1 /* mpn_gcd_1 -- mpn and limb greatest common divisor. 2 3 Copyright 1994, 1996, 2000, 2001, 2009, 2012 Free Software Foundation, Inc. 4 5 This file is part of the GNU MP Library. 6 7 The GNU MP Library is free software; you can redistribute it and/or modify 8 it under the terms of either: 9 10 * the GNU Lesser General Public License as published by the Free 11 Software Foundation; either version 3 of the License, or (at your 12 option) any later version. 13 14 or 15 16 * the GNU General Public License as published by the Free Software 17 Foundation; either version 2 of the License, or (at your option) any 18 later version. 19 20 or both in parallel, as here. 21 22 The GNU MP Library is distributed in the hope that it will be useful, but 23 WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY 24 or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License 25 for more details. 26 27 You should have received copies of the GNU General Public License and the 28 GNU Lesser General Public License along with the GNU MP Library. If not, 29 see https://www.gnu.org/licenses/. */ 30 31 #include "gmp.h" 32 #include "gmp-impl.h" 33 #include "longlong.h" 34 35 #ifndef GCD_1_METHOD 36 #define GCD_1_METHOD 2 37 #endif 38 39 #define USE_ZEROTAB 0 40 41 #if USE_ZEROTAB 42 #define MAXSHIFT 4 43 #define MASK ((1 << MAXSHIFT) - 1) 44 static const unsigned char zerotab[1 << MAXSHIFT] = 45 { 46 #if MAXSHIFT > 4 47 5, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0, 48 #endif 49 4, 0, 1, 0, 2, 0, 1, 0, 3, 0, 1, 0, 2, 0, 1, 0 50 }; 51 #endif 52 53 /* Does not work for U == 0 or V == 0. It would be tough to make it work for 54 V == 0 since gcd(x,0) = x, and U does not generally fit in an mp_limb_t. 55 56 The threshold for doing u%v when size==1 will vary by CPU according to 57 the speed of a division and the code generated for the main loop. Any 58 tuning for this is left to a CPU specific implementation. */ 59 60 mp_limb_t 61 mpn_gcd_1 (mp_srcptr up, mp_size_t size, mp_limb_t vlimb) 62 { 63 mp_limb_t ulimb; 64 unsigned long zero_bits, u_low_zero_bits; 65 66 ASSERT (size >= 1); 67 ASSERT (vlimb != 0); 68 ASSERT_MPN_NONZERO_P (up, size); 69 70 ulimb = up[0]; 71 72 /* Need vlimb odd for modexact, want it odd to get common zeros. */ 73 count_trailing_zeros (zero_bits, vlimb); 74 vlimb >>= zero_bits; 75 76 if (size > 1) 77 { 78 /* Must get common zeros before the mod reduction. If ulimb==0 then 79 vlimb already gives the common zeros. */ 80 if (ulimb != 0) 81 { 82 count_trailing_zeros (u_low_zero_bits, ulimb); 83 zero_bits = MIN (zero_bits, u_low_zero_bits); 84 } 85 86 ulimb = MPN_MOD_OR_MODEXACT_1_ODD (up, size, vlimb); 87 if (ulimb == 0) 88 goto done; 89 90 goto strip_u_maybe; 91 } 92 93 /* size==1, so up[0]!=0 */ 94 count_trailing_zeros (u_low_zero_bits, ulimb); 95 ulimb >>= u_low_zero_bits; 96 zero_bits = MIN (zero_bits, u_low_zero_bits); 97 98 /* make u bigger */ 99 if (vlimb > ulimb) 100 MP_LIMB_T_SWAP (ulimb, vlimb); 101 102 /* if u is much bigger than v, reduce using a division rather than 103 chipping away at it bit-by-bit */ 104 if ((ulimb >> 16) > vlimb) 105 { 106 ulimb %= vlimb; 107 if (ulimb == 0) 108 goto done; 109 goto strip_u_maybe; 110 } 111 112 ASSERT (ulimb & 1); 113 ASSERT (vlimb & 1); 114 115 #if GCD_1_METHOD == 1 116 while (ulimb != vlimb) 117 { 118 ASSERT (ulimb & 1); 119 ASSERT (vlimb & 1); 120 121 if (ulimb > vlimb) 122 { 123 ulimb -= vlimb; 124 do 125 { 126 ulimb >>= 1; 127 ASSERT (ulimb != 0); 128 strip_u_maybe: 129 ; 130 } 131 while ((ulimb & 1) == 0); 132 } 133 else /* vlimb > ulimb. */ 134 { 135 vlimb -= ulimb; 136 do 137 { 138 vlimb >>= 1; 139 ASSERT (vlimb != 0); 140 } 141 while ((vlimb & 1) == 0); 142 } 143 } 144 #else 145 # if GCD_1_METHOD == 2 146 147 ulimb >>= 1; 148 vlimb >>= 1; 149 150 while (ulimb != vlimb) 151 { 152 int c; 153 mp_limb_t t; 154 mp_limb_t vgtu; 155 156 t = ulimb - vlimb; 157 vgtu = LIMB_HIGHBIT_TO_MASK (t); 158 159 /* v <-- min (u, v) */ 160 vlimb += (vgtu & t); 161 162 /* u <-- |u - v| */ 163 ulimb = (t ^ vgtu) - vgtu; 164 165 #if USE_ZEROTAB 166 /* Number of trailing zeros is the same no matter if we look at 167 * t or ulimb, but using t gives more parallelism. */ 168 c = zerotab[t & MASK]; 169 170 while (UNLIKELY (c == MAXSHIFT)) 171 { 172 ulimb >>= MAXSHIFT; 173 if (0) 174 strip_u_maybe: 175 vlimb >>= 1; 176 177 c = zerotab[ulimb & MASK]; 178 } 179 #else 180 if (0) 181 { 182 strip_u_maybe: 183 vlimb >>= 1; 184 t = ulimb; 185 } 186 count_trailing_zeros (c, t); 187 #endif 188 ulimb >>= (c + 1); 189 } 190 191 vlimb = (vlimb << 1) | 1; 192 # else 193 # error Unknown GCD_1_METHOD 194 # endif 195 #endif 196 197 done: 198 return vlimb << zero_bits; 199 } 200