3 * Reed-Solomon encoding and decoding,
4 * by Phil Karn (karn@ka9q.ampr.org) September 1996
5 * Copyright 1999 Phil Karn, KA9Q
6 * Separate CCSDS version create Dec 1998, merged into this version May 1999
8 * This file is derived from my generic RS encoder/decoder, which is
9 * in turn based on the program "new_rs_erasures.c" by Robert
10 * Morelos-Zaragoza (robert@spectra.eng.hawaii.edu) and Hari Thirumoorthy
11 * (harit@spectra.eng.hawaii.edu), Aug 1995
13 * Wireshark - Network traffic analyzer
14 * By Gerald Combs <gerald@wireshark.org>
15 * Copyright 1998 Gerald Combs
17 * SPDX-License-Identifier: GPL-2.0-or-later
20 #include "reedsolomon.h"
21 #include <wsutil/ws_printf.h> /* ws_debug_printf */
24 /* CCSDS field generator polynomial: 1+x+x^2+x^7+x^8 */
25 int Pp[MM+1] = { 1, 1, 1, 0, 0, 0, 0, 1, 1 };
28 /* MM, KK, B0, PRIM are user-defined in rs.h */
30 /* Primitive polynomials - see Lin & Costello, Appendix A,
31 * and Lee & Messerschmitt, p. 453.
33 #if(MM == 2)/* Admittedly silly */
34 int Pp[MM+1] = { 1, 1, 1 };
38 int Pp[MM+1] = { 1, 1, 0, 1 };
42 int Pp[MM+1] = { 1, 1, 0, 0, 1 };
46 int Pp[MM+1] = { 1, 0, 1, 0, 0, 1 };
50 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 1 };
54 int Pp[MM+1] = { 1, 0, 0, 1, 0, 0, 0, 1 };
57 /* 1+x^2+x^3+x^4+x^8 */
58 int Pp[MM+1] = { 1, 0, 1, 1, 1, 0, 0, 0, 1 };
62 int Pp[MM+1] = { 1, 0, 0, 0, 1, 0, 0, 0, 0, 1 };
66 int Pp[MM+1] = { 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1 };
70 int Pp[MM+1] = { 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
73 /* 1+x+x^4+x^6+x^12 */
74 int Pp[MM+1] = { 1, 1, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 1 };
77 /* 1+x+x^3+x^4+x^13 */
78 int Pp[MM+1] = { 1, 1, 0, 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
81 /* 1+x+x^6+x^10+x^14 */
82 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 0, 1 };
86 int Pp[MM+1] = { 1, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1 };
89 /* 1+x+x^3+x^12+x^16 */
90 int Pp[MM+1] = { 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1 };
93 #error "Either CCSDS must be defined, or MM must be set in range 2-16"
98 #ifdef STANDARD_ORDER /* first byte transmitted is index of x**(KK-1) in message poly*/
99 /* definitions used in the encode routine*/
100 #define MESSAGE(i) data[KK-(i)-1]
101 #define REMAINDER(i) bb[NN-KK-(i)-1]
102 /* definitions used in the decode routine*/
103 #define RECEIVED(i) data[NN-1-(i)]
104 #define ERAS_INDEX(i) (NN-1-eras_pos[i])
105 #define INDEX_TO_POS(i) (NN-1-(i))
106 #else /* first byte transmitted is index of x**0 in message polynomial*/
107 /* definitions used in the encode routine*/
108 #define MESSAGE(i) data[i]
109 #define REMAINDER(i) bb[i]
110 /* definitions used in the decode routine*/
111 #define RECEIVED(i) data[i]
112 #define ERAS_INDEX(i) eras_pos[i]
113 #define INDEX_TO_POS(i) i
117 /* This defines the type used to store an element of the Galois Field
118 * used by the code. Make sure this is something larger than a char if
119 * if anything larger than GF(256) is used.
121 * Note: unsigned char will work up to GF(256) but int seems to run
122 * faster on the Pentium.
126 /* index->polynomial form conversion table */
127 static gf Alpha_to[NN + 1];
129 /* Polynomial->index form conversion table */
130 static gf Index_of[NN + 1];
132 /* No legal value in index form represents zero, so
133 * we need a special value for this purpose
137 /* Generator polynomial g(x) in index form */
138 static gf Gg[NN - KK + 1];
140 static int RS_init; /* Initialization flag */
142 /* Compute x % NN, where NN is 2**MM - 1,
143 * without a slow divide
145 /* static inline gf*/
151 x = (x >> MM) + (x & NN);
156 #define min_(a,b) ((a) < (b) ? (a) : (b))
158 #define CLEAR(a,n) {\
160 for(ci=(n)-1;ci >=0;ci--)\
164 #define COPY(a,b,n) {\
166 for(ci=(n)-1;ci >=0;ci--)\
170 #define COPYDOWN(a,b,n) {\
172 for(ci=(n)-1;ci >=0;ci--)\
176 static void init_rs(void);
179 /* Conversion lookup tables from conventional alpha to Berlekamp's
180 * dual-basis representation. Used in the CCSDS version only.
181 * taltab[] -- convert conventional to dual basis
182 * tal1tab[] -- convert dual basis to conventional
184 * Note: the actual RS encoder/decoder works with the conventional basis.
185 * So data is converted from dual to conventional basis before either
186 * encoding or decoding and then converted back.
188 static unsigned char taltab[NN+1],tal1tab[NN+1];
190 static unsigned char tal[] = { 0x8d, 0xef, 0xec, 0x86, 0xfa, 0x99, 0xaf, 0x7b };
192 /* Generate conversion lookup tables between conventional alpha representation
193 * (@**7, @**6, ...@**0)
194 * and Berlekamp's dual basis representation
202 for(i=0;i<256;i++){/* For each value of input */
204 for(j=0;j<8;j++) /* for each column of matrix */
205 for(k=0;k<8;k++){ /* for each row of matrix */
207 taltab[i] ^= tal[7-k] & (1<<j);
209 tal1tab[taltab[i]] = i;
215 static int Ldec;/* Decrement for aux location variable in Chien search */
220 for(Ldec=1;(Ldec % PRIM) != 0;Ldec+= NN)
228 /* generate GF(2**m) from the irreducible polynomial p(X) in Pp[0]..Pp[m]
229 lookup tables: index->polynomial form alpha_to[] contains j=alpha**i;
230 polynomial form -> index form index_of[j=alpha**i] = i
231 alpha=2 is the primitive element of GF(2**m)
232 HARI's COMMENT: (4/13/94) alpha_to[] can be used as follows:
233 Let @ represent the primitive element commonly called "alpha" that
234 is the root of the primitive polynomial p(x). Then in GF(2^m), for any
236 @^i = a(0) + a(1) @ + a(2) @^2 + ... + a(m-1) @^(m-1)
237 where the binary vector (a(0),a(1),a(2),...,a(m-1)) is the representation
238 of the integer "alpha_to[i]" with a(0) being the LSB and a(m-1) the MSB. Thus for
239 example the polynomial representation of @^5 would be given by the binary
240 representation of the integer "alpha_to[5]".
241 Similarily, index_of[] can be used as follows:
242 As above, let @ represent the primitive element of GF(2^m) that is
243 the root of the primitive polynomial p(x). In order to find the power
244 of @ (alpha) that has the polynomial representation
245 a(0) + a(1) @ + a(2) @^2 + ... + a(m-1) @^(m-1)
246 we consider the integer "i" whose binary representation with a(0) being LSB
247 and a(m-1) MSB is (a(0),a(1),...,a(m-1)) and locate the entry
248 "index_of[i]". Now, @^index_of[i] is that element whose polynomial
249 representation is (a(0),a(1),a(2),...,a(m-1)).
251 The element alpha_to[2^m-1] = 0 always signifying that the
252 representation of "@^infinity" = 0 is (0,0,0,...,0).
253 Similarily, the element index_of[0] = A0 always signifying
254 that the power of alpha which has the polynomial representation
255 (0,0,...,0) is "infinity".
262 register int i, mask;
266 for (i = 0; i < MM; i++) {
268 Index_of[Alpha_to[i]] = i;
269 /* If Pp[i] == 1 then, term @^i occurs in poly-repr of @^MM */
271 Alpha_to[MM] ^= mask; /* Bit-wise EXOR operation */
272 mask <<= 1; /* single left-shift */
274 Index_of[Alpha_to[MM]] = MM;
276 * Have obtained poly-repr of @^MM. Poly-repr of @^(i+1) is given by
277 * poly-repr of @^i shifted left one-bit and accounting for any @^MM
278 * term that may occur when poly-repr of @^i is shifted.
281 for (i = MM + 1; i < NN; i++) {
282 if (Alpha_to[i - 1] >= mask)
283 Alpha_to[i] = Alpha_to[MM] ^ ((Alpha_to[i - 1] ^ mask) << 1);
285 Alpha_to[i] = Alpha_to[i - 1] << 1;
286 Index_of[Alpha_to[i]] = i;
293 * Obtain the generator polynomial of the TT-error correcting, length
294 * NN=(2**MM -1) Reed Solomon code from the product of (X+@**(B0+i)), i = 0,
299 * If B0 = 1, TT = 1. deg(g(x)) = 2*TT = 2.
300 * g(x) = (x+@) (x+@**2)
302 * If B0 = 0, TT = 2. deg(g(x)) = 2*TT = 4.
303 * g(x) = (x+1) (x+@) (x+@**2) (x+@**3)
311 for (i = 0; i < NN - KK; i++) {
314 * Below multiply (Gg[0]+Gg[1]*x + ... +Gg[i]x^i) by
315 * (@**(B0+i)*PRIM + x)
317 for (j = i; j > 0; j--)
319 Gg[j] = Gg[j - 1] ^ Alpha_to[modnn((Index_of[Gg[j]]) + (B0 + i) *PRIM)];
322 /* Gg[0] can never be zero */
323 Gg[0] = Alpha_to[modnn(Index_of[Gg[0]] + (B0 + i) * PRIM)];
325 /* convert Gg[] to index form for quicker encoding */
326 for (i = 0; i <= NN - KK; i++)
327 Gg[i] = Index_of[Gg[i]];
332 * take the string of symbols in data[i], i=0..(k-1) and encode
333 * systematically to produce NN-KK parity symbols in bb[0]..bb[NN-KK-1] data[]
334 * is input and bb[] is output in polynomial form. Encoding is done by using
335 * a feedback shift register with appropriate connections specified by the
336 * elements of Gg[], which was generated above. Codeword is c(X) =
337 * data(X)*X**(NN-KK)+ b(X)
341 encode_rs(dtype data[KK], dtype bb[NN-KK])
346 #if DEBUG >= 1 && MM != 8
347 /* Check for illegal input values */
359 /* Convert to conventional basis */
361 MESSAGE(i) = tal1tab[MESSAGE(i)];
364 for(i = KK - 1; i >= 0; i--) {
365 feedback = Index_of[MESSAGE(i) ^ REMAINDER(NN - KK - 1)];
366 if (feedback != A0) { /* feedback term is non-zero */
367 for (j = NN - KK - 1; j > 0; j--)
369 REMAINDER(j) = REMAINDER(j - 1) ^ Alpha_to[modnn(Gg[j] + feedback)];
371 REMAINDER(j) = REMAINDER(j - 1);
372 REMAINDER(0) = Alpha_to[modnn(Gg[0] + feedback)];
373 } else { /* feedback term is zero. encoder becomes a
374 * single-byte shifter */
375 for (j = NN - KK - 1; j > 0; j--)
376 REMAINDER(j) = REMAINDER(j - 1);
381 /* Convert to l-basis */
383 MESSAGE(i) = taltab[MESSAGE(i)];
390 * Performs ERRORS+ERASURES decoding of RS codes. If decoding is successful,
391 * writes the codeword into data[] itself. Otherwise data[] is unaltered.
393 * Return number of symbols corrected, or -1 if codeword is illegal
394 * or uncorrectable. If eras_pos is non-null, the detected error locations
395 * are written back. NOTE! This array must be at least NN-KK elements long.
397 * First "no_eras" erasures are declared by the calling program. Then, the
398 * maximum # of errors correctable is t_after_eras = floor((NN-KK-no_eras)/2).
399 * If the number of channel errors is not greater than "t_after_eras" the
400 * transmitted codeword will be recovered. Details of algorithm can be found
401 * in R. Blahut's "Theory ... of Error-Correcting Codes".
403 * Warning: the eras_pos[] array must not contain duplicate entries; decoder failure
404 * will result. The decoder *could* check for this condition, but it would involve
405 * extra time on every decoding operation.
409 eras_dec_rs(dtype data[NN], int eras_pos[NN-KK], int no_eras)
411 int deg_lambda, el, deg_omega;
413 gf u,q,tmp,num1,num2,den,discr_r;
414 gf lambda[NN-KK + 1], s[NN-KK + 1]; /* Err+Eras Locator poly
415 * and syndrome poly */
416 gf b[NN-KK + 1], t[NN-KK + 1], omega[NN-KK + 1];
417 gf root[NN-KK], reg[NN-KK + 1], loc[NN-KK];
418 int syn_error, count;
424 /* Convert to conventional basis */
426 RECEIVED(i) = tal1tab[RECEIVED(i)];
429 #if DEBUG >= 1 && MM != 8
430 /* Check for illegal input values */
435 /* form the syndromes; i.e., evaluate data(x) at roots of g(x)
436 * namely @**(B0+i)*PRIM, i = 0, ... ,(NN-KK-1)
438 for(i=1;i<=NN-KK;i++){
444 tmp = Index_of[RECEIVED(j)];
446 /* s[i] ^= Alpha_to[modnn(tmp + (B0+i-1)*j)]; */
447 for(i=1;i<=NN-KK;i++)
448 s[i] ^= Alpha_to[modnn(tmp + (B0+i-1)*PRIM*j)];
450 /* Convert syndromes to index form, checking for nonzero condition */
452 for(i=1;i<=NN-KK;i++){
454 /*ws_debug_printf("syndrome %d = %x\n",i,s[i]);*/
455 s[i] = Index_of[s[i]];
459 /* if syndrome is zero, data[] is a codeword and there are no
460 * errors to correct. So return data[] unmodified
465 CLEAR(&lambda[1],NN-KK);
469 /* Init lambda to be the erasure locator polynomial */
470 lambda[1] = Alpha_to[modnn(PRIM * ERAS_INDEX(0))];
471 for (i = 1; i < no_eras; i++) {
472 u = modnn(PRIM*ERAS_INDEX(i));
473 for (j = i+1; j > 0; j--) {
474 tmp = Index_of[lambda[j - 1]];
476 lambda[j] ^= Alpha_to[modnn(u + tmp)];
480 /* Test code that verifies the erasure locator polynomial just constructed
481 Needed only for decoder debugging. */
483 /* find roots of the erasure location polynomial */
484 for(i=1;i<=no_eras;i++)
485 reg[i] = Index_of[lambda[i]];
487 for (i = 1,k=NN-Ldec; i <= NN; i++,k = modnn(NN+k-Ldec)) {
489 for (j = 1; j <= no_eras; j++)
491 reg[j] = modnn(reg[j] + j);
492 q ^= Alpha_to[reg[j]];
496 /* store root and error location number indices */
501 if (count != no_eras) {
502 ws_debug_printf("\n lambda(x) is WRONG\n");
507 ws_debug_printf("\n Erasure positions as determined by roots of Eras Loc Poly:\n");
508 for (i = 0; i < count; i++)
509 ws_debug_printf("%d ", loc[i]);
510 ws_debug_printf("\n");
514 for(i=0;i<NN-KK+1;i++)
515 b[i] = Index_of[lambda[i]];
518 * Begin Berlekamp-Massey algorithm to determine error+erasure
523 while (++r <= NN-KK) { /* r is the step number */
524 /* Compute discrepancy at the r-th step in poly-form */
526 for (i = 0; i < r; i++){
527 if ((lambda[i] != 0) && (s[r - i] != A0)) {
528 discr_r ^= Alpha_to[modnn(Index_of[lambda[i]] + s[r - i])];
531 discr_r = Index_of[discr_r]; /* Index form */
533 /* 2 lines below: B(x) <-- x*B(x) */
534 COPYDOWN(&b[1],b,NN-KK);
537 /* 7 lines below: T(x) <-- lambda(x) - discr_r*x*b(x) */
539 for (i = 0 ; i < NN-KK; i++) {
541 t[i+1] = lambda[i+1] ^ Alpha_to[modnn(discr_r + b[i])];
543 t[i+1] = lambda[i+1];
545 if (2 * el <= r + no_eras - 1) {
546 el = r + no_eras - el;
548 * 2 lines below: B(x) <-- inv(discr_r) *
551 for (i = 0; i <= NN-KK; i++)
552 b[i] = (lambda[i] == 0) ? A0 : modnn(Index_of[lambda[i]] - discr_r + NN);
554 /* 2 lines below: B(x) <-- x*B(x) */
555 COPYDOWN(&b[1],b,NN-KK);
558 COPY(lambda,t,NN-KK+1);
562 /* Convert lambda to index form and compute deg(lambda(x)) */
564 for(i=0;i<NN-KK+1;i++){
565 lambda[i] = Index_of[lambda[i]];
570 * Find roots of the error+erasure locator polynomial by Chien
573 COPY(®[1],&lambda[1],NN-KK);
574 count = 0; /* Number of roots of lambda(x) */
575 for (i = 1,k=NN-Ldec; i <= NN; i++,k = modnn(NN+k-Ldec)) {
577 for (j = deg_lambda; j > 0; j--){
579 reg[j] = modnn(reg[j] + j);
580 q ^= Alpha_to[reg[j]];
585 /* store root (index-form) and error location number */
588 /* If we've already found max possible roots,
589 * abort the search to save time
591 if(++count == deg_lambda)
594 if (deg_lambda != count) {
596 * deg(lambda) unequal to number of roots => uncorrectable
603 * Compute err+eras evaluator poly omega(x) = s(x)*lambda(x) (modulo
604 * x**(NN-KK)). in index form. Also find deg(omega).
607 for (i = 0; i < NN-KK;i++){
609 j = (deg_lambda < i) ? deg_lambda : i;
611 if ((s[i + 1 - j] != A0) && (lambda[j] != A0))
612 tmp ^= Alpha_to[modnn(s[i + 1 - j] + lambda[j])];
616 omega[i] = Index_of[tmp];
621 * Compute error values in poly-form. num1 = omega(inv(X(l))), num2 =
622 * inv(X(l))**(B0-1) and den = lambda_pr(inv(X(l))) all in poly-form
624 for (j = count-1; j >=0; j--) {
626 for (i = deg_omega; i >= 0; i--) {
628 num1 ^= Alpha_to[modnn(omega[i] + i * root[j])];
630 num2 = Alpha_to[modnn(root[j] * (B0 - 1) + NN)];
633 /* lambda[i+1] for i even is the formal derivative lambda_pr of lambda[i] */
634 for (i = min_(deg_lambda,NN-KK-1) & ~1; i >= 0; i -=2) {
635 if(lambda[i+1] != A0)
636 den ^= Alpha_to[modnn(lambda[i+1] + i * root[j])];
640 ws_debug_printf("\n ERROR: denominator = 0\n");
642 /* Convert to dual- basis */
646 /* Apply error to data */
648 RECEIVED(loc[j]) ^= Alpha_to[modnn(Index_of[num1] + Index_of[num2] + NN - Index_of[den])];
653 /* Convert to dual- basis */
655 RECEIVED(i) = taltab[RECEIVED(i)];
657 if(eras_pos != NULL){
658 for(i=0;i<count;i++){
660 eras_pos[i] = INDEX_TO_POS(loc[i]);
665 /* Encoder/decoder initialization - call this first! */
681 * Editor modelines - http://www.wireshark.org/tools/modelines.html
686 * indent-tabs-mode: nil
689 * ex: set shiftwidth=2 tabstop=8 expandtab:
690 * :indentSize=2:tabSize=8:noTabs=true: