xref: /openbsd-src/lib/libc/stdlib/malloc.c (revision e5157e49389faebcb42b7237d55fbf096d9c2523)
1 /*	$OpenBSD: malloc.c,v 1.171 2014/08/18 14:34:58 tedu Exp $	*/
2 /*
3  * Copyright (c) 2008, 2010, 2011 Otto Moerbeek <otto@drijf.net>
4  * Copyright (c) 2012 Matthew Dempsky <matthew@openbsd.org>
5  * Copyright (c) 2008 Damien Miller <djm@openbsd.org>
6  * Copyright (c) 2000 Poul-Henning Kamp <phk@FreeBSD.org>
7  *
8  * Permission to use, copy, modify, and distribute this software for any
9  * purpose with or without fee is hereby granted, provided that the above
10  * copyright notice and this permission notice appear in all copies.
11  *
12  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
13  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
14  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
15  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
16  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
17  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
18  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
19  */
20 
21 /*
22  * If we meet some day, and you think this stuff is worth it, you
23  * can buy me a beer in return. Poul-Henning Kamp
24  */
25 
26 /* #define MALLOC_STATS */
27 
28 #include <sys/types.h>
29 #include <sys/param.h>
30 #include <sys/queue.h>
31 #include <sys/mman.h>
32 #include <sys/uio.h>
33 #include <errno.h>
34 #include <stdint.h>
35 #include <stdlib.h>
36 #include <string.h>
37 #include <stdio.h>
38 #include <unistd.h>
39 
40 #ifdef MALLOC_STATS
41 #include <sys/tree.h>
42 #include <fcntl.h>
43 #endif
44 
45 #include "thread_private.h"
46 
47 #if defined(__sparc__) && !defined(__sparcv9__)
48 #define MALLOC_PAGESHIFT	(13U)
49 #elif defined(__mips64__)
50 #define MALLOC_PAGESHIFT	(14U)
51 #else
52 #define MALLOC_PAGESHIFT	(PAGE_SHIFT)
53 #endif
54 
55 #define MALLOC_MINSHIFT		4
56 #define MALLOC_MAXSHIFT		(MALLOC_PAGESHIFT - 1)
57 #define MALLOC_PAGESIZE		(1UL << MALLOC_PAGESHIFT)
58 #define MALLOC_MINSIZE		(1UL << MALLOC_MINSHIFT)
59 #define MALLOC_PAGEMASK		(MALLOC_PAGESIZE - 1)
60 #define MASK_POINTER(p)		((void *)(((uintptr_t)(p)) & ~MALLOC_PAGEMASK))
61 
62 #define MALLOC_MAXCHUNK		(1 << MALLOC_MAXSHIFT)
63 #define MALLOC_MAXCACHE		256
64 #define MALLOC_DELAYED_CHUNK_MASK	15
65 #define MALLOC_INITIAL_REGIONS	512
66 #define MALLOC_DEFAULT_CACHE	64
67 #define	MALLOC_CHUNK_LISTS	4
68 
69 /*
70  * When the P option is active, we move allocations between half a page
71  * and a whole page towards the end, subject to alignment constraints.
72  * This is the extra headroom we allow. Set to zero to be the most
73  * strict.
74  */
75 #define MALLOC_LEEWAY		0
76 
77 #define PAGEROUND(x)  (((x) + (MALLOC_PAGEMASK)) & ~MALLOC_PAGEMASK)
78 
79 /*
80  * What to use for Junk.  This is the byte value we use to fill with
81  * when the 'J' option is enabled. Use SOME_JUNK right after alloc,
82  * and SOME_FREEJUNK right before free.
83  */
84 #define SOME_JUNK		0xd0	/* as in "Duh" :-) */
85 #define SOME_FREEJUNK		0xdf
86 
87 #define MMAP(sz)	mmap(NULL, (size_t)(sz), PROT_READ | PROT_WRITE, \
88     MAP_ANON | MAP_PRIVATE, -1, (off_t) 0)
89 
90 #define MMAPA(a,sz)	mmap((a), (size_t)(sz), PROT_READ | PROT_WRITE, \
91     MAP_ANON | MAP_PRIVATE, -1, (off_t) 0)
92 
93 #define MQUERY(a, sz)	mquery((a), (size_t)(sz), PROT_READ | PROT_WRITE, \
94     MAP_ANON | MAP_PRIVATE | MAP_FIXED, -1, (off_t)0)
95 
96 #define KERNENTER() if (__isthreaded) do { \
97 	malloc_active--; \
98 	_MALLOC_UNLOCK(); \
99 } while (0)
100 #define KERNEXIT() if (__isthreaded) do { \
101 	_MALLOC_LOCK(); \
102 	malloc_active++; \
103 } while (0)
104 
105 struct region_info {
106 	void *p;		/* page; low bits used to mark chunks */
107 	uintptr_t size;		/* size for pages, or chunk_info pointer */
108 #ifdef MALLOC_STATS
109 	void *f;		/* where allocated from */
110 #endif
111 };
112 
113 LIST_HEAD(chunk_head, chunk_info);
114 
115 struct dir_info {
116 	u_int32_t canary1;
117 	struct region_info *r;		/* region slots */
118 	size_t regions_total;		/* number of region slots */
119 	size_t regions_free;		/* number of free slots */
120 					/* lists of free chunk info structs */
121 	struct chunk_head chunk_info_list[MALLOC_MAXSHIFT + 1];
122 					/* lists of chunks with free slots */
123 	struct chunk_head chunk_dir[MALLOC_MAXSHIFT + 1][MALLOC_CHUNK_LISTS];
124 	size_t free_regions_size;	/* free pages cached */
125 					/* free pages cache */
126 	struct region_info free_regions[MALLOC_MAXCACHE];
127 					/* delayed free chunk slots */
128 	void *delayed_chunks[MALLOC_DELAYED_CHUNK_MASK + 1];
129 	size_t rbytesused;		/* random bytes used */
130 	u_char rbytes[32];		/* random bytes */
131 	u_short chunk_start;
132 #ifdef MALLOC_STATS
133 	size_t inserts;
134 	size_t insert_collisions;
135 	size_t finds;
136 	size_t find_collisions;
137 	size_t deletes;
138 	size_t delete_moves;
139 	size_t cheap_realloc_tries;
140 	size_t cheap_reallocs;
141 	size_t malloc_used;		/* bytes allocated */
142 	size_t malloc_guarded;		/* bytes used for guards */
143 #define STATS_ADD(x,y)	((x) += (y))
144 #define STATS_SUB(x,y)	((x) -= (y))
145 #define STATS_INC(x)	((x)++)
146 #define STATS_ZERO(x)	((x) = 0)
147 #define STATS_SETF(x,y)	((x)->f = (y))
148 #else
149 #define STATS_ADD(x,y)	/* nothing */
150 #define STATS_SUB(x,y)	/* nothing */
151 #define STATS_INC(x)	/* nothing */
152 #define STATS_ZERO(x)	/* nothing */
153 #define STATS_SETF(x,y)	/* nothing */
154 #endif /* MALLOC_STATS */
155 	u_int32_t canary2;
156 };
157 #define DIR_INFO_RSZ	((sizeof(struct dir_info) + MALLOC_PAGEMASK) & \
158 			~MALLOC_PAGEMASK)
159 
160 /*
161  * This structure describes a page worth of chunks.
162  *
163  * How many bits per u_short in the bitmap
164  */
165 #define MALLOC_BITS		(NBBY * sizeof(u_short))
166 struct chunk_info {
167 	LIST_ENTRY(chunk_info) entries;
168 	void *page;			/* pointer to the page */
169 	u_int32_t canary;
170 	u_short size;			/* size of this page's chunks */
171 	u_short shift;			/* how far to shift for this size */
172 	u_short free;			/* how many free chunks */
173 	u_short total;			/* how many chunk */
174 					/* which chunks are free */
175 	u_short bits[1];
176 };
177 
178 struct malloc_readonly {
179 	struct dir_info *malloc_pool;	/* Main bookkeeping information */
180 	int	malloc_abort;		/* abort() on error */
181 	int	malloc_freenow;		/* Free quickly - disable chunk rnd */
182 	int	malloc_freeunmap;	/* mprotect free pages PROT_NONE? */
183 	int	malloc_hint;		/* call madvice on free pages?  */
184 	int	malloc_junk;		/* junk fill? */
185 	int	malloc_move;		/* move allocations to end of page? */
186 	int	malloc_realloc;		/* always realloc? */
187 	int	malloc_xmalloc;		/* xmalloc behaviour? */
188 	size_t	malloc_guard;		/* use guard pages after allocations? */
189 	u_int	malloc_cache;		/* free pages we cache */
190 #ifdef MALLOC_STATS
191 	int	malloc_stats;		/* dump statistics at end */
192 #endif
193 	u_int32_t malloc_canary;	/* Matched against ones in malloc_pool */
194 };
195 
196 /* This object is mapped PROT_READ after initialisation to prevent tampering */
197 static union {
198 	struct malloc_readonly mopts;
199 	u_char _pad[MALLOC_PAGESIZE];
200 } malloc_readonly __attribute__((aligned(MALLOC_PAGESIZE)));
201 #define mopts	malloc_readonly.mopts
202 #define getpool() mopts.malloc_pool
203 
204 char		*malloc_options;	/* compile-time options */
205 static char	*malloc_func;		/* current function */
206 static int	malloc_active;		/* status of malloc */
207 
208 static u_char getrbyte(struct dir_info *d);
209 
210 extern char	*__progname;
211 
212 #ifdef MALLOC_STATS
213 void malloc_dump(int);
214 static void malloc_exit(void);
215 #define CALLER	__builtin_return_address(0)
216 #else
217 #define CALLER	NULL
218 #endif
219 
220 /* low bits of r->p determine size: 0 means >= page size and p->size holding
221  *  real size, otherwise r->size is a shift count, or 1 for malloc(0)
222  */
223 #define REALSIZE(sz, r)						\
224 	(sz) = (uintptr_t)(r)->p & MALLOC_PAGEMASK,		\
225 	(sz) = ((sz) == 0 ? (r)->size : ((sz) == 1 ? 0 : (1 << ((sz)-1))))
226 
227 static inline size_t
228 hash(void *p)
229 {
230 	size_t sum;
231 	uintptr_t u;
232 
233 	u = (uintptr_t)p >> MALLOC_PAGESHIFT;
234 	sum = u;
235 	sum = (sum << 7) - sum + (u >> 16);
236 #ifdef __LP64__
237 	sum = (sum << 7) - sum + (u >> 32);
238 	sum = (sum << 7) - sum + (u >> 48);
239 #endif
240 	return sum;
241 }
242 
243 static void
244 wrterror(char *msg, void *p)
245 {
246 	char		*q = " error: ";
247 	struct iovec	iov[7];
248 	char		pidbuf[20];
249 	char		buf[20];
250 	int		saved_errno = errno;
251 
252 	iov[0].iov_base = __progname;
253 	iov[0].iov_len = strlen(__progname);
254 	iov[1].iov_base = pidbuf;
255 	snprintf(pidbuf, sizeof(pidbuf), "(%d) in ", getpid());
256 	iov[1].iov_len = strlen(pidbuf);
257 	iov[2].iov_base = malloc_func;
258 	iov[2].iov_len = strlen(malloc_func);
259 	iov[3].iov_base = q;
260 	iov[3].iov_len = strlen(q);
261 	iov[4].iov_base = msg;
262 	iov[4].iov_len = strlen(msg);
263 	iov[5].iov_base = buf;
264 	if (p == NULL)
265 		iov[5].iov_len = 0;
266 	else {
267 		snprintf(buf, sizeof(buf), " %p", p);
268 		iov[5].iov_len = strlen(buf);
269 	}
270 	iov[6].iov_base = "\n";
271 	iov[6].iov_len = 1;
272 	writev(STDERR_FILENO, iov, 7);
273 
274 #ifdef MALLOC_STATS
275 	if (mopts.malloc_stats)
276 		malloc_dump(STDERR_FILENO);
277 #endif /* MALLOC_STATS */
278 
279 	errno = saved_errno;
280 	if (mopts.malloc_abort)
281 		abort();
282 }
283 
284 static void
285 rbytes_init(struct dir_info *d)
286 {
287 	arc4random_buf(d->rbytes, sizeof(d->rbytes));
288 	/* add 1 to account for using d->rbytes[0] */
289 	d->rbytesused = 1 + d->rbytes[0] % (sizeof(d->rbytes) / 2);
290 }
291 
292 static inline u_char
293 getrbyte(struct dir_info *d)
294 {
295 	u_char x;
296 
297 	if (d->rbytesused >= sizeof(d->rbytes))
298 		rbytes_init(d);
299 	x = d->rbytes[d->rbytesused++];
300 	return x;
301 }
302 
303 /*
304  * Cache maintenance. We keep at most malloc_cache pages cached.
305  * If the cache is becoming full, unmap pages in the cache for real,
306  * and then add the region to the cache
307  * Opposed to the regular region data structure, the sizes in the
308  * cache are in MALLOC_PAGESIZE units.
309  */
310 static void
311 unmap(struct dir_info *d, void *p, size_t sz)
312 {
313 	size_t psz = sz >> MALLOC_PAGESHIFT;
314 	size_t rsz, tounmap;
315 	struct region_info *r;
316 	u_int i, offset;
317 
318 	if (sz != PAGEROUND(sz)) {
319 		wrterror("munmap round", NULL);
320 		return;
321 	}
322 
323 	if (psz > mopts.malloc_cache) {
324 		i = munmap(p, sz);
325 		if (i)
326 			wrterror("munmap", p);
327 		STATS_SUB(d->malloc_used, sz);
328 		return;
329 	}
330 	tounmap = 0;
331 	rsz = mopts.malloc_cache - d->free_regions_size;
332 	if (psz > rsz)
333 		tounmap = psz - rsz;
334 	offset = getrbyte(d);
335 	for (i = 0; tounmap > 0 && i < mopts.malloc_cache; i++) {
336 		r = &d->free_regions[(i + offset) & (mopts.malloc_cache - 1)];
337 		if (r->p != NULL) {
338 			rsz = r->size << MALLOC_PAGESHIFT;
339 			if (munmap(r->p, rsz))
340 				wrterror("munmap", r->p);
341 			r->p = NULL;
342 			if (tounmap > r->size)
343 				tounmap -= r->size;
344 			else
345 				tounmap = 0;
346 			d->free_regions_size -= r->size;
347 			r->size = 0;
348 			STATS_SUB(d->malloc_used, rsz);
349 		}
350 	}
351 	if (tounmap > 0)
352 		wrterror("malloc cache underflow", NULL);
353 	for (i = 0; i < mopts.malloc_cache; i++) {
354 		r = &d->free_regions[(i + offset) & (mopts.malloc_cache - 1)];
355 		if (r->p == NULL) {
356 			if (mopts.malloc_hint)
357 				madvise(p, sz, MADV_FREE);
358 			if (mopts.malloc_freeunmap)
359 				mprotect(p, sz, PROT_NONE);
360 			r->p = p;
361 			r->size = psz;
362 			d->free_regions_size += psz;
363 			break;
364 		}
365 	}
366 	if (i == mopts.malloc_cache)
367 		wrterror("malloc free slot lost", NULL);
368 	if (d->free_regions_size > mopts.malloc_cache)
369 		wrterror("malloc cache overflow", NULL);
370 }
371 
372 static void
373 zapcacheregion(struct dir_info *d, void *p, size_t len)
374 {
375 	u_int i;
376 	struct region_info *r;
377 	size_t rsz;
378 
379 	for (i = 0; i < mopts.malloc_cache; i++) {
380 		r = &d->free_regions[i];
381 		if (r->p >= p && r->p <= (void *)((char *)p + len)) {
382 			rsz = r->size << MALLOC_PAGESHIFT;
383 			if (munmap(r->p, rsz))
384 				wrterror("munmap", r->p);
385 			r->p = NULL;
386 			d->free_regions_size -= r->size;
387 			r->size = 0;
388 			STATS_SUB(d->malloc_used, rsz);
389 		}
390 	}
391 }
392 
393 static void *
394 map(struct dir_info *d, size_t sz, int zero_fill)
395 {
396 	size_t psz = sz >> MALLOC_PAGESHIFT;
397 	struct region_info *r, *big = NULL;
398 	u_int i, offset;
399 	void *p;
400 
401 	if (mopts.malloc_canary != (d->canary1 ^ (u_int32_t)(uintptr_t)d) ||
402 	    d->canary1 != ~d->canary2)
403 		wrterror("internal struct corrupt", NULL);
404 	if (sz != PAGEROUND(sz)) {
405 		wrterror("map round", NULL);
406 		return MAP_FAILED;
407 	}
408 	if (psz > d->free_regions_size) {
409 		KERNENTER();
410 		p = MMAP(sz);
411 		KERNEXIT();
412 		if (p != MAP_FAILED)
413 			STATS_ADD(d->malloc_used, sz);
414 		/* zero fill not needed */
415 		return p;
416 	}
417 	offset = getrbyte(d);
418 	for (i = 0; i < mopts.malloc_cache; i++) {
419 		r = &d->free_regions[(i + offset) & (mopts.malloc_cache - 1)];
420 		if (r->p != NULL) {
421 			if (r->size == psz) {
422 				p = r->p;
423 				r->p = NULL;
424 				r->size = 0;
425 				d->free_regions_size -= psz;
426 				if (mopts.malloc_freeunmap)
427 					mprotect(p, sz, PROT_READ | PROT_WRITE);
428 				if (mopts.malloc_hint)
429 					madvise(p, sz, MADV_NORMAL);
430 				if (zero_fill)
431 					memset(p, 0, sz);
432 				else if (mopts.malloc_junk == 2 &&
433 				    mopts.malloc_freeunmap)
434 					memset(p, SOME_FREEJUNK, sz);
435 				return p;
436 			} else if (r->size > psz)
437 				big = r;
438 		}
439 	}
440 	if (big != NULL) {
441 		r = big;
442 		p = (char *)r->p + ((r->size - psz) << MALLOC_PAGESHIFT);
443 		if (mopts.malloc_freeunmap)
444 			mprotect(p, sz, PROT_READ | PROT_WRITE);
445 		if (mopts.malloc_hint)
446 			madvise(p, sz, MADV_NORMAL);
447 		r->size -= psz;
448 		d->free_regions_size -= psz;
449 		if (zero_fill)
450 			memset(p, 0, sz);
451 		else if (mopts.malloc_junk == 2 && mopts.malloc_freeunmap)
452 			memset(p, SOME_FREEJUNK, sz);
453 		return p;
454 	}
455 	if (d->free_regions_size > mopts.malloc_cache)
456 		wrterror("malloc cache", NULL);
457 	KERNENTER();
458 	p = MMAP(sz);
459 	KERNEXIT();
460 	if (p != MAP_FAILED)
461 		STATS_ADD(d->malloc_used, sz);
462 	/* zero fill not needed */
463 	return p;
464 }
465 
466 /*
467  * Initialize a dir_info, which should have been cleared by caller
468  */
469 static int
470 omalloc_init(struct dir_info **dp)
471 {
472 	char *p, b[64];
473 	int i, j;
474 	size_t d_avail, regioninfo_size;
475 	struct dir_info *d;
476 
477 	/*
478 	 * Default options
479 	 */
480 	mopts.malloc_abort = 1;
481 	mopts.malloc_junk = 1;
482 	mopts.malloc_move = 1;
483 	mopts.malloc_cache = MALLOC_DEFAULT_CACHE;
484 
485 	for (i = 0; i < 3; i++) {
486 		switch (i) {
487 		case 0:
488 			j = readlink("/etc/malloc.conf", b, sizeof b - 1);
489 			if (j <= 0)
490 				continue;
491 			b[j] = '\0';
492 			p = b;
493 			break;
494 		case 1:
495 			if (issetugid() == 0)
496 				p = getenv("MALLOC_OPTIONS");
497 			else
498 				continue;
499 			break;
500 		case 2:
501 			p = malloc_options;
502 			break;
503 		default:
504 			p = NULL;
505 		}
506 
507 		for (; p != NULL && *p != '\0'; p++) {
508 			switch (*p) {
509 			case '>':
510 				mopts.malloc_cache <<= 1;
511 				if (mopts.malloc_cache > MALLOC_MAXCACHE)
512 					mopts.malloc_cache = MALLOC_MAXCACHE;
513 				break;
514 			case '<':
515 				mopts.malloc_cache >>= 1;
516 				break;
517 			case 'a':
518 				mopts.malloc_abort = 0;
519 				break;
520 			case 'A':
521 				mopts.malloc_abort = 1;
522 				break;
523 #ifdef MALLOC_STATS
524 			case 'd':
525 				mopts.malloc_stats = 0;
526 				break;
527 			case 'D':
528 				mopts.malloc_stats = 1;
529 				break;
530 #endif /* MALLOC_STATS */
531 			case 'f':
532 				mopts.malloc_freenow = 0;
533 				mopts.malloc_freeunmap = 0;
534 				break;
535 			case 'F':
536 				mopts.malloc_freenow = 1;
537 				mopts.malloc_freeunmap = 1;
538 				break;
539 			case 'g':
540 				mopts.malloc_guard = 0;
541 				break;
542 			case 'G':
543 				mopts.malloc_guard = MALLOC_PAGESIZE;
544 				break;
545 			case 'h':
546 				mopts.malloc_hint = 0;
547 				break;
548 			case 'H':
549 				mopts.malloc_hint = 1;
550 				break;
551 			case 'j':
552 				mopts.malloc_junk = 0;
553 				break;
554 			case 'J':
555 				mopts.malloc_junk = 2;
556 				break;
557 			case 'n':
558 			case 'N':
559 				break;
560 			case 'p':
561 				mopts.malloc_move = 0;
562 				break;
563 			case 'P':
564 				mopts.malloc_move = 1;
565 				break;
566 			case 'r':
567 				mopts.malloc_realloc = 0;
568 				break;
569 			case 'R':
570 				mopts.malloc_realloc = 1;
571 				break;
572 			case 's':
573 				mopts.malloc_freeunmap = mopts.malloc_junk = 0;
574 				mopts.malloc_guard = 0;
575 				mopts.malloc_cache = MALLOC_DEFAULT_CACHE;
576 				break;
577 			case 'S':
578 				mopts.malloc_freeunmap = 1;
579 				mopts.malloc_junk = 2;
580 				mopts.malloc_guard = MALLOC_PAGESIZE;
581 				mopts.malloc_cache = 0;
582 				break;
583 			case 'u':
584 				mopts.malloc_freeunmap = 0;
585 				break;
586 			case 'U':
587 				mopts.malloc_freeunmap = 1;
588 				break;
589 			case 'x':
590 				mopts.malloc_xmalloc = 0;
591 				break;
592 			case 'X':
593 				mopts.malloc_xmalloc = 1;
594 				break;
595 			default: {
596 				static const char q[] = "malloc() warning: "
597 				    "unknown char in MALLOC_OPTIONS\n";
598 				write(STDERR_FILENO, q, sizeof(q) - 1);
599 				break;
600 			}
601 			}
602 		}
603 	}
604 
605 #ifdef MALLOC_STATS
606 	if (mopts.malloc_stats && (atexit(malloc_exit) == -1)) {
607 		static const char q[] = "malloc() warning: atexit(2) failed."
608 		    " Will not be able to dump stats on exit\n";
609 		write(STDERR_FILENO, q, sizeof(q) - 1);
610 	}
611 #endif /* MALLOC_STATS */
612 
613 	while ((mopts.malloc_canary = arc4random()) == 0)
614 		;
615 
616 	/*
617 	 * Allocate dir_info with a guard page on either side. Also
618 	 * randomise offset inside the page at which the dir_info
619 	 * lies (subject to alignment by 1 << MALLOC_MINSHIFT)
620 	 */
621 	if ((p = MMAP(DIR_INFO_RSZ + (MALLOC_PAGESIZE * 2))) == MAP_FAILED)
622 		return -1;
623 	mprotect(p, MALLOC_PAGESIZE, PROT_NONE);
624 	mprotect(p + MALLOC_PAGESIZE + DIR_INFO_RSZ,
625 	    MALLOC_PAGESIZE, PROT_NONE);
626 	d_avail = (DIR_INFO_RSZ - sizeof(*d)) >> MALLOC_MINSHIFT;
627 	d = (struct dir_info *)(p + MALLOC_PAGESIZE +
628 	    (arc4random_uniform(d_avail) << MALLOC_MINSHIFT));
629 
630 	rbytes_init(d);
631 	d->regions_free = d->regions_total = MALLOC_INITIAL_REGIONS;
632 	regioninfo_size = d->regions_total * sizeof(struct region_info);
633 	d->r = MMAP(regioninfo_size);
634 	if (d->r == MAP_FAILED) {
635 		wrterror("malloc init mmap failed", NULL);
636 		d->regions_total = 0;
637 		return 1;
638 	}
639 	for (i = 0; i <= MALLOC_MAXSHIFT; i++) {
640 		LIST_INIT(&d->chunk_info_list[i]);
641 		for (j = 0; j < MALLOC_CHUNK_LISTS; j++)
642 			LIST_INIT(&d->chunk_dir[i][j]);
643 	}
644 	STATS_ADD(d->malloc_used, regioninfo_size);
645 	d->canary1 = mopts.malloc_canary ^ (u_int32_t)(uintptr_t)d;
646 	d->canary2 = ~d->canary1;
647 
648 	*dp = d;
649 
650 	/*
651 	 * Options have been set and will never be reset.
652 	 * Prevent further tampering with them.
653 	 */
654 	if (((uintptr_t)&malloc_readonly & MALLOC_PAGEMASK) == 0)
655 		mprotect(&malloc_readonly, sizeof(malloc_readonly), PROT_READ);
656 
657 	return 0;
658 }
659 
660 static int
661 omalloc_grow(struct dir_info *d)
662 {
663 	size_t newtotal;
664 	size_t newsize;
665 	size_t mask;
666 	size_t i;
667 	struct region_info *p;
668 
669 	if (d->regions_total > SIZE_MAX / sizeof(struct region_info) / 2 )
670 		return 1;
671 
672 	newtotal = d->regions_total * 2;
673 	newsize = newtotal * sizeof(struct region_info);
674 	mask = newtotal - 1;
675 
676 	p = MMAP(newsize);
677 	if (p == MAP_FAILED)
678 		return 1;
679 
680 	STATS_ADD(d->malloc_used, newsize);
681 	memset(p, 0, newsize);
682 	STATS_ZERO(d->inserts);
683 	STATS_ZERO(d->insert_collisions);
684 	for (i = 0; i < d->regions_total; i++) {
685 		void *q = d->r[i].p;
686 		if (q != NULL) {
687 			size_t index = hash(q) & mask;
688 			STATS_INC(d->inserts);
689 			while (p[index].p != NULL) {
690 				index = (index - 1) & mask;
691 				STATS_INC(d->insert_collisions);
692 			}
693 			p[index] = d->r[i];
694 		}
695 	}
696 	/* avoid pages containing meta info to end up in cache */
697 	if (munmap(d->r, d->regions_total * sizeof(struct region_info)))
698 		wrterror("munmap", d->r);
699 	else
700 		STATS_SUB(d->malloc_used,
701 		    d->regions_total * sizeof(struct region_info));
702 	d->regions_free = d->regions_free + d->regions_total;
703 	d->regions_total = newtotal;
704 	d->r = p;
705 	return 0;
706 }
707 
708 static struct chunk_info *
709 alloc_chunk_info(struct dir_info *d, int bits)
710 {
711 	struct chunk_info *p;
712 	size_t size, count;
713 
714 	if (bits == 0)
715 		count = MALLOC_PAGESIZE / MALLOC_MINSIZE;
716 	else
717 		count = MALLOC_PAGESIZE >> bits;
718 
719 	size = howmany(count, MALLOC_BITS);
720 	size = sizeof(struct chunk_info) + (size - 1) * sizeof(u_short);
721 	size = ALIGN(size);
722 
723 	if (LIST_EMPTY(&d->chunk_info_list[bits])) {
724 		char *q;
725 		int i;
726 
727 		q = MMAP(MALLOC_PAGESIZE);
728 		if (q == MAP_FAILED)
729 			return NULL;
730 		STATS_ADD(d->malloc_used, MALLOC_PAGESIZE);
731 		count = MALLOC_PAGESIZE / size;
732 		for (i = 0; i < count; i++, q += size)
733 			LIST_INSERT_HEAD(&d->chunk_info_list[bits],
734 			    (struct chunk_info *)q, entries);
735 	}
736 	p = LIST_FIRST(&d->chunk_info_list[bits]);
737 	LIST_REMOVE(p, entries);
738 	memset(p, 0, size);
739 	p->canary = d->canary1;
740 	return p;
741 }
742 
743 
744 /*
745  * The hashtable uses the assumption that p is never NULL. This holds since
746  * non-MAP_FIXED mappings with hint 0 start at BRKSIZ.
747  */
748 static int
749 insert(struct dir_info *d, void *p, size_t sz, void *f)
750 {
751 	size_t index;
752 	size_t mask;
753 	void *q;
754 
755 	if (d->regions_free * 4 < d->regions_total) {
756 		if (omalloc_grow(d))
757 			return 1;
758 	}
759 	mask = d->regions_total - 1;
760 	index = hash(p) & mask;
761 	q = d->r[index].p;
762 	STATS_INC(d->inserts);
763 	while (q != NULL) {
764 		index = (index - 1) & mask;
765 		q = d->r[index].p;
766 		STATS_INC(d->insert_collisions);
767 	}
768 	d->r[index].p = p;
769 	d->r[index].size = sz;
770 #ifdef MALLOC_STATS
771 	d->r[index].f = f;
772 #endif
773 	d->regions_free--;
774 	return 0;
775 }
776 
777 static struct region_info *
778 find(struct dir_info *d, void *p)
779 {
780 	size_t index;
781 	size_t mask = d->regions_total - 1;
782 	void *q, *r;
783 
784 	if (mopts.malloc_canary != (d->canary1 ^ (u_int32_t)(uintptr_t)d) ||
785 	    d->canary1 != ~d->canary2)
786 		wrterror("internal struct corrupt", NULL);
787 	p = MASK_POINTER(p);
788 	index = hash(p) & mask;
789 	r = d->r[index].p;
790 	q = MASK_POINTER(r);
791 	STATS_INC(d->finds);
792 	while (q != p && r != NULL) {
793 		index = (index - 1) & mask;
794 		r = d->r[index].p;
795 		q = MASK_POINTER(r);
796 		STATS_INC(d->find_collisions);
797 	}
798 	return (q == p && r != NULL) ? &d->r[index] : NULL;
799 }
800 
801 static void
802 delete(struct dir_info *d, struct region_info *ri)
803 {
804 	/* algorithm R, Knuth Vol III section 6.4 */
805 	size_t mask = d->regions_total - 1;
806 	size_t i, j, r;
807 
808 	if (d->regions_total & (d->regions_total - 1))
809 		wrterror("regions_total not 2^x", NULL);
810 	d->regions_free++;
811 	STATS_INC(getpool()->deletes);
812 
813 	i = ri - d->r;
814 	for (;;) {
815 		d->r[i].p = NULL;
816 		d->r[i].size = 0;
817 		j = i;
818 		for (;;) {
819 			i = (i - 1) & mask;
820 			if (d->r[i].p == NULL)
821 				return;
822 			r = hash(d->r[i].p) & mask;
823 			if ((i <= r && r < j) || (r < j && j < i) ||
824 			    (j < i && i <= r))
825 				continue;
826 			d->r[j] = d->r[i];
827 			STATS_INC(getpool()->delete_moves);
828 			break;
829 		}
830 
831 	}
832 }
833 
834 /*
835  * Allocate a page of chunks
836  */
837 static struct chunk_info *
838 omalloc_make_chunks(struct dir_info *d, int bits, int listnum)
839 {
840 	struct chunk_info *bp;
841 	void		*pp;
842 	int		i, k;
843 
844 	/* Allocate a new bucket */
845 	pp = map(d, MALLOC_PAGESIZE, 0);
846 	if (pp == MAP_FAILED)
847 		return NULL;
848 
849 	bp = alloc_chunk_info(d, bits);
850 	if (bp == NULL) {
851 		unmap(d, pp, MALLOC_PAGESIZE);
852 		return NULL;
853 	}
854 
855 	/* memory protect the page allocated in the malloc(0) case */
856 	if (bits == 0) {
857 		bp->size = 0;
858 		bp->shift = 1;
859 		i = MALLOC_MINSIZE - 1;
860 		while (i >>= 1)
861 			bp->shift++;
862 		bp->total = bp->free = MALLOC_PAGESIZE >> bp->shift;
863 		bp->page = pp;
864 
865 		k = mprotect(pp, MALLOC_PAGESIZE, PROT_NONE);
866 		if (k < 0) {
867 			unmap(d, pp, MALLOC_PAGESIZE);
868 			LIST_INSERT_HEAD(&d->chunk_info_list[0], bp, entries);
869 			return NULL;
870 		}
871 	} else {
872 		bp->size = 1U << bits;
873 		bp->shift = bits;
874 		bp->total = bp->free = MALLOC_PAGESIZE >> bits;
875 		bp->page = pp;
876 	}
877 
878 	/* set all valid bits in the bitmap */
879 	k = bp->total;
880 	i = 0;
881 
882 	/* Do a bunch at a time */
883 	for (; (k - i) >= MALLOC_BITS; i += MALLOC_BITS)
884 		bp->bits[i / MALLOC_BITS] = (u_short)~0U;
885 
886 	for (; i < k; i++)
887 		bp->bits[i / MALLOC_BITS] |= (u_short)1U << (i % MALLOC_BITS);
888 
889 	LIST_INSERT_HEAD(&d->chunk_dir[bits][listnum], bp, entries);
890 
891 	bits++;
892 	if ((uintptr_t)pp & bits)
893 		wrterror("pp & bits", pp);
894 
895 	insert(d, (void *)((uintptr_t)pp | bits), (uintptr_t)bp, NULL);
896 	return bp;
897 }
898 
899 
900 /*
901  * Allocate a chunk
902  */
903 static void *
904 malloc_bytes(struct dir_info *d, size_t size, void *f)
905 {
906 	int		i, j, listnum;
907 	size_t		k;
908 	u_short		u, *lp;
909 	struct chunk_info *bp;
910 
911 	if (mopts.malloc_canary != (d->canary1 ^ (u_int32_t)(uintptr_t)d) ||
912 	    d->canary1 != ~d->canary2)
913 		wrterror("internal struct corrupt", NULL);
914 	/* Don't bother with anything less than this */
915 	/* unless we have a malloc(0) requests */
916 	if (size != 0 && size < MALLOC_MINSIZE)
917 		size = MALLOC_MINSIZE;
918 
919 	/* Find the right bucket */
920 	if (size == 0)
921 		j = 0;
922 	else {
923 		j = MALLOC_MINSHIFT;
924 		i = (size - 1) >> (MALLOC_MINSHIFT - 1);
925 		while (i >>= 1)
926 			j++;
927 	}
928 
929 	listnum = getrbyte(d) % MALLOC_CHUNK_LISTS;
930 	/* If it's empty, make a page more of that size chunks */
931 	if ((bp = LIST_FIRST(&d->chunk_dir[j][listnum])) == NULL) {
932 		bp = omalloc_make_chunks(d, j, listnum);
933 		if (bp == NULL)
934 			return NULL;
935 	}
936 
937 	if (bp->canary != d->canary1)
938 		wrterror("chunk info corrupted", NULL);
939 
940 	i = d->chunk_start;
941 	if (bp->free > 1)
942 		i += getrbyte(d);
943 	if (i >= bp->total)
944 		i &= bp->total - 1;
945 	for (;;) {
946 		for (;;) {
947 			lp = &bp->bits[i / MALLOC_BITS];
948 			if (!*lp) {
949 				i += MALLOC_BITS;
950 				i &= ~(MALLOC_BITS - 1);
951 				if (i >= bp->total)
952 					i = 0;
953 			} else
954 				break;
955 		}
956 		k = i % MALLOC_BITS;
957 		u = 1 << k;
958 		if (*lp & u)
959 			break;
960 		if (++i >= bp->total)
961 			i = 0;
962 	}
963 	d->chunk_start += i + 1;
964 #ifdef MALLOC_STATS
965 	if (i == 0) {
966 		struct region_info *r = find(d, bp->page);
967 		r->f = f;
968 	}
969 #endif
970 
971 	*lp ^= u;
972 
973 	/* If there are no more free, remove from free-list */
974 	if (!--bp->free)
975 		LIST_REMOVE(bp, entries);
976 
977 	/* Adjust to the real offset of that chunk */
978 	k += (lp - bp->bits) * MALLOC_BITS;
979 	k <<= bp->shift;
980 
981 	if (mopts.malloc_junk == 2 && bp->size > 0)
982 		memset((char *)bp->page + k, SOME_JUNK, bp->size);
983 	return ((char *)bp->page + k);
984 }
985 
986 static uint32_t
987 find_chunknum(struct dir_info *d, struct region_info *r, void *ptr)
988 {
989 	struct chunk_info *info;
990 	uint32_t chunknum;
991 
992 	info = (struct chunk_info *)r->size;
993 	if (info->canary != d->canary1)
994 		wrterror("chunk info corrupted", NULL);
995 
996 	/* Find the chunk number on the page */
997 	chunknum = ((uintptr_t)ptr & MALLOC_PAGEMASK) >> info->shift;
998 
999 	if ((uintptr_t)ptr & ((1U << (info->shift)) - 1)) {
1000 		wrterror("modified chunk-pointer", ptr);
1001 		return -1;
1002 	}
1003 	if (info->bits[chunknum / MALLOC_BITS] &
1004 	    (1U << (chunknum % MALLOC_BITS))) {
1005 		wrterror("chunk is already free", ptr);
1006 		return -1;
1007 	}
1008 	return chunknum;
1009 }
1010 
1011 /*
1012  * Free a chunk, and possibly the page it's on, if the page becomes empty.
1013  */
1014 static void
1015 free_bytes(struct dir_info *d, struct region_info *r, void *ptr)
1016 {
1017 	struct chunk_head *mp;
1018 	struct chunk_info *info;
1019 	uint32_t chunknum;
1020 	int listnum;
1021 
1022 	info = (struct chunk_info *)r->size;
1023 	if ((chunknum = find_chunknum(d, r, ptr)) == -1)
1024 		return;
1025 
1026 	info->bits[chunknum / MALLOC_BITS] |= 1U << (chunknum % MALLOC_BITS);
1027 	info->free++;
1028 
1029 	if (info->free == 1) {
1030 		/* Page became non-full */
1031 		listnum = getrbyte(d) % MALLOC_CHUNK_LISTS;
1032 		if (info->size != 0)
1033 			mp = &d->chunk_dir[info->shift][listnum];
1034 		else
1035 			mp = &d->chunk_dir[0][listnum];
1036 
1037 		LIST_INSERT_HEAD(mp, info, entries);
1038 		return;
1039 	}
1040 
1041 	if (info->free != info->total)
1042 		return;
1043 
1044 	LIST_REMOVE(info, entries);
1045 
1046 	if (info->size == 0 && !mopts.malloc_freeunmap)
1047 		mprotect(info->page, MALLOC_PAGESIZE, PROT_READ | PROT_WRITE);
1048 	unmap(d, info->page, MALLOC_PAGESIZE);
1049 
1050 	delete(d, r);
1051 	if (info->size != 0)
1052 		mp = &d->chunk_info_list[info->shift];
1053 	else
1054 		mp = &d->chunk_info_list[0];
1055 	LIST_INSERT_HEAD(mp, info, entries);
1056 }
1057 
1058 
1059 
1060 static void *
1061 omalloc(size_t sz, int zero_fill, void *f)
1062 {
1063 	struct dir_info *pool = getpool();
1064 	void *p;
1065 	size_t psz;
1066 
1067 	if (sz > MALLOC_MAXCHUNK) {
1068 		if (sz >= SIZE_MAX - mopts.malloc_guard - MALLOC_PAGESIZE) {
1069 			errno = ENOMEM;
1070 			return NULL;
1071 		}
1072 		sz += mopts.malloc_guard;
1073 		psz = PAGEROUND(sz);
1074 		p = map(pool, psz, zero_fill);
1075 		if (p == MAP_FAILED) {
1076 			errno = ENOMEM;
1077 			return NULL;
1078 		}
1079 		if (insert(pool, p, sz, f)) {
1080 			unmap(pool, p, psz);
1081 			errno = ENOMEM;
1082 			return NULL;
1083 		}
1084 		if (mopts.malloc_guard) {
1085 			if (mprotect((char *)p + psz - mopts.malloc_guard,
1086 			    mopts.malloc_guard, PROT_NONE))
1087 				wrterror("mprotect", NULL);
1088 			STATS_ADD(pool->malloc_guarded, mopts.malloc_guard);
1089 		}
1090 
1091 		if (mopts.malloc_move &&
1092 		    sz - mopts.malloc_guard < MALLOC_PAGESIZE -
1093 		    MALLOC_LEEWAY) {
1094 			/* fill whole allocation */
1095 			if (mopts.malloc_junk == 2)
1096 				memset(p, SOME_JUNK, psz - mopts.malloc_guard);
1097 			/* shift towards the end */
1098 			p = ((char *)p) + ((MALLOC_PAGESIZE - MALLOC_LEEWAY -
1099 			    (sz - mopts.malloc_guard)) & ~(MALLOC_MINSIZE-1));
1100 			/* fill zeros if needed and overwritten above */
1101 			if (zero_fill && mopts.malloc_junk == 2)
1102 				memset(p, 0, sz - mopts.malloc_guard);
1103 		} else {
1104 			if (mopts.malloc_junk == 2) {
1105 				if (zero_fill)
1106 					memset((char *)p + sz - mopts.malloc_guard,
1107 					    SOME_JUNK, psz - sz);
1108 				else
1109 					memset(p, SOME_JUNK,
1110 					    psz - mopts.malloc_guard);
1111 			}
1112 		}
1113 
1114 	} else {
1115 		/* takes care of SOME_JUNK */
1116 		p = malloc_bytes(pool, sz, f);
1117 		if (zero_fill && p != NULL && sz > 0)
1118 			memset(p, 0, sz);
1119 	}
1120 
1121 	return p;
1122 }
1123 
1124 /*
1125  * Common function for handling recursion.  Only
1126  * print the error message once, to avoid making the problem
1127  * potentially worse.
1128  */
1129 static void
1130 malloc_recurse(void)
1131 {
1132 	static int noprint;
1133 
1134 	if (noprint == 0) {
1135 		noprint = 1;
1136 		wrterror("recursive call", NULL);
1137 	}
1138 	malloc_active--;
1139 	_MALLOC_UNLOCK();
1140 	errno = EDEADLK;
1141 }
1142 
1143 static int
1144 malloc_init(void)
1145 {
1146 	if (omalloc_init(&mopts.malloc_pool)) {
1147 		_MALLOC_UNLOCK();
1148 		if (mopts.malloc_xmalloc)
1149 			wrterror("out of memory", NULL);
1150 		errno = ENOMEM;
1151 		return -1;
1152 	}
1153 	return 0;
1154 }
1155 
1156 void *
1157 malloc(size_t size)
1158 {
1159 	void *r;
1160 	int saved_errno = errno;
1161 
1162 	_MALLOC_LOCK();
1163 	malloc_func = "malloc():";
1164 	if (getpool() == NULL) {
1165 		if (malloc_init() != 0)
1166 			return NULL;
1167 	}
1168 
1169 	if (malloc_active++) {
1170 		malloc_recurse();
1171 		return NULL;
1172 	}
1173 	r = omalloc(size, 0, CALLER);
1174 	malloc_active--;
1175 	_MALLOC_UNLOCK();
1176 	if (r == NULL && mopts.malloc_xmalloc) {
1177 		wrterror("out of memory", NULL);
1178 		errno = ENOMEM;
1179 	}
1180 	if (r != NULL)
1181 		errno = saved_errno;
1182 	return r;
1183 }
1184 
1185 static void
1186 ofree(void *p)
1187 {
1188 	struct dir_info *pool = getpool();
1189 	struct region_info *r;
1190 	size_t sz;
1191 
1192 	r = find(pool, p);
1193 	if (r == NULL) {
1194 		wrterror("bogus pointer (double free?)", p);
1195 		return;
1196 	}
1197 	REALSIZE(sz, r);
1198 	if (sz > MALLOC_MAXCHUNK) {
1199 		if (sz - mopts.malloc_guard >= MALLOC_PAGESIZE -
1200 		    MALLOC_LEEWAY) {
1201 			if (r->p != p) {
1202 				wrterror("bogus pointer", p);
1203 				return;
1204 			}
1205 		} else {
1206 #if notyetbecause_of_realloc
1207 			/* shifted towards the end */
1208 			if (p != ((char *)r->p) + ((MALLOC_PAGESIZE -
1209 			    MALLOC_MINSIZE - sz - mopts.malloc_guard) &
1210 			    ~(MALLOC_MINSIZE-1))) {
1211 			}
1212 #endif
1213 			p = r->p;
1214 		}
1215 		if (mopts.malloc_guard) {
1216 			if (sz < mopts.malloc_guard)
1217 				wrterror("guard size", NULL);
1218 			if (!mopts.malloc_freeunmap) {
1219 				if (mprotect((char *)p + PAGEROUND(sz) -
1220 				    mopts.malloc_guard, mopts.malloc_guard,
1221 				    PROT_READ | PROT_WRITE))
1222 					wrterror("mprotect", NULL);
1223 			}
1224 			STATS_SUB(pool->malloc_guarded, mopts.malloc_guard);
1225 		}
1226 		if (mopts.malloc_junk && !mopts.malloc_freeunmap) {
1227 			size_t amt = mopts.malloc_junk == 1 ? MALLOC_MAXCHUNK :
1228 			    PAGEROUND(sz) - mopts.malloc_guard;
1229 			memset(p, SOME_FREEJUNK, amt);
1230 		}
1231 		unmap(pool, p, PAGEROUND(sz));
1232 		delete(pool, r);
1233 	} else {
1234 		void *tmp;
1235 		int i;
1236 
1237 		if (mopts.malloc_junk && sz > 0)
1238 			memset(p, SOME_FREEJUNK, sz);
1239 		if (!mopts.malloc_freenow) {
1240 			if (find_chunknum(pool, r, p) == -1)
1241 				return;
1242 			i = getrbyte(pool) & MALLOC_DELAYED_CHUNK_MASK;
1243 			tmp = p;
1244 			p = pool->delayed_chunks[i];
1245 			if (tmp == p) {
1246 				wrterror("double free", p);
1247 				return;
1248 			}
1249 			pool->delayed_chunks[i] = tmp;
1250 		}
1251 		if (p != NULL) {
1252 			r = find(pool, p);
1253 			if (r == NULL) {
1254 				wrterror("bogus pointer (double free?)", p);
1255 				return;
1256 			}
1257 			free_bytes(pool, r, p);
1258 		}
1259 	}
1260 }
1261 
1262 void
1263 free(void *ptr)
1264 {
1265 	int saved_errno = errno;
1266 
1267 	/* This is legal. */
1268 	if (ptr == NULL)
1269 		return;
1270 
1271 	_MALLOC_LOCK();
1272 	malloc_func = "free():";
1273 	if (getpool() == NULL) {
1274 		_MALLOC_UNLOCK();
1275 		wrterror("free() called before allocation", NULL);
1276 		return;
1277 	}
1278 	if (malloc_active++) {
1279 		malloc_recurse();
1280 		return;
1281 	}
1282 	ofree(ptr);
1283 	malloc_active--;
1284 	_MALLOC_UNLOCK();
1285 	errno = saved_errno;
1286 }
1287 
1288 
1289 static void *
1290 orealloc(void *p, size_t newsz, void *f)
1291 {
1292 	struct dir_info *pool = getpool();
1293 	struct region_info *r;
1294 	size_t oldsz, goldsz, gnewsz;
1295 	void *q;
1296 
1297 	if (p == NULL)
1298 		return omalloc(newsz, 0, f);
1299 
1300 	r = find(pool, p);
1301 	if (r == NULL) {
1302 		wrterror("bogus pointer (double free?)", p);
1303 		return NULL;
1304 	}
1305 	if (newsz >= SIZE_MAX - mopts.malloc_guard - MALLOC_PAGESIZE) {
1306 		errno = ENOMEM;
1307 		return NULL;
1308 	}
1309 
1310 	REALSIZE(oldsz, r);
1311 	goldsz = oldsz;
1312 	if (oldsz > MALLOC_MAXCHUNK) {
1313 		if (oldsz < mopts.malloc_guard)
1314 			wrterror("guard size", NULL);
1315 		oldsz -= mopts.malloc_guard;
1316 	}
1317 
1318 	gnewsz = newsz;
1319 	if (gnewsz > MALLOC_MAXCHUNK)
1320 		gnewsz += mopts.malloc_guard;
1321 
1322 	if (newsz > MALLOC_MAXCHUNK && oldsz > MALLOC_MAXCHUNK && p == r->p &&
1323 	    !mopts.malloc_realloc) {
1324 		size_t roldsz = PAGEROUND(goldsz);
1325 		size_t rnewsz = PAGEROUND(gnewsz);
1326 
1327 		if (rnewsz > roldsz) {
1328 			if (!mopts.malloc_guard) {
1329 				void *hint = (char *)p + roldsz;
1330 				size_t needed = rnewsz - roldsz;
1331 
1332 				STATS_INC(pool->cheap_realloc_tries);
1333 				zapcacheregion(pool, hint, needed);
1334 				q = MQUERY(hint, needed);
1335 				if (q == hint)
1336 					q = MMAPA(hint, needed);
1337 				else
1338 					q = MAP_FAILED;
1339 				if (q == hint) {
1340 					STATS_ADD(pool->malloc_used, needed);
1341 					if (mopts.malloc_junk == 2)
1342 						memset(q, SOME_JUNK, needed);
1343 					r->size = newsz;
1344 					STATS_SETF(r, f);
1345 					STATS_INC(pool->cheap_reallocs);
1346 					return p;
1347 				} else if (q != MAP_FAILED) {
1348 					if (munmap(q, needed))
1349 						wrterror("munmap", q);
1350 				}
1351 			}
1352 		} else if (rnewsz < roldsz) {
1353 			if (mopts.malloc_guard) {
1354 				if (mprotect((char *)p + roldsz -
1355 				    mopts.malloc_guard, mopts.malloc_guard,
1356 				    PROT_READ | PROT_WRITE))
1357 					wrterror("mprotect", NULL);
1358 				if (mprotect((char *)p + rnewsz -
1359 				    mopts.malloc_guard, mopts.malloc_guard,
1360 				    PROT_NONE))
1361 					wrterror("mprotect", NULL);
1362 			}
1363 			unmap(pool, (char *)p + rnewsz, roldsz - rnewsz);
1364 			r->size = gnewsz;
1365 			STATS_SETF(r, f);
1366 			return p;
1367 		} else {
1368 			if (newsz > oldsz && mopts.malloc_junk == 2)
1369 				memset((char *)p + newsz, SOME_JUNK,
1370 				    rnewsz - mopts.malloc_guard - newsz);
1371 			r->size = gnewsz;
1372 			STATS_SETF(r, f);
1373 			return p;
1374 		}
1375 	}
1376 	if (newsz <= oldsz && newsz > oldsz / 2 && !mopts.malloc_realloc) {
1377 		if (mopts.malloc_junk == 2 && newsz > 0)
1378 			memset((char *)p + newsz, SOME_JUNK, oldsz - newsz);
1379 		STATS_SETF(r, f);
1380 		return p;
1381 	} else if (newsz != oldsz || mopts.malloc_realloc) {
1382 		q = omalloc(newsz, 0, f);
1383 		if (q == NULL)
1384 			return NULL;
1385 		if (newsz != 0 && oldsz != 0)
1386 			memcpy(q, p, oldsz < newsz ? oldsz : newsz);
1387 		ofree(p);
1388 		return q;
1389 	} else {
1390 		STATS_SETF(r, f);
1391 		return p;
1392 	}
1393 }
1394 
1395 void *
1396 realloc(void *ptr, size_t size)
1397 {
1398 	void *r;
1399 	int saved_errno = errno;
1400 
1401 	_MALLOC_LOCK();
1402 	malloc_func = "realloc():";
1403 	if (getpool() == NULL) {
1404 		if (malloc_init() != 0)
1405 			return NULL;
1406 	}
1407 	if (malloc_active++) {
1408 		malloc_recurse();
1409 		return NULL;
1410 	}
1411 	r = orealloc(ptr, size, CALLER);
1412 
1413 	malloc_active--;
1414 	_MALLOC_UNLOCK();
1415 	if (r == NULL && mopts.malloc_xmalloc) {
1416 		wrterror("out of memory", NULL);
1417 		errno = ENOMEM;
1418 	}
1419 	if (r != NULL)
1420 		errno = saved_errno;
1421 	return r;
1422 }
1423 
1424 
1425 /*
1426  * This is sqrt(SIZE_MAX+1), as s1*s2 <= SIZE_MAX
1427  * if both s1 < MUL_NO_OVERFLOW and s2 < MUL_NO_OVERFLOW
1428  */
1429 #define MUL_NO_OVERFLOW	(1UL << (sizeof(size_t) * 4))
1430 
1431 void *
1432 calloc(size_t nmemb, size_t size)
1433 {
1434 	void *r;
1435 	int saved_errno = errno;
1436 
1437 	_MALLOC_LOCK();
1438 	malloc_func = "calloc():";
1439 	if (getpool() == NULL) {
1440 		if (malloc_init() != 0)
1441 			return NULL;
1442 	}
1443 	if ((nmemb >= MUL_NO_OVERFLOW || size >= MUL_NO_OVERFLOW) &&
1444 	    nmemb > 0 && SIZE_MAX / nmemb < size) {
1445 		_MALLOC_UNLOCK();
1446 		if (mopts.malloc_xmalloc)
1447 			wrterror("out of memory", NULL);
1448 		errno = ENOMEM;
1449 		return NULL;
1450 	}
1451 
1452 	if (malloc_active++) {
1453 		malloc_recurse();
1454 		return NULL;
1455 	}
1456 
1457 	size *= nmemb;
1458 	r = omalloc(size, 1, CALLER);
1459 
1460 	malloc_active--;
1461 	_MALLOC_UNLOCK();
1462 	if (r == NULL && mopts.malloc_xmalloc) {
1463 		wrterror("out of memory", NULL);
1464 		errno = ENOMEM;
1465 	}
1466 	if (r != NULL)
1467 		errno = saved_errno;
1468 	return r;
1469 }
1470 
1471 static void *
1472 mapalign(struct dir_info *d, size_t alignment, size_t sz, int zero_fill)
1473 {
1474 	char *p, *q;
1475 
1476 	if (alignment < MALLOC_PAGESIZE || ((alignment - 1) & alignment) != 0) {
1477 		wrterror("mapalign bad alignment", NULL);
1478 		return MAP_FAILED;
1479 	}
1480 	if (sz != PAGEROUND(sz)) {
1481 		wrterror("mapalign round", NULL);
1482 		return MAP_FAILED;
1483 	}
1484 
1485 	/* Allocate sz + alignment bytes of memory, which must include a
1486 	 * subrange of size bytes that is properly aligned.  Unmap the
1487 	 * other bytes, and then return that subrange.
1488 	 */
1489 
1490 	/* We need sz + alignment to fit into a size_t. */
1491 	if (alignment > SIZE_MAX - sz)
1492 		return MAP_FAILED;
1493 
1494 	p = map(d, sz + alignment, zero_fill);
1495 	if (p == MAP_FAILED)
1496 		return MAP_FAILED;
1497 	q = (char *)(((uintptr_t)p + alignment - 1) & ~(alignment - 1));
1498 	if (q != p) {
1499 		if (munmap(p, q - p))
1500 			wrterror("munmap", p);
1501 	}
1502 	if (munmap(q + sz, alignment - (q - p)))
1503 		wrterror("munmap", q + sz);
1504 	STATS_SUB(d->malloc_used, alignment);
1505 
1506 	return q;
1507 }
1508 
1509 static void *
1510 omemalign(size_t alignment, size_t sz, int zero_fill, void *f)
1511 {
1512 	struct dir_info *pool = getpool();
1513 	size_t psz;
1514 	void *p;
1515 
1516 	if (alignment <= MALLOC_PAGESIZE) {
1517 		/*
1518 		 * max(size, alignment) is enough to assure the requested alignment,
1519 		 * since the allocator always allocates power-of-two blocks.
1520 		 */
1521 		if (sz < alignment)
1522 			sz = alignment;
1523 		return omalloc(sz, zero_fill, f);
1524 	}
1525 
1526 	if (sz >= SIZE_MAX - mopts.malloc_guard - MALLOC_PAGESIZE) {
1527 		errno = ENOMEM;
1528 		return NULL;
1529 	}
1530 
1531 	sz += mopts.malloc_guard;
1532 	psz = PAGEROUND(sz);
1533 
1534 	p = mapalign(pool, alignment, psz, zero_fill);
1535 	if (p == NULL) {
1536 		errno = ENOMEM;
1537 		return NULL;
1538 	}
1539 
1540 	if (insert(pool, p, sz, f)) {
1541 		unmap(pool, p, psz);
1542 		errno = ENOMEM;
1543 		return NULL;
1544 	}
1545 
1546 	if (mopts.malloc_guard) {
1547 		if (mprotect((char *)p + psz - mopts.malloc_guard,
1548 		    mopts.malloc_guard, PROT_NONE))
1549 			wrterror("mprotect", NULL);
1550 		STATS_ADD(pool->malloc_guarded, mopts.malloc_guard);
1551 	}
1552 
1553 	if (mopts.malloc_junk == 2) {
1554 		if (zero_fill)
1555 			memset((char *)p + sz - mopts.malloc_guard,
1556 			    SOME_JUNK, psz - sz);
1557 		else
1558 			memset(p, SOME_JUNK, psz - mopts.malloc_guard);
1559 	}
1560 
1561 	return p;
1562 }
1563 
1564 int
1565 posix_memalign(void **memptr, size_t alignment, size_t size)
1566 {
1567 	int res, saved_errno = errno;
1568 	void *r;
1569 
1570 	/* Make sure that alignment is a large enough power of 2. */
1571 	if (((alignment - 1) & alignment) != 0 || alignment < sizeof(void *))
1572 		return EINVAL;
1573 
1574 	_MALLOC_LOCK();
1575 	malloc_func = "posix_memalign():";
1576 	if (getpool() == NULL) {
1577 		if (malloc_init() != 0)
1578 			goto err;
1579 	}
1580 	if (malloc_active++) {
1581 		malloc_recurse();
1582 		goto err;
1583 	}
1584 	r = omemalign(alignment, size, 0, CALLER);
1585 	malloc_active--;
1586 	_MALLOC_UNLOCK();
1587 	if (r == NULL) {
1588 		if (mopts.malloc_xmalloc) {
1589 			wrterror("out of memory", NULL);
1590 			errno = ENOMEM;
1591 		}
1592 		goto err;
1593 	}
1594 	errno = saved_errno;
1595 	*memptr = r;
1596 	return 0;
1597 
1598 err:
1599 	res = errno;
1600 	errno = saved_errno;
1601 	return res;
1602 }
1603 
1604 #ifdef MALLOC_STATS
1605 
1606 struct malloc_leak {
1607 	void (*f)();
1608 	size_t total_size;
1609 	int count;
1610 };
1611 
1612 struct leaknode {
1613 	RB_ENTRY(leaknode) entry;
1614 	struct malloc_leak d;
1615 };
1616 
1617 static int
1618 leakcmp(struct leaknode *e1, struct leaknode *e2)
1619 {
1620 	return e1->d.f < e2->d.f ? -1 : e1->d.f > e2->d.f;
1621 }
1622 
1623 static RB_HEAD(leaktree, leaknode) leakhead;
1624 RB_GENERATE_STATIC(leaktree, leaknode, entry, leakcmp)
1625 
1626 static void
1627 putleakinfo(void *f, size_t sz, int cnt)
1628 {
1629 	struct leaknode key, *p;
1630 	static struct leaknode *page;
1631 	static int used;
1632 
1633 	if (cnt == 0)
1634 		return;
1635 
1636 	key.d.f = f;
1637 	p = RB_FIND(leaktree, &leakhead, &key);
1638 	if (p == NULL) {
1639 		if (page == NULL ||
1640 		    used >= MALLOC_PAGESIZE / sizeof(struct leaknode)) {
1641 			page = MMAP(MALLOC_PAGESIZE);
1642 			if (page == MAP_FAILED)
1643 				return;
1644 			used = 0;
1645 		}
1646 		p = &page[used++];
1647 		p->d.f = f;
1648 		p->d.total_size = sz * cnt;
1649 		p->d.count = cnt;
1650 		RB_INSERT(leaktree, &leakhead, p);
1651 	} else {
1652 		p->d.total_size += sz * cnt;
1653 		p->d.count += cnt;
1654 	}
1655 }
1656 
1657 static struct malloc_leak *malloc_leaks;
1658 
1659 static void
1660 dump_leaks(int fd)
1661 {
1662 	struct leaknode *p;
1663 	char buf[64];
1664 	int i = 0;
1665 
1666 	snprintf(buf, sizeof(buf), "Leak report\n");
1667 	write(fd, buf, strlen(buf));
1668 	snprintf(buf, sizeof(buf), "                 f     sum      #    avg\n");
1669 	write(fd, buf, strlen(buf));
1670 	/* XXX only one page of summary */
1671 	if (malloc_leaks == NULL)
1672 		malloc_leaks = MMAP(MALLOC_PAGESIZE);
1673 	if (malloc_leaks != MAP_FAILED)
1674 		memset(malloc_leaks, 0, MALLOC_PAGESIZE);
1675 	RB_FOREACH(p, leaktree, &leakhead) {
1676 		snprintf(buf, sizeof(buf), "%18p %7zu %6u %6zu\n", p->d.f,
1677 		    p->d.total_size, p->d.count, p->d.total_size / p->d.count);
1678 		write(fd, buf, strlen(buf));
1679 		if (malloc_leaks == MAP_FAILED ||
1680 		    i >= MALLOC_PAGESIZE / sizeof(struct malloc_leak))
1681 			continue;
1682 		malloc_leaks[i].f = p->d.f;
1683 		malloc_leaks[i].total_size = p->d.total_size;
1684 		malloc_leaks[i].count = p->d.count;
1685 		i++;
1686 	}
1687 }
1688 
1689 static void
1690 dump_chunk(int fd, struct chunk_info *p, void *f, int fromfreelist)
1691 {
1692 	char buf[64];
1693 
1694 	while (p != NULL) {
1695 		snprintf(buf, sizeof(buf), "chunk %18p %18p %4d %d/%d\n",
1696 		    p->page, ((p->bits[0] & 1) ? NULL : f),
1697 		    p->size, p->free, p->total);
1698 		write(fd, buf, strlen(buf));
1699 		if (!fromfreelist) {
1700 			if (p->bits[0] & 1)
1701 				putleakinfo(NULL, p->size, p->total - p->free);
1702 			else {
1703 				putleakinfo(f, p->size, 1);
1704 				putleakinfo(NULL, p->size,
1705 				    p->total - p->free - 1);
1706 			}
1707 			break;
1708 		}
1709 		p = LIST_NEXT(p, entries);
1710 		if (p != NULL) {
1711 			snprintf(buf, sizeof(buf), "        ");
1712 			write(fd, buf, strlen(buf));
1713 		}
1714 	}
1715 }
1716 
1717 static void
1718 dump_free_chunk_info(int fd, struct dir_info *d)
1719 {
1720 	char buf[64];
1721 	int i, j, count;
1722 	struct chunk_info *p;
1723 
1724 	snprintf(buf, sizeof(buf), "Free chunk structs:\n");
1725 	write(fd, buf, strlen(buf));
1726 	for (i = 0; i <= MALLOC_MAXSHIFT; i++) {
1727 		count = 0;
1728 		LIST_FOREACH(p, &d->chunk_info_list[i], entries)
1729 			count++;
1730 		for (j = 0; j < MALLOC_CHUNK_LISTS; j++) {
1731 			p = LIST_FIRST(&d->chunk_dir[i][j]);
1732 			if (p == NULL && count == 0)
1733 				continue;
1734 			snprintf(buf, sizeof(buf), "%2d) %3d ", i, count);
1735 			write(fd, buf, strlen(buf));
1736 			if (p != NULL)
1737 				dump_chunk(fd, p, NULL, 1);
1738 			else
1739 				write(fd, "\n", 1);
1740 		}
1741 	}
1742 
1743 }
1744 
1745 static void
1746 dump_free_page_info(int fd, struct dir_info *d)
1747 {
1748 	char buf[64];
1749 	int i;
1750 
1751 	snprintf(buf, sizeof(buf), "Free pages cached: %zu\n",
1752 	    d->free_regions_size);
1753 	write(fd, buf, strlen(buf));
1754 	for (i = 0; i < mopts.malloc_cache; i++) {
1755 		if (d->free_regions[i].p != NULL) {
1756 			snprintf(buf, sizeof(buf), "%2d) ", i);
1757 			write(fd, buf, strlen(buf));
1758 			snprintf(buf, sizeof(buf), "free at %p: %zu\n",
1759 			    d->free_regions[i].p, d->free_regions[i].size);
1760 			write(fd, buf, strlen(buf));
1761 		}
1762 	}
1763 }
1764 
1765 static void
1766 malloc_dump1(int fd, struct dir_info *d)
1767 {
1768 	char buf[100];
1769 	size_t i, realsize;
1770 
1771 	snprintf(buf, sizeof(buf), "Malloc dir of %s at %p\n", __progname, d);
1772 	write(fd, buf, strlen(buf));
1773 	if (d == NULL)
1774 		return;
1775 	snprintf(buf, sizeof(buf), "Region slots free %zu/%zu\n",
1776 		d->regions_free, d->regions_total);
1777 	write(fd, buf, strlen(buf));
1778 	snprintf(buf, sizeof(buf), "Finds %zu/%zu\n", d->finds,
1779 	    d->find_collisions);
1780 	write(fd, buf, strlen(buf));
1781 	snprintf(buf, sizeof(buf), "Inserts %zu/%zu\n", d->inserts,
1782 	    d->insert_collisions);
1783 	write(fd, buf, strlen(buf));
1784 	snprintf(buf, sizeof(buf), "Deletes %zu/%zu\n", d->deletes,
1785 	    d->delete_moves);
1786 	write(fd, buf, strlen(buf));
1787 	snprintf(buf, sizeof(buf), "Cheap reallocs %zu/%zu\n",
1788 	    d->cheap_reallocs, d->cheap_realloc_tries);
1789 	write(fd, buf, strlen(buf));
1790 	dump_free_chunk_info(fd, d);
1791 	dump_free_page_info(fd, d);
1792 	snprintf(buf, sizeof(buf),
1793 	    "slot)  hash d  type               page                  f size [free/n]\n");
1794 	write(fd, buf, strlen(buf));
1795 	for (i = 0; i < d->regions_total; i++) {
1796 		if (d->r[i].p != NULL) {
1797 			size_t h = hash(d->r[i].p) &
1798 			    (d->regions_total - 1);
1799 			snprintf(buf, sizeof(buf), "%4zx) #%4zx %zd ",
1800 			    i, h, h - i);
1801 			write(fd, buf, strlen(buf));
1802 			REALSIZE(realsize, &d->r[i]);
1803 			if (realsize > MALLOC_MAXCHUNK) {
1804 				putleakinfo(d->r[i].f, realsize, 1);
1805 				snprintf(buf, sizeof(buf),
1806 				    "pages %12p %12p %zu\n", d->r[i].p,
1807 				    d->r[i].f, realsize);
1808 				write(fd, buf, strlen(buf));
1809 			} else
1810 				dump_chunk(fd,
1811 				    (struct chunk_info *)d->r[i].size,
1812 				    d->r[i].f, 0);
1813 		}
1814 	}
1815 	snprintf(buf, sizeof(buf), "In use %zu\n", d->malloc_used);
1816 	write(fd, buf, strlen(buf));
1817 	snprintf(buf, sizeof(buf), "Guarded %zu\n", d->malloc_guarded);
1818 	write(fd, buf, strlen(buf));
1819 	dump_leaks(fd);
1820 	write(fd, "\n", 1);
1821 }
1822 
1823 void
1824 malloc_dump(int fd)
1825 {
1826 	struct dir_info *pool = getpool();
1827 	int i;
1828 	void *p;
1829 	struct region_info *r;
1830 	int saved_errno = errno;
1831 
1832 	for (i = 0; i < MALLOC_DELAYED_CHUNK_MASK + 1; i++) {
1833 		p = pool->delayed_chunks[i];
1834 		if (p == NULL)
1835 			continue;
1836 		r = find(pool, p);
1837 		if (r == NULL)
1838 			wrterror("bogus pointer in malloc_dump", p);
1839 		free_bytes(pool, r, p);
1840 		pool->delayed_chunks[i] = NULL;
1841 	}
1842 	/* XXX leak when run multiple times */
1843 	RB_INIT(&leakhead);
1844 	malloc_dump1(fd, pool);
1845 	errno = saved_errno;
1846 }
1847 
1848 static void
1849 malloc_exit(void)
1850 {
1851 	static const char q[] = "malloc() warning: Couldn't dump stats\n";
1852 	int save_errno = errno, fd;
1853 
1854 	fd = open("malloc.out", O_RDWR|O_APPEND);
1855 	if (fd != -1) {
1856 		malloc_dump(fd);
1857 		close(fd);
1858 	} else
1859 		write(STDERR_FILENO, q, sizeof(q) - 1);
1860 	errno = save_errno;
1861 }
1862 
1863 #endif /* MALLOC_STATS */
1864