 32-bit ecc words
121   */
122  static void load_ecc8(struct bch_control *bch, uint32_t *dst,
123  		      const uint8_t *src)
124  {
125  	uint8_t pad[4] = {0, 0, 0, 0};
126  	unsigned int i, nwords = BCH_ECC_WORDS(bch)-1;
127  
128  	for (i = 0; i < nwords; i++, src += 4)
129  		dst[i] = (src[0] << 24)|(src[1] << 16)|(src[2] << 8)|src[3];
130  
131  	memcpy(pad, src, BCH_ECC_BYTES(bch)-4*nwords);
132  	dst[nwords] = (pad[0] << 24)|(pad[1] << 16)|(pad[2] << 8)|pad[3];
133  }
134  
135  /*
136   * convert 32-bit ecc words to ecc bytes
137   */
138  static void store_ecc8(struct bch_control *bch, uint8_t *dst,
139  		       const uint32_t *src)
140  {
141  	uint8_t pad[4];
142  	unsigned int i, nwords = BCH_ECC_WORDS(bch)-1;
143  
144  	for (i = 0; i < nwords; i++) {
145  		*dst++ = (src[i] >> 24);
146  		*dst++ = (src[i] >> 16) & 0xff;
147  		*dst++ = (src[i] >>  8) & 0xff;
148  		*dst++ = (src[i] >>  0) & 0xff;
149  	}
150  	pad[0] = (src[nwords] >> 24);
151  	pad[1] = (src[nwords] >> 16) & 0xff;
152  	pad[2] = (src[nwords] >>  8) & 0xff;
153  	pad[3] = (src[nwords] >>  0) & 0xff;
154  	memcpy(dst, pad, BCH_ECC_BYTES(bch)-4*nwords);
155  }
156  
157  /**
158   * encode_bch - calculate BCH ecc parity of data
159   * @bch:   BCH control structure
160   * @data:  data to encode
161   * @len:   data length in bytes
162   * @ecc:   ecc parity data, must be initialized by caller
163   *
164   * The @ecc parity array is used both as input and output parameter, in order to
165   * allow incremental computations. It should be of the size indicated by member
166   * @ecc_bytes of @bch, and should be initialized to 0 before the first call.
167   *
168   * The exact number of computed ecc parity bits is given by member @ecc_bits of
169   * @bch; it may be less than m*t for large values of t.
170   */
171  void encode_bch(struct bch_control *bch, const uint8_t *data,
172  		unsigned int len, uint8_t *ecc)
173  {
174  	const unsigned int l = BCH_ECC_WORDS(bch)-1;
175  	unsigned int i, mlen;
176  	unsigned long m;
177  	uint32_t w, r[l+1];
178  	const uint32_t * const tab0 = bch->mod8_tab;
179  	const uint32_t * const tab1 = tab0 + 256*(l+1);
180  	const uint32_t * const tab2 = tab1 + 256*(l+1);
181  	const uint32_t * const tab3 = tab2 + 256*(l+1);
182  	const uint32_t *pdata, *p0, *p1, *p2, *p3;
183  
184  	if (ecc) {
185  		/* load ecc parity bytes into internal 32-bit buffer */
186  		load_ecc8(bch, bch->ecc_buf, ecc);
187  	} else {
188  		memset(bch->ecc_buf, 0, sizeof(r));
189  	}
190  
191  	/* process first unaligned data bytes */
192  	m = ((unsigned long)data) & 3;
193  	if (m) {
194  		mlen = (len < (4-m)) ? len : 4-m;
195  		encode_bch_unaligned(bch, data, mlen, bch->ecc_buf);
196  		data += mlen;
197  		len  -= mlen;
198  	}
199  
200  	/* process 32-bit aligned data words */
201  	pdata = (uint32_t *)data;
202  	mlen  = len/4;
203  	data += 4*mlen;
204  	len  -= 4*mlen;
205  	memcpy(r, bch->ecc_buf, sizeof(r));
206  
207  	/*
208  	 * split each 32-bit word into 4 polynomials of weight 8 as follows:
209  	 *
210  	 * 31 ...24  23 ...16  15 ... 8  7 ... 0
211  	 * xxxxxxxx  yyyyyyyy  zzzzzzzz  tttttttt
212  	 *                               tttttttt  mod g = r0 (precomputed)
213  	 *                     zzzzzzzz  00000000  mod g = r1 (precomputed)
214  	 *           yyyyyyyy  00000000  00000000  mod g = r2 (precomputed)
215  	 * xxxxxxxx  00000000  00000000  00000000  mod g = r3 (precomputed)
216  	 * xxxxxxxx  yyyyyyyy  zzzzzzzz  tttttttt  mod g = r0^r1^r2^r3
217  	 */
218  	while (mlen--) {
219  		/* input data is read in big-endian format */
220  		w = r[0]^cpu_to_be32(*pdata++);
221  		p0 = tab0 + (l+1)*((w >>  0) & 0xff);
222  		p1 = tab1 + (l+1)*((w >>  8) & 0xff);
223  		p2 = tab2 + (l+1)*((w >> 16) & 0xff);
224  		p3 = tab3 + (l+1)*((w >> 24) & 0xff);
225  
226  		for (i = 0; i < l; i++)
227  			r[i] = r[i+1]^p0[i]^p1[i]^p2[i]^p3[i];
228  
229  		r[l] = p0[l]^p1[l]^p2[l]^p3[l];
230  	}
231  	memcpy(bch->ecc_buf, r, sizeof(r));
232  
233  	/* process last unaligned bytes */
234  	if (len)
235  		encode_bch_unaligned(bch, data, len, bch->ecc_buf);
236  
237  	/* store ecc parity bytes into original parity buffer */
238  	if (ecc)
239  		store_ecc8(bch, ecc, bch->ecc_buf);
240  }
241  
242  static inline int modulo(struct bch_control *bch, unsigned int v)
243  {
244  	const unsigned int n = GF_N(bch);
245  	while (v >= n) {
246  		v -= n;
247  		v = (v & n) + (v >> GF_M(bch));
248  	}
249  	return v;
250  }
251  
252  /*
253   * shorter and faster modulo function, only works when v < 2N.
254   */
255  static inline int mod_s(struct bch_control *bch, unsigned int v)
256  {
257  	const unsigned int n = GF_N(bch);
258  	return (v < n) ? v : v-n;
259  }
260  
261  static inline int deg(unsigned int poly)
262  {
263  	/* polynomial degree is the most-significant bit index */
264  	return fls(poly)-1;
265  }
266  
267  static inline int parity(unsigned int x)
268  {
269  	/*
270  	 * public domain code snippet, lifted from
271  	 * http://www-graphics.stanford.edu/~seander/bithacks.html
272  	 */
273  	x ^= x >> 1;
274  	x ^= x >> 2;
275  	x = (x & 0x11111111U) * 0x11111111U;
276  	return (x >> 28) & 1;
277  }
278  
279  /* Galois field basic operations: multiply, divide, inverse, etc. */
280  
281  static inline unsigned int gf_mul(struct bch_control *bch, unsigned int a,
282  				  unsigned int b)
283  {
284  	return (a && b) ? bch->a_pow_tab[mod_s(bch, bch->a_log_tab[a]+
285  					       bch->a_log_tab[b])] : 0;
286  }
287  
288  static inline unsigned int gf_sqr(struct bch_control *bch, unsigned int a)
289  {
290  	return a ? bch->a_pow_tab[mod_s(bch, 2*bch->a_log_tab[a])] : 0;
291  }
292  
293  static inline unsigned int gf_div(struct bch_control *bch, unsigned int a,
294  				  unsigned int b)
295  {
296  	return a ? bch->a_pow_tab[mod_s(bch, bch->a_log_tab[a]+
297  					GF_N(bch)-bch->a_log_tab[b])] : 0;
298  }
299  
300  static inline unsigned int gf_inv(struct bch_control *bch, unsigned int a)
301  {
302  	return bch->a_pow_tab[GF_N(bch)-bch->a_log_tab[a]];
303  }
304  
305  static inline unsigned int a_pow(struct bch_control *bch, int i)
306  {
307  	return bch->a_pow_tab[modulo(bch, i)];
308  }
309  
310  static inline int a_log(struct bch_control *bch, unsigned int x)
311  {
312  	return bch->a_log_tab[x];
313  }
314  
315  static inline int a_ilog(struct bch_control *bch, unsigned int x)
316  {
317  	return mod_s(bch, GF_N(bch)-bch->a_log_tab[x]);
318  }
319  
320  /*
321   * compute 2t syndromes of ecc polynomial, i.e. ecc(a^j) for j=1..2t
322   */
323  static void compute_syndromes(struct bch_control *bch, uint32_t *ecc,
324  			      unsigned int *syn)
325  {
326  	int i, j, s;
327  	unsigned int m;
328  	uint32_t poly;
329  	const int t = GF_T(bch);
330  
331  	s = bch->ecc_bits;
332  
333  	/* make sure extra bits in last ecc word are cleared */
334  	m = ((unsigned int)s) & 31;
335  	if (m)
336  		ecc[s/32] &= ~((1u << (32-m))-1);
337  	memset(syn, 0, 2*t*sizeof(*syn));
338  
339  	/* compute v(a^j) for j=1 .. 2t-1 */
340  	do {
341  		poly = *ecc++;
342  		s -= 32;
343  		while (poly) {
344  			i = deg(poly);
345  			for (j = 0; j < 2*t; j += 2)
346  				syn[j] ^= a_pow(bch, (j+1)*(i+s));
347  
348  			poly ^= (1 << i);
349  		}
350  	} while (s > 0);
351  
352  	/* v(a^(2j)) = v(a^j)^2 */
353  	for (j = 0; j < t; j++)
354  		syn[2*j+1] = gf_sqr(bch, syn[j]);
355  }
356  
357  static void gf_poly_copy(struct gf_poly *dst, struct gf_poly *src)
358  {
359  	memcpy(dst, src, GF_POLY_SZ(src->deg));
360  }
361  
362  static int compute_error_locator_polynomial(struct bch_control *bch,
363  					    const unsigned int *syn)
364  {
365  	const unsigned int t = GF_T(bch);
366  	const unsigned int n = GF_N(bch);
367  	unsigned int i, j, tmp, l, pd = 1, d = syn[0];
368  	struct gf_poly *elp = bch->elp;
369  	struct gf_poly *pelp = bch->poly_2t[0];
370  	struct gf_poly *elp_copy = bch->poly_2t[1];
371  	int k, pp = -1;
372  
373  	memset(pelp, 0, GF_POLY_SZ(2*t));
374  	memset(elp, 0, GF_POLY_SZ(2*t));
375  
376  	pelp->deg = 0;
377  	pelp->c[0] = 1;
378  	elp->deg = 0;
379  	elp->c[0] = 1;
380  
381  	/* use simplified binary Berlekamp-Massey algorithm */
382  	for (i = 0; (i < t) && (elp->deg <= t); i++) {
383  		if (d) {
384  			k = 2*i-pp;
385  			gf_poly_copy(elp_copy, elp);
386  			/* e[i+1](X) = e[i](X)+di*dp^-1*X^2(i-p)*e[p](X) */
387  			tmp = a_log(bch, d)+n-a_log(bch, pd);
388  			for (j = 0; j <= pelp->deg; j++) {
389  				if (pelp->c[j]) {
390  					l = a_log(bch, pelp->c[j]);
391  					elp->c[j+k] ^= a_pow(bch, tmp+l);
392  				}
393  			}
394  			/* compute l[i+1] = max(l[i]->c[l[p]+2*(i-p]) */
395  			tmp = pelp->deg+k;
396  			if (tmp > elp->deg) {
397  				elp->deg = tmp;
398  				gf_poly_copy(pelp, elp_copy);
399  				pd = d;
400  				pp = 2*i;
401  			}
402  		}
403  		/* di+1 = S(2i+3)+elp[i+1].1*S(2i+2)+...+elp[i+1].lS(2i+3-l) */
404  		if (i < t-1) {
405  			d = syn[2*i+2];
406  			for (j = 1; j <= elp->deg; j++)
407  				d ^= gf_mul(bch, elp->c[j], syn[2*i+2-j]);
408  		}
409  	}
410  	dbg("elp=%s\n", gf_poly_str(elp));
411  	return (elp->deg > t) ? -1 : (int)elp->deg;
412  }
413  
414  /*
415   * solve a m x m linear system in GF(2) with an expected number of solutions,
416   * and return the number of found solutions
417   */
418  static int solve_linear_system(struct bch_control *bch, unsigned int *rows,
419  			       unsigned int *sol, int nsol)
420  {
421  	const int m = GF_M(bch);
422  	unsigned int tmp, mask;
423  	int rem, c, r, p, k, param[m];
424  
425  	k = 0;
426  	mask = 1 << m;
427  
428  	/* Gaussian elimination */
429  	for (c = 0; c < m; c++) {
430  		rem = 0;
431  		p = c-k;
432  		/* find suitable row for elimination */
433  		for (r = p; r < m; r++) {
434  			if (rows[r] & mask) {
435  				if (r != p) {
436  					tmp = rows[r];
437  					rows[r] = rows[p];
438  					rows[p] = tmp;
439  				}
440  				rem = r+1;
441  				break;
442  			}
443  		}
444  		if (rem) {
445  			/* perform elimination on remaining rows */
446  			tmp = rows[p];
447  			for (r = rem; r < m; r++) {
448  				if (rows[r] & mask)
449  					rows[r] ^= tmp;
450  			}
451  		} else {
452  			/* elimination not needed, store defective row index */
453  			param[k++] = c;
454  		}
455  		mask >>= 1;
456  	}
457  	/* rewrite system, inserting fake parameter rows */
458  	if (k > 0) {
459  		p = k;
460  		for (r = m-1; r >= 0; r--) {
461  			if ((r > m-1-k) && rows[r])
462  				/* system has no solution */
463  				return 0;
464  
465  			rows[r] = (p && (r == param[p-1])) ?
466  				p--, 1u << (m-r) : rows[r-p];
467  		}
468  	}
469  
470  	if (nsol != (1 << k))
471  		/* unexpected number of solutions */
472  		return 0;
473  
474  	for (p = 0; p < nsol; p++) {
475  		/* set parameters for p-th solution */
476  		for (c = 0; c < k; c++)
477  			rows[param[c]] = (rows[param[c]] & ~1)|((p >> c) & 1);
478  
479  		/* compute unique solution */
480  		tmp = 0;
481  		for (r = m-1; r >= 0; r--) {
482  			mask = rows[r] & (tmp|1);
483  			tmp |= parity(mask) << (m-r);
484  		}
485  		sol[p] = tmp >> 1;
486  	}
487  	return nsol;
488  }
489  
490  /*
491   * this function builds and solves a linear system for finding roots of a degree
492   * 4 affine monic polynomial X^4+aX^2+bX+c over GF(2^m).
493   */
494  static int find_affine4_roots(struct bch_control *bch, unsigned int a,
495  			      unsigned int b, unsigned int c,
496  			      unsigned int *roots)
497  {
498  	int i, j, k;
499  	const int m = GF_M(bch);
500  	unsigned int mask = 0xff, t, rows[16] = {0,};
501  
502  	j = a_log(bch, b);
503  	k = a_log(bch, a);
504  	rows[0] = c;
505  
506  	/* buid linear system to solve X^4+aX^2+bX+c = 0 */
507  	for (i = 0; i < m; i++) {
508  		rows[i+1] = bch->a_pow_tab[4*i]^
509  			(a ? bch->a_pow_tab[mod_s(bch, k)] : 0)^
510  			(b ? bch->a_pow_tab[mod_s(bch, j)] : 0);
511  		j++;
512  		k += 2;
513  	}
514  	/*
515  	 * transpose 16x16 matrix before passing it to linear solver
516  	 * warning: this code assumes m < 16
517  	 */
518  	for (j = 8; j != 0; j >>= 1, mask ^= (mask << j)) {
519  		for (k = 0; k < 16; k = (k+j+1) & ~j) {
520  			t = ((rows[k] >> j)^rows[k+j]) & mask;
521  			rows[k] ^= (t << j);
522  			rows[k+j] ^= t;
523  		}
524  	}
525  	return solve_linear_system(bch, rows, roots, 4);
526  }
527  
528  /*
529   * compute root r of a degree 1 polynomial over GF(2^m) (returned as log(1/r))
530   */
531  static int find_poly_deg1_roots(struct bch_control *bch, struct gf_poly *poly,
532  				unsigned int *roots)
533  {
534  	int n = 0;
535  
536  	if (poly->c[0])
537  		/* poly[X] = bX+c with c!=0, root=c/b */
538  		roots[n++] = mod_s(bch, GF_N(bch)-bch->a_log_tab[poly->c[0]]+
539  				   bch->a_log_tab[poly->c[1]]);
540  	return n;
541  }
542  
543  /*
544   * compute roots of a degree 2 polynomial over GF(2^m)
545   */
546  static int find_poly_deg2_roots(struct bch_control *bch, struct gf_poly *poly,
547  				unsigned int *roots)
548  {
549  	int n = 0, i, l0, l1, l2;
550  	unsigned int u, v, r;
551  
552  	if (poly->c[0] && poly->c[1]) {
553  
554  		l0 = bch->a_log_tab[poly->c[0]];
555  		l1 = bch->a_log_tab[poly->c[1]];
556  		l2 = bch->a_log_tab[poly->c[2]];
557  
558  		/* using z=a/bX, transform aX^2+bX+c into z^2+z+u (u=ac/b^2) */
559  		u = a_pow(bch, l0+l2+2*(GF_N(bch)-l1));
560  		/*
561  		 * let u = sum(li.a^i) i=0..m-1; then compute r = sum(li.xi):
562  		 * r^2+r = sum(li.(xi^2+xi)) = sum(li.(a^i+Tr(a^i).a^k)) =
563  		 * u + sum(li.Tr(a^i).a^k) = u+a^k.Tr(sum(li.a^i)) = u+a^k.Tr(u)
564  		 * i.e. r and r+1 are roots iff Tr(u)=0
565  		 */
566  		r = 0;
567  		v = u;
568  		while (v) {
569  			i = deg(v);
570  			r ^= bch->xi_tab[i];
571  			v ^= (1 << i);
572  		}
573  		/* verify root */
574  		if ((gf_sqr(bch, r)^r) == u) {
575  			/* reverse z=a/bX transformation and compute log(1/r) */
576  			roots[n++] = modulo(bch, 2*GF_N(bch)-l1-
577  					    bch->a_log_tab[r]+l2);
578  			roots[n++] = modulo(bch, 2*GF_N(bch)-l1-
579  					    bch->a_log_tab[r^1]+l2);
580  		}
581  	}
582  	return n;
583  }
584  
585  /*
586   * compute roots of a degree 3 polynomial over GF(2^m)
587   */
588  static int find_poly_deg3_roots(struct bch_control *bch, struct gf_poly *poly,
589  				unsigned int *roots)
590  {
591  	int i, n = 0;
592  	unsigned int a, b, c, a2, b2, c2, e3, tmp[4];
593  
594  	if (poly->c[0]) {
595  		/* transform polynomial into monic X^3 + a2X^2 + b2X + c2 */
596  		e3 = poly->c[3];
597  		c2 = gf_div(bch, poly->c[0], e3);
598  		b2 = gf_div(bch, poly->c[1], e3);
599  		a2 = gf_div(bch, poly->c[2], e3);
600  
601  		/* (X+a2)(X^3+a2X^2+b2X+c2) = X^4+aX^2+bX+c (affine) */
602  		c = gf_mul(bch, a2, c2);           /* c = a2c2      */
603  		b = gf_mul(bch, a2, b2)^c2;        /* b = a2b2 + c2 */
604  		a = gf_sqr(bch, a2)^b2;            /* a = a2^2 + b2 */
605  
606  		/* find the 4 roots of this affine polynomial */
607  		if (find_affine4_roots(bch, a, b, c, tmp) == 4) {
608  			/* remove a2 from final list of roots */
609  			for (i = 0; i < 4; i++) {
610  				if (tmp[i] != a2)
611  					roots[n++] = a_ilog(bch, tmp[i]);
612  			}
613  		}
614  	}
615  	return n;
616  }
617  
618  /*
619   * compute roots of a degree 4 polynomial over GF(2^m)
620   */
621  static int find_poly_deg4_roots(struct bch_control *bch, struct gf_poly *poly,
622  				unsigned int *roots)
623  {
624  	int i, l, n = 0;
625  	unsigned int a, b, c, d, e = 0, f, a2, b2, c2, e4;
626  
627  	if (poly->c[0] == 0)
628  		return 0;
629  
630  	/* transform polynomial into monic X^4 + aX^3 + bX^2 + cX + d */
631  	e4 = poly->c[4];
632  	d = gf_div(bch, poly->c[0], e4);
633  	c = gf_div(bch, poly->c[1], e4);
634  	b = gf_div(bch, poly->c[2], e4);
635  	a = gf_div(bch, poly->c[3], e4);
636  
637  	/* use Y=1/X transformation to get an affine polynomial */
638  	if (a) {
639  		/* first, eliminate cX by using z=X+e with ae^2+c=0 */
640  		if (c) {
641  			/* compute e such that e^2 = c/a */
642  			f = gf_div(bch, c, a);
643  			l = a_log(bch, f);
644  			l += (l & 1) ? GF_N(bch) : 0;
645  			e = a_pow(bch, l/2);
646  			/*
647  			 * use transformation z=X+e:
648  			 * z^4+e^4 + a(z^3+ez^2+e^2z+e^3) + b(z^2+e^2) +cz+ce+d
649  			 * z^4 + az^3 + (ae+b)z^2 + (ae^2+c)z+e^4+be^2+ae^3+ce+d
650  			 * z^4 + az^3 + (ae+b)z^2 + e^4+be^2+d
651  			 * z^4 + az^3 +     b'z^2 + d'
652  			 */
653  			d = a_pow(bch, 2*l)^gf_mul(bch, b, f)^d;
654  			b = gf_mul(bch, a, e)^b;
655  		}
656  		/* now, use Y=1/X to get Y^4 + b/dY^2 + a/dY + 1/d */
657  		if (d == 0)
658  			/* assume all roots have multiplicity 1 */
659  			return 0;
660  
661  		c2 = gf_inv(bch, d);
662  		b2 = gf_div(bch, a, d);
663  		a2 = gf_div(bch, b, d);
664  	} else {
665  		/* polynomial is already affine */
666  		c2 = d;
667  		b2 = c;
668  		a2 = b;
669  	}
670  	/* find the 4 roots of this affine polynomial */
671  	if (find_affine4_roots(bch, a2, b2, c2, roots) == 4) {
672  		for (i = 0; i < 4; i++) {
673  			/* post-process roots (reverse transformations) */
674  			f = a ? gf_inv(bch, roots[i]) : roots[i];
675  			roots[i] = a_ilog(bch, f^e);
676  		}
677  		n = 4;
678  	}
679  	return n;
680  }
681  
682  /*
683   * build monic, log-based representation of a polynomial
684   */
685  static void gf_poly_logrep(struct bch_control *bch,
686  			   const struct gf_poly *a, int *rep)
687  {
688  	int i, d = a->deg, l = GF_N(bch)-a_log(bch, a->c[a->deg]);
689  
690  	/* represent 0 values with -1; warning, rep[d] is not set to 1 */
691  	for (i = 0; i < d; i++)
692  		rep[i] = a->c[i] ? mod_s(bch, a_log(bch, a->c[i])+l) : -1;
693  }
694  
695  /*
696   * compute polynomial Euclidean division remainder in GF(2^m)[X]
697   */
698  static void gf_poly_mod(struct bch_control *bch, struct gf_poly *a,
699  			const struct gf_poly *b, int *rep)
700  {
701  	int la, p, m;
702  	unsigned int i, j, *c = a->c;
703  	const unsigned int d = b->deg;
704  
705  	if (a->deg < d)
706  		return;
707  
708  	/* reuse or compute log representation of denominator */
709  	if (!rep) {
710  		rep = bch->cache;
711  		gf_poly_logrep(bch, b, rep);
712  	}
713  
714  	for (j = a->deg; j >= d; j--) {
715  		if (c[j]) {
716  			la = a_log(bch, c[j]);
717  			p = j-d;
718  			for (i = 0; i < d; i++, p++) {
719  				m = rep[i];
720  				if (m >= 0)
721  					c[p] ^= bch->a_pow_tab[mod_s(bch,
722  								     m+la)];
723  			}
724  		}
725  	}
726  	a->deg = d-1;
727  	while (!c[a->deg] && a->deg)
728  		a->deg--;
729  }
730  
731  /*
732   * compute polynomial Euclidean division quotient in GF(2^m)[X]
733   */
734  static void gf_poly_div(struct bch_control *bch, struct gf_poly *a,
735  			const struct gf_poly *b, struct gf_poly *q)
736  {
737  	if (a->deg >= b->deg) {
738  		q->deg = a->deg-b->deg;
739  		/* compute a mod b (modifies a) */
740  		gf_poly_mod(bch, a, b, NULL);
741  		/* quotient is stored in upper part of polynomial a */
742  		memcpy(q->c, &a->c[b->deg], (1+q->deg)*sizeof(unsigned int));
743  	} else {
744  		q->deg = 0;
745  		q->c[0] = 0;
746  	}
747  }
748  
749  /*
750   * compute polynomial GCD (Greatest Common Divisor) in GF(2^m)[X]
751   */
752  static struct gf_poly *gf_poly_gcd(struct bch_control *bch, struct gf_poly *a,
753  				   struct gf_poly *b)
754  {
755  	struct gf_poly *tmp;
756  
757  	dbg("gcd(%s,%s)=", gf_poly_str(a), gf_poly_str(b));
758  
759  	if (a->deg < b->deg) {
760  		tmp = b;
761  		b = a;
762  		a = tmp;
763  	}
764  
765  	while (b->deg > 0) {
766  		gf_poly_mod(bch, a, b, NULL);
767  		tmp = b;
768  		b = a;
769  		a = tmp;
770  	}
771  
772  	dbg("%s\n", gf_poly_str(a));
773  
774  	return a;
775  }
776  
777  /*
778   * Given a polynomial f and an integer k, compute Tr(a^kX) mod f
779   * This is used in Berlekamp Trace algorithm for splitting polynomials
780   */
781  static void compute_trace_bk_mod(struct bch_control *bch, int k,
782  				 const struct gf_poly *f, struct gf_poly *z,
783  				 struct gf_poly *out)
784  {
785  	const int m = GF_M(bch);
786  	int i, j;
787  
788  	/* z contains z^2j mod f */
789  	z->deg = 1;
790  	z->c[0] = 0;
791  	z->c[1] = bch->a_pow_tab[k];
792  
793  	out->deg = 0;
794  	memset(out, 0, GF_POLY_SZ(f->deg));
795  
796  	/* compute f log representation only once */
797  	gf_poly_logrep(bch, f, bch->cache);
798  
799  	for (i = 0; i < m; i++) {
800  		/* add a^(k*2^i)(z^(2^i) mod f) and compute (z^(2^i) mod f)^2 */
801  		for (j = z->deg; j >= 0; j--) {
802  			out->c[j] ^= z->c[j];
803  			z->c[2*j] = gf_sqr(bch, z->c[j]);
804  			z->c[2*j+1] = 0;
805  		}
806  		if (z->deg > out->deg)
807  			out->deg = z->deg;
808  
809  		if (i < m-1) {
810  			z->deg *= 2;
811  			/* z^(2(i+1)) mod f = (z^(2^i) mod f)^2 mod f */
812  			gf_poly_mod(bch, z, f, bch->cache);
813  		}
814  	}
815  	while (!out->c[out->deg] && out->deg)
816  		out->deg--;
817  
818  	dbg("Tr(a^%d.X) mod f = %s\n", k, gf_poly_str(out));
819  }
820  
821  /*
822   * factor a polynomial using Berlekamp Trace algorithm (BTA)
823   */
824  static void factor_polynomial(struct bch_control *bch, int k, struct gf_poly *f,
825  			      struct gf_poly **g, struct gf_poly **h)
826  {
827  	struct gf_poly *f2 = bch->poly_2t[0];
828  	struct gf_poly *q  = bch->poly_2t[1];
829  	struct gf_poly *tk = bch->poly_2t[2];
830  	struct gf_poly *z  = bch->poly_2t[3];
831  	struct gf_poly *gcd;
832  
833  	dbg("factoring %s...\n", gf_poly_str(f));
834  
835  	*g = f;
836  	*h = NULL;
837  
838  	/* tk = Tr(a^k.X) mod f */
839  	compute_trace_bk_mod(bch, k, f, z, tk);
840  
841  	if (tk->deg > 0) {
842  		/* compute g = gcd(f, tk) (destructive operation) */
843  		gf_poly_copy(f2, f);
844  		gcd = gf_poly_gcd(bch, f2, tk);
845  		if (gcd->deg < f->deg) {
846  			/* compute h=f/gcd(f,tk); this will modify f and q */
847  			gf_poly_div(bch, f, gcd, q);
848  			/* store g and h in-place (clobbering f) */
849  			*h = &((struct gf_poly_deg1 *)f)[gcd->deg].poly;
850  			gf_poly_copy(*g, gcd);
851  			gf_poly_copy(*h, q);
852  		}
853  	}
854  }
855  
856  /*
857   * find roots of a polynomial, using BTZ algorithm; see the beginning of this
858   * file for details
859   */
860  static int find_poly_roots(struct bch_control *bch, unsigned int k,
861  			   struct gf_poly *poly, unsigned int *roots)
862  {
863  	int cnt;
864  	struct gf_poly *f1, *f2;
865  
866  	switch (poly->deg) {
867  		/* handle low degree polynomials with ad hoc techniques */
868  	case 1:
869  		cnt = find_poly_deg1_roots(bch, poly, roots);
870  		break;
871  	case 2:
872  		cnt = find_poly_deg2_roots(bch, poly, roots);
873  		break;
874  	case 3:
875  		cnt = find_poly_deg3_roots(bch, poly, roots);
876  		break;
877  	case 4:
878  		cnt = find_poly_deg4_roots(bch, poly, roots);
879  		break;
880  	default:
881  		/* factor polynomial using Berlekamp Trace Algorithm (BTA) */
882  		cnt = 0;
883  		if (poly->deg && (k <= GF_M(bch))) {
884  			factor_polynomial(bch, k, poly, &f1, &f2);
885  			if (f1)
886  				cnt += find_poly_roots(bch, k+1, f1, roots);
887  			if (f2)
888  				cnt += find_poly_roots(bch, k+1, f2, roots+cnt);
889  		}
890  		break;
891  	}
892  	return cnt;
893  }
894  
895  #if defined(USE_CHIEN_SEARCH)
896  /*
897   * exhaustive root search (Chien) implementation - not used, included only for
898   * reference/comparison tests
899   */
900  static int chien_search(struct bch_control *bch, unsigned int len,
901  			struct gf_poly *p, unsigned int *roots)
902  {
903  	int m;
904  	unsigned int i, j, syn, syn0, count = 0;
905  	const unsigned int k = 8*len+bch->ecc_bits;
906  
907  	/* use a log-based representation of polynomial */
908  	gf_poly_logrep(bch, p, bch->cache);
909  	bch->cache[p->deg] = 0;
910  	syn0 = gf_div(bch, p->c[0], p->c[p->deg]);
911  
912  	for (i = GF_N(bch)-k+1; i <= GF_N(bch); i++) {
913  		/* compute elp(a^i) */
914  		for (j = 1, syn = syn0; j <= p->deg; j++) {
915  			m = bch->cache[j];
916  			if (m >= 0)
917  				syn ^= a_pow(bch, m+j*i);
918  		}
919  		if (syn == 0) {
920  			roots[count++] = GF_N(bch)-i;
921  			if (count == p->deg)
922  				break;
923  		}
924  	}
925  	return (count == p->deg) ? count : 0;
926  }
927  #define find_poly_roots(_p, _k, _elp, _loc) chien_search(_p, len, _elp, _loc)
928  #endif /* USE_CHIEN_SEARCH */
929  
930  /**
931   * decode_bch - decode received codeword and find bit error locations
932   * @bch:      BCH control structure
933   * @data:     received data, ignored if @calc_ecc is provided
934   * @len:      data length in bytes, must always be provided
935   * @recv_ecc: received ecc, if NULL then assume it was XORed in @calc_ecc
936   * @calc_ecc: calculated ecc, if NULL then calc_ecc is computed from @data
937   * @syn:      hw computed syndrome data (if NULL, syndrome is calculated)
938   * @errloc:   output array of error locations
939   *
940   * Returns:
941   *  The number of errors found, or -EBADMSG if decoding failed, or -EINVAL if
942   *  invalid parameters were provided
943   *
944   * Depending on the available hw BCH support and the need to compute @calc_ecc
945   * separately (using encode_bch()), this function should be called with one of
946   * the following parameter configurations -
947   *
948   * by providing @data and @recv_ecc only:
949   *   decode_bch(@bch, @data, @len, @recv_ecc, NULL, NULL, @errloc)
950   *
951   * by providing @recv_ecc and @calc_ecc:
952   *   decode_bch(@bch, NULL, @len, @recv_ecc, @calc_ecc, NULL, @errloc)
953   *
954   * by providing ecc = recv_ecc XOR calc_ecc:
955   *   decode_bch(@bch, NULL, @len, NULL, ecc, NULL, @errloc)
956   *
957   * by providing syndrome results @syn:
958   *   decode_bch(@bch, NULL, @len, NULL, NULL, @syn, @errloc)
959   *
960   * Once decode_bch() has successfully returned with a positive value, error
961   * locations returned in array @errloc should be interpreted as follows -
962   *
963   * if (errloc[n] >= 8*len), then n-th error is located in ecc (no need for
964   * data correction)
965   *
966   * if (errloc[n] < 8*len), then n-th error is located in data and can be
967   * corrected with statement data[errloc[n]/8] ^= 1 << (errloc[n] % 8);
968   *
969   * Note that this function does not perform any data correction by itself, it
970   * merely indicates error locations.
971   */
972  int decode_bch(struct bch_control *bch, const uint8_t *data, unsigned int len,
973  	       const uint8_t *recv_ecc, const uint8_t *calc_ecc,
974  	       const unsigned int *syn, unsigned int *errloc)
975  {
976  	const unsigned int ecc_words = BCH_ECC_WORDS(bch);
977  	unsigned int nbits;
978  	int i, err, nroots;
979  	uint32_t sum;
980  
981  	/* sanity check: make sure data length can be handled */
982  	if (8*len > (bch->n-bch->ecc_bits))
983  		return -EINVAL;
984  
985  	/* if caller does not provide syndromes, compute them */
986  	if (!syn) {
987  		if (!calc_ecc) {
988  			/* compute received data ecc into an internal buffer */
989  			if (!data || !recv_ecc)
990  				return -EINVAL;
991  			encode_bch(bch, data, len, NULL);
992  		} else {
993  			/* load provided calculated ecc */
994  			load_ecc8(bch, bch->ecc_buf, calc_ecc);
995  		}
996  		/* load received ecc or assume it was XORed in calc_ecc */
997  		if (recv_ecc) {
998  			load_ecc8(bch, bch->ecc_buf2, recv_ecc);
999  			/* XOR received and calculated ecc */
1000  			for (i = 0, sum = 0; i < (int)ecc_words; i++) {
1001  				bch->ecc_buf[i] ^= bch->ecc_buf2[i];
1002  				sum |= bch->ecc_buf[i];
1003  			}
1004  			if (!sum)
1005  				/* no error found */
1006  				return 0;
1007  		}
1008  		compute_syndromes(bch, bch->ecc_buf, bch->syn);
1009  		syn = bch->syn;
1010  	}
1011  
1012  	err = compute_error_locator_polynomial(bch, syn);
1013  	if (err > 0) {
1014  		nroots = find_poly_roots(bch, 1, bch->elp, errloc);
1015  		if (err != nroots)
1016  			err = -1;
1017  	}
1018  	if (err > 0) {
1019  		/* post-process raw error locations for easier correction */
1020  		nbits = (len*8)+bch->ecc_bits;
1021  		for (i = 0; i < err; i++) {
1022  			if (errloc[i] >= nbits) {
1023  				err = -1;
1024  				break;
1025  			}
1026  			errloc[i] = nbits-1-errloc[i];
1027  			errloc[i] = (errloc[i] & ~7)|(7-(errloc[i] & 7));
1028  		}
1029  	}
1030  	return (err >= 0) ? err : -EBADMSG;
1031  }
1032  
1033  /*
1034   * generate Galois field lookup tables
1035   */
1036  static int build_gf_tables(struct bch_control *bch, unsigned int poly)
1037  {
1038  	unsigned int i, x = 1;
1039  	const unsigned int k = 1 << deg(poly);
1040  
1041  	/* primitive polynomial must be of degree m */
1042  	if (k != (1u << GF_M(bch)))
1043  		return -1;
1044  
1045  	for (i = 0; i < GF_N(bch); i++) {
1046  		bch->a_pow_tab[i] = x;
1047  		bch->a_log_tab[x] = i;
1048  		if (i && (x == 1))
1049  			/* polynomial is not primitive (a^i=1 with 0<i<2^m-1) */
1050  			return -1;
1051  		x <<= 1;
1052  		if (x & k)
1053  			x ^= poly;
1054  	}
1055  	bch->a_pow_tab[GF_N(bch)] = 1;
1056  	bch->a_log_tab[0] = 0;
1057  
1058  	return 0;
1059  }
1060  
1061  /*
1062   * compute generator polynomial remainder tables for fast encoding
1063   */
1064  static void build_mod8_tables(struct bch_control *bch, const uint32_t *g)
1065  {
1066  	int i, j, b, d;
1067  	uint32_t data, hi, lo, *tab;
1068  	const int l = BCH_ECC_WORDS(bch);
1069  	const int plen = DIV_ROUND_UP(bch->ecc_bits+1, 32);
1070  	const int ecclen = DIV_ROUND_UP(bch->ecc_bits, 32);
1071  
1072  	memset(bch->mod8_tab, 0, 4*256*l*sizeof(*bch->mod8_tab));
1073  
1074  	for (i = 0; i < 256; i++) {
1075  		/* p(X)=i is a small polynomial of weight <= 8 */
1076  		for (b = 0; b < 4; b++) {
1077  			/* we want to compute (p(X).X^(8*b+deg(g))) mod g(X) */
1078  			tab = bch->mod8_tab + (b*256+i)*l;
1079  			data = i << (8*b);
1080  			while (data) {
1081  				d = deg(data);
1082  				/* subtract X^d.g(X) from p(X).X^(8*b+deg(g)) */
1083  				data ^= g[0] >> (31-d);
1084  				for (j = 0; j < ecclen; j++) {
1085  					hi = (d < 31) ? g[j] << (d+1) : 0;
1086  					lo = (j+1 < plen) ?
1087  						g[j+1] >> (31-d) : 0;
1088  					tab[j] ^= hi|lo;
1089  				}
1090  			}
1091  		}
1092  	}
1093  }
1094  
1095  /*
1096   * build a base for factoring degree 2 polynomials
1097   */
1098  static int build_deg2_base(struct bch_control *bch)
1099  {
1100  	const int m = GF_M(bch);
1101  	int i, j, r;
1102  	unsigned int sum, x, y, remaining, ak = 0, xi[m];
1103  
1104  	/* find k s.t. Tr(a^k) = 1 and 0 <= k < m */
1105  	for (i = 0; i < m; i++) {
1106  		for (j = 0, sum = 0; j < m; j++)
1107  			sum ^= a_pow(bch, i*(1 << j));
1108  
1109  		if (sum) {
1110  			ak = bch->a_pow_tab[i];
1111  			break;
1112  		}
1113  	}
1114  	/* find xi, i=0..m-1 such that xi^2+xi = a^i+Tr(a^i).a^k */
1115  	remaining = m;
1116  	memset(xi, 0, sizeof(xi));
1117  
1118  	for (x = 0; (x <= GF_N(bch)) && remaining; x++) {
1119  		y = gf_sqr(bch, x)^x;
1120  		for (i = 0; i < 2; i++) {
1121  			r = a_log(bch, y);
1122  			if (y && (r < m) && !xi[r]) {
1123  				bch->xi_tab[r] = x;
1124  				xi[r] = 1;
1125  				remaining--;
1126  				dbg("x%d = %x\n", r, x);
1127  				break;
1128  			}
1129  			y ^= ak;
1130  		}
1131  	}
1132  	/* should not happen but check anyway */
1133  	return remaining ? -1 : 0;
1134  }
1135  
1136  static void *bch_alloc(size_t size, int *err)
1137  {
1138  	void *ptr;
1139  
1140  	ptr = kmalloc(size, GFP_KERNEL);
1141  	if (ptr == NULL)
1142  		*err = 1;
1143  	return ptr;
1144  }
1145  
1146  /*
1147   * compute generator polynomial for given (m,t) parameters.
1148   */
1149  static uint32_t *compute_generator_polynomial(struct bch_control *bch)
1150  {
1151  	const unsigned int m = GF_M(bch);
1152  	const unsigned int t = GF_T(bch);
1153  	int n, err = 0;
1154  	unsigned int i, j, nbits, r, word, *roots;
1155  	struct gf_poly *g;
1156  	uint32_t *genpoly;
1157  
1158  	g = bch_alloc(GF_POLY_SZ(m*t), &err);
1159  	roots = bch_alloc((bch->n+1)*sizeof(*roots), &err);
1160  	genpoly = bch_alloc(DIV_ROUND_UP(m*t+1, 32)*sizeof(*genpoly), &err);
1161  
1162  	if (err) {
1163  		kfree(genpoly);
1164  		genpoly = NULL;
1165  		goto finish;
1166  	}
1167  
1168  	/* enumerate all roots of g(X) */
1169  	memset(roots , 0, (bch->n+1)*sizeof(*roots));
1170  	for (i = 0; i < t; i++) {
1171  		for (j = 0, r = 2*i+1; j < m; j++) {
1172  			roots[r] = 1;
1173  			r = mod_s(bch, 2*r);
1174  		}
1175  	}
1176  	/* build generator polynomial g(X) */
1177  	g->deg = 0;
1178  	g->c[0] = 1;
1179  	for (i = 0; i < GF_N(bch); i++) {
1180  		if (roots[i]) {
1181  			/* multiply g(X) by (X+root) */
1182  			r = bch->a_pow_tab[i];
1183  			g->c[g->deg+1] = 1;
1184  			for (j = g->deg; j > 0; j--)
1185  				g->c[j] = gf_mul(bch, g->c[j], r)^g->c[j-1];
1186  
1187  			g->c[0] = gf_mul(bch, g->c[0], r);
1188  			g->deg++;
1189  		}
1190  	}
1191  	/* store left-justified binary representation of g(X) */
1192  	n = g->deg+1;
1193  	i = 0;
1194  
1195  	while (n > 0) {
1196  		nbits = (n > 32) ? 32 : n;
1197  		for (j = 0, word = 0; j < nbits; j++) {
1198  			if (g->c[n-1-j])
1199  				word |= 1u << (31-j);
1200  		}
1201  		genpoly[i++] = word;
1202  		n -= nbits;
1203  	}
1204  	bch->ecc_bits = g->deg;
1205  
1206  finish:
1207  	kfree(g);
1208  	kfree(roots);
1209  
1210  	return genpoly;
1211  }
1212  
1213  /**
1214   * init_bch - initialize a BCH encoder/decoder
1215   * @m:          Galois field order, should be in the range 5-15
1216   * @t:          maximum error correction capability, in bits
1217   * @prim_poly:  user-provided primitive polynomial (or 0 to use default)
1218   *
1219   * Returns:
1220   *  a newly allocated BCH control structure if successful, NULL otherwise
1221   *
1222   * This initialization can take some time, as lookup tables are built for fast
1223   * encoding/decoding; make sure not to call this function from a time critical
1224   * path. Usually, init_bch() should be called on module/driver init and
1225   * free_bch() should be called to release memory on exit.
1226   *
1227   * You may provide your own primitive polynomial of degree @m in argument
1228   * @prim_poly, or let init_bch() use its default polynomial.
1229   *
1230   * Once init_bch() has successfully returned a pointer to a newly allocated
1231   * BCH control structure, ecc length in bytes is given by member @ecc_bytes of
1232   * the structure.
1233   */
1234  struct bch_control *init_bch(int m, int t, unsigned int prim_poly)
1235  {
1236  	int err = 0;
1237  	unsigned int i, words;
1238  	uint32_t *genpoly;
1239  	struct bch_control *bch = NULL;
1240  
1241  	const int min_m = 5;
1242  	const int max_m = 15;
1243  
1244  	/* default primitive polynomials */
1245  	static const unsigned int prim_poly_tab[] = {
1246  		0x25, 0x43, 0x83, 0x11d, 0x211, 0x409, 0x805, 0x1053, 0x201b,
1247  		0x402b, 0x8003,
1248  	};
1249  
1250  #if defined(CONFIG_BCH_CONST_PARAMS)
1251  	if ((m != (CONFIG_BCH_CONST_M)) || (t != (CONFIG_BCH_CONST_T))) {
1252  		printk(KERN_ERR "bch encoder/decoder was configured to support "
1253  		       "parameters m=%d, t=%d only!\n",
1254  		       CONFIG_BCH_CONST_M, CONFIG_BCH_CONST_T);
1255  		goto fail;
1256  	}
1257  #endif
1258  	if ((m < min_m) || (m > max_m))
1259  		/*
1260  		 * values of m greater than 15 are not currently supported;
1261  		 * supporting m > 15 would require changing table base type
1262  		 * (uint16_t) and a small patch in matrix transposition
1263  		 */
1264  		goto fail;
1265  
1266  	/* sanity checks */
1267  	if ((t < 1) || (m*t >= ((1 << m)-1)))
1268  		/* invalid t value */
1269  		goto fail;
1270  
1271  	/* select a primitive polynomial for generating GF(2^m) */
1272  	if (prim_poly == 0)
1273  		prim_poly = prim_poly_tab[m-min_m];
1274  
1275  	bch = kzalloc(sizeof(*bch), GFP_KERNEL);
1276  	if (bch == NULL)
1277  		goto fail;
1278  
1279  	bch->m = m;
1280  	bch->t = t;
1281  	bch->n = (1 << m)-1;
1282  	words  = DIV_ROUND_UP(m*t, 32);
1283  	bch->ecc_bytes = DIV_ROUND_UP(m*t, 8);
1284  	bch->a_pow_tab = bch_alloc((1+bch->n)*sizeof(*bch->a_pow_tab), &err);
1285  	bch->a_log_tab = bch_alloc((1+bch->n)*sizeof(*bch->a_log_tab), &err);
1286  	bch->mod8_tab  = bch_alloc(words*1024*sizeof(*bch->mod8_tab), &err);
1287  	bch->ecc_buf   = bch_alloc(words*sizeof(*bch->ecc_buf), &err);
1288  	bch->ecc_buf2  = bch_alloc(words*sizeof(*bch->ecc_buf2), &err);
1289  	bch->xi_tab    = bch_alloc(m*sizeof(*bch->xi_tab), &err);
1290  	bch->syn       = bch_alloc(2*t*sizeof(*bch->syn), &err);
1291  	bch->cache     = bch_alloc(2*t*sizeof(*bch->cache), &err);
1292  	bch->elp       = bch_alloc((t+1)*sizeof(struct gf_poly_deg1), &err);
1293  
1294  	for (i = 0; i < ARRAY_SIZE(bch->poly_2t); i++)
1295  		bch->poly_2t[i] = bch_alloc(GF_POLY_SZ(2*t), &err);
1296  
1297  	if (err)
1298  		goto fail;
1299  
1300  	err = build_gf_tables(bch, prim_poly);
1301  	if (err)
1302  		goto fail;
1303  
1304  	/* use generator polynomial for computing encoding tables */
1305  	genpoly = compute_generator_polynomial(bch);
1306  	if (genpoly == NULL)
1307  		goto fail;
1308  
1309  	build_mod8_tables(bch, genpoly);
1310  	kfree(genpoly);
1311  
1312  	err = build_deg2_base(bch);
1313  	if (err)
1314  		goto fail;
1315  
1316  	return bch;
1317  
1318  fail:
1319  	free_bch(bch);
1320  	return NULL;
1321  }
1322  
1323  /**
1324   *  free_bch - free the BCH control structure
1325   *  @bch:    BCH control structure to release
1326   */
1327  void free_bch(struct bch_control *bch)
1328  {
1329  	unsigned int i;
1330  
1331  	if (bch) {
1332  		kfree(bch->a_pow_tab);
1333  		kfree(bch->a_log_tab);
1334  		kfree(bch->mod8_tab);
1335  		kfree(bch->ecc_buf);
1336  		kfree(bch->ecc_buf2);
1337  		kfree(bch->xi_tab);
1338  		kfree(bch->syn);
1339  		kfree(bch->cache);
1340  		kfree(bch->elp);
1341  
1342  		for (i = 0; i < ARRAY_SIZE(bch->poly_2t); i++)
1343  			kfree(bch->poly_2t[i]);
1344  
1345  		kfree(bch);
1346  	}
1347  }
1348