Coverage Report

Created: 2024-06-28 06:19

/src/gmp-6.2.1/mpn/gcdext_lehmer.c
Line
Count
Source (jump to first uncovered line)
1
/* mpn_gcdext -- Extended Greatest Common Divisor.
2
3
Copyright 1996, 1998, 2000-2005, 2008, 2009, 2012 Free Software Foundation,
4
Inc.
5
6
This file is part of the GNU MP Library.
7
8
The GNU MP Library is free software; you can redistribute it and/or modify
9
it under the terms of either:
10
11
  * the GNU Lesser General Public License as published by the Free
12
    Software Foundation; either version 3 of the License, or (at your
13
    option) any later version.
14
15
or
16
17
  * the GNU General Public License as published by the Free Software
18
    Foundation; either version 2 of the License, or (at your option) any
19
    later version.
20
21
or both in parallel, as here.
22
23
The GNU MP Library is distributed in the hope that it will be useful, but
24
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
25
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
26
for more details.
27
28
You should have received copies of the GNU General Public License and the
29
GNU Lesser General Public License along with the GNU MP Library.  If not,
30
see https://www.gnu.org/licenses/.  */
31
32
#include "gmp-impl.h"
33
#include "longlong.h"
34
35
/* Here, d is the index of the cofactor to update. FIXME: Could use qn
36
   = 0 for the common case q = 1. */
37
void
38
mpn_gcdext_hook (void *p, mp_srcptr gp, mp_size_t gn,
39
     mp_srcptr qp, mp_size_t qn, int d)
40
6.35k
{
41
6.35k
  struct gcdext_ctx *ctx = (struct gcdext_ctx *) p;
42
6.35k
  mp_size_t un = ctx->un;
43
44
6.35k
  if (gp)
45
405
    {
46
405
      mp_srcptr up;
47
48
405
      ASSERT (gn > 0);
49
405
      ASSERT (gp[gn-1] > 0);
50
51
405
      MPN_COPY (ctx->gp, gp, gn);
52
405
      ctx->gn = gn;
53
54
405
      if (d < 0)
55
341
  {
56
341
    int c;
57
58
    /* Must return the smallest cofactor, +u1 or -u0 */
59
341
    MPN_CMP (c, ctx->u0, ctx->u1, un);
60
341
    ASSERT (c != 0 || (un == 1 && ctx->u0[0] == 1 && ctx->u1[0] == 1));
61
62
341
    d = c < 0;
63
341
  }
64
65
405
      up = d ? ctx->u0 : ctx->u1;
66
67
405
      MPN_NORMALIZE (up, un);
68
405
      MPN_COPY (ctx->up, up, un);
69
70
405
      *ctx->usize = d ? -un : un;
71
405
    }
72
5.94k
  else
73
5.94k
    {
74
5.94k
      mp_limb_t cy;
75
5.94k
      mp_ptr u0 = ctx->u0;
76
5.94k
      mp_ptr u1 = ctx->u1;
77
78
5.94k
      ASSERT (d >= 0);
79
80
5.94k
      if (d)
81
2.75k
  MP_PTR_SWAP (u0, u1);
82
83
5.94k
      qn -= (qp[qn-1] == 0);
84
85
      /* Update u0 += q  * u1 */
86
5.94k
      if (qn == 1)
87
3.41k
  {
88
3.41k
    mp_limb_t q = qp[0];
89
90
3.41k
    if (q == 1)
91
      /* A common case. */
92
3.00k
      cy = mpn_add_n (u0, u0, u1, un);
93
407
    else
94
407
      cy = mpn_addmul_1 (u0, u1, un, q);
95
3.41k
  }
96
2.53k
      else
97
2.53k
  {
98
2.53k
    mp_size_t u1n;
99
2.53k
    mp_ptr tp;
100
101
2.53k
    u1n = un;
102
2.53k
    MPN_NORMALIZE (u1, u1n);
103
104
2.53k
    if (u1n == 0)
105
0
      return;
106
107
    /* Should always have u1n == un here, and u1 >= u0. The
108
       reason is that we alternate adding u0 to u1 and u1 to u0
109
       (corresponding to subtractions a - b and b - a), and we
110
       can get a large quotient only just after a switch, which
111
       means that we'll add (a multiple of) the larger u to the
112
       smaller. */
113
114
2.53k
    tp = ctx->tp;
115
116
2.53k
    if (qn > u1n)
117
442
      mpn_mul (tp, qp, qn, u1, u1n);
118
2.09k
    else
119
2.09k
      mpn_mul (tp, u1, u1n, qp, qn);
120
121
2.53k
    u1n += qn;
122
2.53k
    u1n -= tp[u1n-1] == 0;
123
124
2.53k
    if (u1n >= un)
125
2.53k
      {
126
2.53k
        cy = mpn_add (u0, tp, u1n, u0, un);
127
2.53k
        un = u1n;
128
2.53k
      }
129
0
    else
130
      /* Note: Unlikely case, maybe never happens? */
131
0
      cy = mpn_add (u0, u0, un, tp, u1n);
132
133
2.53k
  }
134
5.94k
      u0[un] = cy;
135
5.94k
      ctx->un = un + (cy > 0);
136
5.94k
    }
137
6.35k
}
138
139
/* Temporary storage: 3*(n+1) for u. If hgcd2 succeeds, we need n for
140
   the matrix-vector multiplication adjusting a, b. If hgcd fails, we
141
   need at most n for the quotient and n+1 for the u update (reusing
142
   the extra u). In all, 4n + 3. */
143
144
mp_size_t
145
mpn_gcdext_lehmer_n (mp_ptr gp, mp_ptr up, mp_size_t *usize,
146
         mp_ptr ap, mp_ptr bp, mp_size_t n,
147
         mp_ptr tp)
148
2.05k
{
149
2.05k
  mp_size_t ualloc = n + 1;
150
151
  /* Keeps track of the second row of the reduction matrix
152
   *
153
   *   M = (v0, v1 ; u0, u1)
154
   *
155
   * which correspond to the first column of the inverse
156
   *
157
   *   M^{-1} = (u1, -v1; -u0, v0)
158
   *
159
   * This implies that
160
   *
161
   *   a =  u1 A (mod B)
162
   *   b = -u0 A (mod B)
163
   *
164
   * where A, B denotes the input values.
165
   */
166
167
2.05k
  struct gcdext_ctx ctx;
168
2.05k
  mp_size_t un;
169
2.05k
  mp_ptr u0;
170
2.05k
  mp_ptr u1;
171
2.05k
  mp_ptr u2;
172
173
2.05k
  MPN_ZERO (tp, 3*ualloc);
174
2.05k
  u0 = tp; tp += ualloc;
175
2.05k
  u1 = tp; tp += ualloc;
176
2.05k
  u2 = tp; tp += ualloc;
177
178
2.05k
  u1[0] = 1; un = 1;
179
180
2.05k
  ctx.gp = gp;
181
2.05k
  ctx.up = up;
182
2.05k
  ctx.usize = usize;
183
184
  /* FIXME: Handle n == 2 differently, after the loop? */
185
64.8k
  while (n >= 2)
186
63.1k
    {
187
63.1k
      struct hgcd_matrix1 M;
188
63.1k
      mp_limb_t ah, al, bh, bl;
189
63.1k
      mp_limb_t mask;
190
191
63.1k
      mask = ap[n-1] | bp[n-1];
192
63.1k
      ASSERT (mask > 0);
193
194
63.1k
      if (mask & GMP_NUMB_HIGHBIT)
195
1.39k
  {
196
1.39k
    ah = ap[n-1]; al = ap[n-2];
197
1.39k
    bh = bp[n-1]; bl = bp[n-2];
198
1.39k
  }
199
61.7k
      else if (n == 2)
200
1.39k
  {
201
    /* We use the full inputs without truncation, so we can
202
       safely shift left. */
203
1.39k
    int shift;
204
205
1.39k
    count_leading_zeros (shift, mask);
206
1.39k
    ah = MPN_EXTRACT_NUMB (shift, ap[1], ap[0]);
207
1.39k
    al = ap[0] << shift;
208
1.39k
    bh = MPN_EXTRACT_NUMB (shift, bp[1], bp[0]);
209
1.39k
    bl = bp[0] << shift;
210
1.39k
  }
211
60.3k
      else
212
60.3k
  {
213
60.3k
    int shift;
214
215
60.3k
    count_leading_zeros (shift, mask);
216
60.3k
    ah = MPN_EXTRACT_NUMB (shift, ap[n-1], ap[n-2]);
217
60.3k
    al = MPN_EXTRACT_NUMB (shift, ap[n-2], ap[n-3]);
218
60.3k
    bh = MPN_EXTRACT_NUMB (shift, bp[n-1], bp[n-2]);
219
60.3k
    bl = MPN_EXTRACT_NUMB (shift, bp[n-2], bp[n-3]);
220
60.3k
  }
221
222
      /* Try an mpn_nhgcd2 step */
223
63.1k
      if (mpn_hgcd2 (ah, al, bh, bl, &M))
224
59.8k
  {
225
59.8k
    n = mpn_matrix22_mul1_inverse_vector (&M, tp, ap, bp, n);
226
59.8k
    MP_PTR_SWAP (ap, tp);
227
59.8k
    un = mpn_hgcd_mul_matrix1_vector(&M, u2, u0, u1, un);
228
59.8k
    MP_PTR_SWAP (u0, u2);
229
59.8k
  }
230
3.34k
      else
231
3.34k
  {
232
    /* mpn_hgcd2 has failed. Then either one of a or b is very
233
       small, or the difference is very small. Perform one
234
       subtraction followed by one division. */
235
3.34k
    ctx.u0 = u0;
236
3.34k
    ctx.u1 = u1;
237
3.34k
    ctx.tp = u2;
238
3.34k
    ctx.un = un;
239
240
    /* Temporary storage n for the quotient and ualloc for the
241
       new cofactor. */
242
3.34k
    n = mpn_gcd_subdiv_step (ap, bp, n, 0, mpn_gcdext_hook, &ctx, tp);
243
3.34k
    if (n == 0)
244
405
      return ctx.gn;
245
246
2.94k
    un = ctx.un;
247
2.94k
  }
248
63.1k
    }
249
1.64k
  ASSERT_ALWAYS (ap[0] > 0);
250
1.64k
  ASSERT_ALWAYS (bp[0] > 0);
251
252
1.64k
  if (ap[0] == bp[0])
253
142
    {
254
142
      int c;
255
256
      /* Which cofactor to return now? Candidates are +u1 and -u0,
257
   depending on which of a and b was most recently reduced,
258
   which we don't keep track of. So compare and get the smallest
259
   one. */
260
261
142
      gp[0] = ap[0];
262
263
142
      MPN_CMP (c, u0, u1, un);
264
142
      ASSERT (c != 0 || (un == 1 && u0[0] == 1 && u1[0] == 1));
265
142
      if (c < 0)
266
55
  {
267
55
    MPN_NORMALIZE (u0, un);
268
55
    MPN_COPY (up, u0, un);
269
55
    *usize = -un;
270
55
  }
271
87
      else
272
87
  {
273
87
    MPN_NORMALIZE_NOT_ZERO (u1, un);
274
87
    MPN_COPY (up, u1, un);
275
87
    *usize = un;
276
87
  }
277
142
      return 1;
278
142
    }
279
1.50k
  else
280
1.50k
    {
281
1.50k
      mp_limb_t uh, vh;
282
1.50k
      mp_limb_signed_t u;
283
1.50k
      mp_limb_signed_t v;
284
1.50k
      int negate;
285
286
1.50k
      gp[0] = mpn_gcdext_1 (&u, &v, ap[0], bp[0]);
287
288
      /* Set up = u u1 - v u0. Keep track of size, un grows by one or
289
   two limbs. */
290
291
1.50k
      if (u == 0)
292
12
  {
293
12
    ASSERT (v == 1);
294
12
    MPN_NORMALIZE (u0, un);
295
12
    MPN_COPY (up, u0, un);
296
12
    *usize = -un;
297
12
    return 1;
298
12
  }
299
1.49k
      else if (v == 0)
300
59
  {
301
59
    ASSERT (u == 1);
302
59
    MPN_NORMALIZE (u1, un);
303
59
    MPN_COPY (up, u1, un);
304
59
    *usize = un;
305
59
    return 1;
306
59
  }
307
1.43k
      else if (u > 0)
308
699
  {
309
699
    negate = 0;
310
699
    ASSERT (v < 0);
311
699
    v = -v;
312
699
  }
313
736
      else
314
736
  {
315
736
    negate = 1;
316
736
    ASSERT (v > 0);
317
736
    u = -u;
318
736
  }
319
320
1.43k
      uh = mpn_mul_1 (up, u1, un, u);
321
1.43k
      vh = mpn_addmul_1 (up, u0, un, v);
322
323
1.43k
      if ( (uh | vh) > 0)
324
403
  {
325
403
    uh += vh;
326
403
    up[un++] = uh;
327
403
    if (uh < vh)
328
0
      up[un++] = 1;
329
403
  }
330
331
1.43k
      MPN_NORMALIZE_NOT_ZERO (up, un);
332
333
1.43k
      *usize = negate ? -un : un;
334
1.43k
      return 1;
335
1.43k
    }
336
1.64k
}