1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
|
/*-
* Copyright (c) 1992, 1993
* The Regents of the University of California. All rights reserved.
*
* This software was developed by the Computer Systems Engineering group
* at Lawrence Berkeley Laboratory under DARPA contract BG 91-66 and
* contributed to Berkeley.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* From: Id: qdivrem.c,v 1.7 1997/11/07 09:20:40 phk Exp
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
/*
* Multiprecision divide. This algorithm is from Knuth vol. 2 (2nd ed),
* section 4.3.1, pp. 257--259.
*/
#include "quad.h"
#define B (1 << HALF_BITS) /* digit base */
/* Combine two `digits' to make a single two-digit number. */
#define COMBINE(a, b) (((u_long)(a) << HALF_BITS) | (b))
/* select a type for digits in base B: use unsigned short if they fit */
#if ULONG_MAX == 0xffffffff && USHRT_MAX >= 0xffff
typedef unsigned short digit;
#else
typedef u_long digit;
#endif
/*
* Shift p[0]..p[len] left `sh' bits, ignoring any bits that
* `fall out' the left (there never will be any such anyway).
* We may assume len >= 0. NOTE THAT THIS WRITES len+1 DIGITS.
*/
static void
shl(digit *p, int len, int sh)
{
int i;
for (i = 0; i < len; i++)
p[i] = LHALF(p[i] << sh) | (p[i + 1] >> (HALF_BITS - sh));
p[i] = LHALF(p[i] << sh);
}
/*
* __qdivrem(u, v, rem) returns u/v and, optionally, sets *rem to u%v.
*
* We do this in base 2-sup-HALF_BITS, so that all intermediate products
* fit within u_long. As a consequence, the maximum length dividend and
* divisor are 4 `digits' in this base (they are shorter if they have
* leading zeros).
*/
u_quad_t
__qdivrem(uq, vq, arq)
u_quad_t uq, vq, *arq;
{
union uu tmp;
digit *u, *v, *q;
digit v1, v2;
u_long qhat, rhat, t;
int m, n, d, j, i;
digit uspace[5], vspace[5], qspace[5];
/*
* Take care of special cases: divide by zero, and u < v.
*/
if (vq == 0) {
/* divide by zero. */
static volatile const unsigned int zero = 0;
tmp.ul[H] = tmp.ul[L] = 1 / zero;
if (arq)
*arq = uq;
return (tmp.q);
}
if (uq < vq) {
if (arq)
*arq = uq;
return (0);
}
u = &uspace[0];
v = &vspace[0];
q = &qspace[0];
/*
* Break dividend and divisor into digits in base B, then
* count leading zeros to determine m and n. When done, we
* will have:
* u = (u[1]u[2]...u[m+n]) sub B
* v = (v[1]v[2]...v[n]) sub B
* v[1] != 0
* 1 < n <= 4 (if n = 1, we use a different division algorithm)
* m >= 0 (otherwise u < v, which we already checked)
* m + n = 4
* and thus
* m = 4 - n <= 2
*/
tmp.uq = uq;
u[0] = 0;
u[1] = HHALF(tmp.ul[H]);
u[2] = LHALF(tmp.ul[H]);
u[3] = HHALF(tmp.ul[L]);
u[4] = LHALF(tmp.ul[L]);
tmp.uq = vq;
v[1] = HHALF(tmp.ul[H]);
v[2] = LHALF(tmp.ul[H]);
v[3] = HHALF(tmp.ul[L]);
v[4] = LHALF(tmp.ul[L]);
for (n = 4; v[1] == 0; v++) {
if (--n == 1) {
u_long rbj; /* r*B+u[j] (not root boy jim) */
digit q1, q2, q3, q4;
/*
* Change of plan, per exercise 16.
* r = 0;
* for j = 1..4:
* q[j] = floor((r*B + u[j]) / v),
* r = (r*B + u[j]) % v;
* We unroll this completely here.
*/
t = v[2]; /* nonzero, by definition */
q1 = u[1] / t;
rbj = COMBINE(u[1] % t, u[2]);
q2 = rbj / t;
rbj = COMBINE(rbj % t, u[3]);
q3 = rbj / t;
rbj = COMBINE(rbj % t, u[4]);
q4 = rbj / t;
if (arq)
*arq = rbj % t;
tmp.ul[H] = COMBINE(q1, q2);
tmp.ul[L] = COMBINE(q3, q4);
return (tmp.q);
}
}
/*
* By adjusting q once we determine m, we can guarantee that
* there is a complete four-digit quotient at &qspace[1] when
* we finally stop.
*/
for (m = 4 - n; u[1] == 0; u++)
m--;
for (i = 4 - m; --i >= 0;)
q[i] = 0;
q += 4 - m;
/*
* Here we run Program D, translated from MIX to C and acquiring
* a few minor changes.
*
* D1: choose multiplier 1 << d to ensure v[1] >= B/2.
*/
d = 0;
for (t = v[1]; t < B / 2; t <<= 1)
d++;
if (d > 0) {
shl(&u[0], m + n, d); /* u <<= d */
shl(&v[1], n - 1, d); /* v <<= d */
}
/*
* D2: j = 0.
*/
j = 0;
v1 = v[1]; /* for D3 -- note that v[1..n] are constant */
v2 = v[2]; /* for D3 */
do {
digit uj0, uj1, uj2;
/*
* D3: Calculate qhat (\^q, in TeX notation).
* Let qhat = min((u[j]*B + u[j+1])/v[1], B-1), and
* let rhat = (u[j]*B + u[j+1]) mod v[1].
* While rhat < B and v[2]*qhat > rhat*B+u[j+2],
* decrement qhat and increase rhat correspondingly.
* Note that if rhat >= B, v[2]*qhat < rhat*B.
*/
uj0 = u[j + 0]; /* for D3 only -- note that u[j+...] change */
uj1 = u[j + 1]; /* for D3 only */
uj2 = u[j + 2]; /* for D3 only */
if (uj0 == v1) {
qhat = B;
rhat = uj1;
goto qhat_too_big;
} else {
u_long nn = COMBINE(uj0, uj1);
qhat = nn / v1;
rhat = nn % v1;
}
while (v2 * qhat > COMBINE(rhat, uj2)) {
qhat_too_big:
qhat--;
if ((rhat += v1) >= B)
break;
}
/*
* D4: Multiply and subtract.
* The variable `t' holds any borrows across the loop.
* We split this up so that we do not require v[0] = 0,
* and to eliminate a final special case.
*/
for (t = 0, i = n; i > 0; i--) {
t = u[i + j] - v[i] * qhat - t;
u[i + j] = LHALF(t);
t = (B - HHALF(t)) & (B - 1);
}
t = u[j] - t;
u[j] = LHALF(t);
/*
* D5: test remainder.
* There is a borrow if and only if HHALF(t) is nonzero;
* in that (rare) case, qhat was too large (by exactly 1).
* Fix it by adding v[1..n] to u[j..j+n].
*/
if (HHALF(t)) {
qhat--;
for (t = 0, i = n; i > 0; i--) { /* D6: add back. */
t += u[i + j] + v[i];
u[i + j] = LHALF(t);
t = HHALF(t);
}
u[j] = LHALF(u[j] + t);
}
q[j] = qhat;
} while (++j <= m); /* D7: loop on j. */
/*
* If caller wants the remainder, we have to calculate it as
* u[m..m+n] >> d (this is at most n digits and thus fits in
* u[m+1..m+n], but we may need more source digits).
*/
if (arq) {
if (d) {
for (i = m + n; i > m; --i)
u[i] = (u[i] >> d) |
LHALF(u[i - 1] << (HALF_BITS - d));
u[i] = 0;
}
tmp.ul[H] = COMBINE(uspace[1], uspace[2]);
tmp.ul[L] = COMBINE(uspace[3], uspace[4]);
*arq = tmp.q;
}
tmp.ul[H] = COMBINE(qspace[1], qspace[2]);
tmp.ul[L] = COMBINE(qspace[3], qspace[4]);
return (tmp.q);
}
/*
* Divide two unsigned quads.
*/
u_quad_t
__udivdi3(a, b)
u_quad_t a, b;
{
return (__qdivrem(a, b, (u_quad_t *)0));
}
/*
* Return remainder after dividing two unsigned quads.
*/
u_quad_t
__umoddi3(a, b)
u_quad_t a, b;
{
u_quad_t r;
(void)__qdivrem(a, b, &r);
return (r);
}
|