Coverage Report

Created: 2025-11-16 06:46

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/src/gmp/mpn/tdiv_qr.c
Line
Count
Source
1
/* mpn_tdiv_qr -- Divide the numerator (np,nn) by the denominator (dp,dn) and
2
   write the nn-dn+1 quotient limbs at qp and the dn remainder limbs at rp.  If
3
   qxn is non-zero, generate that many fraction limbs and append them after the
4
   other quotient limbs, and update the remainder accordingly.  The input
5
   operands are unaffected.
6
7
   Preconditions:
8
   1. The most significant limb of the divisor must be non-zero.
9
   2. nn >= dn, even if qxn is non-zero.  (??? relax this ???)
10
11
   The time complexity of this is O(qn*qn+M(dn,qn)), where M(m,n) is the time
12
   complexity of multiplication.
13
14
Copyright 1997, 2000-2002, 2005, 2009, 2015 Free Software Foundation, Inc.
15
16
This file is part of the GNU MP Library.
17
18
The GNU MP Library is free software; you can redistribute it and/or modify
19
it under the terms of either:
20
21
  * the GNU Lesser General Public License as published by the Free
22
    Software Foundation; either version 3 of the License, or (at your
23
    option) any later version.
24
25
or
26
27
  * the GNU General Public License as published by the Free Software
28
    Foundation; either version 2 of the License, or (at your option) any
29
    later version.
30
31
or both in parallel, as here.
32
33
The GNU MP Library is distributed in the hope that it will be useful, but
34
WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
35
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
36
for more details.
37
38
You should have received copies of the GNU General Public License and the
39
GNU Lesser General Public License along with the GNU MP Library.  If not,
40
see https://www.gnu.org/licenses/.  */
41
42
#include "gmp-impl.h"
43
#include "longlong.h"
44
45
46
void
47
mpn_tdiv_qr (mp_ptr qp, mp_ptr rp, mp_size_t qxn,
48
       mp_srcptr np, mp_size_t nn, mp_srcptr dp, mp_size_t dn)
49
123k
{
50
123k
  ASSERT_ALWAYS (qxn == 0);
51
52
123k
  ASSERT (nn >= 0);
53
123k
  ASSERT (dn >= 0);
54
123k
  ASSERT (dn == 0 || dp[dn - 1] != 0);
55
123k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, np, nn));
56
123k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, dp, dn));
57
58
123k
  switch (dn)
59
123k
    {
60
0
    case 0:
61
0
      DIVIDE_BY_ZERO;
62
63
9.50k
    case 1:
64
9.50k
      {
65
9.50k
  rp[0] = mpn_divrem_1 (qp, (mp_size_t) 0, np, nn, dp[0]);
66
9.50k
  return;
67
0
      }
68
69
4.16k
    case 2:
70
4.16k
      {
71
4.16k
  mp_ptr n2p;
72
4.16k
  mp_limb_t qhl, cy;
73
4.16k
  TMP_DECL;
74
4.16k
  TMP_MARK;
75
4.16k
  if ((dp[1] & GMP_NUMB_HIGHBIT) == 0)
76
3.37k
    {
77
3.37k
      int cnt;
78
3.37k
      mp_limb_t d2p[2];
79
3.37k
      count_leading_zeros (cnt, dp[1]);
80
3.37k
      cnt -= GMP_NAIL_BITS;
81
3.37k
      d2p[1] = (dp[1] << cnt) | (dp[0] >> (GMP_NUMB_BITS - cnt));
82
3.37k
      d2p[0] = (dp[0] << cnt) & GMP_NUMB_MASK;
83
3.37k
      n2p = TMP_ALLOC_LIMBS (nn + 1);
84
3.37k
      cy = mpn_lshift (n2p, np, nn, cnt);
85
3.37k
      n2p[nn] = cy;
86
3.37k
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn + (cy != 0), d2p);
87
3.37k
      if (cy == 0)
88
1.22k
        qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
89
3.37k
      rp[0] = (n2p[0] >> cnt)
90
3.37k
        | ((n2p[1] << (GMP_NUMB_BITS - cnt)) & GMP_NUMB_MASK);
91
3.37k
      rp[1] = (n2p[1] >> cnt);
92
3.37k
    }
93
790
  else
94
790
    {
95
790
      n2p = TMP_ALLOC_LIMBS (nn);
96
790
      MPN_COPY (n2p, np, nn);
97
790
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn, dp);
98
790
      qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
99
790
      rp[0] = n2p[0];
100
790
      rp[1] = n2p[1];
101
790
    }
102
4.16k
  TMP_FREE;
103
4.16k
  return;
104
0
      }
105
106
109k
    default:
107
109k
      {
108
109k
  int adjust;
109
109k
  gmp_pi1_t dinv;
110
109k
  TMP_DECL;
111
109k
  TMP_MARK;
112
109k
  adjust = np[nn - 1] >= dp[dn - 1];  /* conservative tests for quotient size */
113
109k
  if (nn + adjust >= 2 * dn)
114
71.8k
    {
115
71.8k
      mp_ptr n2p, d2p;
116
71.8k
      mp_limb_t cy;
117
71.8k
      int cnt;
118
119
71.8k
      qp[nn - dn] = 0;        /* zero high quotient limb */
120
71.8k
      if ((dp[dn - 1] & GMP_NUMB_HIGHBIT) == 0) /* normalize divisor */
121
4.95k
        {
122
4.95k
    count_leading_zeros (cnt, dp[dn - 1]);
123
4.95k
    cnt -= GMP_NAIL_BITS;
124
4.95k
    d2p = TMP_ALLOC_LIMBS (dn);
125
4.95k
    mpn_lshift (d2p, dp, dn, cnt);
126
4.95k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
127
4.95k
    cy = mpn_lshift (n2p, np, nn, cnt);
128
4.95k
    n2p[nn] = cy;
129
4.95k
    nn += adjust;
130
4.95k
        }
131
66.9k
      else
132
66.9k
        {
133
66.9k
    cnt = 0;
134
66.9k
    d2p = (mp_ptr) dp;
135
66.9k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
136
66.9k
    MPN_COPY (n2p, np, nn);
137
66.9k
    n2p[nn] = 0;
138
66.9k
    nn += adjust;
139
66.9k
        }
140
141
71.8k
      invert_pi1 (dinv, d2p[dn - 1], d2p[dn - 2]);
142
71.8k
      if (BELOW_THRESHOLD (dn, DC_DIV_QR_THRESHOLD))
143
71.6k
        mpn_sbpi1_div_qr (qp, n2p, nn, d2p, dn, dinv.inv32);
144
262
      else if (BELOW_THRESHOLD (dn, MUPI_DIV_QR_THRESHOLD) ||   /* fast condition */
145
0
         BELOW_THRESHOLD (nn, 2 * MU_DIV_QR_THRESHOLD) || /* fast condition */
146
0
         (double) (2 * (MU_DIV_QR_THRESHOLD - MUPI_DIV_QR_THRESHOLD)) * dn /* slow... */
147
0
         + (double) MUPI_DIV_QR_THRESHOLD * nn > (double) dn * nn)    /* ...condition */
148
262
        mpn_dcpi1_div_qr (qp, n2p, nn, d2p, dn, &dinv);
149
0
      else
150
0
        {
151
0
    mp_size_t itch = mpn_mu_div_qr_itch (nn, dn, 0);
152
0
    mp_ptr scratch = TMP_ALLOC_LIMBS (itch);
153
0
    mpn_mu_div_qr (qp, rp, n2p, nn, d2p, dn, scratch);
154
0
    n2p = rp;
155
0
        }
156
157
71.8k
      if (cnt != 0)
158
4.95k
        mpn_rshift (rp, n2p, dn, cnt);
159
66.9k
      else
160
66.9k
        MPN_COPY (rp, n2p, dn);
161
71.8k
      TMP_FREE;
162
71.8k
      return;
163
71.8k
    }
164
165
  /* When we come here, the numerator/partial remainder is less
166
     than twice the size of the denominator.  */
167
168
37.6k
    {
169
      /* Problem:
170
171
         Divide a numerator N with nn limbs by a denominator D with dn
172
         limbs forming a quotient of qn=nn-dn+1 limbs.  When qn is small
173
         compared to dn, conventional division algorithms perform poorly.
174
         We want an algorithm that has an expected running time that is
175
         dependent only on qn.
176
177
         Algorithm (very informally stated):
178
179
         1) Divide the 2 x qn most significant limbs from the numerator
180
      by the qn most significant limbs from the denominator.  Call
181
      the result qest.  This is either the correct quotient, but
182
      might be 1 or 2 too large.  Compute the remainder from the
183
      division.  (This step is implemented by an mpn_divrem call.)
184
185
         2) Is the most significant limb from the remainder < p, where p
186
      is the product of the most significant limb from the quotient
187
      and the next(d)?  (Next(d) denotes the next ignored limb from
188
      the denominator.)  If it is, decrement qest, and adjust the
189
      remainder accordingly.
190
191
         3) Is the remainder >= qest?  If it is, qest is the desired
192
      quotient.  The algorithm terminates.
193
194
         4) Subtract qest x next(d) from the remainder.  If there is
195
      borrow out, decrement qest, and adjust the remainder
196
      accordingly.
197
198
         5) Skip one word from the denominator (i.e., let next(d) denote
199
      the next less significant limb.  */
200
201
37.6k
      mp_size_t qn;
202
37.6k
      mp_ptr n2p, d2p;
203
37.6k
      mp_ptr tp;
204
37.6k
      mp_limb_t cy;
205
37.6k
      mp_size_t in, rn;
206
37.6k
      mp_limb_t quotient_too_large;
207
37.6k
      unsigned int cnt;
208
209
37.6k
      qn = nn - dn;
210
37.6k
      qp[qn] = 0;       /* zero high quotient limb */
211
37.6k
      qn += adjust;     /* qn cannot become bigger */
212
213
37.6k
      if (qn == 0)
214
210
        {
215
210
    MPN_COPY (rp, np, dn);
216
210
    TMP_FREE;
217
210
    return;
218
210
        }
219
220
37.4k
      in = dn - qn;   /* (at least partially) ignored # of limbs in ops */
221
      /* Normalize denominator by shifting it to the left such that its
222
         most significant bit is set.  Then shift the numerator the same
223
         amount, to mathematically preserve quotient.  */
224
37.4k
      if ((dp[dn - 1] & GMP_NUMB_HIGHBIT) == 0)
225
24.0k
        {
226
24.0k
    count_leading_zeros (cnt, dp[dn - 1]);
227
24.0k
    cnt -= GMP_NAIL_BITS;
228
229
24.0k
    d2p = TMP_ALLOC_LIMBS (qn);
230
24.0k
    mpn_lshift (d2p, dp + in, qn, cnt);
231
24.0k
    d2p[0] |= dp[in - 1] >> (GMP_NUMB_BITS - cnt);
232
233
24.0k
    n2p = TMP_ALLOC_LIMBS (2 * qn + 1);
234
24.0k
    cy = mpn_lshift (n2p, np + nn - 2 * qn, 2 * qn, cnt);
235
24.0k
    if (adjust)
236
13.2k
      {
237
13.2k
        n2p[2 * qn] = cy;
238
13.2k
        n2p++;
239
13.2k
      }
240
10.7k
    else
241
10.7k
      {
242
10.7k
        n2p[0] |= np[nn - 2 * qn - 1] >> (GMP_NUMB_BITS - cnt);
243
10.7k
      }
244
24.0k
        }
245
13.3k
      else
246
13.3k
        {
247
13.3k
    cnt = 0;
248
13.3k
    d2p = (mp_ptr) dp + in;
249
250
13.3k
    n2p = TMP_ALLOC_LIMBS (2 * qn + 1);
251
13.3k
    MPN_COPY (n2p, np + nn - 2 * qn, 2 * qn);
252
13.3k
    if (adjust)
253
1.11k
      {
254
1.11k
        n2p[2 * qn] = 0;
255
1.11k
        n2p++;
256
1.11k
      }
257
13.3k
        }
258
259
      /* Get an approximate quotient using the extracted operands.  */
260
37.4k
      if (qn == 1)
261
9.24k
        {
262
9.24k
    mp_limb_t q0, r0;
263
9.24k
    udiv_qrnnd (q0, r0, n2p[1], n2p[0] << GMP_NAIL_BITS, d2p[0] << GMP_NAIL_BITS);
264
9.24k
    n2p[0] = r0 >> GMP_NAIL_BITS;
265
9.24k
    qp[0] = q0;
266
9.24k
        }
267
28.1k
      else if (qn == 2)
268
8.37k
        mpn_divrem_2 (qp, 0L, n2p, 4L, d2p); /* FIXME: obsolete function */
269
19.7k
      else
270
19.7k
        {
271
19.7k
    invert_pi1 (dinv, d2p[qn - 1], d2p[qn - 2]);
272
19.7k
    if (BELOW_THRESHOLD (qn, DC_DIV_QR_THRESHOLD))
273
18.6k
      mpn_sbpi1_div_qr (qp, n2p, 2 * qn, d2p, qn, dinv.inv32);
274
1.14k
    else if (BELOW_THRESHOLD (qn, MU_DIV_QR_THRESHOLD))
275
1.14k
      mpn_dcpi1_div_qr (qp, n2p, 2 * qn, d2p, qn, &dinv);
276
0
    else
277
0
      {
278
0
        mp_size_t itch = mpn_mu_div_qr_itch (2 * qn, qn, 0);
279
0
        mp_ptr scratch = TMP_ALLOC_LIMBS (itch);
280
0
        mp_ptr r2p = rp;
281
0
        if (np == r2p) /* If N and R share space, put ... */
282
0
          r2p += nn - qn; /* intermediate remainder at N's upper end. */
283
0
        mpn_mu_div_qr (qp, r2p, n2p, 2 * qn, d2p, qn, scratch);
284
0
        MPN_COPY (n2p, r2p, qn);
285
0
      }
286
19.7k
        }
287
288
37.4k
      rn = qn;
289
      /* Multiply the first ignored divisor limb by the most significant
290
         quotient limb.  If that product is > the partial remainder's
291
         most significant limb, we know the quotient is too large.  This
292
         test quickly catches most cases where the quotient is too large;
293
         it catches all cases where the quotient is 2 too large.  */
294
37.4k
      {
295
37.4k
        mp_limb_t dl, x;
296
37.4k
        mp_limb_t h, dummy;
297
298
37.4k
        if (in - 2 < 0)
299
1.95k
    dl = 0;
300
35.4k
        else
301
35.4k
    dl = dp[in - 2];
302
303
37.4k
#if GMP_NAIL_BITS == 0
304
37.4k
        x = (dp[in - 1] << cnt) | ((dl >> 1) >> ((~cnt) % GMP_LIMB_BITS));
305
#else
306
        x = (dp[in - 1] << cnt) & GMP_NUMB_MASK;
307
        if (cnt != 0)
308
    x |= dl >> (GMP_NUMB_BITS - cnt);
309
#endif
310
37.4k
        umul_ppmm (h, dummy, x, qp[qn - 1] << GMP_NAIL_BITS);
311
312
37.4k
        if (n2p[qn - 1] < h)
313
2.37k
    {
314
2.37k
      mp_limb_t cy;
315
316
2.37k
      mpn_decr_u (qp, (mp_limb_t) 1);
317
2.37k
      cy = mpn_add_n (n2p, n2p, d2p, qn);
318
2.37k
      if (cy)
319
988
        {
320
          /* The partial remainder is safely large.  */
321
988
          n2p[qn] = cy;
322
988
          ++rn;
323
988
        }
324
2.37k
    }
325
37.4k
      }
326
327
37.4k
      quotient_too_large = 0;
328
37.4k
      if (cnt != 0)
329
24.0k
        {
330
24.0k
    mp_limb_t cy1, cy2;
331
332
    /* Append partially used numerator limb to partial remainder.  */
333
24.0k
    cy1 = mpn_lshift (n2p, n2p, rn, GMP_NUMB_BITS - cnt);
334
24.0k
    n2p[0] |= np[in - 1] & (GMP_NUMB_MASK >> cnt);
335
336
    /* Update partial remainder with partially used divisor limb.  */
337
24.0k
    cy2 = mpn_submul_1 (n2p, qp, qn, dp[in - 1] & (GMP_NUMB_MASK >> cnt));
338
24.0k
    if (qn != rn)
339
578
      {
340
578
        ASSERT_ALWAYS (n2p[qn] >= cy2);
341
578
        n2p[qn] -= cy2;
342
578
      }
343
23.4k
    else
344
23.4k
      {
345
23.4k
        n2p[qn] = cy1 - cy2; /* & GMP_NUMB_MASK; */
346
347
23.4k
        quotient_too_large = (cy1 < cy2);
348
23.4k
        ++rn;
349
23.4k
      }
350
24.0k
    --in;
351
24.0k
        }
352
      /* True: partial remainder now is neutral, i.e., it is not shifted up.  */
353
354
37.4k
      tp = TMP_ALLOC_LIMBS (dn);
355
356
37.4k
      if (in < qn)
357
12.0k
        {
358
12.0k
    if (in == 0)
359
1.60k
      {
360
1.60k
        MPN_COPY (rp, n2p, rn);
361
1.60k
        ASSERT_ALWAYS (rn == dn);
362
1.60k
        goto foo;
363
1.60k
      }
364
10.4k
    mpn_mul (tp, qp, qn, dp, in);
365
10.4k
        }
366
25.3k
      else
367
25.3k
        mpn_mul (tp, dp, in, qp, qn);
368
369
35.7k
      cy = mpn_sub (n2p, n2p, rn, tp + in, qn);
370
35.7k
      MPN_COPY (rp + in, n2p, dn - in);
371
35.7k
      quotient_too_large |= cy;
372
35.7k
      cy = mpn_sub_n (rp, np, tp, in);
373
35.7k
      cy = mpn_sub_1 (rp + in, rp + in, rn, cy);
374
35.7k
      quotient_too_large |= cy;
375
37.4k
    foo:
376
37.4k
      if (quotient_too_large)
377
1.60k
        {
378
1.60k
    mpn_decr_u (qp, (mp_limb_t) 1);
379
1.60k
    mpn_add_n (rp, rp, dp, dn);
380
1.60k
        }
381
37.4k
    }
382
37.4k
  TMP_FREE;
383
37.4k
  return;
384
35.7k
      }
385
123k
    }
386
123k
}