• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* Copyright (C) 2011 by Valentin Ochs
2  *
3  * Permission is hereby granted, free of charge, to any person obtaining a copy
4  * of this software and associated documentation files (the "Software"), to
5  * deal in the Software without restriction, including without limitation the
6  * rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
7  * sell copies of the Software, and to permit persons to whom the Software is
8  * furnished to do so, subject to the following conditions:
9  *
10  * The above copyright notice and this permission notice shall be included in
11  * all copies or substantial portions of the Software.
12  *
13  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
14  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
15  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
16  * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
17  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
18  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
19  * IN THE SOFTWARE.
20  */
21 
22 /* Minor changes by Rich Felker for integration in musl, 2011-04-27. */
23 
24 /* Smoothsort, an adaptive variant of Heapsort.  Memory usage: O(1).
25    Run time: Worst case O(n log n), close to O(n) in the mostly-sorted case. */
26 
27 #define _BSD_SOURCE
28 #include <stdint.h>
29 #include <stdlib.h>
30 #include <string.h>
31 
32 #include "atomic.h"
33 #define ntz(x) a_ctz_l((x))
34 
35 typedef int (*cmpfun)(const void *, const void *, void *);
36 
pntz(size_t p[2])37 static inline int pntz(size_t p[2]) {
38 	int r = ntz(p[0] - 1);
39 	if(r != 0 || (r = 8*sizeof(size_t) + ntz(p[1])) != 8*sizeof(size_t)) {
40 		return r;
41 	}
42 	return 0;
43 }
44 
cycle(size_t width,unsigned char * ar[],int n)45 static void cycle(size_t width, unsigned char* ar[], int n)
46 {
47 	unsigned char tmp[256];
48 	size_t l;
49 	int i;
50 
51 	if(n < 2) {
52 		return;
53 	}
54 
55 	ar[n] = tmp;
56 	while(width) {
57 		l = sizeof(tmp) < width ? sizeof(tmp) : width;
58 		memcpy(ar[n], ar[0], l);
59 		for(i = 0; i < n; i++) {
60 			memcpy(ar[i], ar[i + 1], l);
61 			ar[i] += l;
62 		}
63 		width -= l;
64 	}
65 }
66 
67 /* shl() and shr() need n > 0 */
shl(size_t p[2],int n)68 static inline void shl(size_t p[2], int n)
69 {
70 	if(n >= 8 * sizeof(size_t)) {
71 		n -= 8 * sizeof(size_t);
72 		p[1] = p[0];
73 		p[0] = 0;
74 	}
75 	p[1] <<= n;
76 	p[1] |= p[0] >> (sizeof(size_t) * 8 - n);
77 	p[0] <<= n;
78 }
79 
shr(size_t p[2],int n)80 static inline void shr(size_t p[2], int n)
81 {
82 	if(n >= 8 * sizeof(size_t)) {
83 		n -= 8 * sizeof(size_t);
84 		p[0] = p[1];
85 		p[1] = 0;
86 	}
87 	p[0] >>= n;
88 	p[0] |= p[1] << (sizeof(size_t) * 8 - n);
89 	p[1] >>= n;
90 }
91 
sift(unsigned char * head,size_t width,cmpfun cmp,void * arg,int pshift,size_t lp[])92 static void sift(unsigned char *head, size_t width, cmpfun cmp, void *arg, int pshift, size_t lp[])
93 {
94 	unsigned char *rt, *lf;
95 	unsigned char *ar[14 * sizeof(size_t) + 1];
96 	int i = 1;
97 
98 	ar[0] = head;
99 	while(pshift > 1) {
100 		rt = head - width;
101 		lf = head - width - lp[pshift - 2];
102 
103 		if(cmp(ar[0], lf, arg) >= 0 && cmp(ar[0], rt, arg) >= 0) {
104 			break;
105 		}
106 		if(cmp(lf, rt, arg) >= 0) {
107 			ar[i++] = lf;
108 			head = lf;
109 			pshift -= 1;
110 		} else {
111 			ar[i++] = rt;
112 			head = rt;
113 			pshift -= 2;
114 		}
115 	}
116 	cycle(width, ar, i);
117 }
118 
trinkle(unsigned char * head,size_t width,cmpfun cmp,void * arg,size_t pp[2],int pshift,int trusty,size_t lp[])119 static void trinkle(unsigned char *head, size_t width, cmpfun cmp, void *arg, size_t pp[2], int pshift, int trusty, size_t lp[])
120 {
121 	unsigned char *stepson,
122 	              *rt, *lf;
123 	size_t p[2];
124 	unsigned char *ar[14 * sizeof(size_t) + 1];
125 	int i = 1;
126 	int trail;
127 
128 	p[0] = pp[0];
129 	p[1] = pp[1];
130 
131 	ar[0] = head;
132 	while(p[0] != 1 || p[1] != 0) {
133 		stepson = head - lp[pshift];
134 		if(cmp(stepson, ar[0], arg) <= 0) {
135 			break;
136 		}
137 		if(!trusty && pshift > 1) {
138 			rt = head - width;
139 			lf = head - width - lp[pshift - 2];
140 			if(cmp(rt, stepson, arg) >= 0 || cmp(lf, stepson, arg) >= 0) {
141 				break;
142 			}
143 		}
144 
145 		ar[i++] = stepson;
146 		head = stepson;
147 		trail = pntz(p);
148 		shr(p, trail);
149 		pshift += trail;
150 		trusty = 0;
151 	}
152 	if(!trusty) {
153 		cycle(width, ar, i);
154 		sift(head, width, cmp, arg, pshift, lp);
155 	}
156 }
157 
__qsort_r(void * base,size_t nel,size_t width,cmpfun cmp,void * arg)158 void __qsort_r(void *base, size_t nel, size_t width, cmpfun cmp, void *arg)
159 {
160 	size_t lp[12*sizeof(size_t)];
161 	size_t i, size = width * nel;
162 	unsigned char *head, *high;
163 	size_t p[2] = {1, 0};
164 	int pshift = 1;
165 	int trail;
166 
167 	if (!size) return;
168 
169 	head = base;
170 	high = head + size - width;
171 
172 	/* Precompute Leonardo numbers, scaled by element width */
173 	for(lp[0]=lp[1]=width, i=2; (lp[i]=lp[i-2]+lp[i-1]+width) < size; i++);
174 
175 	while(head < high) {
176 		if((p[0] & 3) == 3) {
177 			sift(head, width, cmp, arg, pshift, lp);
178 			shr(p, 2);
179 			pshift += 2;
180 		} else {
181 			if(lp[pshift - 1] >= high - head) {
182 				trinkle(head, width, cmp, arg, p, pshift, 0, lp);
183 			} else {
184 				sift(head, width, cmp, arg, pshift, lp);
185 			}
186 
187 			if(pshift == 1) {
188 				shl(p, 1);
189 				pshift = 0;
190 			} else {
191 				shl(p, pshift - 1);
192 				pshift = 1;
193 			}
194 		}
195 
196 		p[0] |= 1;
197 		head += width;
198 	}
199 
200 	trinkle(head, width, cmp, arg, p, pshift, 0, lp);
201 
202 	while(pshift != 1 || p[0] != 1 || p[1] != 0) {
203 		if(pshift <= 1) {
204 			trail = pntz(p);
205 			shr(p, trail);
206 			pshift += trail;
207 		} else {
208 			shl(p, 2);
209 			pshift -= 2;
210 			p[0] ^= 7;
211 			shr(p, 1);
212 			trinkle(head - lp[pshift] - width, width, cmp, arg, p, pshift + 1, 1, lp);
213 			shl(p, 1);
214 			p[0] |= 1;
215 			trinkle(head - width, width, cmp, arg, p, pshift, 1, lp);
216 		}
217 		head -= width;
218 	}
219 }
220 
221 weak_alias(__qsort_r, qsort_r);
222