1 From 57700c26d73cf7fa6f5cfaec1145eccf388acab9 Mon Sep 17 00:00:00 2001
2 From: Nikos Mavrogiannopoulos <nmav@gnutls.org>
3 Date: Sun, 9 Mar 2014 11:27:42 +0100
4 Subject: [PATCH 1/5] Updated mini-gmp
7 mini-gmp.c | 890 +++++++++++++++++++++++++++++++++++++++----------------------
9 2 files changed, 618 insertions(+), 327 deletions(-)
11 diff --git a/mini-gmp.c b/mini-gmp.c
12 index 8b6f070..766df30 100644
17 Contributed to the GNU project by Niels Möller
19 -Copyright 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1999, 2000, 2001,
20 -2002, 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 2013
21 -Free Software Foundation, Inc.
22 +Copyright 1991-1997, 1999-2014 Free Software Foundation, Inc.
24 This file is part of the GNU MP Library.
26 The GNU MP Library is free software; you can redistribute it and/or modify
27 -it under the terms of the GNU Lesser General Public License as published by
28 -the Free Software Foundation; either version 3 of the License, or (at your
29 -option) any later version.
30 +it under the terms of either:
32 + * the GNU Lesser General Public License as published by the Free
33 + Software Foundation; either version 3 of the License, or (at your
34 + option) any later version.
38 + * the GNU General Public License as published by the Free Software
39 + Foundation; either version 2 of the License, or (at your option) any
42 +or both in parallel, as here.
44 The GNU MP Library is distributed in the hope that it will be useful, but
45 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
46 -or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
47 -License for more details.
48 +or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
51 -You should have received a copy of the GNU Lesser General Public License
52 -along with the GNU MP Library. If not, see http://www.gnu.org/licenses/. */
53 +You should have received copies of the GNU General Public License and the
54 +GNU Lesser General Public License along with the GNU MP Library. If not,
55 +see https://www.gnu.org/licenses/. */
57 /* NOTE: All functions in this file which are not declared in
58 mini-gmp.h are internal, and are not intended to be compatible
59 @@ -222,11 +231,13 @@ along with the GNU MP Library. If not, see http://www.gnu.org/licenses/. */
61 #define MPZ_SRCPTR_SWAP(x, y) \
63 - mpz_srcptr __mpz_srcptr_swap__tmp = (x); \
64 + mpz_srcptr __mpz_srcptr_swap__tmp = (x); \
66 (y) = __mpz_srcptr_swap__tmp; \
69 +const int mp_bits_per_limb = GMP_LIMB_BITS;
72 /* Memory allocation and other helper functions. */
74 @@ -342,12 +353,10 @@ mpn_copyd (mp_ptr d, mp_srcptr s, mp_size_t n)
76 mpn_cmp (mp_srcptr ap, mp_srcptr bp, mp_size_t n)
81 - if (ap[n-1] < bp[n-1])
83 - else if (ap[n-1] > bp[n-1])
86 + return ap[n] > bp[n] ? 1 : -1;
90 @@ -355,10 +364,8 @@ mpn_cmp (mp_srcptr ap, mp_srcptr bp, mp_size_t n)
92 mpn_cmp4 (mp_srcptr ap, mp_size_t an, mp_srcptr bp, mp_size_t bn)
99 + return an < bn ? -1 : 1;
101 return mpn_cmp (ap, bp, an);
103 @@ -373,20 +380,31 @@ mpn_normalized_size (mp_srcptr xp, mp_size_t n)
105 #define mpn_zero_p(xp, n) (mpn_normalized_size ((xp), (n)) == 0)
108 +mpn_zero (mp_ptr rp, mp_size_t n)
112 + for (i = 0; i < n; i++)
117 mpn_add_1 (mp_ptr rp, mp_srcptr ap, mp_size_t n, mp_limb_t b)
123 - for (i = 0; i < n; i++)
127 mp_limb_t r = ap[i] + b;
137 @@ -429,7 +447,8 @@ mpn_sub_1 (mp_ptr rp, mp_srcptr ap, mp_size_t n, mp_limb_t b)
141 - for (i = 0; i < n; i++)
147 @@ -437,6 +456,8 @@ mpn_sub_1 (mp_ptr rp, mp_srcptr ap, mp_size_t n, mp_limb_t b)
156 @@ -602,7 +623,7 @@ mpn_lshift (mp_ptr rp, mp_srcptr up, mp_size_t n, unsigned int cnt)
157 retval = low_limb >> tnc;
158 high_limb = (low_limb << cnt);
160 - for (i = n - 1; i != 0; i--)
161 + for (i = n; --i != 0;)
164 *--rp = high_limb | (low_limb >> tnc);
165 @@ -630,7 +651,7 @@ mpn_rshift (mp_ptr rp, mp_srcptr up, mp_size_t n, unsigned int cnt)
166 retval = (high_limb << tnc);
167 low_limb = high_limb >> cnt;
169 - for (i = n - 1; i != 0; i--)
170 + for (i = n; --i != 0;)
173 *rp++ = low_limb | (high_limb << tnc);
174 @@ -641,6 +662,46 @@ mpn_rshift (mp_ptr rp, mp_srcptr up, mp_size_t n, unsigned int cnt)
179 +mpn_common_scan (mp_limb_t limb, mp_size_t i, mp_srcptr up, mp_size_t un,
184 + assert (ux == 0 || ux == GMP_LIMB_MAX);
185 + assert (0 <= i && i <= un );
191 + return (ux == 0 ? ~(mp_bitcnt_t) 0 : un * GMP_LIMB_BITS);
194 + gmp_ctz (cnt, limb);
195 + return (mp_bitcnt_t) i * GMP_LIMB_BITS + cnt;
199 +mpn_scan1 (mp_srcptr ptr, mp_bitcnt_t bit)
202 + i = bit / GMP_LIMB_BITS;
204 + return mpn_common_scan ( ptr[i] & (GMP_LIMB_MAX << (bit % GMP_LIMB_BITS)),
209 +mpn_scan0 (mp_srcptr ptr, mp_bitcnt_t bit)
212 + i = bit / GMP_LIMB_BITS;
214 + return mpn_common_scan (~ptr[i] & (GMP_LIMB_MAX << (bit % GMP_LIMB_BITS)),
215 + i, ptr, i, GMP_LIMB_MAX);
219 /* MPN division interface. */
221 @@ -715,8 +776,7 @@ mpn_invert_3by2 (mp_limb_t u1, mp_limb_t u0)
225 - if (r > u1 || (r == u1 && tl > u0))
227 + m -= ((r > u1) | ((r == u1) & (tl > u0)));
231 @@ -836,14 +896,20 @@ mpn_div_qr_1 (mp_ptr qp, mp_srcptr np, mp_size_t nn, mp_limb_t d)
234 /* Special case for powers of two. */
235 - if (d > 1 && (d & (d-1)) == 0)
236 + if ((d & (d-1)) == 0)
239 mp_limb_t r = np[0] & (d-1);
240 - gmp_ctz (shift, d);
242 - mpn_rshift (qp, np, nn, shift);
246 + mpn_copyi (qp, np, nn);
250 + gmp_ctz (shift, d);
251 + mpn_rshift (qp, np, nn, shift);
257 @@ -880,7 +946,8 @@ mpn_div_qr_2_preinv (mp_ptr qp, mp_ptr rp, mp_srcptr np, mp_size_t nn,
261 - for (i = nn - 2; i >= 0; i--)
267 @@ -889,6 +956,7 @@ mpn_div_qr_2_preinv (mp_ptr qp, mp_ptr rp, mp_srcptr np, mp_size_t nn,
275 @@ -930,18 +998,19 @@ mpn_div_qr_pi1 (mp_ptr qp,
279 - assert ((dp[dn-1] & GMP_LIMB_HIGHBIT) != 0);
284 + assert ((d1 & GMP_LIMB_HIGHBIT) != 0);
285 /* Iteration variable is the index of the q limb.
287 * We divide <n1, np[dn-1+i], np[dn-2+i], np[dn-3+i],..., np[i]>
288 * by <d1, d0, dp[dn-3], ..., dp[0] >
291 - for (i = nn - dn; i >= 0; i--)
295 mp_limb_t n0 = np[dn-1+i];
297 @@ -973,6 +1042,7 @@ mpn_div_qr_pi1 (mp_ptr qp,
305 @@ -994,7 +1064,9 @@ mpn_div_qr_preinv (mp_ptr qp, mp_ptr np, mp_size_t nn,
309 - assert (dp[dn-1] & GMP_LIMB_HIGHBIT);
310 + assert (inv->d1 == dp[dn-1]);
311 + assert (inv->d0 == dp[dn-2]);
312 + assert ((inv->d1 & GMP_LIMB_HIGHBIT) != 0);
316 @@ -1002,9 +1074,6 @@ mpn_div_qr_preinv (mp_ptr qp, mp_ptr np, mp_size_t nn,
320 - assert (inv->d1 == dp[dn-1]);
321 - assert (inv->d0 == dp[dn-2]);
323 mpn_div_qr_pi1 (qp, np, nn, nh, dp, dn, inv->di);
326 @@ -1238,15 +1307,14 @@ mpn_set_str_other (mp_ptr rp, const unsigned char *sp, size_t sn,
334 - first = 1 + (sn - 1) % info->exp;
335 + k = 1 + (sn - 1) % info->exp;
339 - for (k = 1; k < first; k++)
344 @@ -1300,7 +1368,7 @@ mpz_init (mpz_t r)
347 /* The utility of this function is a bit limited, since many functions
348 - assings the result variable using mpz_swap. */
349 + assigns the result variable using mpz_swap. */
351 mpz_init2 (mpz_t r, mp_bitcnt_t bits)
353 @@ -1422,7 +1490,7 @@ mpz_fits_ulong_p (const mpz_t u)
355 mp_size_t us = u->_mp_size;
357 - return us == 0 || us == 1;
358 + return (us == (us > 0));
362 @@ -1459,6 +1527,48 @@ mpz_getlimbn (const mpz_t u, mp_size_t n)
367 +mpz_realloc2 (mpz_t x, mp_bitcnt_t n)
369 + mpz_realloc (x, 1 + (n - (n != 0)) / GMP_LIMB_BITS);
373 +mpz_limbs_read (mpz_srcptr x)
379 +mpz_limbs_modify (mpz_t x, mp_size_t n)
382 + return MPZ_REALLOC (x, n);
386 +mpz_limbs_write (mpz_t x, mp_size_t n)
388 + return mpz_limbs_modify (x, n);
392 +mpz_limbs_finish (mpz_t x, mp_size_t xs)
395 + xn = mpn_normalized_size (x->_mp_d, GMP_ABS (xs));
396 + x->_mp_size = xs < 0 ? -xn : xn;
400 +mpz_roinit_n (mpz_t x, mp_srcptr xp, mp_size_t xs)
403 + x->_mp_d = (mp_ptr) xp;
404 + mpz_limbs_finish (x, xs);
409 /* Conversions and comparison to double. */
411 @@ -1473,19 +1583,15 @@ mpz_set_d (mpz_t r, double x)
413 /* x != x is true when x is a NaN, and x == x * 0.5 is true when x is
415 - if (x == 0.0 || x != x || x == x * 0.5)
416 + if (x != x || x == x * 0.5)
435 @@ -1502,8 +1608,9 @@ mpz_set_d (mpz_t r, double x)
440 - for (i = rn-1; i-- > 0; )
447 @@ -1611,12 +1718,7 @@ mpz_sgn (const mpz_t u)
449 mp_size_t usize = u->_mp_size;
453 - else if (usize < 0)
457 + return (usize > 0) - (usize < 0);
461 @@ -1635,10 +1737,9 @@ mpz_cmp_si (const mpz_t u, long v)
462 mp_limb_t ul = u->_mp_d[0];
463 if ((mp_limb_t)GMP_NEG_CAST (unsigned long int, v) < ul)
465 - else if ( (mp_limb_t)GMP_NEG_CAST (unsigned long int, v) > ul)
468 + return (mp_limb_t)GMP_NEG_CAST (unsigned long int, v) > ul;
474 @@ -1653,12 +1754,8 @@ mpz_cmp_ui (const mpz_t u, unsigned long v)
477 mp_limb_t ul = (usize > 0) ? u->_mp_d[0] : 0;
482 + return (ul > v) - (ul < v);
488 @@ -1667,16 +1764,12 @@ mpz_cmp (const mpz_t a, const mpz_t b)
489 mp_size_t asize = a->_mp_size;
490 mp_size_t bsize = b->_mp_size;
494 - else if (asize < bsize)
496 - else if (asize > 0)
497 + if (asize != bsize)
498 + return (asize < bsize) ? -1 : 1;
499 + else if (asize >= 0)
500 return mpn_cmp (a->_mp_d, b->_mp_d, asize);
501 - else if (asize < 0)
502 - return -mpn_cmp (a->_mp_d, b->_mp_d, -asize);
505 + return mpn_cmp (b->_mp_d, a->_mp_d, -asize);
509 @@ -1690,12 +1783,7 @@ mpz_cmpabs_ui (const mpz_t u, unsigned long v)
511 ul = (un == 1) ? u->_mp_d[0] : 0;
519 + return (ul > v) - (ul < v);
523 @@ -1753,7 +1841,7 @@ mpz_abs_add_ui (mpz_t r, const mpz_t a, unsigned long b)
525 cy = mpn_add_1 (rp, a->_mp_d, an, b);
532 @@ -1815,20 +1903,21 @@ mpz_abs_add (mpz_t r, const mpz_t a, const mpz_t b)
534 mp_size_t an = GMP_ABS (a->_mp_size);
535 mp_size_t bn = GMP_ABS (b->_mp_size);
540 - rn = GMP_MAX (an, bn);
541 - rp = MPZ_REALLOC (r, rn + 1);
543 - cy = mpn_add (rp, a->_mp_d, an, b->_mp_d, bn);
545 - cy = mpn_add (rp, b->_mp_d, bn, a->_mp_d, an);
548 + MPZ_SRCPTR_SWAP (a, b);
549 + MP_SIZE_T_SWAP (an, bn);
553 + rp = MPZ_REALLOC (r, an + 1);
554 + cy = mpn_add (rp, a->_mp_d, an, b->_mp_d, bn);
558 - return rn + (cy > 0);
563 @@ -1899,31 +1988,26 @@ mpz_mul_si (mpz_t r, const mpz_t u, long int v)
565 mpz_mul_ui (mpz_t r, const mpz_t u, unsigned long int v)
573 - un = GMP_ABS (u->_mp_size);
576 - if (un == 0 || v == 0)
577 + if (us == 0 || v == 0)
583 - mpz_init2 (t, (un + 1) * GMP_LIMB_BITS);
587 + tp = MPZ_REALLOC (r, un + 1);
588 cy = mpn_mul_1 (tp, u->_mp_d, un, v);
591 - t->_mp_size = un + (cy > 0);
592 - if (u->_mp_size < 0)
593 - t->_mp_size = - t->_mp_size;
598 + r->_mp_size = (us < 0) ? - un : un;
602 @@ -1934,8 +2018,8 @@ mpz_mul (mpz_t r, const mpz_t u, const mpz_t v)
606 - un = GMP_ABS (u->_mp_size);
607 - vn = GMP_ABS (v->_mp_size);
611 if (un == 0 || vn == 0)
613 @@ -1943,7 +2027,10 @@ mpz_mul (mpz_t r, const mpz_t u, const mpz_t v)
617 - sign = (u->_mp_size ^ v->_mp_size) < 0;
618 + sign = (un ^ vn) < 0;
623 mpz_init2 (t, (un + vn) * GMP_LIMB_BITS);
625 @@ -1996,6 +2083,46 @@ mpz_mul_2exp (mpz_t r, const mpz_t u, mp_bitcnt_t bits)
626 r->_mp_size = (u->_mp_size < 0) ? - rn : rn;
630 +mpz_addmul_ui (mpz_t r, const mpz_t u, unsigned long int v)
634 + mpz_mul_ui (t, u, v);
640 +mpz_submul_ui (mpz_t r, const mpz_t u, unsigned long int v)
644 + mpz_mul_ui (t, u, v);
650 +mpz_addmul (mpz_t r, const mpz_t u, const mpz_t v)
660 +mpz_submul (mpz_t r, const mpz_t u, const mpz_t v)
671 enum mpz_div_round_mode { GMP_DIV_FLOOR, GMP_DIV_CEIL, GMP_DIV_TRUNC };
672 @@ -2060,8 +2187,7 @@ mpz_div_qr (mpz_t q, mpz_t r,
678 + mpz_init_set (tr, n);
682 @@ -2171,10 +2297,7 @@ mpz_tdiv_r (mpz_t r, const mpz_t n, const mpz_t d)
684 mpz_mod (mpz_t r, const mpz_t n, const mpz_t d)
686 - if (d->_mp_size >= 0)
687 - mpz_div_qr (NULL, r, n, d, GMP_DIV_FLOOR);
689 - mpz_div_qr (NULL, r, n, d, GMP_DIV_CEIL);
690 + mpz_div_qr (NULL, r, n, d, d->_mp_size >= 0 ? GMP_DIV_FLOOR : GMP_DIV_CEIL);
694 @@ -2184,7 +2307,7 @@ mpz_div_q_2exp (mpz_t q, const mpz_t u, mp_bitcnt_t bit_index,
703 @@ -2226,7 +2349,8 @@ mpz_div_q_2exp (mpz_t q, const mpz_t u, mp_bitcnt_t bit_index,
707 - mpz_add_ui (q, q, adjust);
709 + mpz_add_ui (q, q, 1);
713 @@ -2303,7 +2427,7 @@ mpz_div_r_2exp (mpz_t r, const mpz_t u, mp_bitcnt_t bit_index,
715 /* r > 0, need to flip sign. */
717 - for (i++; i < rn; i++)
722 @@ -2366,6 +2490,24 @@ mpz_divisible_p (const mpz_t n, const mpz_t d)
723 return mpz_div_qr (NULL, NULL, n, d, GMP_DIV_TRUNC) == 0;
727 +mpz_congruent_p (const mpz_t a, const mpz_t b, const mpz_t m)
732 + /* a == b (mod 0) iff a == b */
733 + if (mpz_sgn (m) == 0)
734 + return (mpz_cmp (a, b) == 0);
738 + res = mpz_divisible_p (t, m);
745 mpz_div_qr_ui (mpz_t q, mpz_t r,
746 const mpz_t n, unsigned long d, enum mpz_div_round_mode mode)
747 @@ -2579,32 +2721,16 @@ mpz_gcd_ui (mpz_t g, const mpz_t u, unsigned long v)
751 -mpz_make_odd (mpz_t r, const mpz_t u)
752 +mpz_make_odd (mpz_t r)
754 - mp_size_t un, rn, i;
758 - un = GMP_ABS (u->_mp_size);
762 - for (i = 0; u->_mp_d[i] == 0; i++)
765 - gmp_ctz (shift, u->_mp_d[i]);
768 - rp = MPZ_REALLOC (r, rn);
771 - mpn_rshift (rp, u->_mp_d + i, rn, shift);
772 - rn -= (rp[rn-1] == 0);
775 - mpn_copyi (rp, u->_mp_d + i, rn);
776 + assert (r->_mp_size > 0);
777 + /* Count trailing zeros, equivalent to mpn_scan1, because we know that there is a 1 */
778 + shift = mpn_common_scan (r->_mp_d[0], 0, r->_mp_d, 0, 0);
779 + mpz_tdiv_q_2exp (r, r, shift);
782 - return i * GMP_LIMB_BITS + shift;
787 @@ -2627,8 +2753,10 @@ mpz_gcd (mpz_t g, const mpz_t u, const mpz_t v)
791 - uz = mpz_make_odd (tu, u);
792 - vz = mpz_make_odd (tv, v);
794 + uz = mpz_make_odd (tu);
796 + vz = mpz_make_odd (tv);
797 gz = GMP_MIN (uz, vz);
799 if (tu->_mp_size < tv->_mp_size)
800 @@ -2644,7 +2772,7 @@ mpz_gcd (mpz_t g, const mpz_t u, const mpz_t v)
804 - mpz_make_odd (tu, tu);
806 c = mpz_cmp (tu, tv);
809 @@ -2706,8 +2834,10 @@ mpz_gcdext (mpz_t g, mpz_t s, mpz_t t, const mpz_t u, const mpz_t v)
813 - uz = mpz_make_odd (tu, u);
814 - vz = mpz_make_odd (tv, v);
816 + uz = mpz_make_odd (tu);
818 + vz = mpz_make_odd (tv);
819 gz = GMP_MIN (uz, vz);
822 @@ -2755,7 +2885,7 @@ mpz_gcdext (mpz_t g, mpz_t s, mpz_t t, const mpz_t u, const mpz_t v)
823 if (tu->_mp_size > 0)
826 - shift = mpz_make_odd (tu, tu);
827 + shift = mpz_make_odd (tu);
828 mpz_mul_2exp (t0, t0, shift);
829 mpz_mul_2exp (s0, s0, shift);
831 @@ -2778,7 +2908,7 @@ mpz_gcdext (mpz_t g, mpz_t s, mpz_t t, const mpz_t u, const mpz_t v)
832 mpz_add (t0, t0, t1);
833 mpz_add (s0, s0, s1);
835 - shift = mpz_make_odd (tv, tv);
836 + shift = mpz_make_odd (tv);
837 mpz_mul_2exp (t1, t1, shift);
838 mpz_mul_2exp (s1, s1, shift);
840 @@ -2788,7 +2918,7 @@ mpz_gcdext (mpz_t g, mpz_t s, mpz_t t, const mpz_t u, const mpz_t v)
841 mpz_add (t1, t0, t1);
842 mpz_add (s1, s0, s1);
844 - shift = mpz_make_odd (tu, tu);
845 + shift = mpz_make_odd (tu);
846 mpz_mul_2exp (t0, t0, shift);
847 mpz_mul_2exp (s0, s0, shift);
849 @@ -2926,12 +3056,16 @@ mpz_pow_ui (mpz_t r, const mpz_t b, unsigned long e)
851 mpz_init_set_ui (tr, 1);
853 - for (bit = GMP_ULONG_HIGHBIT; bit > 0; bit >>= 1)
854 + bit = GMP_ULONG_HIGHBIT;
857 mpz_mul (tr, tr, tr);
867 @@ -2987,7 +3121,7 @@ mpz_powm (mpz_t r, const mpz_t b, const mpz_t e, const mpz_t m)
870 if (!mpz_invert (base, b, m))
871 - gmp_die ("mpz_powm: Negative exponent and non-invertibe base.");
872 + gmp_die ("mpz_powm: Negative exponent and non-invertible base.");
876 @@ -3019,7 +3153,8 @@ mpz_powm (mpz_t r, const mpz_t b, const mpz_t e, const mpz_t m)
877 mp_limb_t w = e->_mp_d[en];
880 - for (bit = GMP_LIMB_HIGHBIT; bit > 0; bit >>= 1)
881 + bit = GMP_LIMB_HIGHBIT;
884 mpz_mul (tr, tr, tr);
886 @@ -3029,7 +3164,9 @@ mpz_powm (mpz_t r, const mpz_t b, const mpz_t e, const mpz_t m)
887 mpn_div_qr_preinv (NULL, tr->_mp_d, tr->_mp_size, mp, mn, &minv);
888 tr->_mp_size = mpn_normalized_size (tr->_mp_d, mn);
895 /* Final reduction */
896 @@ -3064,21 +3201,26 @@ mpz_rootrem (mpz_t x, mpz_t r, const mpz_t y, unsigned long z)
899 sgn = y->_mp_size < 0;
900 - if (sgn && (z & 1) == 0)
901 + if ((~z & sgn) != 0)
902 gmp_die ("mpz_rootrem: Negative argument, with even root.");
904 gmp_die ("mpz_rootrem: Zeroth root.");
906 if (mpz_cmpabs_ui (y, 1) <= 0) {
917 - mpz_setbit (t, mpz_sizeinbase (y, 2) / z + 1);
920 + tb = mpz_sizeinbase (y, 2) / z + 1;
922 + mpz_setbit (t, tb);
925 if (z == 2) /* simplify sqrt loop: z-1 == 1 */
927 @@ -3110,7 +3252,8 @@ mpz_rootrem (mpz_t x, mpz_t r, const mpz_t y, unsigned long z)
928 mpz_pow_ui (t, u, z);
937 @@ -3142,19 +3285,56 @@ mpz_sqrt (mpz_t s, const mpz_t u)
938 mpz_rootrem (s, NULL, u, 2);
942 +mpz_perfect_square_p (const mpz_t u)
944 + if (u->_mp_size <= 0)
945 + return (u->_mp_size == 0);
947 + return mpz_root (NULL, u, 2);
951 +mpn_perfect_square_p (mp_srcptr p, mp_size_t n)
956 + assert (p [n-1] != 0);
957 + return mpz_root (NULL, mpz_roinit_n (t, p, n), 2);
961 +mpn_sqrtrem (mp_ptr sp, mp_ptr rp, mp_srcptr p, mp_size_t n)
967 + assert (p [n-1] != 0);
971 + mpz_rootrem (s, r, mpz_roinit_n (u, p, n), 2);
973 + assert (s->_mp_size == (n+1)/2);
974 + mpn_copyd (sp, s->_mp_d, s->_mp_size);
978 + mpn_copyd (rp, r->_mp_d, res);
986 mpz_fac_ui (mpz_t x, unsigned long n)
994 - mpz_mul_ui (x, x, n);
995 + mpz_set_ui (x, n + (n == 0));
997 + mpz_mul_ui (x, x, --n);
1001 @@ -3162,25 +3342,120 @@ mpz_bin_uiui (mpz_t r, unsigned long n, unsigned long k)
1009 - mpz_fac_ui (r, n);
1010 + mpz_set_ui (r, k <= n);
1013 + k = (k <= n) ? n - k : 0;
1017 - mpz_divexact (r, r, t);
1018 - mpz_fac_ui (t, n - k);
1020 + for (; k > 0; k--)
1021 + mpz_mul_ui (r, r, n--);
1023 mpz_divexact (r, r, t);
1028 +/* Primality testing */
1030 +gmp_millerrabin (const mpz_t n, const mpz_t nm1, mpz_t y,
1031 + const mpz_t q, mp_bitcnt_t k)
1035 + /* Caller must initialize y to the base. */
1036 + mpz_powm (y, y, q, n);
1038 + if (mpz_cmp_ui (y, 1) == 0 || mpz_cmp (y, nm1) == 0)
1041 + for (i = 1; i < k; i++)
1043 + mpz_powm_ui (y, y, 2, n);
1044 + if (mpz_cmp (y, nm1) == 0)
1046 + if (mpz_cmp_ui (y, 1) == 0)
1052 +/* This product is 0xc0cfd797, and fits in 32 bits. */
1053 +#define GMP_PRIME_PRODUCT \
1054 + (3UL*5UL*7UL*11UL*13UL*17UL*19UL*23UL*29UL)
1056 +/* Bit (p+1)/2 is set, for each odd prime <= 61 */
1057 +#define GMP_PRIME_MASK 0xc96996dcUL
1060 +mpz_probab_prime_p (const mpz_t n, int reps)
1069 + /* Note that we use the absolute value of n only, for compatibility
1070 + with the real GMP. */
1071 + if (mpz_even_p (n))
1072 + return (mpz_cmpabs_ui (n, 2) == 0) ? 2 : 0;
1074 + /* Above test excludes n == 0 */
1075 + assert (n->_mp_size != 0);
1077 + if (mpz_cmpabs_ui (n, 64) < 0)
1078 + return (GMP_PRIME_MASK >> (n->_mp_d[0] >> 1)) & 2;
1080 + if (mpz_gcd_ui (NULL, n, GMP_PRIME_PRODUCT) != 1)
1083 + /* All prime factors are >= 31. */
1084 + if (mpz_cmpabs_ui (n, 31*31) < 0)
1087 + /* Use Miller-Rabin, with a deterministic sequence of bases, a[j] =
1088 + j^2 + j + 41 using Euler's polynomial. We potentially stop early,
1089 + if a[j] >= n - 1. Since n >= 31*31, this can happen only if reps >
1090 + 30 (a[30] == 971 > 31*31 == 961). */
1096 + /* Find q and k, where q is odd and n = 1 + 2**k * q. */
1098 + mpz_sub_ui (nm1, nm1, 1);
1099 + k = mpz_scan1 (nm1, 0);
1100 + mpz_tdiv_q_2exp (q, nm1, k);
1102 + for (j = 0, is_prime = 1; is_prime && j < reps; j++)
1104 + mpz_set_ui (y, (unsigned long) j*j+j+41);
1105 + if (mpz_cmp (y, nm1) >= 0)
1107 + /* Don't try any further bases. */
1111 + is_prime &= gmp_millerrabin (n, nm1, y, q, k);
1121 /* Logical operations and bit manipulation. */
1123 /* Numbers are treated as if represented in two's complement (and
1124 infinitely sign extended). For a negative values we get the two's
1125 - complement from -x = ~x + 1, where ~ is bitwise complementt.
1126 + complement from -x = ~x + 1, where ~ is bitwise complement.
1130 @@ -3374,7 +3649,8 @@ mpz_and (mpz_t r, const mpz_t u, const mpz_t v)
1134 - for (i = 0; i < vn; i++)
1138 ul = (up[i] ^ ux) + uc;
1140 @@ -3386,6 +3662,7 @@ mpz_and (mpz_t r, const mpz_t u, const mpz_t v)
1148 @@ -3445,7 +3722,8 @@ mpz_ior (mpz_t r, const mpz_t u, const mpz_t v)
1152 - for (i = 0; i < vn; i++)
1156 ul = (up[i] ^ ux) + uc;
1158 @@ -3457,6 +3735,7 @@ mpz_ior (mpz_t r, const mpz_t u, const mpz_t v)
1166 @@ -3512,7 +3791,8 @@ mpz_xor (mpz_t r, const mpz_t u, const mpz_t v)
1170 - for (i = 0; i < vn; i++)
1174 ul = (up[i] ^ ux) + uc;
1176 @@ -3524,6 +3804,7 @@ mpz_xor (mpz_t r, const mpz_t u, const mpz_t v)
1184 @@ -3561,20 +3842,28 @@ gmp_popcount_limb (mp_limb_t x)
1188 -mpz_popcount (const mpz_t u)
1189 +mpn_popcount (mp_srcptr p, mp_size_t n)
1195 + for (c = 0, i = 0; i < n; i++)
1196 + c += gmp_popcount_limb (p[i]);
1202 +mpz_popcount (const mpz_t u)
1209 return ~(mp_bitcnt_t) 0;
1211 - for (c = 0, i = 0; i < un; i++)
1212 - c += gmp_popcount_limb (u->_mp_d[i]);
1215 + return mpn_popcount (u->_mp_d, un);
1219 @@ -3591,16 +3880,13 @@ mpz_hamdist (const mpz_t u, const mpz_t v)
1221 return ~(mp_bitcnt_t) 0;
1224 + comp = - (uc = vc = (un < 0));
1231 - comp = - (mp_limb_t) 1;
1234 - uc = vc = comp = 0;
1238 @@ -3636,10 +3922,8 @@ mpz_scan1 (const mpz_t u, mp_bitcnt_t starting_bit)
1241 mp_size_t us, un, i;
1242 - mp_limb_t limb, ux, uc;
1244 + mp_limb_t limb, ux;
1249 i = starting_bit / GMP_LIMB_BITS;
1250 @@ -3649,36 +3933,24 @@ mpz_scan1 (const mpz_t u, mp_bitcnt_t starting_bit)
1252 return (us >= 0 ? ~(mp_bitcnt_t) 0 : starting_bit);
1256 - ux = GMP_LIMB_MAX;
1257 - uc = mpn_zero_p (up, i);
1262 - limb = (ux ^ up[i]) + uc;
1265 - /* Mask to 0 all bits before starting_bit, thus ignoring them. */
1266 - limb &= (GMP_LIMB_MAX << (starting_bit % GMP_LIMB_BITS));
1272 + if (starting_bit != 0)
1279 - /* For the u > 0 case, this can happen only for the first
1280 - masked limb. For the u < 0 case, it happens when the
1281 - highest limbs of the absolute value are all ones. */
1282 - return (us >= 0 ? ~(mp_bitcnt_t) 0 : un * GMP_LIMB_BITS);
1283 + ux = mpn_zero_p (up, i);
1284 + limb = ~ limb + ux;
1285 + ux = - (mp_limb_t) (limb >= ux);
1287 - limb = (ux ^ up[i]) + uc;
1290 + /* Mask to 0 all bits before starting_bit, thus ignoring them. */
1291 + limb &= (GMP_LIMB_MAX << (starting_bit % GMP_LIMB_BITS));
1293 - gmp_ctz (cnt, limb);
1294 - return (mp_bitcnt_t) i * GMP_LIMB_BITS + cnt;
1296 + return mpn_common_scan (limb, i, up, un, ux);
1300 @@ -3686,46 +3958,28 @@ mpz_scan0 (const mpz_t u, mp_bitcnt_t starting_bit)
1303 mp_size_t us, un, i;
1304 - mp_limb_t limb, ux, uc;
1306 + mp_limb_t limb, ux;
1310 + ux = - (mp_limb_t) (us >= 0);
1312 i = starting_bit / GMP_LIMB_BITS;
1314 /* When past end, there's an immediate 0 bit for u>=0, or no 0 bits for
1315 u<0. Notice this test picks up all cases of u==0 too. */
1317 - return (us >= 0 ? starting_bit : ~(mp_bitcnt_t) 0);
1318 + return (ux ? starting_bit : ~(mp_bitcnt_t) 0);
1322 - ux = GMP_LIMB_MAX;
1323 - uc = mpn_zero_p (up, i);
1328 + limb = up[i] ^ ux;
1330 - limb = (ux ^ up[i]) + uc;
1333 + limb -= mpn_zero_p (up, i); /* limb = ~(~limb + zero_p) */
1335 - /* Mask to 1 all bits before starting_bit, thus ignoring them. */
1336 - limb |= ((mp_limb_t) 1 << (starting_bit % GMP_LIMB_BITS)) - 1;
1337 + /* Mask all bits before starting_bit, thus ignoring them. */
1338 + limb &= (GMP_LIMB_MAX << (starting_bit % GMP_LIMB_BITS));
1340 - while (limb == GMP_LIMB_MAX)
1346 - return (us >= 0 ? un * GMP_LIMB_BITS : ~(mp_bitcnt_t) 0);
1348 - limb = (ux ^ up[i]) + uc;
1351 - gmp_ctz (cnt, ~limb);
1352 - return (mp_bitcnt_t) i * GMP_LIMB_BITS + cnt;
1353 + return mpn_common_scan (limb, i, up, un, ux);
1357 @@ -3771,11 +4025,15 @@ mpz_sizeinbase (const mpz_t u, int base)
1358 mpn_copyi (tp, up, un);
1359 mpn_div_qr_1_invert (&bi, base);
1361 - for (ndigits = 0; un > 0; ndigits++)
1366 mpn_div_qr_1_preinv (tp, tp, un, &bi);
1367 un -= (tp[un-1] == 0);
1374 @@ -3852,7 +4110,6 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1375 mp_size_t rn, alloc;
1382 @@ -3861,13 +4118,8 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1383 while (isspace( (unsigned char) *sp))
1393 + sign = (*sp == '-');
1398 @@ -3894,7 +4146,7 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1400 dp = gmp_xalloc (sn + (sn == 0));
1402 - for (dn = 0; *sp; sp++)
1403 + for (sn = 0; *sp; sp++)
1407 @@ -3916,7 +4168,7 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1415 bits = mpn_base_power_of_two_p (base);
1416 @@ -3925,7 +4177,7 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1418 alloc = (sn * bits + GMP_LIMB_BITS - 1) / GMP_LIMB_BITS;
1419 rp = MPZ_REALLOC (r, alloc);
1420 - rn = mpn_set_str_bits (rp, dp, dn, bits);
1421 + rn = mpn_set_str_bits (rp, dp, sn, bits);
1425 @@ -3933,7 +4185,7 @@ mpz_set_str (mpz_t r, const char *sp, int base)
1426 mpn_get_base_info (&info, base);
1427 alloc = (sn + info.exp - 1) / info.exp;
1428 rp = MPZ_REALLOC (r, alloc);
1429 - rn = mpn_set_str_other (rp, dp, dn, base, &info);
1430 + rn = mpn_set_str_other (rp, dp, sn, base, &info);
1432 assert (rn <= alloc);
1434 @@ -3967,14 +4219,9 @@ mpz_out_str (FILE *stream, int base, const mpz_t x)
1436 gmp_detect_endian (void)
1438 - static const int i = 1;
1439 + static const int i = 2;
1440 const unsigned char *p = (const unsigned char *) &i;
1442 - /* Little endian */
1450 /* Import and export. Does not support nails. */
1451 @@ -4037,29 +4284,22 @@ mpz_import (mpz_t r, size_t count, int order, size_t size, int endian,
1456 + assert (i + (bytes > 0) == rn);
1461 + i = mpn_normalized_size (rp, i);
1463 - r->_mp_size = mpn_normalized_size (rp, i);
1468 mpz_export (void *r, size_t *countp, int order, size_t size, int endian,
1469 size_t nails, const mpz_t u)
1472 - ptrdiff_t word_step;
1477 - /* The current (partial) limb. */
1479 - /* The number of bytes left to to in this limb. */
1481 - /* The index where the limb was read. */
1485 gmp_die ("mpz_import: Nails not supported.");
1487 @@ -4067,62 +4307,74 @@ mpz_export (void *r, size_t *countp, int order, size_t size, int endian,
1488 assert (endian >= -1 && endian <= 1);
1489 assert (size > 0 || u->_mp_size == 0);
1491 - un = GMP_ABS (u->_mp_size);
1504 + ptrdiff_t word_step;
1505 + /* The current (partial) limb. */
1507 + /* The number of bytes left to to in this limb. */
1509 + /* The index where the limb was read. */
1512 - /* Count bytes in top limb. */
1513 - for (limb = u->_mp_d[un-1], k = 0; limb > 0; k++, limb >>= CHAR_BIT)
1515 + un = GMP_ABS (un);
1518 + /* Count bytes in top limb. */
1519 + limb = u->_mp_d[un-1];
1520 + assert (limb != 0);
1522 - count = (k + (un-1) * sizeof (mp_limb_t) + size - 1) / size;
1525 + k++; limb >>= CHAR_BIT;
1526 + } while (limb != 0);
1529 - r = gmp_xalloc (count * size);
1530 + count = (k + (un-1) * sizeof (mp_limb_t) + size - 1) / size;
1533 - endian = gmp_detect_endian ();
1535 + r = gmp_xalloc (count * size);
1537 - p = (unsigned char *) r;
1539 + endian = gmp_detect_endian ();
1541 - word_step = (order != endian) ? 2 * size : 0;
1542 + p = (unsigned char *) r;
1544 - /* Process bytes from the least significant end, so point p at the
1545 - least significant word. */
1548 - p += size * (count - 1);
1549 - word_step = - word_step;
1551 + word_step = (order != endian) ? 2 * size : 0;
1553 - /* And at least significant byte of that word. */
1556 + /* Process bytes from the least significant end, so point p at the
1557 + least significant word. */
1560 + p += size * (count - 1);
1561 + word_step = - word_step;
1564 - for (bytes = 0, i = 0, k = 0; k < count; k++, p += word_step)
1567 - for (j = 0; j < size; j++, p -= (ptrdiff_t) endian)
1572 - limb = u->_mp_d[i++];
1573 - bytes = sizeof (mp_limb_t);
1576 - limb >>= CHAR_BIT;
1581 - assert (k == count);
1582 + /* And at least significant byte of that word. */
1586 + for (bytes = 0, i = 0, k = 0; k < count; k++, p += word_step)
1589 + for (j = 0; j < size; j++, p -= (ptrdiff_t) endian)
1594 + limb = u->_mp_d[i++];
1595 + bytes = sizeof (mp_limb_t);
1598 + limb >>= CHAR_BIT;
1603 + assert (k == count);
1608 diff --git a/mini-gmp.h b/mini-gmp.h
1609 index 8c94ca2..d8f691f 100644
1613 /* mini-gmp, a minimalistic implementation of a GNU GMP subset.
1615 -Copyright 2011, 2012, 2013 Free Software Foundation, Inc.
1616 +Copyright 2011-2014 Free Software Foundation, Inc.
1618 This file is part of the GNU MP Library.
1620 The GNU MP Library is free software; you can redistribute it and/or modify
1621 -it under the terms of the GNU Lesser General Public License as published by
1622 -the Free Software Foundation; either version 3 of the License, or (at your
1623 -option) any later version.
1624 +it under the terms of either:
1626 + * the GNU Lesser General Public License as published by the Free
1627 + Software Foundation; either version 3 of the License, or (at your
1628 + option) any later version.
1632 + * the GNU General Public License as published by the Free Software
1633 + Foundation; either version 2 of the License, or (at your option) any
1636 +or both in parallel, as here.
1638 The GNU MP Library is distributed in the hope that it will be useful, but
1639 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
1640 -or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public
1641 -License for more details.
1642 +or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
1645 -You should have received a copy of the GNU Lesser General Public License
1646 -along with the GNU MP Library. If not, see http://www.gnu.org/licenses/. */
1647 +You should have received copies of the GNU General Public License and the
1648 +GNU Lesser General Public License along with the GNU MP Library. If not,
1649 +see https://www.gnu.org/licenses/. */
1651 /* About mini-gmp: This is a minimal implementation of a subset of the
1652 GMP interface. It is intended for inclusion into applications which
1653 @@ -64,8 +75,11 @@ typedef __mpz_struct mpz_t[1];
1654 typedef __mpz_struct *mpz_ptr;
1655 typedef const __mpz_struct *mpz_srcptr;
1657 +extern const int mp_bits_per_limb;
1659 void mpn_copyi (mp_ptr, mp_srcptr, mp_size_t);
1660 void mpn_copyd (mp_ptr, mp_srcptr, mp_size_t);
1661 +void mpn_zero (mp_ptr, mp_size_t);
1663 int mpn_cmp (mp_srcptr, mp_srcptr, mp_size_t);
1665 @@ -84,10 +98,17 @@ mp_limb_t mpn_submul_1 (mp_ptr, mp_srcptr, mp_size_t, mp_limb_t);
1666 mp_limb_t mpn_mul (mp_ptr, mp_srcptr, mp_size_t, mp_srcptr, mp_size_t);
1667 void mpn_mul_n (mp_ptr, mp_srcptr, mp_srcptr, mp_size_t);
1668 void mpn_sqr (mp_ptr, mp_srcptr, mp_size_t);
1669 +int mpn_perfect_square_p (mp_srcptr, mp_size_t);
1670 +mp_size_t mpn_sqrtrem (mp_ptr, mp_ptr, mp_srcptr, mp_size_t);
1672 mp_limb_t mpn_lshift (mp_ptr, mp_srcptr, mp_size_t, unsigned int);
1673 mp_limb_t mpn_rshift (mp_ptr, mp_srcptr, mp_size_t, unsigned int);
1675 +mp_bitcnt_t mpn_scan0 (mp_srcptr, mp_bitcnt_t);
1676 +mp_bitcnt_t mpn_scan1 (mp_srcptr, mp_bitcnt_t);
1678 +mp_bitcnt_t mpn_popcount (mp_srcptr, mp_size_t);
1680 mp_limb_t mpn_invert_3by2 (mp_limb_t, mp_limb_t);
1681 #define mpn_invert_limb(x) mpn_invert_3by2 ((x), 0)
1683 @@ -124,6 +145,10 @@ void mpz_mul_si (mpz_t, const mpz_t, long int);
1684 void mpz_mul_ui (mpz_t, const mpz_t, unsigned long int);
1685 void mpz_mul (mpz_t, const mpz_t, const mpz_t);
1686 void mpz_mul_2exp (mpz_t, const mpz_t, mp_bitcnt_t);
1687 +void mpz_addmul_ui (mpz_t, const mpz_t, unsigned long int);
1688 +void mpz_addmul (mpz_t, const mpz_t, const mpz_t);
1689 +void mpz_submul_ui (mpz_t, const mpz_t, unsigned long int);
1690 +void mpz_submul (mpz_t, const mpz_t, const mpz_t);
1692 void mpz_cdiv_qr (mpz_t, mpz_t, const mpz_t, const mpz_t);
1693 void mpz_fdiv_qr (mpz_t, mpz_t, const mpz_t, const mpz_t);
1694 @@ -147,6 +172,7 @@ void mpz_mod (mpz_t, const mpz_t, const mpz_t);
1695 void mpz_divexact (mpz_t, const mpz_t, const mpz_t);
1697 int mpz_divisible_p (const mpz_t, const mpz_t);
1698 +int mpz_congruent_p (const mpz_t, const mpz_t, const mpz_t);
1700 unsigned long mpz_cdiv_qr_ui (mpz_t, mpz_t, const mpz_t, unsigned long);
1701 unsigned long mpz_fdiv_qr_ui (mpz_t, mpz_t, const mpz_t, unsigned long);
1702 @@ -176,6 +202,7 @@ int mpz_invert (mpz_t, const mpz_t, const mpz_t);
1704 void mpz_sqrtrem (mpz_t, mpz_t, const mpz_t);
1705 void mpz_sqrt (mpz_t, const mpz_t);
1706 +int mpz_perfect_square_p (const mpz_t);
1708 void mpz_pow_ui (mpz_t, const mpz_t, unsigned long);
1709 void mpz_ui_pow_ui (mpz_t, unsigned long, unsigned long);
1710 @@ -188,6 +215,9 @@ int mpz_root (mpz_t, const mpz_t, unsigned long);
1711 void mpz_fac_ui (mpz_t, unsigned long);
1712 void mpz_bin_uiui (mpz_t, unsigned long, unsigned long);
1715 +mpz_probab_prime_p (const mpz_t, int);
1717 int mpz_tstbit (const mpz_t, mp_bitcnt_t);
1718 void mpz_setbit (mpz_t, mp_bitcnt_t);
1719 void mpz_clrbit (mpz_t, mp_bitcnt_t);
1720 @@ -211,6 +241,15 @@ double mpz_get_d (const mpz_t);
1721 size_t mpz_size (const mpz_t);
1722 mp_limb_t mpz_getlimbn (const mpz_t, mp_size_t);
1724 +void mpz_realloc2 (mpz_t, mp_bitcnt_t);
1725 +mp_srcptr mpz_limbs_read (mpz_srcptr);
1726 +mp_ptr mpz_limbs_modify (mpz_t, mp_size_t);
1727 +mp_ptr mpz_limbs_write (mpz_t, mp_size_t);
1728 +void mpz_limbs_finish (mpz_t, mp_size_t);
1729 +mpz_srcptr mpz_roinit_n (mpz_t, mp_srcptr, mp_size_t);
1731 +#define MPZ_ROINIT_N(xp, xs) {{0, (xs),(xp) }}
1733 void mpz_set_si (mpz_t, signed long int);
1734 void mpz_set_ui (mpz_t, unsigned long int);
1735 void mpz_set (mpz_t, const mpz_t);