David Brazdil | 0f672f6 | 2019-12-10 10:32:29 +0000 | [diff] [blame] | 1 | // SPDX-License-Identifier: GPL-2.0-or-later |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 2 | /* mpihelp-div.c - MPI helper functions |
| 3 | * Copyright (C) 1994, 1996 Free Software Foundation, Inc. |
| 4 | * Copyright (C) 1998, 1999 Free Software Foundation, Inc. |
| 5 | * |
| 6 | * This file is part of GnuPG. |
| 7 | * |
Andrew Scull | b4b6d4a | 2019-01-02 15:54:55 +0000 | [diff] [blame] | 8 | * Note: This code is heavily based on the GNU MP Library. |
| 9 | * Actually it's the same code with only minor changes in the |
| 10 | * way the data is stored; this is to support the abstraction |
| 11 | * of an optional secure memory allocation which may be used |
| 12 | * to avoid revealing of sensitive data due to paging etc. |
| 13 | * The GNU MP Library itself is published under the LGPL; |
| 14 | * however I decided to publish this code under the plain GPL. |
| 15 | */ |
| 16 | |
| 17 | #include "mpi-internal.h" |
| 18 | #include "longlong.h" |
| 19 | |
| 20 | #ifndef UMUL_TIME |
| 21 | #define UMUL_TIME 1 |
| 22 | #endif |
| 23 | #ifndef UDIV_TIME |
| 24 | #define UDIV_TIME UMUL_TIME |
| 25 | #endif |
| 26 | |
| 27 | /* Divide num (NP/NSIZE) by den (DP/DSIZE) and write |
| 28 | * the NSIZE-DSIZE least significant quotient limbs at QP |
| 29 | * and the DSIZE long remainder at NP. If QEXTRA_LIMBS is |
| 30 | * non-zero, generate that many fraction bits and append them after the |
| 31 | * other quotient limbs. |
| 32 | * Return the most significant limb of the quotient, this is always 0 or 1. |
| 33 | * |
| 34 | * Preconditions: |
| 35 | * 0. NSIZE >= DSIZE. |
| 36 | * 1. The most significant bit of the divisor must be set. |
| 37 | * 2. QP must either not overlap with the input operands at all, or |
| 38 | * QP + DSIZE >= NP must hold true. (This means that it's |
| 39 | * possible to put the quotient in the high part of NUM, right after the |
| 40 | * remainder in NUM. |
| 41 | * 3. NSIZE >= DSIZE, even if QEXTRA_LIMBS is non-zero. |
| 42 | */ |
| 43 | |
| 44 | mpi_limb_t |
| 45 | mpihelp_divrem(mpi_ptr_t qp, mpi_size_t qextra_limbs, |
| 46 | mpi_ptr_t np, mpi_size_t nsize, mpi_ptr_t dp, mpi_size_t dsize) |
| 47 | { |
| 48 | mpi_limb_t most_significant_q_limb = 0; |
| 49 | |
| 50 | switch (dsize) { |
| 51 | case 0: |
| 52 | /* We are asked to divide by zero, so go ahead and do it! (To make |
| 53 | the compiler not remove this statement, return the value.) */ |
| 54 | /* |
| 55 | * existing clients of this function have been modified |
| 56 | * not to call it with dsize == 0, so this should not happen |
| 57 | */ |
| 58 | return 1 / dsize; |
| 59 | |
| 60 | case 1: |
| 61 | { |
| 62 | mpi_size_t i; |
| 63 | mpi_limb_t n1; |
| 64 | mpi_limb_t d; |
| 65 | |
| 66 | d = dp[0]; |
| 67 | n1 = np[nsize - 1]; |
| 68 | |
| 69 | if (n1 >= d) { |
| 70 | n1 -= d; |
| 71 | most_significant_q_limb = 1; |
| 72 | } |
| 73 | |
| 74 | qp += qextra_limbs; |
| 75 | for (i = nsize - 2; i >= 0; i--) |
| 76 | udiv_qrnnd(qp[i], n1, n1, np[i], d); |
| 77 | qp -= qextra_limbs; |
| 78 | |
| 79 | for (i = qextra_limbs - 1; i >= 0; i--) |
| 80 | udiv_qrnnd(qp[i], n1, n1, 0, d); |
| 81 | |
| 82 | np[0] = n1; |
| 83 | } |
| 84 | break; |
| 85 | |
| 86 | case 2: |
| 87 | { |
| 88 | mpi_size_t i; |
| 89 | mpi_limb_t n1, n0, n2; |
| 90 | mpi_limb_t d1, d0; |
| 91 | |
| 92 | np += nsize - 2; |
| 93 | d1 = dp[1]; |
| 94 | d0 = dp[0]; |
| 95 | n1 = np[1]; |
| 96 | n0 = np[0]; |
| 97 | |
| 98 | if (n1 >= d1 && (n1 > d1 || n0 >= d0)) { |
| 99 | sub_ddmmss(n1, n0, n1, n0, d1, d0); |
| 100 | most_significant_q_limb = 1; |
| 101 | } |
| 102 | |
| 103 | for (i = qextra_limbs + nsize - 2 - 1; i >= 0; i--) { |
| 104 | mpi_limb_t q; |
| 105 | mpi_limb_t r; |
| 106 | |
| 107 | if (i >= qextra_limbs) |
| 108 | np--; |
| 109 | else |
| 110 | np[0] = 0; |
| 111 | |
| 112 | if (n1 == d1) { |
| 113 | /* Q should be either 111..111 or 111..110. Need special |
| 114 | * treatment of this rare case as normal division would |
| 115 | * give overflow. */ |
| 116 | q = ~(mpi_limb_t) 0; |
| 117 | |
| 118 | r = n0 + d1; |
| 119 | if (r < d1) { /* Carry in the addition? */ |
| 120 | add_ssaaaa(n1, n0, r - d0, |
| 121 | np[0], 0, d0); |
| 122 | qp[i] = q; |
| 123 | continue; |
| 124 | } |
| 125 | n1 = d0 - (d0 != 0 ? 1 : 0); |
| 126 | n0 = -d0; |
| 127 | } else { |
| 128 | udiv_qrnnd(q, r, n1, n0, d1); |
| 129 | umul_ppmm(n1, n0, d0, q); |
| 130 | } |
| 131 | |
| 132 | n2 = np[0]; |
| 133 | q_test: |
| 134 | if (n1 > r || (n1 == r && n0 > n2)) { |
| 135 | /* The estimated Q was too large. */ |
| 136 | q--; |
| 137 | sub_ddmmss(n1, n0, n1, n0, 0, d0); |
| 138 | r += d1; |
| 139 | if (r >= d1) /* If not carry, test Q again. */ |
| 140 | goto q_test; |
| 141 | } |
| 142 | |
| 143 | qp[i] = q; |
| 144 | sub_ddmmss(n1, n0, r, n2, n1, n0); |
| 145 | } |
| 146 | np[1] = n1; |
| 147 | np[0] = n0; |
| 148 | } |
| 149 | break; |
| 150 | |
| 151 | default: |
| 152 | { |
| 153 | mpi_size_t i; |
| 154 | mpi_limb_t dX, d1, n0; |
| 155 | |
| 156 | np += nsize - dsize; |
| 157 | dX = dp[dsize - 1]; |
| 158 | d1 = dp[dsize - 2]; |
| 159 | n0 = np[dsize - 1]; |
| 160 | |
| 161 | if (n0 >= dX) { |
| 162 | if (n0 > dX |
| 163 | || mpihelp_cmp(np, dp, dsize - 1) >= 0) { |
| 164 | mpihelp_sub_n(np, np, dp, dsize); |
| 165 | n0 = np[dsize - 1]; |
| 166 | most_significant_q_limb = 1; |
| 167 | } |
| 168 | } |
| 169 | |
| 170 | for (i = qextra_limbs + nsize - dsize - 1; i >= 0; i--) { |
| 171 | mpi_limb_t q; |
| 172 | mpi_limb_t n1, n2; |
| 173 | mpi_limb_t cy_limb; |
| 174 | |
| 175 | if (i >= qextra_limbs) { |
| 176 | np--; |
| 177 | n2 = np[dsize]; |
| 178 | } else { |
| 179 | n2 = np[dsize - 1]; |
| 180 | MPN_COPY_DECR(np + 1, np, dsize - 1); |
| 181 | np[0] = 0; |
| 182 | } |
| 183 | |
| 184 | if (n0 == dX) { |
| 185 | /* This might over-estimate q, but it's probably not worth |
| 186 | * the extra code here to find out. */ |
| 187 | q = ~(mpi_limb_t) 0; |
| 188 | } else { |
| 189 | mpi_limb_t r; |
| 190 | |
| 191 | udiv_qrnnd(q, r, n0, np[dsize - 1], dX); |
| 192 | umul_ppmm(n1, n0, d1, q); |
| 193 | |
| 194 | while (n1 > r |
| 195 | || (n1 == r |
| 196 | && n0 > np[dsize - 2])) { |
| 197 | q--; |
| 198 | r += dX; |
| 199 | if (r < dX) /* I.e. "carry in previous addition?" */ |
| 200 | break; |
| 201 | n1 -= n0 < d1; |
| 202 | n0 -= d1; |
| 203 | } |
| 204 | } |
| 205 | |
| 206 | /* Possible optimization: We already have (q * n0) and (1 * n1) |
| 207 | * after the calculation of q. Taking advantage of that, we |
| 208 | * could make this loop make two iterations less. */ |
| 209 | cy_limb = mpihelp_submul_1(np, dp, dsize, q); |
| 210 | |
| 211 | if (n2 != cy_limb) { |
| 212 | mpihelp_add_n(np, np, dp, dsize); |
| 213 | q--; |
| 214 | } |
| 215 | |
| 216 | qp[i] = q; |
| 217 | n0 = np[dsize - 1]; |
| 218 | } |
| 219 | } |
| 220 | } |
| 221 | |
| 222 | return most_significant_q_limb; |
| 223 | } |