1 /*
2 * mergesort() implementation for systems that don't have it.
3 *
4 * Copyright (c) 1992, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from software contributed to Berkeley by
8 * Peter McIlroy.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 */
34 #include "util.h"
35
36 #if defined(LIBC_SCCS) && !defined(lint)
37 static char sccsid[] = "@(#)merge.c 8.2 (Berkeley) 2/14/94";
38 #endif /* LIBC_SCCS and not lint */
39
40 #ifdef HAVE_MERGESORT
41 #undef yasm__mergesort
42 #endif
43
44 #ifndef HAVE_MERGESORT
45 /*
46 * Hybrid exponential search/linear search merge sort with hybrid
47 * natural/pairwise first pass. Requires about .3% more comparisons
48 * for random data than LSMS with pairwise first pass alone.
49 * It works for objects as small as two bytes.
50 */
51
52 #define NATURAL
53 #define THRESHOLD 16 /* Best choice for natural merge cut-off. */
54
55 /* #define NATURAL to get hybrid natural merge.
56 * (The default is pairwise merging.)
57 */
58
59 #include <errno.h>
60 #include <string.h>
61
62 static void setup(unsigned char *, unsigned char *, size_t, size_t,
63 int (*)(const void *, const void *));
64 static void insertionsort(unsigned char *, size_t, size_t,
65 int (*)(const void *, const void *));
66
67 #define ISIZE sizeof(int)
68 #define PSIZE sizeof(unsigned char *)
69 #define ICOPY_LIST(src, dst, last) \
70 do \
71 *(int*)dst = *(int*)src, src += ISIZE, dst += ISIZE; \
72 while(src < last)
73 #define ICOPY_ELT(src, dst, i) \
74 do \
75 *(int*) dst = *(int*) src, src += ISIZE, dst += ISIZE; \
76 while (i -= ISIZE)
77
78 #define CCOPY_LIST(src, dst, last) \
79 do \
80 *dst++ = *src++; \
81 while (src < last)
82 #define CCOPY_ELT(src, dst, i) \
83 do \
84 *dst++ = *src++; \
85 while (i -= 1)
86
87 /*
88 * Find the next possible pointer head. (Trickery for forcing an array
89 * to do double duty as a linked list when objects do not align with word
90 * boundaries.
91 */
92 /* Assumption: PSIZE is a power of 2. */
93 #define EVAL(p) (unsigned char **) \
94 ((unsigned char *)0 + \
95 (((unsigned char *)p + PSIZE - 1 - (unsigned char *) 0) & ~(PSIZE - 1)))
96 #endif /*HAVE_MERGESORT*/
97
98 /*
99 * Arguments are as for qsort.
100 */
101 int
yasm__mergesort(void * base,size_t nmemb,size_t size,int (* cmp)(const void *,const void *))102 yasm__mergesort(void *base, size_t nmemb, size_t size,
103 int (*cmp)(const void *, const void *))
104 {
105 #ifdef HAVE_MERGESORT
106 return mergesort(base, nmemb, size, cmp);
107 #else
108 size_t i;
109 int sense;
110 int big, iflag;
111 unsigned char *f1, *f2, *t, *b, *tp2, *q, *l1, *l2;
112 unsigned char *list2, *list1, *p2, *p, *last, **p1;
113
114 if (size < PSIZE / 2) { /* Pointers must fit into 2 * size. */
115 #ifdef EINVAL
116 errno = EINVAL;
117 #endif
118 return (-1);
119 }
120
121 if (nmemb == 0)
122 return (0);
123
124 /*
125 * XXX
126 * Stupid subtraction for the Cray.
127 */
128 iflag = 0;
129 if (!(size % ISIZE) && !(((char *)base - (char *)0) % ISIZE))
130 iflag = 1;
131
132 if ((list2 = yasm_xmalloc(nmemb * size + PSIZE)) == NULL)
133 return (-1);
134
135 list1 = base;
136 setup(list1, list2, nmemb, size, cmp);
137 last = list2 + nmemb * size;
138 i = 0;
139 big = 0;
140 while (*EVAL(list2) != last) {
141 l2 = list1;
142 p1 = EVAL(list1);
143 for (tp2 = p2 = list2; p2 != last; p1 = EVAL(l2)) {
144 p2 = *EVAL(p2);
145 f1 = l2;
146 f2 = l1 = list1 + (p2 - list2);
147 if (p2 != last)
148 p2 = *EVAL(p2);
149 l2 = list1 + (p2 - list2);
150 while (f1 < l1 && f2 < l2) {
151 if ((*cmp)(f1, f2) <= 0) {
152 q = f2;
153 b = f1, t = l1;
154 sense = -1;
155 } else {
156 q = f1;
157 b = f2, t = l2;
158 sense = 0;
159 }
160 if (!big) { /* here i = 0 */
161 while ((b += size) < t && cmp(q, b) >sense)
162 if (++i == 6) {
163 big = 1;
164 goto EXPONENTIAL;
165 }
166 } else {
167 EXPONENTIAL: for (i = size; ; i <<= 1)
168 if ((p = (b + i)) >= t) {
169 if ((p = t - size) > b &&
170 (*cmp)(q, p) <= sense)
171 t = p;
172 else
173 b = p;
174 break;
175 } else if ((*cmp)(q, p) <= sense) {
176 t = p;
177 if (i == size)
178 big = 0;
179 goto FASTCASE;
180 } else
181 b = p;
182 while (t > b+size) {
183 i = (((t - b) / size) >> 1) * size;
184 if ((*cmp)(q, p = b + i) <= sense)
185 t = p;
186 else
187 b = p;
188 }
189 goto COPY;
190 FASTCASE: while (i > size)
191 if ((*cmp)(q,
192 p = b + (i >>= 1)) <= sense)
193 t = p;
194 else
195 b = p;
196 COPY: b = t;
197 }
198 i = size;
199 if (q == f1) {
200 if (iflag) {
201 ICOPY_LIST(f2, tp2, b);
202 ICOPY_ELT(f1, tp2, i);
203 } else {
204 CCOPY_LIST(f2, tp2, b);
205 CCOPY_ELT(f1, tp2, i);
206 }
207 } else {
208 if (iflag) {
209 ICOPY_LIST(f1, tp2, b);
210 ICOPY_ELT(f2, tp2, i);
211 } else {
212 CCOPY_LIST(f1, tp2, b);
213 CCOPY_ELT(f2, tp2, i);
214 }
215 }
216 }
217 if (f2 < l2) {
218 if (iflag)
219 ICOPY_LIST(f2, tp2, l2);
220 else
221 CCOPY_LIST(f2, tp2, l2);
222 } else if (f1 < l1) {
223 if (iflag)
224 ICOPY_LIST(f1, tp2, l1);
225 else
226 CCOPY_LIST(f1, tp2, l1);
227 }
228 *p1 = l2;
229 }
230 tp2 = list1; /* swap list1, list2 */
231 list1 = list2;
232 list2 = tp2;
233 last = list2 + nmemb*size;
234 }
235 if (base == list2) {
236 memmove(list2, list1, nmemb*size);
237 list2 = list1;
238 }
239 yasm_xfree(list2);
240 return (0);
241 #endif /*HAVE_MERGESORT*/
242 }
243
244 #ifndef HAVE_MERGESORT
245
246 #define swap(a, b) { \
247 s = b; \
248 i = size; \
249 do { \
250 tmp = *a; *a++ = *s; *s++ = tmp; \
251 } while (--i); \
252 a -= size; \
253 }
254 #define reverse(bot, top) { \
255 s = top; \
256 do { \
257 i = size; \
258 do { \
259 tmp = *bot; *bot++ = *s; *s++ = tmp; \
260 } while (--i); \
261 s -= size2; \
262 } while(bot < s); \
263 }
264
265 /*
266 * Optional hybrid natural/pairwise first pass. Eats up list1 in runs of
267 * increasing order, list2 in a corresponding linked list. Checks for runs
268 * when THRESHOLD/2 pairs compare with same sense. (Only used when NATURAL
269 * is defined. Otherwise simple pairwise merging is used.)
270 */
271 void
setup(unsigned char * list1,unsigned char * list2,size_t n,size_t size,int (* cmp)(const void *,const void *))272 setup(unsigned char *list1, unsigned char *list2, size_t n, size_t size,
273 int (*cmp)(const void *, const void *))
274 {
275 size_t i;
276 unsigned int tmp;
277 int length, sense;
278 size_t size2;
279 unsigned char *f1, *f2, *s, *l2, *last, *p2;
280
281 size2 = size*2;
282 if (n <= 5) {
283 insertionsort(list1, n, size, cmp);
284 *EVAL(list2) = (unsigned char*) list2 + n*size;
285 return;
286 }
287 /*
288 * Avoid running pointers out of bounds; limit n to evens
289 * for simplicity.
290 */
291 i = 4 + (n & 1);
292 insertionsort(list1 + (n - i) * size, i, size, cmp);
293 last = list1 + size * (n - i);
294 *EVAL(list2 + (last - list1)) = list2 + n * size;
295
296 #ifdef NATURAL
297 p2 = list2;
298 f1 = list1;
299 sense = (cmp(f1, f1 + size) > 0);
300 for (; f1 < last; sense = !sense) {
301 length = 2;
302 /* Find pairs with same sense. */
303 for (f2 = f1 + size2; f2 < last; f2 += size2) {
304 if ((cmp(f2, f2+ size) > 0) != sense)
305 break;
306 length += 2;
307 }
308 if (length < THRESHOLD) { /* Pairwise merge */
309 do {
310 p2 = *EVAL(p2) = f1 + size2 - list1 + list2;
311 if (sense > 0)
312 swap (f1, f1 + size);
313 } while ((f1 += size2) < f2);
314 } else { /* Natural merge */
315 l2 = f2;
316 for (f2 = f1 + size2; f2 < l2; f2 += size2) {
317 if ((cmp(f2-size, f2) > 0) != sense) {
318 p2 = *EVAL(p2) = f2 - list1 + list2;
319 if (sense > 0)
320 reverse(f1, f2-size);
321 f1 = f2;
322 }
323 }
324 if (sense > 0)
325 reverse (f1, f2-size);
326 f1 = f2;
327 if (f2 < last || cmp(f2 - size, f2) > 0)
328 p2 = *EVAL(p2) = f2 - list1 + list2;
329 else
330 p2 = *EVAL(p2) = list2 + n*size;
331 }
332 }
333 #else /* pairwise merge only. */
334 for (f1 = list1, p2 = list2; f1 < last; f1 += size2) {
335 p2 = *EVAL(p2) = p2 + size2;
336 if (cmp (f1, f1 + size) > 0)
337 swap(f1, f1 + size);
338 }
339 #endif /* NATURAL */
340 }
341
342 /*
343 * This is to avoid out-of-bounds addresses in sorting the
344 * last 4 elements.
345 */
346 static void
insertionsort(unsigned char * a,size_t n,size_t size,int (* cmp)(const void *,const void *))347 insertionsort(unsigned char *a, size_t n, size_t size,
348 int (*cmp)(const void *, const void *))
349 {
350 unsigned char *ai, *s, *t, *u, tmp;
351 size_t i;
352
353 for (ai = a+size; --n >= 1; ai += size)
354 for (t = ai; t > a; t -= size) {
355 u = t - size;
356 if (cmp(u, t) <= 0)
357 break;
358 swap(u, t);
359 }
360 }
361 #endif /*HAVE_MERGESORT*/
362