/src/gmp-6.2.1/mpn/sbpi1_div_q.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* mpn_sbpi1_div_q -- Schoolbook division using the Möller-Granlund 3/2 |
2 | | division algorithm. |
3 | | |
4 | | Contributed to the GNU project by Torbjorn Granlund. |
5 | | |
6 | | THE FUNCTION IN THIS FILE IS INTERNAL WITH A MUTABLE INTERFACE. IT IS ONLY |
7 | | SAFE TO REACH IT THROUGH DOCUMENTED INTERFACES. IN FACT, IT IS ALMOST |
8 | | GUARANTEED THAT IT WILL CHANGE OR DISAPPEAR IN A FUTURE GMP RELEASE. |
9 | | |
10 | | Copyright 2007, 2009 Free Software Foundation, Inc. |
11 | | |
12 | | This file is part of the GNU MP Library. |
13 | | |
14 | | The GNU MP Library is free software; you can redistribute it and/or modify |
15 | | it under the terms of either: |
16 | | |
17 | | * the GNU Lesser General Public License as published by the Free |
18 | | Software Foundation; either version 3 of the License, or (at your |
19 | | option) any later version. |
20 | | |
21 | | or |
22 | | |
23 | | * the GNU General Public License as published by the Free Software |
24 | | Foundation; either version 2 of the License, or (at your option) any |
25 | | later version. |
26 | | |
27 | | or both in parallel, as here. |
28 | | |
29 | | The GNU MP Library is distributed in the hope that it will be useful, but |
30 | | WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY |
31 | | or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License |
32 | | for more details. |
33 | | |
34 | | You should have received copies of the GNU General Public License and the |
35 | | GNU Lesser General Public License along with the GNU MP Library. If not, |
36 | | see https://www.gnu.org/licenses/. */ |
37 | | |
38 | | |
39 | | #include "gmp-impl.h" |
40 | | #include "longlong.h" |
41 | | |
42 | | mp_limb_t |
43 | | mpn_sbpi1_div_q (mp_ptr qp, |
44 | | mp_ptr np, mp_size_t nn, |
45 | | mp_srcptr dp, mp_size_t dn, |
46 | | mp_limb_t dinv) |
47 | 0 | { |
48 | 0 | mp_limb_t qh; |
49 | 0 | mp_size_t qn, i; |
50 | 0 | mp_limb_t n1, n0; |
51 | 0 | mp_limb_t d1, d0; |
52 | 0 | mp_limb_t cy, cy1; |
53 | 0 | mp_limb_t q; |
54 | 0 | mp_limb_t flag; |
55 | |
|
56 | 0 | mp_size_t dn_orig = dn; |
57 | 0 | mp_srcptr dp_orig = dp; |
58 | 0 | mp_ptr np_orig = np; |
59 | |
|
60 | 0 | ASSERT (dn > 2); |
61 | 0 | ASSERT (nn >= dn); |
62 | 0 | ASSERT ((dp[dn-1] & GMP_NUMB_HIGHBIT) != 0); |
63 | | |
64 | 0 | np += nn; |
65 | |
|
66 | 0 | qn = nn - dn; |
67 | 0 | if (qn + 1 < dn) |
68 | 0 | { |
69 | 0 | dp += dn - (qn + 1); |
70 | 0 | dn = qn + 1; |
71 | 0 | } |
72 | |
|
73 | 0 | qh = mpn_cmp (np - dn, dp, dn) >= 0; |
74 | 0 | if (qh != 0) |
75 | 0 | mpn_sub_n (np - dn, np - dn, dp, dn); |
76 | |
|
77 | 0 | qp += qn; |
78 | |
|
79 | 0 | dn -= 2; /* offset dn by 2 for main division loops, |
80 | | saving two iterations in mpn_submul_1. */ |
81 | 0 | d1 = dp[dn + 1]; |
82 | 0 | d0 = dp[dn + 0]; |
83 | |
|
84 | 0 | np -= 2; |
85 | |
|
86 | 0 | n1 = np[1]; |
87 | |
|
88 | 0 | for (i = qn - (dn + 2); i >= 0; i--) |
89 | 0 | { |
90 | 0 | np--; |
91 | 0 | if (UNLIKELY (n1 == d1) && np[1] == d0) |
92 | 0 | { |
93 | 0 | q = GMP_NUMB_MASK; |
94 | 0 | mpn_submul_1 (np - dn, dp, dn + 2, q); |
95 | 0 | n1 = np[1]; /* update n1, last loop's value will now be invalid */ |
96 | 0 | } |
97 | 0 | else |
98 | 0 | { |
99 | 0 | udiv_qr_3by2 (q, n1, n0, n1, np[1], np[0], d1, d0, dinv); |
100 | |
|
101 | 0 | cy = mpn_submul_1 (np - dn, dp, dn, q); |
102 | |
|
103 | 0 | cy1 = n0 < cy; |
104 | 0 | n0 = (n0 - cy) & GMP_NUMB_MASK; |
105 | 0 | cy = n1 < cy1; |
106 | 0 | n1 -= cy1; |
107 | 0 | np[0] = n0; |
108 | |
|
109 | 0 | if (UNLIKELY (cy != 0)) |
110 | 0 | { |
111 | 0 | n1 += d1 + mpn_add_n (np - dn, np - dn, dp, dn + 1); |
112 | 0 | q--; |
113 | 0 | } |
114 | 0 | } |
115 | |
|
116 | 0 | *--qp = q; |
117 | 0 | } |
118 | |
|
119 | 0 | flag = ~CNST_LIMB(0); |
120 | |
|
121 | 0 | if (dn >= 0) |
122 | 0 | { |
123 | 0 | for (i = dn; i > 0; i--) |
124 | 0 | { |
125 | 0 | np--; |
126 | 0 | if (UNLIKELY (n1 >= (d1 & flag))) |
127 | 0 | { |
128 | 0 | q = GMP_NUMB_MASK; |
129 | 0 | cy = mpn_submul_1 (np - dn, dp, dn + 2, q); |
130 | |
|
131 | 0 | if (UNLIKELY (n1 != cy)) |
132 | 0 | { |
133 | 0 | if (n1 < (cy & flag)) |
134 | 0 | { |
135 | 0 | q--; |
136 | 0 | mpn_add_n (np - dn, np - dn, dp, dn + 2); |
137 | 0 | } |
138 | 0 | else |
139 | 0 | flag = 0; |
140 | 0 | } |
141 | 0 | n1 = np[1]; |
142 | 0 | } |
143 | 0 | else |
144 | 0 | { |
145 | 0 | udiv_qr_3by2 (q, n1, n0, n1, np[1], np[0], d1, d0, dinv); |
146 | |
|
147 | 0 | cy = mpn_submul_1 (np - dn, dp, dn, q); |
148 | |
|
149 | 0 | cy1 = n0 < cy; |
150 | 0 | n0 = (n0 - cy) & GMP_NUMB_MASK; |
151 | 0 | cy = n1 < cy1; |
152 | 0 | n1 -= cy1; |
153 | 0 | np[0] = n0; |
154 | |
|
155 | 0 | if (UNLIKELY (cy != 0)) |
156 | 0 | { |
157 | 0 | n1 += d1 + mpn_add_n (np - dn, np - dn, dp, dn + 1); |
158 | 0 | q--; |
159 | 0 | } |
160 | 0 | } |
161 | |
|
162 | 0 | *--qp = q; |
163 | | |
164 | | /* Truncate operands. */ |
165 | 0 | dn--; |
166 | 0 | dp++; |
167 | 0 | } |
168 | |
|
169 | 0 | np--; |
170 | 0 | if (UNLIKELY (n1 >= (d1 & flag))) |
171 | 0 | { |
172 | 0 | q = GMP_NUMB_MASK; |
173 | 0 | cy = mpn_submul_1 (np, dp, 2, q); |
174 | |
|
175 | 0 | if (UNLIKELY (n1 != cy)) |
176 | 0 | { |
177 | 0 | if (n1 < (cy & flag)) |
178 | 0 | { |
179 | 0 | q--; |
180 | 0 | add_ssaaaa (np[1], np[0], np[1], np[0], dp[1], dp[0]); |
181 | 0 | } |
182 | 0 | else |
183 | 0 | flag = 0; |
184 | 0 | } |
185 | 0 | n1 = np[1]; |
186 | 0 | } |
187 | 0 | else |
188 | 0 | { |
189 | 0 | udiv_qr_3by2 (q, n1, n0, n1, np[1], np[0], d1, d0, dinv); |
190 | |
|
191 | 0 | np[0] = n0; |
192 | 0 | np[1] = n1; |
193 | 0 | } |
194 | |
|
195 | 0 | *--qp = q; |
196 | 0 | } |
197 | 0 | ASSERT_ALWAYS (np[1] == n1); |
198 | 0 | np += 2; |
199 | | |
200 | |
|
201 | 0 | dn = dn_orig; |
202 | 0 | if (UNLIKELY (n1 < (dn & flag))) |
203 | 0 | { |
204 | 0 | mp_limb_t q, x; |
205 | | |
206 | | /* The quotient may be too large if the remainder is small. Recompute |
207 | | for above ignored operand parts, until the remainder spills. |
208 | | |
209 | | FIXME: The quality of this code isn't the same as the code above. |
210 | | 1. We don't compute things in an optimal order, high-to-low, in order |
211 | | to terminate as quickly as possible. |
212 | | 2. We mess with pointers and sizes, adding and subtracting and |
213 | | adjusting to get things right. It surely could be streamlined. |
214 | | 3. The only termination criteria are that we determine that the |
215 | | quotient needs to be adjusted, or that we have recomputed |
216 | | everything. We should stop when the remainder is so large |
217 | | that no additional subtracting could make it spill. |
218 | | 4. If nothing else, we should not do two loops of submul_1 over the |
219 | | data, instead handle both the triangularization and chopping at |
220 | | once. */ |
221 | |
|
222 | 0 | x = n1; |
223 | |
|
224 | 0 | if (dn > 2) |
225 | 0 | { |
226 | | /* Compensate for triangularization. */ |
227 | 0 | mp_limb_t y; |
228 | |
|
229 | 0 | dp = dp_orig; |
230 | 0 | if (qn + 1 < dn) |
231 | 0 | { |
232 | 0 | dp += dn - (qn + 1); |
233 | 0 | dn = qn + 1; |
234 | 0 | } |
235 | |
|
236 | 0 | y = np[-2]; |
237 | |
|
238 | 0 | for (i = dn - 3; i >= 0; i--) |
239 | 0 | { |
240 | 0 | q = qp[i]; |
241 | 0 | cy = mpn_submul_1 (np - (dn - i), dp, dn - i - 2, q); |
242 | |
|
243 | 0 | if (y < cy) |
244 | 0 | { |
245 | 0 | if (x == 0) |
246 | 0 | { |
247 | 0 | cy = mpn_sub_1 (qp, qp, qn, 1); |
248 | 0 | ASSERT_ALWAYS (cy == 0); |
249 | 0 | return qh - cy; |
250 | 0 | } |
251 | 0 | x--; |
252 | 0 | } |
253 | 0 | y -= cy; |
254 | 0 | } |
255 | 0 | np[-2] = y; |
256 | 0 | } |
257 | | |
258 | 0 | dn = dn_orig; |
259 | 0 | if (qn + 1 < dn) |
260 | 0 | { |
261 | | /* Compensate for ignored dividend and divisor tails. */ |
262 | |
|
263 | 0 | dp = dp_orig; |
264 | 0 | np = np_orig; |
265 | |
|
266 | 0 | if (qh != 0) |
267 | 0 | { |
268 | 0 | cy = mpn_sub_n (np + qn, np + qn, dp, dn - (qn + 1)); |
269 | 0 | if (cy != 0) |
270 | 0 | { |
271 | 0 | if (x == 0) |
272 | 0 | { |
273 | 0 | if (qn != 0) |
274 | 0 | cy = mpn_sub_1 (qp, qp, qn, 1); |
275 | 0 | return qh - cy; |
276 | 0 | } |
277 | 0 | x--; |
278 | 0 | } |
279 | 0 | } |
280 | | |
281 | 0 | if (qn == 0) |
282 | 0 | return qh; |
283 | | |
284 | 0 | for (i = dn - qn - 2; i >= 0; i--) |
285 | 0 | { |
286 | 0 | cy = mpn_submul_1 (np + i, qp, qn, dp[i]); |
287 | 0 | cy = mpn_sub_1 (np + qn + i, np + qn + i, dn - qn - i - 1, cy); |
288 | 0 | if (cy != 0) |
289 | 0 | { |
290 | 0 | if (x == 0) |
291 | 0 | { |
292 | 0 | cy = mpn_sub_1 (qp, qp, qn, 1); |
293 | 0 | return qh; |
294 | 0 | } |
295 | 0 | x--; |
296 | 0 | } |
297 | 0 | } |
298 | 0 | } |
299 | 0 | } |
300 | | |
301 | 0 | return qh; |
302 | 0 | } |