Coverage Report

Created: 2026-06-30 06:42

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
88.0k
{
50
88.0k
  ASSERT_ALWAYS (qxn == 0);
51
52
88.0k
  ASSERT (nn >= 0);
53
88.0k
  ASSERT (dn >= 0);
54
88.0k
  ASSERT (dn == 0 || dp[dn - 1] != 0);
55
88.0k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, np, nn));
56
88.0k
  ASSERT (! MPN_OVERLAP_P (qp, nn - dn + 1 + qxn, dp, dn));
57
58
88.0k
  switch (dn)
59
88.0k
    {
60
0
    case 0:
61
0
      DIVIDE_BY_ZERO;
62
63
8.84k
    case 1:
64
8.84k
      {
65
8.84k
  rp[0] = mpn_divrem_1 (qp, (mp_size_t) 0, np, nn, dp[0]);
66
8.84k
  return;
67
0
      }
68
69
3.89k
    case 2:
70
3.89k
      {
71
3.89k
  mp_ptr n2p;
72
3.89k
  mp_limb_t qhl, cy;
73
3.89k
  TMP_DECL;
74
3.89k
  TMP_MARK;
75
3.89k
  if ((dp[1] & GMP_NUMB_HIGHBIT) == 0)
76
3.12k
    {
77
3.12k
      int cnt;
78
3.12k
      mp_limb_t d2p[2];
79
3.12k
      count_leading_zeros (cnt, dp[1]);
80
3.12k
      cnt -= GMP_NAIL_BITS;
81
3.12k
      d2p[1] = (dp[1] << cnt) | (dp[0] >> (GMP_NUMB_BITS - cnt));
82
3.12k
      d2p[0] = (dp[0] << cnt) & GMP_NUMB_MASK;
83
3.12k
      n2p = TMP_ALLOC_LIMBS (nn + 1);
84
3.12k
      cy = mpn_lshift (n2p, np, nn, cnt);
85
3.12k
      n2p[nn] = cy;
86
3.12k
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn + (cy != 0), d2p);
87
3.12k
      if (cy == 0)
88
1.11k
        qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
89
3.12k
      rp[0] = (n2p[0] >> cnt)
90
3.12k
        | ((n2p[1] << (GMP_NUMB_BITS - cnt)) & GMP_NUMB_MASK);
91
3.12k
      rp[1] = (n2p[1] >> cnt);
92
3.12k
    }
93
776
  else
94
776
    {
95
776
      n2p = TMP_ALLOC_LIMBS (nn);
96
776
      MPN_COPY (n2p, np, nn);
97
776
      qhl = mpn_divrem_2 (qp, 0L, n2p, nn, dp);
98
776
      qp[nn - 2] = qhl; /* always store nn-2+1 quotient limbs */
99
776
      rp[0] = n2p[0];
100
776
      rp[1] = n2p[1];
101
776
    }
102
3.89k
  TMP_FREE;
103
3.89k
  return;
104
0
      }
105
106
75.3k
    default:
107
75.3k
      {
108
75.3k
  int adjust;
109
75.3k
  gmp_pi1_t dinv;
110
75.3k
  TMP_DECL;
111
75.3k
  TMP_MARK;
112
75.3k
  adjust = np[nn - 1] >= dp[dn - 1];  /* conservative tests for quotient size */
113
75.3k
  if (nn + adjust >= 2 * dn)
114
43.0k
    {
115
43.0k
      mp_ptr n2p, d2p;
116
43.0k
      mp_limb_t cy;
117
43.0k
      int cnt;
118
119
43.0k
      qp[nn - dn] = 0;        /* zero high quotient limb */
120
43.0k
      if ((dp[dn - 1] & GMP_NUMB_HIGHBIT) == 0) /* normalize divisor */
121
5.19k
        {
122
5.19k
    count_leading_zeros (cnt, dp[dn - 1]);
123
5.19k
    cnt -= GMP_NAIL_BITS;
124
5.19k
    d2p = TMP_ALLOC_LIMBS (dn);
125
5.19k
    mpn_lshift (d2p, dp, dn, cnt);
126
5.19k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
127
5.19k
    cy = mpn_lshift (n2p, np, nn, cnt);
128
5.19k
    n2p[nn] = cy;
129
5.19k
    nn += adjust;
130
5.19k
        }
131
37.8k
      else
132
37.8k
        {
133
37.8k
    cnt = 0;
134
37.8k
    d2p = (mp_ptr) dp;
135
37.8k
    n2p = TMP_ALLOC_LIMBS (nn + 1);
136
37.8k
    MPN_COPY (n2p, np, nn);
137
37.8k
    n2p[nn] = 0;
138
37.8k
    nn += adjust;
139
37.8k
        }
140
141
43.0k
      invert_pi1 (dinv, d2p[dn - 1], d2p[dn - 2]);
142
43.0k
      if (BELOW_THRESHOLD (dn, DC_DIV_QR_THRESHOLD))
143
42.8k
        mpn_sbpi1_div_qr (qp, n2p, nn, d2p, dn, dinv.inv32);
144
252
      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
252
        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
43.0k
      if (cnt != 0)
158
5.19k
        mpn_rshift (rp, n2p, dn, cnt);
159
37.8k
      else
160
37.8k
        MPN_COPY (rp, n2p, dn);
161
43.0k
      TMP_FREE;
162
43.0k
      return;
163
43.0k
    }
164
165
  /* When we come here, the numerator/partial remainder is less
166
     than twice the size of the denominator.  */
167
168
32.2k
    {
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
32.2k
      mp_size_t qn;
202
32.2k
      mp_ptr n2p, d2p;
203
32.2k
      mp_ptr tp;
204
32.2k
      mp_limb_t cy;
205
32.2k
      mp_size_t in, rn;
206
32.2k
      mp_limb_t quotient_too_large;
207
32.2k
      unsigned int cnt;
208
209
32.2k
      qn = nn - dn;
210
32.2k
      qp[qn] = 0;       /* zero high quotient limb */
211
32.2k
      qn += adjust;     /* qn cannot become bigger */
212
213
32.2k
      if (qn == 0)
214
249
        {
215
249
    MPN_COPY (rp, np, dn);
216
249
    TMP_FREE;
217
249
    return;
218
249
        }
219
220
32.0k
      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
32.0k
      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
13.7k
      {
237
13.7k
        n2p[2 * qn] = cy;
238
13.7k
        n2p++;
239
13.7k
      }
240
10.3k
    else
241
10.3k
      {
242
10.3k
        n2p[0] |= np[nn - 2 * qn - 1] >> (GMP_NUMB_BITS - cnt);
243
10.3k
      }
244
24.0k
        }
245
7.93k
      else
246
7.93k
        {
247
7.93k
    cnt = 0;
248
7.93k
    d2p = (mp_ptr) dp + in;
249
250
7.93k
    n2p = TMP_ALLOC_LIMBS (2 * qn + 1);
251
7.93k
    MPN_COPY (n2p, np + nn - 2 * qn, 2 * qn);
252
7.93k
    if (adjust)
253
840
      {
254
840
        n2p[2 * qn] = 0;
255
840
        n2p++;
256
840
      }
257
7.93k
        }
258
259
      /* Get an approximate quotient using the extracted operands.  */
260
32.0k
      if (qn == 1)
261
7.67k
        {
262
7.67k
    mp_limb_t q0, r0;
263
7.67k
    udiv_qrnnd (q0, r0, n2p[1], n2p[0] << GMP_NAIL_BITS, d2p[0] << GMP_NAIL_BITS);
264
7.67k
    n2p[0] = r0 >> GMP_NAIL_BITS;
265
7.67k
    qp[0] = q0;
266
7.67k
        }
267
24.3k
      else if (qn == 2)
268
8.33k
        mpn_divrem_2 (qp, 0L, n2p, 4L, d2p); /* FIXME: obsolete function */
269
15.9k
      else
270
15.9k
        {
271
15.9k
    invert_pi1 (dinv, d2p[qn - 1], d2p[qn - 2]);
272
15.9k
    if (BELOW_THRESHOLD (qn, DC_DIV_QR_THRESHOLD))
273
14.9k
      mpn_sbpi1_div_qr (qp, n2p, 2 * qn, d2p, qn, dinv.inv32);
274
1.05k
    else if (BELOW_THRESHOLD (qn, MU_DIV_QR_THRESHOLD))
275
1.05k
      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
15.9k
        }
287
288
32.0k
      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
32.0k
      {
295
32.0k
        mp_limb_t dl, x;
296
32.0k
        mp_limb_t h, dummy;
297
298
32.0k
        if (in - 2 < 0)
299
1.94k
    dl = 0;
300
30.0k
        else
301
30.0k
    dl = dp[in - 2];
302
303
32.0k
#if GMP_NAIL_BITS == 0
304
32.0k
        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
32.0k
        umul_ppmm (h, dummy, x, qp[qn - 1] << GMP_NAIL_BITS);
311
312
32.0k
        if (n2p[qn - 1] < h)
313
2.28k
    {
314
2.28k
      mp_limb_t cy;
315
316
2.28k
      mpn_decr_u (qp, (mp_limb_t) 1);
317
2.28k
      cy = mpn_add_n (n2p, n2p, d2p, qn);
318
2.28k
      if (cy)
319
750
        {
320
          /* The partial remainder is safely large.  */
321
750
          n2p[qn] = cy;
322
750
          ++rn;
323
750
        }
324
2.28k
    }
325
32.0k
      }
326
327
32.0k
      quotient_too_large = 0;
328
32.0k
      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
538
      {
340
538
        ASSERT_ALWAYS (n2p[qn] >= cy2);
341
538
        n2p[qn] -= cy2;
342
538
      }
343
23.5k
    else
344
23.5k
      {
345
23.5k
        n2p[qn] = cy1 - cy2; /* & GMP_NUMB_MASK; */
346
347
23.5k
        quotient_too_large = (cy1 < cy2);
348
23.5k
        ++rn;
349
23.5k
      }
350
24.0k
    --in;
351
24.0k
        }
352
      /* True: partial remainder now is neutral, i.e., it is not shifted up.  */
353
354
32.0k
      tp = TMP_ALLOC_LIMBS (dn);
355
356
32.0k
      if (in < qn)
357
9.36k
        {
358
9.36k
    if (in == 0)
359
1.64k
      {
360
1.64k
        MPN_COPY (rp, n2p, rn);
361
1.64k
        ASSERT_ALWAYS (rn == dn);
362
1.64k
        goto foo;
363
1.64k
      }
364
7.71k
    mpn_mul (tp, qp, qn, dp, in);
365
7.71k
        }
366
22.6k
      else
367
22.6k
        mpn_mul (tp, dp, in, qp, qn);
368
369
30.3k
      cy = mpn_sub (n2p, n2p, rn, tp + in, qn);
370
30.3k
      MPN_COPY (rp + in, n2p, dn - in);
371
30.3k
      quotient_too_large |= cy;
372
30.3k
      cy = mpn_sub_n (rp, np, tp, in);
373
30.3k
      cy = mpn_sub_1 (rp + in, rp + in, rn, cy);
374
30.3k
      quotient_too_large |= cy;
375
32.0k
    foo:
376
32.0k
      if (quotient_too_large)
377
1.63k
        {
378
1.63k
    mpn_decr_u (qp, (mp_limb_t) 1);
379
1.63k
    mpn_add_n (rp, rp, dp, dn);
380
1.63k
        }
381
32.0k
    }
382
32.0k
  TMP_FREE;
383
32.0k
  return;
384
30.3k
      }
385
88.0k
    }
386
88.0k
}