1 /* Malloc implementation for multiple threads without lock contention.
2 Copyright (C) 1996, 1997 Free Software Foundation, Inc.
3 This file is part of the GNU C Library.
4 Contributed by Wolfram Gloger <wmglo@dent.med.uni-muenchen.de>
5 and Doug Lea <dl@cs.oswego.edu>, 1996.
7 The GNU C Library is free software; you can redistribute it and/or
8 modify it under the terms of the GNU Library General Public License as
9 published by the Free Software Foundation; either version 2 of the
10 License, or (at your option) any later version.
12 The GNU C Library is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 Library General Public License for more details.
17 You should have received a copy of the GNU Library General Public
18 License along with the GNU C Library; see the file COPYING.LIB. If not,
19 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA. */
22 /* V2.6.4-pt3 Thu Feb 20 1997
24 This work is mainly derived from malloc-2.6.4 by Doug Lea
25 <dl@cs.oswego.edu>, which is available from:
27 ftp://g.oswego.edu/pub/misc/malloc.c
29 Most of the original comments are reproduced in the code below.
31 * Why use this malloc?
33 This is not the fastest, most space-conserving, most portable, or
34 most tunable malloc ever written. However it is among the fastest
35 while also being among the most space-conserving, portable and tunable.
36 Consistent balance across these factors results in a good general-purpose
37 allocator. For a high-level description, see
38 http://g.oswego.edu/dl/html/malloc.html
40 On many systems, the standard malloc implementation is by itself not
41 thread-safe, and therefore wrapped with a single global lock around
42 all malloc-related functions. In some applications, especially with
43 multiple available processors, this can lead to contention problems
44 and bad performance. This malloc version was designed with the goal
45 to avoid waiting for locks as much as possible. Statistics indicate
46 that this goal is achieved in many cases.
48 * Synopsis of public routines
50 (Much fuller descriptions are contained in the program documentation below.)
53 Initialize global configuration. When compiled for multiple threads,
54 this function must be called once before any other function in the
55 package. It is not required otherwise. It is called automatically
56 in the Linux/GNU C libray or when compiling with MALLOC_HOOKS.
58 Return a pointer to a newly allocated chunk of at least n bytes, or null
59 if no space is available.
61 Release the chunk of memory pointed to by p, or no effect if p is null.
62 realloc(Void_t* p, size_t n);
63 Return a pointer to a chunk of size n that contains the same data
64 as does chunk p up to the minimum of (n, p's size) bytes, or null
65 if no space is available. The returned pointer may or may not be
66 the same as p. If p is null, equivalent to malloc. Unless the
67 #define REALLOC_ZERO_BYTES_FREES below is set, realloc with a
68 size argument of zero (re)allocates a minimum-sized chunk.
69 memalign(size_t alignment, size_t n);
70 Return a pointer to a newly allocated chunk of n bytes, aligned
71 in accord with the alignment argument, which must be a power of
74 Equivalent to memalign(pagesize, n), where pagesize is the page
75 size of the system (or as near to this as can be figured out from
76 all the includes/defines below.)
78 Equivalent to valloc(minimum-page-that-holds(n)), that is,
79 round up n to nearest pagesize.
80 calloc(size_t unit, size_t quantity);
81 Returns a pointer to quantity * unit bytes, with all locations
84 Equivalent to free(p).
85 malloc_trim(size_t pad);
86 Release all but pad bytes of freed top-most memory back
87 to the system. Return 1 if successful, else 0.
88 malloc_usable_size(Void_t* p);
89 Report the number usable allocated bytes associated with allocated
90 chunk p. This may or may not report more bytes than were requested,
91 due to alignment and minimum size constraints.
93 Prints brief summary statistics on stderr.
95 Returns (by copy) a struct containing various summary statistics.
96 mallopt(int parameter_number, int parameter_value)
97 Changes one of the tunable parameters described below. Returns
98 1 if successful in changing the parameter, else 0.
103 8 byte alignment is currently hardwired into the design. This
104 seems to suffice for all current machines and C compilers.
106 Assumed pointer representation: 4 or 8 bytes
107 Code for 8-byte pointers is untested by me but has worked
108 reliably by Wolfram Gloger, who contributed most of the
109 changes supporting this.
111 Assumed size_t representation: 4 or 8 bytes
112 Note that size_t is allowed to be 4 bytes even if pointers are 8.
114 Minimum overhead per allocated chunk: 4 or 8 bytes
115 Each malloced chunk has a hidden overhead of 4 bytes holding size
116 and status information.
118 Minimum allocated size: 4-byte ptrs: 16 bytes (including 4 overhead)
119 8-byte ptrs: 24/32 bytes (including, 4/8 overhead)
121 When a chunk is freed, 12 (for 4byte ptrs) or 20 (for 8 byte
122 ptrs but 4 byte size) or 24 (for 8/8) additional bytes are
123 needed; 4 (8) for a trailing size field
124 and 8 (16) bytes for free list pointers. Thus, the minimum
125 allocatable size is 16/24/32 bytes.
127 Even a request for zero bytes (i.e., malloc(0)) returns a
128 pointer to something of the minimum allocatable size.
130 Maximum allocated size: 4-byte size_t: 2^31 - 8 bytes
131 8-byte size_t: 2^63 - 16 bytes
133 It is assumed that (possibly signed) size_t bit values suffice to
134 represent chunk sizes. `Possibly signed' is due to the fact
135 that `size_t' may be defined on a system as either a signed or
136 an unsigned type. To be conservative, values that would appear
137 as negative numbers are avoided.
138 Requests for sizes with a negative sign bit will return a
141 Maximum overhead wastage per allocated chunk: normally 15 bytes
143 Alignment demands, plus the minimum allocatable size restriction
144 make the normal worst-case wastage 15 bytes (i.e., up to 15
145 more bytes will be allocated than were requested in malloc), with
147 1. Because requests for zero bytes allocate non-zero space,
148 the worst case wastage for a request of zero bytes is 24 bytes.
149 2. For requests >= mmap_threshold that are serviced via
150 mmap(), the worst case wastage is 8 bytes plus the remainder
151 from a system page (the minimal mmap unit); typically 4096 bytes.
155 Here are some features that are NOT currently supported
157 * No automated mechanism for fully checking that all accesses
158 to malloced memory stay within their bounds.
159 * No support for compaction.
161 * Synopsis of compile-time options:
163 People have reported using previous versions of this malloc on all
164 versions of Unix, sometimes by tweaking some of the defines
165 below. It has been tested most extensively on Solaris and
166 Linux. People have also reported adapting this malloc for use in
167 stand-alone embedded systems.
169 The implementation is in straight, hand-tuned ANSI C. Among other
170 consequences, it uses a lot of macros. Because of this, to be at
171 all usable, this code should be compiled using an optimizing compiler
172 (for example gcc -O2) that can simplify expressions and control
175 __STD_C (default: derived from C compiler defines)
176 Nonzero if using ANSI-standard C compiler, a C++ compiler, or
177 a C compiler sufficiently close to ANSI to get away with it.
178 MALLOC_DEBUG (default: NOT defined)
179 Define to enable debugging. Adds fairly extensive assertion-based
180 checking to help track down memory errors, but noticeably slows down
182 MALLOC_HOOKS (default: NOT defined)
183 Define to enable support run-time replacement of the allocation
184 functions through user-defined `hooks'.
185 REALLOC_ZERO_BYTES_FREES (default: NOT defined)
186 Define this if you think that realloc(p, 0) should be equivalent
187 to free(p). Otherwise, since malloc returns a unique pointer for
188 malloc(0), so does realloc(p, 0).
189 HAVE_MEMCPY (default: defined)
190 Define if you are not otherwise using ANSI STD C, but still
191 have memcpy and memset in your C library and want to use them.
192 Otherwise, simple internal versions are supplied.
193 USE_MEMCPY (default: 1 if HAVE_MEMCPY is defined, 0 otherwise)
194 Define as 1 if you want the C library versions of memset and
195 memcpy called in realloc and calloc (otherwise macro versions are used).
196 At least on some platforms, the simple macro versions usually
197 outperform libc versions.
198 HAVE_MMAP (default: defined as 1)
199 Define to non-zero to optionally make malloc() use mmap() to
200 allocate very large blocks.
201 HAVE_MREMAP (default: defined as 0 unless Linux libc set)
202 Define to non-zero to optionally make realloc() use mremap() to
203 reallocate very large blocks.
204 malloc_getpagesize (default: derived from system #includes)
205 Either a constant or routine call returning the system page size.
206 HAVE_USR_INCLUDE_MALLOC_H (default: NOT defined)
207 Optionally define if you are on a system with a /usr/include/malloc.h
208 that declares struct mallinfo. It is not at all necessary to
209 define this even if you do, but will ensure consistency.
210 INTERNAL_SIZE_T (default: size_t)
211 Define to a 32-bit type (probably `unsigned int') if you are on a
212 64-bit machine, yet do not want or need to allow malloc requests of
213 greater than 2^31 to be handled. This saves space, especially for
215 _LIBC (default: NOT defined)
216 Defined only when compiled as part of the Linux libc/glibc.
217 Also note that there is some odd internal name-mangling via defines
218 (for example, internally, `malloc' is named `mALLOc') needed
219 when compiling in this case. These look funny but don't otherwise
221 LACKS_UNISTD_H (default: undefined)
222 Define this if your system does not have a <unistd.h>.
223 MORECORE (default: sbrk)
224 The name of the routine to call to obtain more memory from the system.
225 MORECORE_FAILURE (default: -1)
226 The value returned upon failure of MORECORE.
227 MORECORE_CLEARS (default 1)
228 True (1) if the routine mapped to MORECORE zeroes out memory (which
230 DEFAULT_TRIM_THRESHOLD
232 DEFAULT_MMAP_THRESHOLD
234 Default values of tunable parameters (described in detail below)
235 controlling interaction with host system routines (sbrk, mmap, etc).
236 These values may also be changed dynamically via mallopt(). The
237 preset defaults are those that give best performance for typical
240 When the standard debugging hooks are in place, and a pointer is
241 detected as corrupt, do nothing (0), print an error message (1),
249 * Compile-time options for multiple threads:
251 USE_PTHREADS, USE_THR, USE_SPROC
252 Define one of these as 1 to select the thread interface:
253 POSIX threads, Solaris threads or SGI sproc's, respectively.
254 If none of these is defined as non-zero, you get a `normal'
255 malloc implementation which is not thread-safe. Support for
256 multiple threads requires HAVE_MMAP=1. As an exception, when
257 compiling for GNU libc, i.e. when _LIBC is defined, then none of
258 the USE_... symbols have to be defined.
262 When thread support is enabled, additional `heap's are created
263 with mmap calls. These are limited in size; HEAP_MIN_SIZE should
264 be a multiple of the page size, while HEAP_MAX_SIZE must be a power
265 of two for alignment reasons. HEAP_MAX_SIZE should be at least
266 twice as large as the mmap threshold.
268 When this is defined as non-zero, some statistics on mutex locking
279 #if defined (__STDC__)
286 #endif /*__cplusplus*/
299 # include <stddef.h> /* for size_t */
300 # if defined(_LIBC) || defined(MALLOC_HOOKS)
301 # include <stdlib.h> /* for getenv(), abort() */
304 # include <sys/types.h>
307 /* Macros for handling mutexes and thread-specific data. This is
308 included early, because some thread-related header files (such as
309 pthread.h) should be included before any others. */
310 #include "thread-m.h"
316 #include <stdio.h> /* needed for malloc_stats */
327 Because freed chunks may be overwritten with link fields, this
328 malloc will often die when freed memory is overwritten by user
329 programs. This can be very effective (albeit in an annoying way)
330 in helping track down dangling pointers.
332 If you compile with -DMALLOC_DEBUG, a number of assertion checks are
333 enabled that will catch more memory errors. You probably won't be
334 able to make much sense of the actual assertion errors, but they
335 should help you locate incorrectly overwritten memory. The
336 checking is fairly extensive, and will slow down execution
337 noticeably. Calling malloc_stats or mallinfo with MALLOC_DEBUG set will
338 attempt to check every non-mmapped allocated and free chunk in the
339 course of computing the summaries. (By nature, mmapped regions
340 cannot be checked very much automatically.)
342 Setting MALLOC_DEBUG may also be helpful if you are trying to modify
343 this code. The assertions in the check routines spell out in more
344 detail the assumptions and invariants underlying the algorithms.
351 #define assert(x) ((void)0)
356 INTERNAL_SIZE_T is the word-size used for internal bookkeeping
357 of chunk sizes. On a 64-bit machine, you can reduce malloc
358 overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int'
359 at the expense of not being able to handle requests greater than
360 2^31. This limitation is hardly ever a concern; you are encouraged
361 to set this. However, the default version is the same as size_t.
364 #ifndef INTERNAL_SIZE_T
365 #define INTERNAL_SIZE_T size_t
369 REALLOC_ZERO_BYTES_FREES should be set if a call to
370 realloc with zero bytes should be the same as a call to free.
371 Some people think it should. Otherwise, since this malloc
372 returns a unique pointer for malloc(0), so does realloc(p, 0).
376 /* #define REALLOC_ZERO_BYTES_FREES */
380 HAVE_MEMCPY should be defined if you are not otherwise using
381 ANSI STD C, but still have memcpy and memset in your C library
382 and want to use them in calloc and realloc. Otherwise simple
383 macro versions are defined here.
385 USE_MEMCPY should be defined as 1 if you actually want to
386 have memset and memcpy called. People report that the macro
387 versions are often enough faster than libc versions on many
388 systems that it is better to use them.
392 #define HAVE_MEMCPY 1
402 #if (__STD_C || defined(HAVE_MEMCPY))
405 void* memset(void*, int, size_t);
406 void* memcpy(void*, const void*, size_t);
415 /* The following macros are only invoked with (2n+1)-multiples of
416 INTERNAL_SIZE_T units, with a positive integer n. This is exploited
417 for fast inline execution when n is small. */
419 #define MALLOC_ZERO(charp, nbytes) \
421 INTERNAL_SIZE_T mzsz = (nbytes); \
422 if(mzsz <= 9*sizeof(mzsz)) { \
423 INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp); \
424 if(mzsz >= 5*sizeof(mzsz)) { *mz++ = 0; \
426 if(mzsz >= 7*sizeof(mzsz)) { *mz++ = 0; \
428 if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0; \
433 } else memset((charp), 0, mzsz); \
436 #define MALLOC_COPY(dest,src,nbytes) \
438 INTERNAL_SIZE_T mcsz = (nbytes); \
439 if(mcsz <= 9*sizeof(mcsz)) { \
440 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src); \
441 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest); \
442 if(mcsz >= 5*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
443 *mcdst++ = *mcsrc++; \
444 if(mcsz >= 7*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
445 *mcdst++ = *mcsrc++; \
446 if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
447 *mcdst++ = *mcsrc++; }}} \
448 *mcdst++ = *mcsrc++; \
449 *mcdst++ = *mcsrc++; \
451 } else memcpy(dest, src, mcsz); \
454 #else /* !USE_MEMCPY */
456 /* Use Duff's device for good zeroing/copying performance. */
458 #define MALLOC_ZERO(charp, nbytes) \
460 INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp); \
461 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
462 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
464 case 0: for(;;) { *mzp++ = 0; \
465 case 7: *mzp++ = 0; \
466 case 6: *mzp++ = 0; \
467 case 5: *mzp++ = 0; \
468 case 4: *mzp++ = 0; \
469 case 3: *mzp++ = 0; \
470 case 2: *mzp++ = 0; \
471 case 1: *mzp++ = 0; if(mcn <= 0) break; mcn--; } \
475 #define MALLOC_COPY(dest,src,nbytes) \
477 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src; \
478 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest; \
479 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
480 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
482 case 0: for(;;) { *mcdst++ = *mcsrc++; \
483 case 7: *mcdst++ = *mcsrc++; \
484 case 6: *mcdst++ = *mcsrc++; \
485 case 5: *mcdst++ = *mcsrc++; \
486 case 4: *mcdst++ = *mcsrc++; \
487 case 3: *mcdst++ = *mcsrc++; \
488 case 2: *mcdst++ = *mcsrc++; \
489 case 1: *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; } \
497 Define HAVE_MMAP to optionally make malloc() use mmap() to
498 allocate very large blocks. These will be returned to the
499 operating system immediately after a free().
507 Define HAVE_MREMAP to make realloc() use mremap() to re-allocate
508 large blocks. This is currently only possible on Linux with
509 kernel versions newer than 1.3.77.
513 #define HAVE_MREMAP defined(__linux__)
520 #include <sys/mman.h>
522 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
523 #define MAP_ANONYMOUS MAP_ANON
526 #endif /* HAVE_MMAP */
529 Access to system page size. To the extent possible, this malloc
530 manages memory from the system in page-size units.
532 The following mechanics for getpagesize were adapted from
533 bsd/gnu getpagesize.h
536 #ifndef LACKS_UNISTD_H
540 #ifndef malloc_getpagesize
541 # ifdef _SC_PAGESIZE /* some SVR4 systems omit an underscore */
542 # ifndef _SC_PAGE_SIZE
543 # define _SC_PAGE_SIZE _SC_PAGESIZE
546 # ifdef _SC_PAGE_SIZE
547 # define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
549 # if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
550 extern size_t getpagesize();
551 # define malloc_getpagesize getpagesize()
553 # include <sys/param.h>
554 # ifdef EXEC_PAGESIZE
555 # define malloc_getpagesize EXEC_PAGESIZE
559 # define malloc_getpagesize NBPG
561 # define malloc_getpagesize (NBPG * CLSIZE)
565 # define malloc_getpagesize NBPC
568 # define malloc_getpagesize PAGESIZE
570 # define malloc_getpagesize (4096) /* just guess */
583 This version of malloc supports the standard SVID/XPG mallinfo
584 routine that returns a struct containing the same kind of
585 information you can get from malloc_stats. It should work on
586 any SVID/XPG compliant system that has a /usr/include/malloc.h
587 defining struct mallinfo. (If you'd like to install such a thing
588 yourself, cut out the preliminary declarations as described above
589 and below and save them in a malloc.h file. But there's no
590 compelling reason to bother to do this.)
592 The main declaration needed is the mallinfo struct that is returned
593 (by-copy) by mallinfo(). The SVID/XPG malloinfo struct contains a
594 bunch of fields, most of which are not even meaningful in this
595 version of malloc. Some of these fields are are instead filled by
596 mallinfo() with other numbers that might possibly be of interest.
598 HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
599 /usr/include/malloc.h file that includes a declaration of struct
600 mallinfo. If so, it is included; else an SVID2/XPG2 compliant
601 version is declared below. These must be precisely the same for
606 /* #define HAVE_USR_INCLUDE_MALLOC_H */
608 #if HAVE_USR_INCLUDE_MALLOC_H
609 # include "/usr/include/malloc.h"
614 # include "ptmalloc.h"
620 #ifndef DEFAULT_TRIM_THRESHOLD
621 #define DEFAULT_TRIM_THRESHOLD (128 * 1024)
625 M_TRIM_THRESHOLD is the maximum amount of unused top-most memory
626 to keep before releasing via malloc_trim in free().
628 Automatic trimming is mainly useful in long-lived programs.
629 Because trimming via sbrk can be slow on some systems, and can
630 sometimes be wasteful (in cases where programs immediately
631 afterward allocate more large chunks) the value should be high
632 enough so that your overall system performance would improve by
635 The trim threshold and the mmap control parameters (see below)
636 can be traded off with one another. Trimming and mmapping are
637 two different ways of releasing unused memory back to the
638 system. Between these two, it is often possible to keep
639 system-level demands of a long-lived program down to a bare
640 minimum. For example, in one test suite of sessions measuring
641 the XF86 X server on Linux, using a trim threshold of 128K and a
642 mmap threshold of 192K led to near-minimal long term resource
645 If you are using this malloc in a long-lived program, it should
646 pay to experiment with these values. As a rough guide, you
647 might set to a value close to the average size of a process
648 (program) running on your system. Releasing this much memory
649 would allow such a process to run in memory. Generally, it's
650 worth it to tune for trimming rather than memory mapping when a
651 program undergoes phases where several large chunks are
652 allocated and released in ways that can reuse each other's
653 storage, perhaps mixed with phases where there are no such
654 chunks at all. And in well-behaved long-lived programs,
655 controlling release of large blocks via trimming versus mapping
658 However, in most programs, these parameters serve mainly as
659 protection against the system-level effects of carrying around
660 massive amounts of unneeded memory. Since frequent calls to
661 sbrk, mmap, and munmap otherwise degrade performance, the default
662 parameters are set to relatively high values that serve only as
665 The default trim value is high enough to cause trimming only in
666 fairly extreme (by current memory consumption standards) cases.
667 It must be greater than page size to have any useful effect. To
668 disable trimming completely, you can set to (unsigned long)(-1);
674 #ifndef DEFAULT_TOP_PAD
675 #define DEFAULT_TOP_PAD (0)
679 M_TOP_PAD is the amount of extra `padding' space to allocate or
680 retain whenever sbrk is called. It is used in two ways internally:
682 * When sbrk is called to extend the top of the arena to satisfy
683 a new malloc request, this much padding is added to the sbrk
686 * When malloc_trim is called automatically from free(),
687 it is used as the `pad' argument.
689 In both cases, the actual amount of padding is rounded
690 so that the end of the arena is always a system page boundary.
692 The main reason for using padding is to avoid calling sbrk so
693 often. Having even a small pad greatly reduces the likelihood
694 that nearly every malloc request during program start-up (or
695 after trimming) will invoke sbrk, which needlessly wastes
698 Automatic rounding-up to page-size units is normally sufficient
699 to avoid measurable overhead, so the default is 0. However, in
700 systems where sbrk is relatively slow, it can pay to increase
701 this value, at the expense of carrying around more memory than
707 #ifndef DEFAULT_MMAP_THRESHOLD
708 #define DEFAULT_MMAP_THRESHOLD (128 * 1024)
713 M_MMAP_THRESHOLD is the request size threshold for using mmap()
714 to service a request. Requests of at least this size that cannot
715 be allocated using already-existing space will be serviced via mmap.
716 (If enough normal freed space already exists it is used instead.)
718 Using mmap segregates relatively large chunks of memory so that
719 they can be individually obtained and released from the host
720 system. A request serviced through mmap is never reused by any
721 other request (at least not directly; the system may just so
722 happen to remap successive requests to the same locations).
724 Segregating space in this way has the benefit that mmapped space
725 can ALWAYS be individually released back to the system, which
726 helps keep the system level memory demands of a long-lived
727 program low. Mapped memory can never become `locked' between
728 other chunks, as can happen with normally allocated chunks, which
729 menas that even trimming via malloc_trim would not release them.
731 However, it has the disadvantages that:
733 1. The space cannot be reclaimed, consolidated, and then
734 used to service later requests, as happens with normal chunks.
735 2. It can lead to more wastage because of mmap page alignment
737 3. It causes malloc performance to be more dependent on host
738 system memory management support routines which may vary in
739 implementation quality and may impose arbitrary
740 limitations. Generally, servicing a request via normal
741 malloc steps is faster than going through a system's mmap.
743 All together, these considerations should lead you to use mmap
744 only for relatively large requests.
751 #ifndef DEFAULT_MMAP_MAX
753 #define DEFAULT_MMAP_MAX (1024)
755 #define DEFAULT_MMAP_MAX (0)
760 M_MMAP_MAX is the maximum number of requests to simultaneously
761 service using mmap. This parameter exists because:
763 1. Some systems have a limited number of internal tables for
765 2. In most systems, overreliance on mmap can degrade overall
767 3. If a program allocates many large regions, it is probably
768 better off using normal sbrk-based allocation routines that
769 can reclaim and reallocate normal heap memory. Using a
770 small value allows transition into this mode after the
771 first few allocations.
773 Setting to 0 disables all use of mmap. If HAVE_MMAP is not set,
774 the default value is 0, and attempts to set it to non-zero values
775 in mallopt will fail.
780 #ifndef DEFAULT_CHECK_ACTION
781 #define DEFAULT_CHECK_ACTION 1
784 /* What to do if the standard debugging hooks are in place and a
785 corrupt pointer is detected: do nothing (0), print an error message
786 (1), or call abort() (2). */
790 #define HEAP_MIN_SIZE (32*1024)
791 #define HEAP_MAX_SIZE (1024*1024) /* must be a power of two */
793 /* HEAP_MIN_SIZE and HEAP_MAX_SIZE limit the size of mmap()ed heaps
794 that are dynamically created for multi-threaded programs. The
795 maximum size must be a power of two, for fast determination of
796 which heap belongs to a chunk. It should be much larger than
797 the mmap threshold, so that requests with a size just below that
798 threshold can be fulfilled without creating too many heaps.
804 #define THREAD_STATS 0
807 /* If THREAD_STATS is non-zero, some statistics on mutex locking are
813 Special defines for the Linux/GNU C library.
822 Void_t
* __default_morecore (ptrdiff_t);
823 Void_t
*(*__morecore
)(ptrdiff_t) = __default_morecore
;
827 Void_t
* __default_morecore ();
828 Void_t
*(*__morecore
)() = __default_morecore
;
832 #define MORECORE (*__morecore)
833 #define MORECORE_FAILURE 0
834 #define MORECORE_CLEARS 1
836 #define munmap __munmap
837 #define mremap __mremap
838 #define mprotect __mprotect
839 #undef malloc_getpagesize
840 #define malloc_getpagesize __getpagesize()
845 extern Void_t
* sbrk(ptrdiff_t);
847 extern Void_t
* sbrk();
851 #define MORECORE sbrk
854 #ifndef MORECORE_FAILURE
855 #define MORECORE_FAILURE -1
858 #ifndef MORECORE_CLEARS
859 #define MORECORE_CLEARS 1
866 #define cALLOc __libc_calloc
867 #define fREe __libc_free
868 #define mALLOc __libc_malloc
869 #define mEMALIGn __libc_memalign
870 #define rEALLOc __libc_realloc
871 #define vALLOc __libc_valloc
872 #define pvALLOc __libc_pvalloc
873 #define mALLINFo __libc_mallinfo
874 #define mALLOPt __libc_mallopt
875 #define mALLOC_STATs __malloc_stats
876 #define mALLOC_USABLE_SIZe __malloc_usable_size
877 #define mALLOC_TRIm __malloc_trim
878 #define mALLOC_GET_STATe __malloc_get_state
879 #define mALLOC_SET_STATe __malloc_set_state
883 #define cALLOc calloc
885 #define mALLOc malloc
886 #define mEMALIGn memalign
887 #define rEALLOc realloc
888 #define vALLOc valloc
889 #define pvALLOc pvalloc
890 #define mALLINFo mallinfo
891 #define mALLOPt mallopt
892 #define mALLOC_STATs malloc_stats
893 #define mALLOC_USABLE_SIZe malloc_usable_size
894 #define mALLOC_TRIm malloc_trim
895 #define mALLOC_GET_STATe malloc_get_state
896 #define mALLOC_SET_STATe malloc_set_state
900 /* Public routines */
905 void ptmalloc_init(void);
907 Void_t
* mALLOc(size_t);
909 Void_t
* rEALLOc(Void_t
*, size_t);
910 Void_t
* mEMALIGn(size_t, size_t);
911 Void_t
* vALLOc(size_t);
912 Void_t
* pvALLOc(size_t);
913 Void_t
* cALLOc(size_t, size_t);
915 int mALLOC_TRIm(size_t);
916 size_t mALLOC_USABLE_SIZe(Void_t
*);
917 void mALLOC_STATs(void);
918 int mALLOPt(int, int);
919 struct mallinfo
mALLINFo(void);
920 Void_t
* mALLOC_GET_STATe(void);
921 int mALLOC_SET_STATe(Void_t
*);
926 void ptmalloc_init();
937 size_t mALLOC_USABLE_SIZe();
940 struct mallinfo
mALLINFo();
941 Void_t
* mALLOC_GET_STATe();
942 int mALLOC_SET_STATe();
948 }; /* end of extern "C" */
951 #if !defined(NO_THREADS) && !HAVE_MMAP
952 "Can't have threads support without mmap"
963 INTERNAL_SIZE_T prev_size
; /* Size of previous chunk (if free). */
964 INTERNAL_SIZE_T size
; /* Size in bytes, including overhead. */
965 struct malloc_chunk
* fd
; /* double links -- used only if free. */
966 struct malloc_chunk
* bk
;
969 typedef struct malloc_chunk
* mchunkptr
;
973 malloc_chunk details:
975 (The following includes lightly edited explanations by Colin Plumb.)
977 Chunks of memory are maintained using a `boundary tag' method as
978 described in e.g., Knuth or Standish. (See the paper by Paul
979 Wilson ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a
980 survey of such techniques.) Sizes of free chunks are stored both
981 in the front of each chunk and at the end. This makes
982 consolidating fragmented chunks into bigger chunks very fast. The
983 size fields also hold bits representing whether chunks are free or
986 An allocated chunk looks like this:
989 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
990 | Size of previous chunk, if allocated | |
991 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
992 | Size of chunk, in bytes |P|
993 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
994 | User data starts here... .
996 . (malloc_usable_space() bytes) .
998 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1000 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1003 Where "chunk" is the front of the chunk for the purpose of most of
1004 the malloc code, but "mem" is the pointer that is returned to the
1005 user. "Nextchunk" is the beginning of the next contiguous chunk.
1007 Chunks always begin on even word boundaries, so the mem portion
1008 (which is returned to the user) is also on an even word boundary, and
1009 thus double-word aligned.
1011 Free chunks are stored in circular doubly-linked lists, and look like this:
1013 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1014 | Size of previous chunk |
1015 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1016 `head:' | Size of chunk, in bytes |P|
1017 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1018 | Forward pointer to next chunk in list |
1019 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1020 | Back pointer to previous chunk in list |
1021 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1022 | Unused space (may be 0 bytes long) .
1025 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1026 `foot:' | Size of chunk, in bytes |
1027 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1029 The P (PREV_INUSE) bit, stored in the unused low-order bit of the
1030 chunk size (which is always a multiple of two words), is an in-use
1031 bit for the *previous* chunk. If that bit is *clear*, then the
1032 word before the current chunk size contains the previous chunk
1033 size, and can be used to find the front of the previous chunk.
1034 (The very first chunk allocated always has this bit set,
1035 preventing access to non-existent (or non-owned) memory.)
1037 Note that the `foot' of the current chunk is actually represented
1038 as the prev_size of the NEXT chunk. (This makes it easier to
1039 deal with alignments etc).
1041 The two exceptions to all this are
1043 1. The special chunk `top', which doesn't bother using the
1044 trailing size field since there is no
1045 next contiguous chunk that would have to index off it. (After
1046 initialization, `top' is forced to always exist. If it would
1047 become less than MINSIZE bytes long, it is replenished via
1050 2. Chunks allocated via mmap, which have the second-lowest-order
1051 bit (IS_MMAPPED) set in their size fields. Because they are
1052 never merged or traversed from any other chunk, they have no
1053 foot size or inuse information.
1055 Available chunks are kept in any of several places (all declared below):
1057 * `av': An array of chunks serving as bin headers for consolidated
1058 chunks. Each bin is doubly linked. The bins are approximately
1059 proportionally (log) spaced. There are a lot of these bins
1060 (128). This may look excessive, but works very well in
1061 practice. All procedures maintain the invariant that no
1062 consolidated chunk physically borders another one. Chunks in
1063 bins are kept in size order, with ties going to the
1064 approximately least recently used chunk.
1066 The chunks in each bin are maintained in decreasing sorted order by
1067 size. This is irrelevant for the small bins, which all contain
1068 the same-sized chunks, but facilitates best-fit allocation for
1069 larger chunks. (These lists are just sequential. Keeping them in
1070 order almost never requires enough traversal to warrant using
1071 fancier ordered data structures.) Chunks of the same size are
1072 linked with the most recently freed at the front, and allocations
1073 are taken from the back. This results in LRU or FIFO allocation
1074 order, which tends to give each chunk an equal opportunity to be
1075 consolidated with adjacent freed chunks, resulting in larger free
1076 chunks and less fragmentation.
1078 * `top': The top-most available chunk (i.e., the one bordering the
1079 end of available memory) is treated specially. It is never
1080 included in any bin, is used only if no other chunk is
1081 available, and is released back to the system if it is very
1082 large (see M_TRIM_THRESHOLD).
1084 * `last_remainder': A bin holding only the remainder of the
1085 most recently split (non-top) chunk. This bin is checked
1086 before other non-fitting chunks, so as to provide better
1087 locality for runs of sequentially allocated chunks.
1089 * Implicitly, through the host system's memory mapping tables.
1090 If supported, requests greater than a threshold are usually
1091 serviced via calls to mmap, and then later released via munmap.
1098 The bins are an array of pairs of pointers serving as the
1099 heads of (initially empty) doubly-linked lists of chunks, laid out
1100 in a way so that each pair can be treated as if it were in a
1101 malloc_chunk. (This way, the fd/bk offsets for linking bin heads
1102 and chunks are the same).
1104 Bins for sizes < 512 bytes contain chunks of all the same size, spaced
1105 8 bytes apart. Larger bins are approximately logarithmically
1106 spaced. (See the table below.)
1114 4 bins of size 32768
1115 2 bins of size 262144
1116 1 bin of size what's left
1118 There is actually a little bit of slop in the numbers in bin_index
1119 for the sake of speed. This makes no difference elsewhere.
1121 The special chunks `top' and `last_remainder' get their own bins,
1122 (this is implemented via yet more trickery with the av array),
1123 although `top' is never properly linked to its bin since it is
1124 always handled specially.
1128 #define NAV 128 /* number of bins */
1130 typedef struct malloc_chunk
* mbinptr
;
1132 /* An arena is a configuration of malloc_chunks together with an array
1133 of bins. With multiple threads, it must be locked via a mutex
1134 before changing its data structures. One or more `heaps' are
1135 associated with each arena, except for the main_arena, which is
1136 associated only with the `main heap', i.e. the conventional free
1137 store obtained with calls to MORECORE() (usually sbrk). The `av'
1138 array is never mentioned directly in the code, but instead used via
1139 bin access macros. */
1141 typedef struct _arena
{
1142 mbinptr av
[2*NAV
+ 2];
1143 struct _arena
*next
;
1146 long stat_lock_direct
, stat_lock_loop
, stat_lock_wait
;
1152 /* A heap is a single contiguous memory region holding (coalesceable)
1153 malloc_chunks. It is allocated with mmap() and always starts at an
1154 address aligned to HEAP_MAX_SIZE. Not used unless compiling for
1155 multiple threads. */
1157 typedef struct _heap_info
{
1158 arena
*ar_ptr
; /* Arena for this heap. */
1159 struct _heap_info
*prev
; /* Previous heap. */
1160 size_t size
; /* Current size in bytes. */
1161 size_t pad
; /* Make sure the following data is properly aligned. */
1166 Static functions (forward declarations)
1171 static void chunk_free(arena
*ar_ptr
, mchunkptr p
);
1172 static mchunkptr
chunk_alloc(arena
*ar_ptr
, INTERNAL_SIZE_T size
);
1173 static mchunkptr
chunk_realloc(arena
*ar_ptr
, mchunkptr oldp
,
1174 INTERNAL_SIZE_T oldsize
, INTERNAL_SIZE_T nb
);
1175 static mchunkptr
chunk_align(arena
*ar_ptr
, INTERNAL_SIZE_T nb
,
1177 static int main_trim(size_t pad
);
1179 static int heap_trim(heap_info
*heap
, size_t pad
);
1181 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1182 static Void_t
* malloc_check(size_t sz
);
1183 static void free_check(Void_t
* mem
);
1184 static Void_t
* realloc_check(Void_t
* oldmem
, size_t bytes
);
1185 static Void_t
* memalign_check(size_t alignment
, size_t bytes
);
1186 static Void_t
* malloc_starter(size_t sz
);
1187 static void free_starter(Void_t
* mem
);
1192 static void chunk_free();
1193 static mchunkptr
chunk_alloc();
1194 static mchunkptr
chunk_realloc();
1195 static mchunkptr
chunk_align();
1196 static int main_trim();
1198 static int heap_trim();
1200 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1201 static Void_t
* malloc_check();
1202 static void free_check();
1203 static Void_t
* realloc_check();
1204 static Void_t
* memalign_check();
1205 static Void_t
* malloc_starter();
1206 static void free_starter();
1213 /* sizes, alignments */
1215 #define SIZE_SZ (sizeof(INTERNAL_SIZE_T))
1216 #define MALLOC_ALIGNMENT (SIZE_SZ + SIZE_SZ)
1217 #define MALLOC_ALIGN_MASK (MALLOC_ALIGNMENT - 1)
1218 #define MINSIZE (sizeof(struct malloc_chunk))
1220 /* conversion from malloc headers to user pointers, and back */
1222 #define chunk2mem(p) ((Void_t*)((char*)(p) + 2*SIZE_SZ))
1223 #define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ))
1225 /* pad request bytes into a usable size */
1227 #define request2size(req) \
1228 (((long)((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) < \
1229 (long)(MINSIZE + MALLOC_ALIGN_MASK)) ? MINSIZE : \
1230 (((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) & ~(MALLOC_ALIGN_MASK)))
1232 /* Check if m has acceptable alignment */
1234 #define aligned_OK(m) (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
1240 Physical chunk operations
1244 /* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
1246 #define PREV_INUSE 0x1
1248 /* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
1250 #define IS_MMAPPED 0x2
1252 /* Bits to mask off when extracting size */
1254 #define SIZE_BITS (PREV_INUSE|IS_MMAPPED)
1257 /* Ptr to next physical malloc_chunk. */
1259 #define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~PREV_INUSE) ))
1261 /* Ptr to previous physical malloc_chunk */
1263 #define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) ))
1266 /* Treat space at ptr + offset as a chunk */
1268 #define chunk_at_offset(p, s) ((mchunkptr)(((char*)(p)) + (s)))
1274 Dealing with use bits
1277 /* extract p's inuse bit */
1280 ((((mchunkptr)(((char*)(p))+((p)->size & ~PREV_INUSE)))->size) & PREV_INUSE)
1282 /* extract inuse bit of previous chunk */
1284 #define prev_inuse(p) ((p)->size & PREV_INUSE)
1286 /* check for mmap()'ed chunk */
1288 #define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
1290 /* set/clear chunk as in use without otherwise disturbing */
1292 #define set_inuse(p) \
1293 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size |= PREV_INUSE
1295 #define clear_inuse(p) \
1296 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size &= ~(PREV_INUSE)
1298 /* check/set/clear inuse bits in known places */
1300 #define inuse_bit_at_offset(p, s)\
1301 (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE)
1303 #define set_inuse_bit_at_offset(p, s)\
1304 (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE)
1306 #define clear_inuse_bit_at_offset(p, s)\
1307 (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE))
1313 Dealing with size fields
1316 /* Get size, ignoring use bits */
1318 #define chunksize(p) ((p)->size & ~(SIZE_BITS))
1320 /* Set size at head, without disturbing its use bit */
1322 #define set_head_size(p, s) ((p)->size = (((p)->size & PREV_INUSE) | (s)))
1324 /* Set size/use ignoring previous bits in header */
1326 #define set_head(p, s) ((p)->size = (s))
1328 /* Set size at footer (only when chunk is not in use) */
1330 #define set_foot(p, s) (((mchunkptr)((char*)(p) + (s)))->prev_size = (s))
1338 #define bin_at(a, i) ((mbinptr)((char*)&(((a)->av)[2*(i) + 2]) - 2*SIZE_SZ))
1339 #define init_bin(a, i) ((a)->av[2*i+2] = (a)->av[2*i+3] = bin_at((a), i))
1340 #define next_bin(b) ((mbinptr)((char*)(b) + 2 * sizeof(mbinptr)))
1341 #define prev_bin(b) ((mbinptr)((char*)(b) - 2 * sizeof(mbinptr)))
1344 The first 2 bins are never indexed. The corresponding av cells are instead
1345 used for bookkeeping. This is not to save space, but to simplify
1346 indexing, maintain locality, and avoid some initialization tests.
1349 #define binblocks(a) (bin_at(a,0)->size)/* bitvector of nonempty blocks */
1350 #define top(a) (bin_at(a,0)->fd) /* The topmost chunk */
1351 #define last_remainder(a) (bin_at(a,1)) /* remainder from last split */
1354 Because top initially points to its own bin with initial
1355 zero size, thus forcing extension on the first malloc request,
1356 we avoid having any special code in malloc to check whether
1357 it even exists yet. But we still need to in malloc_extend_top.
1360 #define initial_top(a) ((mchunkptr)bin_at(a, 0))
1364 /* field-extraction macros */
1366 #define first(b) ((b)->fd)
1367 #define last(b) ((b)->bk)
1373 #define bin_index(sz) \
1374 (((((unsigned long)(sz)) >> 9) == 0) ? (((unsigned long)(sz)) >> 3): \
1375 ((((unsigned long)(sz)) >> 9) <= 4) ? 56 + (((unsigned long)(sz)) >> 6): \
1376 ((((unsigned long)(sz)) >> 9) <= 20) ? 91 + (((unsigned long)(sz)) >> 9): \
1377 ((((unsigned long)(sz)) >> 9) <= 84) ? 110 + (((unsigned long)(sz)) >> 12): \
1378 ((((unsigned long)(sz)) >> 9) <= 340) ? 119 + (((unsigned long)(sz)) >> 15): \
1379 ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18): \
1382 bins for chunks < 512 are all spaced 8 bytes apart, and hold
1383 identically sized chunks. This is exploited in malloc.
1386 #define MAX_SMALLBIN 63
1387 #define MAX_SMALLBIN_SIZE 512
1388 #define SMALLBIN_WIDTH 8
1390 #define smallbin_index(sz) (((unsigned long)(sz)) >> 3)
1393 Requests are `small' if both the corresponding and the next bin are small
1396 #define is_small_request(nb) ((nb) < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH)
1401 To help compensate for the large number of bins, a one-level index
1402 structure is used for bin-by-bin searching. `binblocks' is a
1403 one-word bitvector recording whether groups of BINBLOCKWIDTH bins
1404 have any (possibly) non-empty bins, so they can be skipped over
1405 all at once during during traversals. The bits are NOT always
1406 cleared as soon as all bins in a block are empty, but instead only
1407 when all are noticed to be empty during traversal in malloc.
1410 #define BINBLOCKWIDTH 4 /* bins per block */
1412 /* bin<->block macros */
1414 #define idx2binblock(ix) ((unsigned)1 << ((ix) / BINBLOCKWIDTH))
1415 #define mark_binblock(a, ii) (binblocks(a) |= idx2binblock(ii))
1416 #define clear_binblock(a, ii) (binblocks(a) &= ~(idx2binblock(ii)))
1421 /* Static bookkeeping data */
1423 /* Helper macro to initialize bins */
1424 #define IAV(i) bin_at(&main_arena, i), bin_at(&main_arena, i)
1426 static arena main_arena
= {
1429 IAV(0), IAV(1), IAV(2), IAV(3), IAV(4), IAV(5), IAV(6), IAV(7),
1430 IAV(8), IAV(9), IAV(10), IAV(11), IAV(12), IAV(13), IAV(14), IAV(15),
1431 IAV(16), IAV(17), IAV(18), IAV(19), IAV(20), IAV(21), IAV(22), IAV(23),
1432 IAV(24), IAV(25), IAV(26), IAV(27), IAV(28), IAV(29), IAV(30), IAV(31),
1433 IAV(32), IAV(33), IAV(34), IAV(35), IAV(36), IAV(37), IAV(38), IAV(39),
1434 IAV(40), IAV(41), IAV(42), IAV(43), IAV(44), IAV(45), IAV(46), IAV(47),
1435 IAV(48), IAV(49), IAV(50), IAV(51), IAV(52), IAV(53), IAV(54), IAV(55),
1436 IAV(56), IAV(57), IAV(58), IAV(59), IAV(60), IAV(61), IAV(62), IAV(63),
1437 IAV(64), IAV(65), IAV(66), IAV(67), IAV(68), IAV(69), IAV(70), IAV(71),
1438 IAV(72), IAV(73), IAV(74), IAV(75), IAV(76), IAV(77), IAV(78), IAV(79),
1439 IAV(80), IAV(81), IAV(82), IAV(83), IAV(84), IAV(85), IAV(86), IAV(87),
1440 IAV(88), IAV(89), IAV(90), IAV(91), IAV(92), IAV(93), IAV(94), IAV(95),
1441 IAV(96), IAV(97), IAV(98), IAV(99), IAV(100), IAV(101), IAV(102), IAV(103),
1442 IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111),
1443 IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119),
1444 IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127)
1446 &main_arena
, /* next */
1449 0, 0, 0, /* stat_lock_direct, stat_lock_loop, stat_lock_wait */
1451 MUTEX_INITIALIZER
/* mutex */
1456 /* Thread specific data */
1459 static tsd_key_t arena_key
;
1460 static mutex_t list_lock
= MUTEX_INITIALIZER
;
1464 static int stat_n_heaps
= 0;
1465 #define THREAD_STAT(x) x
1467 #define THREAD_STAT(x) do ; while(0)
1470 /* variables holding tunable values */
1472 static unsigned long trim_threshold
= DEFAULT_TRIM_THRESHOLD
;
1473 static unsigned long top_pad
= DEFAULT_TOP_PAD
;
1474 static unsigned int n_mmaps_max
= DEFAULT_MMAP_MAX
;
1475 static unsigned long mmap_threshold
= DEFAULT_MMAP_THRESHOLD
;
1476 static int check_action
= DEFAULT_CHECK_ACTION
;
1478 /* The first value returned from sbrk */
1479 static char* sbrk_base
= (char*)(-1);
1481 /* The maximum memory obtained from system via sbrk */
1482 static unsigned long max_sbrked_mem
= 0;
1484 /* The maximum via either sbrk or mmap (too difficult to track with threads) */
1486 static unsigned long max_total_mem
= 0;
1489 /* The total memory obtained from system via sbrk */
1490 #define sbrked_mem (main_arena.size)
1492 /* Tracking mmaps */
1494 static unsigned int n_mmaps
= 0;
1495 static unsigned int max_n_mmaps
= 0;
1496 static unsigned long mmapped_mem
= 0;
1497 static unsigned long max_mmapped_mem
= 0;
1502 #define weak_variable
1504 /* In GNU libc we want the hook variables to be weak definitions to
1505 avoid a problem with Emacs. */
1506 #define weak_variable weak_function
1509 /* Already initialized? */
1510 int __malloc_initialized
= 0;
1513 /* Initialization routine. */
1516 static void ptmalloc_init
__MALLOC_P ((void)) __attribute__ ((constructor
));
1520 ptmalloc_init
__MALLOC_P((void))
1523 ptmalloc_init
__MALLOC_P((void))
1526 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1527 __malloc_ptr_t (*save_malloc_hook
) __MALLOC_P ((size_t __size
));
1528 void (*save_free_hook
) __MALLOC_P ((__malloc_ptr_t __ptr
));
1532 if(__malloc_initialized
) return;
1533 __malloc_initialized
= 1;
1534 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1535 /* With some threads implementations, creating thread-specific data
1536 or initializing a mutex may call malloc() itself. Provide a
1537 simple starter version (realloc() won't work). */
1538 save_malloc_hook
= __malloc_hook
;
1539 save_free_hook
= __free_hook
;
1540 __malloc_hook
= malloc_starter
;
1541 __free_hook
= free_starter
;
1543 #if defined(_LIBC) && !defined (NO_THREADS)
1544 /* Initialize the pthreads interface. */
1545 if (__pthread_initialize
!= NULL
)
1546 __pthread_initialize();
1549 mutex_init(&main_arena
.mutex
);
1550 mutex_init(&list_lock
);
1551 tsd_key_create(&arena_key
, NULL
);
1552 tsd_setspecific(arena_key
, (Void_t
*)&main_arena
);
1554 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1555 if((s
= getenv("MALLOC_TRIM_THRESHOLD_")))
1556 mALLOPt(M_TRIM_THRESHOLD
, atoi(s
));
1557 if((s
= getenv("MALLOC_TOP_PAD_")))
1558 mALLOPt(M_TOP_PAD
, atoi(s
));
1559 if((s
= getenv("MALLOC_MMAP_THRESHOLD_")))
1560 mALLOPt(M_MMAP_THRESHOLD
, atoi(s
));
1561 if((s
= getenv("MALLOC_MMAP_MAX_")))
1562 mALLOPt(M_MMAP_MAX
, atoi(s
));
1563 s
= getenv("MALLOC_CHECK_");
1564 __malloc_hook
= save_malloc_hook
;
1565 __free_hook
= save_free_hook
;
1567 if(s
[0]) mALLOPt(M_CHECK_ACTION
, (int)(s
[0] - '0'));
1568 __malloc_check_init();
1570 if(__malloc_initialize_hook
!= NULL
)
1571 (*__malloc_initialize_hook
)();
1575 #if defined(_LIBC) || defined(MALLOC_HOOKS)
1577 /* Hooks for debugging versions. The initial hooks just call the
1578 initialization routine, then do the normal work. */
1582 malloc_hook_ini(size_t sz
)
1584 malloc_hook_ini(sz
) size_t sz
;
1587 __malloc_hook
= NULL
;
1588 __realloc_hook
= NULL
;
1589 __memalign_hook
= NULL
;
1596 realloc_hook_ini(Void_t
* ptr
, size_t sz
)
1598 realloc_hook_ini(ptr
, sz
) Void_t
* ptr
; size_t sz
;
1601 __malloc_hook
= NULL
;
1602 __realloc_hook
= NULL
;
1603 __memalign_hook
= NULL
;
1605 return rEALLOc(ptr
, sz
);
1610 memalign_hook_ini(size_t sz
, size_t alignment
)
1612 memalign_hook_ini(sz
, alignment
) size_t sz
; size_t alignment
;
1615 __malloc_hook
= NULL
;
1616 __realloc_hook
= NULL
;
1617 __memalign_hook
= NULL
;
1619 return mEMALIGn(sz
, alignment
);
1622 void weak_variable (*__malloc_initialize_hook
) __MALLOC_P ((void)) = NULL
;
1623 void weak_variable (*__free_hook
) __MALLOC_P ((__malloc_ptr_t __ptr
)) = NULL
;
1624 __malloc_ptr_t
weak_variable (*__malloc_hook
)
1625 __MALLOC_P ((size_t __size
)) = malloc_hook_ini
;
1626 __malloc_ptr_t
weak_variable (*__realloc_hook
)
1627 __MALLOC_P ((__malloc_ptr_t __ptr
, size_t __size
)) = realloc_hook_ini
;
1628 __malloc_ptr_t
weak_variable (*__memalign_hook
)
1629 __MALLOC_P ((size_t __size
, size_t __alignment
)) = memalign_hook_ini
;
1630 void weak_variable (*__after_morecore_hook
) __MALLOC_P ((void)) = NULL
;
1632 /* Activate a standard set of debugging hooks. */
1634 __malloc_check_init()
1636 __malloc_hook
= malloc_check
;
1637 __free_hook
= free_check
;
1638 __realloc_hook
= realloc_check
;
1639 __memalign_hook
= memalign_check
;
1640 if(check_action
== 1)
1641 fprintf(stderr
, "malloc: using debugging hooks\n");
1650 /* Routines dealing with mmap(). */
1654 #ifndef MAP_ANONYMOUS
1656 static int dev_zero_fd
= -1; /* Cached file descriptor for /dev/zero. */
1658 #define MMAP(size, prot) ((dev_zero_fd < 0) ? \
1659 (dev_zero_fd = open("/dev/zero", O_RDWR), \
1660 mmap(0, (size), (prot), MAP_PRIVATE, dev_zero_fd, 0)) : \
1661 mmap(0, (size), (prot), MAP_PRIVATE, dev_zero_fd, 0))
1665 #define MMAP(size, prot) \
1666 (mmap(0, (size), (prot), MAP_PRIVATE|MAP_ANONYMOUS, -1, 0))
1671 static mchunkptr
mmap_chunk(size_t size
)
1673 static mchunkptr
mmap_chunk(size
) size_t size
;
1676 size_t page_mask
= malloc_getpagesize
- 1;
1679 if(n_mmaps
>= n_mmaps_max
) return 0; /* too many regions */
1681 /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because
1682 * there is no following chunk whose prev_size field could be used.
1684 size
= (size
+ SIZE_SZ
+ page_mask
) & ~page_mask
;
1686 p
= (mchunkptr
)MMAP(size
, PROT_READ
|PROT_WRITE
);
1687 if(p
== (mchunkptr
)-1) return 0;
1690 if (n_mmaps
> max_n_mmaps
) max_n_mmaps
= n_mmaps
;
1692 /* We demand that eight bytes into a page must be 8-byte aligned. */
1693 assert(aligned_OK(chunk2mem(p
)));
1695 /* The offset to the start of the mmapped region is stored
1696 * in the prev_size field of the chunk; normally it is zero,
1697 * but that can be changed in memalign().
1700 set_head(p
, size
|IS_MMAPPED
);
1702 mmapped_mem
+= size
;
1703 if ((unsigned long)mmapped_mem
> (unsigned long)max_mmapped_mem
)
1704 max_mmapped_mem
= mmapped_mem
;
1706 if ((unsigned long)(mmapped_mem
+ sbrked_mem
) > (unsigned long)max_total_mem
)
1707 max_total_mem
= mmapped_mem
+ sbrked_mem
;
1713 static void munmap_chunk(mchunkptr p
)
1715 static void munmap_chunk(p
) mchunkptr p
;
1718 INTERNAL_SIZE_T size
= chunksize(p
);
1721 assert (chunk_is_mmapped(p
));
1722 assert(! ((char*)p
>= sbrk_base
&& (char*)p
< sbrk_base
+ sbrked_mem
));
1723 assert((n_mmaps
> 0));
1724 assert(((p
->prev_size
+ size
) & (malloc_getpagesize
-1)) == 0);
1727 mmapped_mem
-= (size
+ p
->prev_size
);
1729 ret
= munmap((char *)p
- p
->prev_size
, size
+ p
->prev_size
);
1731 /* munmap returns non-zero on failure */
1738 static mchunkptr
mremap_chunk(mchunkptr p
, size_t new_size
)
1740 static mchunkptr
mremap_chunk(p
, new_size
) mchunkptr p
; size_t new_size
;
1743 size_t page_mask
= malloc_getpagesize
- 1;
1744 INTERNAL_SIZE_T offset
= p
->prev_size
;
1745 INTERNAL_SIZE_T size
= chunksize(p
);
1748 assert (chunk_is_mmapped(p
));
1749 assert(! ((char*)p
>= sbrk_base
&& (char*)p
< sbrk_base
+ sbrked_mem
));
1750 assert((n_mmaps
> 0));
1751 assert(((size
+ offset
) & (malloc_getpagesize
-1)) == 0);
1753 /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
1754 new_size
= (new_size
+ offset
+ SIZE_SZ
+ page_mask
) & ~page_mask
;
1756 cp
= (char *)mremap((char *)p
- offset
, size
+ offset
, new_size
,
1759 if (cp
== (char *)-1) return 0;
1761 p
= (mchunkptr
)(cp
+ offset
);
1763 assert(aligned_OK(chunk2mem(p
)));
1765 assert((p
->prev_size
== offset
));
1766 set_head(p
, (new_size
- offset
)|IS_MMAPPED
);
1768 mmapped_mem
-= size
+ offset
;
1769 mmapped_mem
+= new_size
;
1770 if ((unsigned long)mmapped_mem
> (unsigned long)max_mmapped_mem
)
1771 max_mmapped_mem
= mmapped_mem
;
1773 if ((unsigned long)(mmapped_mem
+ sbrked_mem
) > (unsigned long)max_total_mem
)
1774 max_total_mem
= mmapped_mem
+ sbrked_mem
;
1779 #endif /* HAVE_MREMAP */
1781 #endif /* HAVE_MMAP */
1785 /* Managing heaps and arenas (for concurrent threads) */
1789 /* Create a new heap. size is automatically rounded up to a multiple
1790 of the page size. */
1794 new_heap(size_t size
)
1796 new_heap(size
) size_t size
;
1799 size_t page_mask
= malloc_getpagesize
- 1;
1804 if(size
< HEAP_MIN_SIZE
)
1805 size
= HEAP_MIN_SIZE
;
1806 size
= (size
+ page_mask
) & ~page_mask
;
1807 if(size
> HEAP_MAX_SIZE
)
1809 p1
= (char *)MMAP(HEAP_MAX_SIZE
<<1, PROT_NONE
);
1810 if(p1
== (char *)-1)
1812 p2
= (char *)(((unsigned long)p1
+ HEAP_MAX_SIZE
) & ~(HEAP_MAX_SIZE
-1));
1815 munmap(p2
+ HEAP_MAX_SIZE
, HEAP_MAX_SIZE
- ul
);
1816 if(mprotect(p2
, size
, PROT_READ
|PROT_WRITE
) != 0) {
1817 munmap(p2
, HEAP_MAX_SIZE
);
1820 h
= (heap_info
*)p2
;
1822 THREAD_STAT(stat_n_heaps
++);
1826 /* Grow or shrink a heap. size is automatically rounded up to a
1827 multiple of the page size if it is positive. */
1831 grow_heap(heap_info
*h
, long diff
)
1833 grow_heap(h
, diff
) heap_info
*h
; long diff
;
1836 size_t page_mask
= malloc_getpagesize
- 1;
1840 diff
= (diff
+ page_mask
) & ~page_mask
;
1841 new_size
= (long)h
->size
+ diff
;
1842 if(new_size
> HEAP_MAX_SIZE
)
1844 if(mprotect((char *)h
+ h
->size
, diff
, PROT_READ
|PROT_WRITE
) != 0)
1847 new_size
= (long)h
->size
+ diff
;
1848 if(new_size
< (long)sizeof(*h
))
1850 if(mprotect((char *)h
+ new_size
, -diff
, PROT_NONE
) != 0)
1857 /* Delete a heap. */
1859 #define delete_heap(heap) munmap((char*)(heap), HEAP_MAX_SIZE)
1861 /* arena_get() acquires an arena and locks the corresponding mutex.
1862 First, try the one last locked successfully by this thread. (This
1863 is the common case and handled with a macro for speed.) Then, loop
1864 once over the circularly linked list of arenas. If no arena is
1865 readily available, create a new one. */
1867 #define arena_get(ptr, size) do { \
1868 Void_t *vptr = NULL; \
1869 ptr = (arena *)tsd_getspecific(arena_key, vptr); \
1870 if(ptr && !mutex_trylock(&ptr->mutex)) { \
1871 THREAD_STAT(++(ptr->stat_lock_direct)); \
1873 ptr = arena_get2(ptr, (size)); \
1878 arena_get2(arena
*a_tsd
, size_t size
)
1880 arena_get2(a_tsd
, size
) arena
*a_tsd
; size_t size
;
1887 unsigned long misalign
;
1890 a
= a_tsd
= &main_arena
;
1894 /* This can only happen while initializing the new arena. */
1895 (void)mutex_lock(&main_arena
.mutex
);
1896 THREAD_STAT(++(main_arena
.stat_lock_wait
));
1901 /* Check the global, circularly linked list for available arenas. */
1903 if(!mutex_trylock(&a
->mutex
)) {
1904 THREAD_STAT(++(a
->stat_lock_loop
));
1905 tsd_setspecific(arena_key
, (Void_t
*)a
);
1909 } while(a
!= a_tsd
);
1911 /* Nothing immediately available, so generate a new arena. */
1912 h
= new_heap(size
+ (sizeof(*h
) + sizeof(*a
) + MALLOC_ALIGNMENT
));
1915 a
= h
->ar_ptr
= (arena
*)(h
+1);
1916 for(i
=0; i
<NAV
; i
++)
1920 tsd_setspecific(arena_key
, (Void_t
*)a
);
1921 mutex_init(&a
->mutex
);
1922 i
= mutex_lock(&a
->mutex
); /* remember result */
1924 /* Set up the top chunk, with proper alignment. */
1925 ptr
= (char *)(a
+ 1);
1926 misalign
= (unsigned long)chunk2mem(ptr
) & MALLOC_ALIGN_MASK
;
1928 ptr
+= MALLOC_ALIGNMENT
- misalign
;
1929 top(a
) = (mchunkptr
)ptr
;
1930 set_head(top(a
), (((char*)h
+ h
->size
) - ptr
) | PREV_INUSE
);
1932 /* Add the new arena to the list. */
1933 (void)mutex_lock(&list_lock
);
1934 a
->next
= main_arena
.next
;
1935 main_arena
.next
= a
;
1936 (void)mutex_unlock(&list_lock
);
1938 if(i
) /* locking failed; keep arena for further attempts later */
1941 THREAD_STAT(++(a
->stat_lock_loop
));
1945 /* find the heap and corresponding arena for a given ptr */
1947 #define heap_for_ptr(ptr) \
1948 ((heap_info *)((unsigned long)(ptr) & ~(HEAP_MAX_SIZE-1)))
1949 #define arena_for_ptr(ptr) \
1950 (((mchunkptr)(ptr) < top(&main_arena) && (char *)(ptr) >= sbrk_base) ? \
1951 &main_arena : heap_for_ptr(ptr)->ar_ptr)
1953 #else /* defined(NO_THREADS) */
1955 /* Without concurrent threads, there is only one arena. */
1957 #define arena_get(ptr, sz) (ptr = &main_arena)
1958 #define arena_for_ptr(ptr) (&main_arena)
1960 #endif /* !defined(NO_THREADS) */
1972 These routines make a number of assertions about the states
1973 of data structures that should be true at all times. If any
1974 are not true, it's very likely that a user program has somehow
1975 trashed memory. (It's also possible that there is a coding error
1976 in malloc. In which case, please report it!)
1980 static void do_check_chunk(arena
*ar_ptr
, mchunkptr p
)
1982 static void do_check_chunk(ar_ptr
, p
) arena
*ar_ptr
; mchunkptr p
;
1985 INTERNAL_SIZE_T sz
= p
->size
& ~PREV_INUSE
;
1987 /* No checkable chunk is mmapped */
1988 assert(!chunk_is_mmapped(p
));
1991 if(ar_ptr
!= &main_arena
) {
1992 heap_info
*heap
= heap_for_ptr(p
);
1993 assert(heap
->ar_ptr
== ar_ptr
);
1994 assert((char *)p
+ sz
<= (char *)heap
+ heap
->size
);
1999 /* Check for legal address ... */
2000 assert((char*)p
>= sbrk_base
);
2001 if (p
!= top(ar_ptr
))
2002 assert((char*)p
+ sz
<= (char*)top(ar_ptr
));
2004 assert((char*)p
+ sz
<= sbrk_base
+ sbrked_mem
);
2010 static void do_check_free_chunk(arena
*ar_ptr
, mchunkptr p
)
2012 static void do_check_free_chunk(ar_ptr
, p
) arena
*ar_ptr
; mchunkptr p
;
2015 INTERNAL_SIZE_T sz
= p
->size
& ~PREV_INUSE
;
2016 mchunkptr next
= chunk_at_offset(p
, sz
);
2018 do_check_chunk(ar_ptr
, p
);
2020 /* Check whether it claims to be free ... */
2023 /* Must have OK size and fields */
2024 assert((long)sz
>= (long)MINSIZE
);
2025 assert((sz
& MALLOC_ALIGN_MASK
) == 0);
2026 assert(aligned_OK(chunk2mem(p
)));
2027 /* ... matching footer field */
2028 assert(next
->prev_size
== sz
);
2029 /* ... and is fully consolidated */
2030 assert(prev_inuse(p
));
2031 assert (next
== top(ar_ptr
) || inuse(next
));
2033 /* ... and has minimally sane links */
2034 assert(p
->fd
->bk
== p
);
2035 assert(p
->bk
->fd
== p
);
2039 static void do_check_inuse_chunk(arena
*ar_ptr
, mchunkptr p
)
2041 static void do_check_inuse_chunk(ar_ptr
, p
) arena
*ar_ptr
; mchunkptr p
;
2044 mchunkptr next
= next_chunk(p
);
2045 do_check_chunk(ar_ptr
, p
);
2047 /* Check whether it claims to be in use ... */
2050 /* ... whether its size is OK (it might be a fencepost) ... */
2051 assert(chunksize(p
) >= MINSIZE
|| next
->size
== (0|PREV_INUSE
));
2053 /* ... and is surrounded by OK chunks.
2054 Since more things can be checked with free chunks than inuse ones,
2055 if an inuse chunk borders them and debug is on, it's worth doing them.
2059 mchunkptr prv
= prev_chunk(p
);
2060 assert(next_chunk(prv
) == p
);
2061 do_check_free_chunk(ar_ptr
, prv
);
2063 if (next
== top(ar_ptr
))
2065 assert(prev_inuse(next
));
2066 assert(chunksize(next
) >= MINSIZE
);
2068 else if (!inuse(next
))
2069 do_check_free_chunk(ar_ptr
, next
);
2074 static void do_check_malloced_chunk(arena
*ar_ptr
,
2075 mchunkptr p
, INTERNAL_SIZE_T s
)
2077 static void do_check_malloced_chunk(ar_ptr
, p
, s
)
2078 arena
*ar_ptr
; mchunkptr p
; INTERNAL_SIZE_T s
;
2081 INTERNAL_SIZE_T sz
= p
->size
& ~PREV_INUSE
;
2084 do_check_inuse_chunk(ar_ptr
, p
);
2086 /* Legal size ... */
2087 assert((long)sz
>= (long)MINSIZE
);
2088 assert((sz
& MALLOC_ALIGN_MASK
) == 0);
2090 assert(room
< (long)MINSIZE
);
2092 /* ... and alignment */
2093 assert(aligned_OK(chunk2mem(p
)));
2096 /* ... and was allocated at front of an available chunk */
2097 assert(prev_inuse(p
));
2102 #define check_free_chunk(A,P) do_check_free_chunk(A,P)
2103 #define check_inuse_chunk(A,P) do_check_inuse_chunk(A,P)
2104 #define check_chunk(A,P) do_check_chunk(A,P)
2105 #define check_malloced_chunk(A,P,N) do_check_malloced_chunk(A,P,N)
2107 #define check_free_chunk(A,P)
2108 #define check_inuse_chunk(A,P)
2109 #define check_chunk(A,P)
2110 #define check_malloced_chunk(A,P,N)
2116 Macro-based internal utilities
2121 Linking chunks in bin lists.
2122 Call these only with variables, not arbitrary expressions, as arguments.
2126 Place chunk p of size s in its bin, in size order,
2127 putting it ahead of others of same size.
2131 #define frontlink(A, P, S, IDX, BK, FD) \
2133 if (S < MAX_SMALLBIN_SIZE) \
2135 IDX = smallbin_index(S); \
2136 mark_binblock(A, IDX); \
2137 BK = bin_at(A, IDX); \
2141 FD->bk = BK->fd = P; \
2145 IDX = bin_index(S); \
2146 BK = bin_at(A, IDX); \
2148 if (FD == BK) mark_binblock(A, IDX); \
2151 while (FD != BK && S < chunksize(FD)) FD = FD->fd; \
2156 FD->bk = BK->fd = P; \
2161 /* take a chunk off a list */
2163 #define unlink(P, BK, FD) \
2171 /* Place p as the last remainder */
2173 #define link_last_remainder(A, P) \
2175 last_remainder(A)->fd = last_remainder(A)->bk = P; \
2176 P->fd = P->bk = last_remainder(A); \
2179 /* Clear the last_remainder bin */
2181 #define clear_last_remainder(A) \
2182 (last_remainder(A)->fd = last_remainder(A)->bk = last_remainder(A))
2189 Extend the top-most chunk by obtaining memory from system.
2190 Main interface to sbrk (but see also malloc_trim).
2194 static void malloc_extend_top(arena
*ar_ptr
, INTERNAL_SIZE_T nb
)
2196 static void malloc_extend_top(ar_ptr
, nb
) arena
*ar_ptr
; INTERNAL_SIZE_T nb
;
2199 unsigned long pagesz
= malloc_getpagesize
;
2200 mchunkptr old_top
= top(ar_ptr
); /* Record state of old top */
2201 INTERNAL_SIZE_T old_top_size
= chunksize(old_top
);
2202 INTERNAL_SIZE_T top_size
; /* new size of top chunk */
2205 if(ar_ptr
== &main_arena
) {
2208 char* brk
; /* return value from sbrk */
2209 INTERNAL_SIZE_T front_misalign
; /* unusable bytes at front of sbrked space */
2210 INTERNAL_SIZE_T correction
; /* bytes for 2nd sbrk call */
2211 char* new_brk
; /* return of 2nd sbrk call */
2212 char* old_end
= (char*)(chunk_at_offset(old_top
, old_top_size
));
2214 /* Pad request with top_pad plus minimal overhead */
2215 INTERNAL_SIZE_T sbrk_size
= nb
+ top_pad
+ MINSIZE
;
2217 /* If not the first time through, round to preserve page boundary */
2218 /* Otherwise, we need to correct to a page size below anyway. */
2219 /* (We also correct below if an intervening foreign sbrk call.) */
2221 if (sbrk_base
!= (char*)(-1))
2222 sbrk_size
= (sbrk_size
+ (pagesz
- 1)) & ~(pagesz
- 1);
2224 brk
= (char*)(MORECORE (sbrk_size
));
2226 /* Fail if sbrk failed or if a foreign sbrk call killed our space */
2227 if (brk
== (char*)(MORECORE_FAILURE
) ||
2228 (brk
< old_end
&& old_top
!= initial_top(&main_arena
)))
2231 /* Call the `morecore' hook if necessary. */
2232 if (__after_morecore_hook
)
2233 (*__after_morecore_hook
) ();
2235 sbrked_mem
+= sbrk_size
;
2237 if (brk
== old_end
) { /* can just add bytes to current top */
2238 top_size
= sbrk_size
+ old_top_size
;
2239 set_head(old_top
, top_size
| PREV_INUSE
);
2240 old_top
= 0; /* don't free below */
2242 if (sbrk_base
== (char*)(-1)) /* First time through. Record base */
2245 /* Someone else called sbrk(). Count those bytes as sbrked_mem. */
2246 sbrked_mem
+= brk
- (char*)old_end
;
2248 /* Guarantee alignment of first new chunk made from this space */
2249 front_misalign
= (unsigned long)chunk2mem(brk
) & MALLOC_ALIGN_MASK
;
2250 if (front_misalign
> 0) {
2251 correction
= (MALLOC_ALIGNMENT
) - front_misalign
;
2256 /* Guarantee the next brk will be at a page boundary */
2257 correction
+= pagesz
- ((unsigned long)(brk
+ sbrk_size
) & (pagesz
- 1));
2259 /* Allocate correction */
2260 new_brk
= (char*)(MORECORE (correction
));
2261 if (new_brk
== (char*)(MORECORE_FAILURE
)) return;
2263 /* Call the `morecore' hook if necessary. */
2264 if (__after_morecore_hook
)
2265 (*__after_morecore_hook
) ();
2267 sbrked_mem
+= correction
;
2269 top(&main_arena
) = (mchunkptr
)brk
;
2270 top_size
= new_brk
- brk
+ correction
;
2271 set_head(top(&main_arena
), top_size
| PREV_INUSE
);
2273 if (old_top
== initial_top(&main_arena
))
2274 old_top
= 0; /* don't free below */
2277 if ((unsigned long)sbrked_mem
> (unsigned long)max_sbrked_mem
)
2278 max_sbrked_mem
= sbrked_mem
;
2280 if ((unsigned long)(mmapped_mem
+ sbrked_mem
) >
2281 (unsigned long)max_total_mem
)
2282 max_total_mem
= mmapped_mem
+ sbrked_mem
;
2286 } else { /* ar_ptr != &main_arena */
2287 heap_info
*old_heap
, *heap
;
2288 size_t old_heap_size
;
2290 if(old_top_size
< MINSIZE
) /* this should never happen */
2293 /* First try to extend the current heap. */
2294 if(MINSIZE
+ nb
<= old_top_size
)
2296 old_heap
= heap_for_ptr(old_top
);
2297 old_heap_size
= old_heap
->size
;
2298 if(grow_heap(old_heap
, MINSIZE
+ nb
- old_top_size
) == 0) {
2299 ar_ptr
->size
+= old_heap
->size
- old_heap_size
;
2300 top_size
= ((char *)old_heap
+ old_heap
->size
) - (char *)old_top
;
2301 set_head(old_top
, top_size
| PREV_INUSE
);
2305 /* A new heap must be created. */
2306 heap
= new_heap(nb
+ top_pad
+ (MINSIZE
+ sizeof(*heap
)));
2309 heap
->ar_ptr
= ar_ptr
;
2310 heap
->prev
= old_heap
;
2311 ar_ptr
->size
+= heap
->size
;
2313 /* Set up the new top, so we can safely use chunk_free() below. */
2314 top(ar_ptr
) = chunk_at_offset(heap
, sizeof(*heap
));
2315 top_size
= heap
->size
- sizeof(*heap
);
2316 set_head(top(ar_ptr
), top_size
| PREV_INUSE
);
2318 #endif /* !defined(NO_THREADS) */
2320 /* We always land on a page boundary */
2321 assert(((unsigned long)((char*)top(ar_ptr
) + top_size
) & (pagesz
-1)) == 0);
2323 /* Setup fencepost and free the old top chunk. */
2325 /* The fencepost takes at least MINSIZE bytes, because it might
2326 become the top chunk again later. Note that a footer is set
2327 up, too, although the chunk is marked in use. */
2328 old_top_size
-= MINSIZE
;
2329 set_head(chunk_at_offset(old_top
, old_top_size
+ 2*SIZE_SZ
), 0|PREV_INUSE
);
2330 if(old_top_size
>= MINSIZE
) {
2331 set_head(chunk_at_offset(old_top
, old_top_size
), (2*SIZE_SZ
)|PREV_INUSE
);
2332 set_foot(chunk_at_offset(old_top
, old_top_size
), (2*SIZE_SZ
));
2333 set_head_size(old_top
, old_top_size
);
2334 chunk_free(ar_ptr
, old_top
);
2336 set_head(old_top
, (old_top_size
+ 2*SIZE_SZ
)|PREV_INUSE
);
2337 set_foot(old_top
, (old_top_size
+ 2*SIZE_SZ
));
2345 /* Main public routines */
2351 The requested size is first converted into a usable form, `nb'.
2352 This currently means to add 4 bytes overhead plus possibly more to
2353 obtain 8-byte alignment and/or to obtain a size of at least
2354 MINSIZE (currently 16, 24, or 32 bytes), the smallest allocatable
2355 size. (All fits are considered `exact' if they are within MINSIZE
2358 From there, the first successful of the following steps is taken:
2360 1. The bin corresponding to the request size is scanned, and if
2361 a chunk of exactly the right size is found, it is taken.
2363 2. The most recently remaindered chunk is used if it is big
2364 enough. This is a form of (roving) first fit, used only in
2365 the absence of exact fits. Runs of consecutive requests use
2366 the remainder of the chunk used for the previous such request
2367 whenever possible. This limited use of a first-fit style
2368 allocation strategy tends to give contiguous chunks
2369 coextensive lifetimes, which improves locality and can reduce
2370 fragmentation in the long run.
2372 3. Other bins are scanned in increasing size order, using a
2373 chunk big enough to fulfill the request, and splitting off
2374 any remainder. This search is strictly by best-fit; i.e.,
2375 the smallest (with ties going to approximately the least
2376 recently used) chunk that fits is selected.
2378 4. If large enough, the chunk bordering the end of memory
2379 (`top') is split off. (This use of `top' is in accord with
2380 the best-fit search rule. In effect, `top' is treated as
2381 larger (and thus less well fitting) than any other available
2382 chunk since it can be extended to be as large as necessary
2383 (up to system limitations).
2385 5. If the request size meets the mmap threshold and the
2386 system supports mmap, and there are few enough currently
2387 allocated mmapped regions, and a call to mmap succeeds,
2388 the request is allocated via direct memory mapping.
2390 6. Otherwise, the top of memory is extended by
2391 obtaining more space from the system (normally using sbrk,
2392 but definable to anything else via the MORECORE macro).
2393 Memory is gathered from the system (in system page-sized
2394 units) in a way that allows chunks obtained across different
2395 sbrk calls to be consolidated, but does not require
2396 contiguous memory. Thus, it should be safe to intersperse
2397 mallocs with other sbrk calls.
2400 All allocations are made from the the `lowest' part of any found
2401 chunk. (The implementation invariant is that prev_inuse is
2402 always true of any allocated chunk; i.e., that each allocated
2403 chunk borders either a previously allocated and still in-use chunk,
2404 or the base of its memory arena.)
2409 Void_t
* mALLOc(size_t bytes
)
2411 Void_t
* mALLOc(bytes
) size_t bytes
;
2415 INTERNAL_SIZE_T nb
; /* padded request size */
2418 #if defined(_LIBC) || defined(MALLOC_HOOKS)
2419 if (__malloc_hook
!= NULL
) {
2422 result
= (*__malloc_hook
)(bytes
);
2427 nb
= request2size(bytes
);
2428 arena_get(ar_ptr
, nb
+ top_pad
);
2431 victim
= chunk_alloc(ar_ptr
, nb
);
2432 (void)mutex_unlock(&ar_ptr
->mutex
);
2433 return victim
? chunk2mem(victim
) : 0;
2438 chunk_alloc(arena
*ar_ptr
, INTERNAL_SIZE_T nb
)
2440 chunk_alloc(ar_ptr
, nb
) arena
*ar_ptr
; INTERNAL_SIZE_T nb
;
2443 mchunkptr victim
; /* inspected/selected chunk */
2444 INTERNAL_SIZE_T victim_size
; /* its size */
2445 int idx
; /* index for bin traversal */
2446 mbinptr bin
; /* associated bin */
2447 mchunkptr remainder
; /* remainder from a split */
2448 long remainder_size
; /* its size */
2449 int remainder_index
; /* its bin index */
2450 unsigned long block
; /* block traverser bit */
2451 int startidx
; /* first bin of a traversed block */
2452 mchunkptr fwd
; /* misc temp for linking */
2453 mchunkptr bck
; /* misc temp for linking */
2454 mbinptr q
; /* misc temp */
2457 /* Check for exact match in a bin */
2459 if (is_small_request(nb
)) /* Faster version for small requests */
2461 idx
= smallbin_index(nb
);
2463 /* No traversal or size check necessary for small bins. */
2465 q
= bin_at(ar_ptr
, idx
);
2468 /* Also scan the next one, since it would have a remainder < MINSIZE */
2476 victim_size
= chunksize(victim
);
2477 unlink(victim
, bck
, fwd
);
2478 set_inuse_bit_at_offset(victim
, victim_size
);
2479 check_malloced_chunk(ar_ptr
, victim
, nb
);
2483 idx
+= 2; /* Set for bin scan below. We've already scanned 2 bins. */
2488 idx
= bin_index(nb
);
2489 bin
= bin_at(ar_ptr
, idx
);
2491 for (victim
= last(bin
); victim
!= bin
; victim
= victim
->bk
)
2493 victim_size
= chunksize(victim
);
2494 remainder_size
= victim_size
- nb
;
2496 if (remainder_size
>= (long)MINSIZE
) /* too big */
2498 --idx
; /* adjust to rescan below after checking last remainder */
2502 else if (remainder_size
>= 0) /* exact fit */
2504 unlink(victim
, bck
, fwd
);
2505 set_inuse_bit_at_offset(victim
, victim_size
);
2506 check_malloced_chunk(ar_ptr
, victim
, nb
);
2515 /* Try to use the last split-off remainder */
2517 if ( (victim
= last_remainder(ar_ptr
)->fd
) != last_remainder(ar_ptr
))
2519 victim_size
= chunksize(victim
);
2520 remainder_size
= victim_size
- nb
;
2522 if (remainder_size
>= (long)MINSIZE
) /* re-split */
2524 remainder
= chunk_at_offset(victim
, nb
);
2525 set_head(victim
, nb
| PREV_INUSE
);
2526 link_last_remainder(ar_ptr
, remainder
);
2527 set_head(remainder
, remainder_size
| PREV_INUSE
);
2528 set_foot(remainder
, remainder_size
);
2529 check_malloced_chunk(ar_ptr
, victim
, nb
);
2533 clear_last_remainder(ar_ptr
);
2535 if (remainder_size
>= 0) /* exhaust */
2537 set_inuse_bit_at_offset(victim
, victim_size
);
2538 check_malloced_chunk(ar_ptr
, victim
, nb
);
2542 /* Else place in bin */
2544 frontlink(ar_ptr
, victim
, victim_size
, remainder_index
, bck
, fwd
);
2548 If there are any possibly nonempty big-enough blocks,
2549 search for best fitting chunk by scanning bins in blockwidth units.
2552 if ( (block
= idx2binblock(idx
)) <= binblocks(ar_ptr
))
2555 /* Get to the first marked block */
2557 if ( (block
& binblocks(ar_ptr
)) == 0)
2559 /* force to an even block boundary */
2560 idx
= (idx
& ~(BINBLOCKWIDTH
- 1)) + BINBLOCKWIDTH
;
2562 while ((block
& binblocks(ar_ptr
)) == 0)
2564 idx
+= BINBLOCKWIDTH
;
2569 /* For each possibly nonempty block ... */
2572 startidx
= idx
; /* (track incomplete blocks) */
2573 q
= bin
= bin_at(ar_ptr
, idx
);
2575 /* For each bin in this block ... */
2578 /* Find and use first big enough chunk ... */
2580 for (victim
= last(bin
); victim
!= bin
; victim
= victim
->bk
)
2582 victim_size
= chunksize(victim
);
2583 remainder_size
= victim_size
- nb
;
2585 if (remainder_size
>= (long)MINSIZE
) /* split */
2587 remainder
= chunk_at_offset(victim
, nb
);
2588 set_head(victim
, nb
| PREV_INUSE
);
2589 unlink(victim
, bck
, fwd
);
2590 link_last_remainder(ar_ptr
, remainder
);
2591 set_head(remainder
, remainder_size
| PREV_INUSE
);
2592 set_foot(remainder
, remainder_size
);
2593 check_malloced_chunk(ar_ptr
, victim
, nb
);
2597 else if (remainder_size
>= 0) /* take */
2599 set_inuse_bit_at_offset(victim
, victim_size
);
2600 unlink(victim
, bck
, fwd
);
2601 check_malloced_chunk(ar_ptr
, victim
, nb
);
2607 bin
= next_bin(bin
);
2609 } while ((++idx
& (BINBLOCKWIDTH
- 1)) != 0);
2611 /* Clear out the block bit. */
2613 do /* Possibly backtrack to try to clear a partial block */
2615 if ((startidx
& (BINBLOCKWIDTH
- 1)) == 0)
2617 binblocks(ar_ptr
) &= ~block
;
2622 } while (first(q
) == q
);
2624 /* Get to the next possibly nonempty block */
2626 if ( (block
<<= 1) <= binblocks(ar_ptr
) && (block
!= 0) )
2628 while ((block
& binblocks(ar_ptr
)) == 0)
2630 idx
+= BINBLOCKWIDTH
;
2640 /* Try to use top chunk */
2642 /* Require that there be a remainder, ensuring top always exists */
2643 if ( (remainder_size
= chunksize(top(ar_ptr
)) - nb
) < (long)MINSIZE
)
2647 /* If big and would otherwise need to extend, try to use mmap instead */
2648 if ((unsigned long)nb
>= (unsigned long)mmap_threshold
&&
2649 (victim
= mmap_chunk(nb
)) != 0)
2654 malloc_extend_top(ar_ptr
, nb
);
2655 if ((remainder_size
= chunksize(top(ar_ptr
)) - nb
) < (long)MINSIZE
)
2656 return 0; /* propagate failure */
2659 victim
= top(ar_ptr
);
2660 set_head(victim
, nb
| PREV_INUSE
);
2661 top(ar_ptr
) = chunk_at_offset(victim
, nb
);
2662 set_head(top(ar_ptr
), remainder_size
| PREV_INUSE
);
2663 check_malloced_chunk(ar_ptr
, victim
, nb
);
2677 1. free(0) has no effect.
2679 2. If the chunk was allocated via mmap, it is released via munmap().
2681 3. If a returned chunk borders the current high end of memory,
2682 it is consolidated into the top, and if the total unused
2683 topmost memory exceeds the trim threshold, malloc_trim is
2686 4. Other chunks are consolidated as they arrive, and
2687 placed in corresponding bins. (This includes the case of
2688 consolidating with the current `last_remainder').
2694 void fREe(Void_t
* mem
)
2696 void fREe(mem
) Void_t
* mem
;
2700 mchunkptr p
; /* chunk corresponding to mem */
2702 #if defined(_LIBC) || defined(MALLOC_HOOKS)
2703 if (__free_hook
!= NULL
) {
2704 (*__free_hook
)(mem
);
2709 if (mem
== 0) /* free(0) has no effect */
2715 if (chunk_is_mmapped(p
)) /* release mmapped memory. */
2722 ar_ptr
= arena_for_ptr(p
);
2724 if(!mutex_trylock(&ar_ptr
->mutex
))
2725 ++(ar_ptr
->stat_lock_direct
);
2727 (void)mutex_lock(&ar_ptr
->mutex
);
2728 ++(ar_ptr
->stat_lock_wait
);
2731 (void)mutex_lock(&ar_ptr
->mutex
);
2733 chunk_free(ar_ptr
, p
);
2734 (void)mutex_unlock(&ar_ptr
->mutex
);
2739 chunk_free(arena
*ar_ptr
, mchunkptr p
)
2741 chunk_free(ar_ptr
, p
) arena
*ar_ptr
; mchunkptr p
;
2744 INTERNAL_SIZE_T hd
= p
->size
; /* its head field */
2745 INTERNAL_SIZE_T sz
; /* its size */
2746 int idx
; /* its bin index */
2747 mchunkptr next
; /* next contiguous chunk */
2748 INTERNAL_SIZE_T nextsz
; /* its size */
2749 INTERNAL_SIZE_T prevsz
; /* size of previous contiguous chunk */
2750 mchunkptr bck
; /* misc temp for linking */
2751 mchunkptr fwd
; /* misc temp for linking */
2752 int islr
; /* track whether merging with last_remainder */
2754 check_inuse_chunk(ar_ptr
, p
);
2756 sz
= hd
& ~PREV_INUSE
;
2757 next
= chunk_at_offset(p
, sz
);
2758 nextsz
= chunksize(next
);
2760 if (next
== top(ar_ptr
)) /* merge with top */
2764 if (!(hd
& PREV_INUSE
)) /* consolidate backward */
2766 prevsz
= p
->prev_size
;
2767 p
= chunk_at_offset(p
, -prevsz
);
2769 unlink(p
, bck
, fwd
);
2772 set_head(p
, sz
| PREV_INUSE
);
2776 if(ar_ptr
== &main_arena
) {
2778 if ((unsigned long)(sz
) >= (unsigned long)trim_threshold
)
2782 heap_info
*heap
= heap_for_ptr(p
);
2784 assert(heap
->ar_ptr
== ar_ptr
);
2786 /* Try to get rid of completely empty heaps, if possible. */
2787 if((unsigned long)(sz
) >= (unsigned long)trim_threshold
||
2788 p
== chunk_at_offset(heap
, sizeof(*heap
)))
2789 heap_trim(heap
, top_pad
);
2795 set_head(next
, nextsz
); /* clear inuse bit */
2799 if (!(hd
& PREV_INUSE
)) /* consolidate backward */
2801 prevsz
= p
->prev_size
;
2802 p
= chunk_at_offset(p
, -prevsz
);
2805 if (p
->fd
== last_remainder(ar_ptr
)) /* keep as last_remainder */
2808 unlink(p
, bck
, fwd
);
2811 if (!(inuse_bit_at_offset(next
, nextsz
))) /* consolidate forward */
2815 if (!islr
&& next
->fd
== last_remainder(ar_ptr
))
2816 /* re-insert last_remainder */
2819 link_last_remainder(ar_ptr
, p
);
2822 unlink(next
, bck
, fwd
);
2825 set_head(p
, sz
| PREV_INUSE
);
2828 frontlink(ar_ptr
, p
, sz
, idx
, bck
, fwd
);
2839 Chunks that were obtained via mmap cannot be extended or shrunk
2840 unless HAVE_MREMAP is defined, in which case mremap is used.
2841 Otherwise, if their reallocation is for additional space, they are
2842 copied. If for less, they are just left alone.
2844 Otherwise, if the reallocation is for additional space, and the
2845 chunk can be extended, it is, else a malloc-copy-free sequence is
2846 taken. There are several different ways that a chunk could be
2847 extended. All are tried:
2849 * Extending forward into following adjacent free chunk.
2850 * Shifting backwards, joining preceding adjacent space
2851 * Both shifting backwards and extending forward.
2852 * Extending into newly sbrked space
2854 Unless the #define REALLOC_ZERO_BYTES_FREES is set, realloc with a
2855 size argument of zero (re)allocates a minimum-sized chunk.
2857 If the reallocation is for less space, and the new request is for
2858 a `small' (<512 bytes) size, then the newly unused space is lopped
2861 The old unix realloc convention of allowing the last-free'd chunk
2862 to be used as an argument to realloc is no longer supported.
2863 I don't know of any programs still relying on this feature,
2864 and allowing it would also allow too many other incorrect
2865 usages of realloc to be sensible.
2872 Void_t
* rEALLOc(Void_t
* oldmem
, size_t bytes
)
2874 Void_t
* rEALLOc(oldmem
, bytes
) Void_t
* oldmem
; size_t bytes
;
2878 INTERNAL_SIZE_T nb
; /* padded request size */
2880 mchunkptr oldp
; /* chunk corresponding to oldmem */
2881 INTERNAL_SIZE_T oldsize
; /* its size */
2883 mchunkptr newp
; /* chunk to return */
2885 #if defined(_LIBC) || defined(MALLOC_HOOKS)
2886 if (__realloc_hook
!= NULL
) {
2889 result
= (*__realloc_hook
)(oldmem
, bytes
);
2894 #ifdef REALLOC_ZERO_BYTES_FREES
2895 if (bytes
== 0) { fREe(oldmem
); return 0; }
2898 /* realloc of null is supposed to be same as malloc */
2899 if (oldmem
== 0) return mALLOc(bytes
);
2901 oldp
= mem2chunk(oldmem
);
2902 oldsize
= chunksize(oldp
);
2904 nb
= request2size(bytes
);
2907 if (chunk_is_mmapped(oldp
))
2912 newp
= mremap_chunk(oldp
, nb
);
2913 if(newp
) return chunk2mem(newp
);
2915 /* Note the extra SIZE_SZ overhead. */
2916 if(oldsize
- SIZE_SZ
>= nb
) return oldmem
; /* do nothing */
2917 /* Must alloc, copy, free. */
2918 newmem
= mALLOc(bytes
);
2919 if (newmem
== 0) return 0; /* propagate failure */
2920 MALLOC_COPY(newmem
, oldmem
, oldsize
- 2*SIZE_SZ
);
2926 ar_ptr
= arena_for_ptr(oldp
);
2928 if(!mutex_trylock(&ar_ptr
->mutex
))
2929 ++(ar_ptr
->stat_lock_direct
);
2931 (void)mutex_lock(&ar_ptr
->mutex
);
2932 ++(ar_ptr
->stat_lock_wait
);
2935 (void)mutex_lock(&ar_ptr
->mutex
);
2939 /* As in malloc(), remember this arena for the next allocation. */
2940 tsd_setspecific(arena_key
, (Void_t
*)ar_ptr
);
2943 newp
= chunk_realloc(ar_ptr
, oldp
, oldsize
, nb
);
2945 (void)mutex_unlock(&ar_ptr
->mutex
);
2946 return newp
? chunk2mem(newp
) : NULL
;
2951 chunk_realloc(arena
* ar_ptr
, mchunkptr oldp
, INTERNAL_SIZE_T oldsize
,
2954 chunk_realloc(ar_ptr
, oldp
, oldsize
, nb
)
2955 arena
* ar_ptr
; mchunkptr oldp
; INTERNAL_SIZE_T oldsize
, nb
;
2958 mchunkptr newp
= oldp
; /* chunk to return */
2959 INTERNAL_SIZE_T newsize
= oldsize
; /* its size */
2961 mchunkptr next
; /* next contiguous chunk after oldp */
2962 INTERNAL_SIZE_T nextsize
; /* its size */
2964 mchunkptr prev
; /* previous contiguous chunk before oldp */
2965 INTERNAL_SIZE_T prevsize
; /* its size */
2967 mchunkptr remainder
; /* holds split off extra space from newp */
2968 INTERNAL_SIZE_T remainder_size
; /* its size */
2970 mchunkptr bck
; /* misc temp for linking */
2971 mchunkptr fwd
; /* misc temp for linking */
2973 check_inuse_chunk(ar_ptr
, oldp
);
2975 if ((long)(oldsize
) < (long)(nb
))
2978 /* Try expanding forward */
2980 next
= chunk_at_offset(oldp
, oldsize
);
2981 if (next
== top(ar_ptr
) || !inuse(next
))
2983 nextsize
= chunksize(next
);
2985 /* Forward into top only if a remainder */
2986 if (next
== top(ar_ptr
))
2988 if ((long)(nextsize
+ newsize
) >= (long)(nb
+ MINSIZE
))
2990 newsize
+= nextsize
;
2991 top(ar_ptr
) = chunk_at_offset(oldp
, nb
);
2992 set_head(top(ar_ptr
), (newsize
- nb
) | PREV_INUSE
);
2993 set_head_size(oldp
, nb
);
2998 /* Forward into next chunk */
2999 else if (((long)(nextsize
+ newsize
) >= (long)(nb
)))
3001 unlink(next
, bck
, fwd
);
3002 newsize
+= nextsize
;
3012 /* Try shifting backwards. */
3014 if (!prev_inuse(oldp
))
3016 prev
= prev_chunk(oldp
);
3017 prevsize
= chunksize(prev
);
3019 /* try forward + backward first to save a later consolidation */
3024 if (next
== top(ar_ptr
))
3026 if ((long)(nextsize
+ prevsize
+ newsize
) >= (long)(nb
+ MINSIZE
))
3028 unlink(prev
, bck
, fwd
);
3030 newsize
+= prevsize
+ nextsize
;
3031 MALLOC_COPY(chunk2mem(newp
), chunk2mem(oldp
), oldsize
- SIZE_SZ
);
3032 top(ar_ptr
) = chunk_at_offset(newp
, nb
);
3033 set_head(top(ar_ptr
), (newsize
- nb
) | PREV_INUSE
);
3034 set_head_size(newp
, nb
);
3039 /* into next chunk */
3040 else if (((long)(nextsize
+ prevsize
+ newsize
) >= (long)(nb
)))
3042 unlink(next
, bck
, fwd
);
3043 unlink(prev
, bck
, fwd
);
3045 newsize
+= nextsize
+ prevsize
;
3046 MALLOC_COPY(chunk2mem(newp
), chunk2mem(oldp
), oldsize
- SIZE_SZ
);
3052 if (prev
!= 0 && (long)(prevsize
+ newsize
) >= (long)nb
)
3054 unlink(prev
, bck
, fwd
);
3056 newsize
+= prevsize
;
3057 MALLOC_COPY(chunk2mem(newp
), chunk2mem(oldp
), oldsize
- SIZE_SZ
);
3064 newp
= chunk_alloc (ar_ptr
, nb
);
3066 if (newp
== 0) /* propagate failure */
3069 /* Avoid copy if newp is next chunk after oldp. */
3070 /* (This can only happen when new chunk is sbrk'ed.) */
3072 if ( newp
== next_chunk(oldp
))
3074 newsize
+= chunksize(newp
);
3079 /* Otherwise copy, free, and exit */
3080 MALLOC_COPY(chunk2mem(newp
), chunk2mem(oldp
), oldsize
- SIZE_SZ
);
3081 chunk_free(ar_ptr
, oldp
);
3086 split
: /* split off extra room in old or expanded chunk */
3088 if (newsize
- nb
>= MINSIZE
) /* split off remainder */
3090 remainder
= chunk_at_offset(newp
, nb
);
3091 remainder_size
= newsize
- nb
;
3092 set_head_size(newp
, nb
);
3093 set_head(remainder
, remainder_size
| PREV_INUSE
);
3094 set_inuse_bit_at_offset(remainder
, remainder_size
);
3095 chunk_free(ar_ptr
, remainder
);
3099 set_head_size(newp
, newsize
);
3100 set_inuse_bit_at_offset(newp
, newsize
);
3103 check_inuse_chunk(ar_ptr
, newp
);
3114 memalign requests more than enough space from malloc, finds a spot
3115 within that chunk that meets the alignment request, and then
3116 possibly frees the leading and trailing space.
3118 The alignment argument must be a power of two. This property is not
3119 checked by memalign, so misuse may result in random runtime errors.
3121 8-byte alignment is guaranteed by normal malloc calls, so don't
3122 bother calling memalign with an argument of 8 or less.
3124 Overreliance on memalign is a sure way to fragment space.
3130 Void_t
* mEMALIGn(size_t alignment
, size_t bytes
)
3132 Void_t
* mEMALIGn(alignment
, bytes
) size_t alignment
; size_t bytes
;
3136 INTERNAL_SIZE_T nb
; /* padded request size */
3139 #if defined(_LIBC) || defined(MALLOC_HOOKS)
3140 if (__memalign_hook
!= NULL
) {
3143 result
= (*__memalign_hook
)(alignment
, bytes
);
3148 /* If need less alignment than we give anyway, just relay to malloc */
3150 if (alignment
<= MALLOC_ALIGNMENT
) return mALLOc(bytes
);
3152 /* Otherwise, ensure that it is at least a minimum chunk size */
3154 if (alignment
< MINSIZE
) alignment
= MINSIZE
;
3156 nb
= request2size(bytes
);
3157 arena_get(ar_ptr
, nb
+ alignment
+ MINSIZE
);
3160 p
= chunk_align(ar_ptr
, nb
, alignment
);
3161 (void)mutex_unlock(&ar_ptr
->mutex
);
3162 return p
? chunk2mem(p
) : NULL
;
3167 chunk_align(arena
* ar_ptr
, INTERNAL_SIZE_T nb
, size_t alignment
)
3169 chunk_align(ar_ptr
, nb
, alignment
)
3170 arena
* ar_ptr
; INTERNAL_SIZE_T nb
; size_t alignment
;
3173 char* m
; /* memory returned by malloc call */
3174 mchunkptr p
; /* corresponding chunk */
3175 char* brk
; /* alignment point within p */
3176 mchunkptr newp
; /* chunk to return */
3177 INTERNAL_SIZE_T newsize
; /* its size */
3178 INTERNAL_SIZE_T leadsize
; /* leading space befor alignment point */
3179 mchunkptr remainder
; /* spare room at end to split off */
3180 long remainder_size
; /* its size */
3182 /* Call chunk_alloc with worst case padding to hit alignment. */
3183 p
= chunk_alloc(ar_ptr
, nb
+ alignment
+ MINSIZE
);
3185 return 0; /* propagate failure */
3189 if ((((unsigned long)(m
)) % alignment
) == 0) /* aligned */
3192 if(chunk_is_mmapped(p
)) {
3193 return p
; /* nothing more to do */
3197 else /* misaligned */
3200 Find an aligned spot inside chunk.
3201 Since we need to give back leading space in a chunk of at
3202 least MINSIZE, if the first calculation places us at
3203 a spot with less than MINSIZE leader, we can move to the
3204 next aligned spot -- we've allocated enough total room so that
3205 this is always possible.
3208 brk
= (char*)mem2chunk(((unsigned long)(m
+ alignment
- 1)) & -alignment
);
3209 if ((long)(brk
- (char*)(p
)) < (long)MINSIZE
) brk
+= alignment
;
3211 newp
= (mchunkptr
)brk
;
3212 leadsize
= brk
- (char*)(p
);
3213 newsize
= chunksize(p
) - leadsize
;
3216 if(chunk_is_mmapped(p
))
3218 newp
->prev_size
= p
->prev_size
+ leadsize
;
3219 set_head(newp
, newsize
|IS_MMAPPED
);
3224 /* give back leader, use the rest */
3226 set_head(newp
, newsize
| PREV_INUSE
);
3227 set_inuse_bit_at_offset(newp
, newsize
);
3228 set_head_size(p
, leadsize
);
3229 chunk_free(ar_ptr
, p
);
3232 assert (newsize
>=nb
&& (((unsigned long)(chunk2mem(p
))) % alignment
) == 0);
3235 /* Also give back spare room at the end */
3237 remainder_size
= chunksize(p
) - nb
;
3239 if (remainder_size
>= (long)MINSIZE
)
3241 remainder
= chunk_at_offset(p
, nb
);
3242 set_head(remainder
, remainder_size
| PREV_INUSE
);
3243 set_head_size(p
, nb
);
3244 chunk_free(ar_ptr
, remainder
);
3247 check_inuse_chunk(ar_ptr
, p
);
3255 valloc just invokes memalign with alignment argument equal
3256 to the page size of the system (or as near to this as can
3257 be figured out from all the includes/defines above.)
3261 Void_t
* vALLOc(size_t bytes
)
3263 Void_t
* vALLOc(bytes
) size_t bytes
;
3266 return mEMALIGn (malloc_getpagesize
, bytes
);
3270 pvalloc just invokes valloc for the nearest pagesize
3271 that will accommodate request
3276 Void_t
* pvALLOc(size_t bytes
)
3278 Void_t
* pvALLOc(bytes
) size_t bytes
;
3281 size_t pagesize
= malloc_getpagesize
;
3282 return mEMALIGn (pagesize
, (bytes
+ pagesize
- 1) & ~(pagesize
- 1));
3287 calloc calls chunk_alloc, then zeroes out the allocated chunk.
3292 Void_t
* cALLOc(size_t n
, size_t elem_size
)
3294 Void_t
* cALLOc(n
, elem_size
) size_t n
; size_t elem_size
;
3298 mchunkptr p
, oldtop
;
3299 INTERNAL_SIZE_T sz
, csz
, oldtopsize
;
3302 #if defined(_LIBC) || defined(MALLOC_HOOKS)
3303 if (__malloc_hook
!= NULL
) {
3305 mem
= (*__malloc_hook
)(sz
);
3309 return memset(mem
, 0, sz
);
3311 while(sz
> 0) ((char*)mem
)[--sz
] = 0; /* rather inefficient */
3317 sz
= request2size(n
* elem_size
);
3318 arena_get(ar_ptr
, sz
);
3322 /* check if expand_top called, in which case don't need to clear */
3324 oldtop
= top(ar_ptr
);
3325 oldtopsize
= chunksize(top(ar_ptr
));
3327 p
= chunk_alloc (ar_ptr
, sz
);
3329 /* Only clearing follows, so we can unlock early. */
3330 (void)mutex_unlock(&ar_ptr
->mutex
);
3338 /* Two optional cases in which clearing not necessary */
3341 if (chunk_is_mmapped(p
)) return mem
;
3347 if (p
== oldtop
&& csz
> oldtopsize
)
3349 /* clear only the bytes from non-freshly-sbrked memory */
3354 MALLOC_ZERO(mem
, csz
- SIZE_SZ
);
3361 cfree just calls free. It is needed/defined on some systems
3362 that pair it with calloc, presumably for odd historical reasons.
3368 void cfree(Void_t
*mem
)
3370 void cfree(mem
) Void_t
*mem
;
3381 Malloc_trim gives memory back to the system (via negative
3382 arguments to sbrk) if there is unused memory at the `high' end of
3383 the malloc pool. You can call this after freeing large blocks of
3384 memory to potentially reduce the system-level memory requirements
3385 of a program. However, it cannot guarantee to reduce memory. Under
3386 some allocation patterns, some large free blocks of memory will be
3387 locked between two used chunks, so they cannot be given back to
3390 The `pad' argument to malloc_trim represents the amount of free
3391 trailing space to leave untrimmed. If this argument is zero,
3392 only the minimum amount of memory to maintain internal data
3393 structures will be left (one page or less). Non-zero arguments
3394 can be supplied to maintain enough trailing space to service
3395 future expected allocations without having to re-obtain memory
3398 Malloc_trim returns 1 if it actually released any memory, else 0.
3403 int mALLOC_TRIm(size_t pad
)
3405 int mALLOC_TRIm(pad
) size_t pad
;
3410 (void)mutex_lock(&main_arena
.mutex
);
3411 res
= main_trim(pad
);
3412 (void)mutex_unlock(&main_arena
.mutex
);
3416 /* Trim the main arena. */
3420 main_trim(size_t pad
)
3422 main_trim(pad
) size_t pad
;
3425 mchunkptr top_chunk
; /* The current top chunk */
3426 long top_size
; /* Amount of top-most memory */
3427 long extra
; /* Amount to release */
3428 char* current_brk
; /* address returned by pre-check sbrk call */
3429 char* new_brk
; /* address returned by negative sbrk call */
3431 unsigned long pagesz
= malloc_getpagesize
;
3433 top_chunk
= top(&main_arena
);
3434 top_size
= chunksize(top_chunk
);
3435 extra
= ((top_size
- pad
- MINSIZE
+ (pagesz
-1)) / pagesz
- 1) * pagesz
;
3437 if (extra
< (long)pagesz
) /* Not enough memory to release */
3440 /* Test to make sure no one else called sbrk */
3441 current_brk
= (char*)(MORECORE (0));
3442 if (current_brk
!= (char*)(top_chunk
) + top_size
)
3443 return 0; /* Apparently we don't own memory; must fail */
3445 new_brk
= (char*)(MORECORE (-extra
));
3447 /* Call the `morecore' hook if necessary. */
3448 if (__after_morecore_hook
)
3449 (*__after_morecore_hook
) ();
3451 if (new_brk
== (char*)(MORECORE_FAILURE
)) { /* sbrk failed? */
3452 /* Try to figure out what we have */
3453 current_brk
= (char*)(MORECORE (0));
3454 top_size
= current_brk
- (char*)top_chunk
;
3455 if (top_size
>= (long)MINSIZE
) /* if not, we are very very dead! */
3457 sbrked_mem
= current_brk
- sbrk_base
;
3458 set_head(top_chunk
, top_size
| PREV_INUSE
);
3460 check_chunk(&main_arena
, top_chunk
);
3463 sbrked_mem
-= extra
;
3465 /* Success. Adjust top accordingly. */
3466 set_head(top_chunk
, (top_size
- extra
) | PREV_INUSE
);
3467 check_chunk(&main_arena
, top_chunk
);
3475 heap_trim(heap_info
*heap
, size_t pad
)
3477 heap_trim(heap
, pad
) heap_info
*heap
; size_t pad
;
3480 unsigned long pagesz
= malloc_getpagesize
;
3481 arena
*ar_ptr
= heap
->ar_ptr
;
3482 mchunkptr top_chunk
= top(ar_ptr
), p
, bck
, fwd
;
3483 heap_info
*prev_heap
;
3484 long new_size
, top_size
, extra
;
3486 /* Can this heap go away completely ? */
3487 while(top_chunk
== chunk_at_offset(heap
, sizeof(*heap
))) {
3488 prev_heap
= heap
->prev
;
3489 p
= chunk_at_offset(prev_heap
, prev_heap
->size
- (MINSIZE
-2*SIZE_SZ
));
3490 assert(p
->size
== (0|PREV_INUSE
)); /* must be fencepost */
3492 new_size
= chunksize(p
) + (MINSIZE
-2*SIZE_SZ
);
3493 assert(new_size
>0 && new_size
<(long)(2*MINSIZE
));
3495 new_size
+= p
->prev_size
;
3496 assert(new_size
>0 && new_size
<HEAP_MAX_SIZE
);
3497 if(new_size
+ (HEAP_MAX_SIZE
- prev_heap
->size
) < pad
+ MINSIZE
+ pagesz
)
3499 ar_ptr
->size
-= heap
->size
;
3502 if(!prev_inuse(p
)) { /* consolidate backward */
3504 unlink(p
, bck
, fwd
);
3506 assert(((unsigned long)((char*)p
+ new_size
) & (pagesz
-1)) == 0);
3507 assert( ((char*)p
+ new_size
) == ((char*)heap
+ heap
->size
) );
3508 top(ar_ptr
) = top_chunk
= p
;
3509 set_head(top_chunk
, new_size
| PREV_INUSE
);
3510 check_chunk(ar_ptr
, top_chunk
);
3512 top_size
= chunksize(top_chunk
);
3513 extra
= ((top_size
- pad
- MINSIZE
+ (pagesz
-1))/pagesz
- 1) * pagesz
;
3514 if(extra
< (long)pagesz
)
3516 /* Try to shrink. */
3517 if(grow_heap(heap
, -extra
) != 0)
3519 ar_ptr
->size
-= extra
;
3521 /* Success. Adjust top accordingly. */
3522 set_head(top_chunk
, (top_size
- extra
) | PREV_INUSE
);
3523 check_chunk(ar_ptr
, top_chunk
);
3534 This routine tells you how many bytes you can actually use in an
3535 allocated chunk, which may be more than you requested (although
3536 often not). You can use this many bytes without worrying about
3537 overwriting other allocated objects. Not a particularly great
3538 programming practice, but still sometimes useful.
3543 size_t mALLOC_USABLE_SIZe(Void_t
* mem
)
3545 size_t mALLOC_USABLE_SIZe(mem
) Void_t
* mem
;
3555 if(!chunk_is_mmapped(p
))
3557 if (!inuse(p
)) return 0;
3558 check_inuse_chunk(arena_for_ptr(mem
), p
);
3559 return chunksize(p
) - SIZE_SZ
;
3561 return chunksize(p
) - 2*SIZE_SZ
;
3568 /* Utility to update mallinfo for malloc_stats() and mallinfo() */
3572 malloc_update_mallinfo(arena
*ar_ptr
, struct mallinfo
*mi
)
3574 malloc_update_mallinfo(ar_ptr
, mi
) arena
*ar_ptr
; struct mallinfo
*mi
;
3583 INTERNAL_SIZE_T avail
;
3585 (void)mutex_lock(&ar_ptr
->mutex
);
3586 avail
= chunksize(top(ar_ptr
));
3587 navail
= ((long)(avail
) >= (long)MINSIZE
)? 1 : 0;
3589 for (i
= 1; i
< NAV
; ++i
)
3591 b
= bin_at(ar_ptr
, i
);
3592 for (p
= last(b
); p
!= b
; p
= p
->bk
)
3595 check_free_chunk(ar_ptr
, p
);
3596 for (q
= next_chunk(p
);
3597 q
!= top(ar_ptr
) && inuse(q
) && (long)chunksize(q
) > 0;
3599 check_inuse_chunk(ar_ptr
, q
);
3601 avail
+= chunksize(p
);
3606 mi
->arena
= ar_ptr
->size
;
3607 mi
->ordblks
= navail
;
3608 mi
->uordblks
= ar_ptr
->size
- avail
;
3609 mi
->fordblks
= avail
;
3610 mi
->hblks
= n_mmaps
;
3611 mi
->hblkhd
= mmapped_mem
;
3612 mi
->keepcost
= chunksize(top(ar_ptr
));
3614 (void)mutex_unlock(&ar_ptr
->mutex
);
3617 #if !defined(NO_THREADS) && MALLOC_DEBUG > 1
3619 /* Print the complete contents of a single heap to stderr. */
3623 dump_heap(heap_info
*heap
)
3625 dump_heap(heap
) heap_info
*heap
;
3631 fprintf(stderr
, "Heap %p, size %10lx:\n", heap
, (long)heap
->size
);
3632 ptr
= (heap
->ar_ptr
!= (arena
*)(heap
+1)) ?
3633 (char*)(heap
+ 1) : (char*)(heap
+ 1) + sizeof(arena
);
3634 p
= (mchunkptr
)(((unsigned long)ptr
+ MALLOC_ALIGN_MASK
) &
3635 ~MALLOC_ALIGN_MASK
);
3637 fprintf(stderr
, "chunk %p size %10lx", p
, (long)p
->size
);
3638 if(p
== top(heap
->ar_ptr
)) {
3639 fprintf(stderr
, " (top)\n");
3641 } else if(p
->size
== (0|PREV_INUSE
)) {
3642 fprintf(stderr
, " (fence)\n");
3645 fprintf(stderr
, "\n");
3658 For all arenas separately and in total, prints on stderr the
3659 amount of space obtained from the system, and the current number
3660 of bytes allocated via malloc (or realloc, etc) but not yet
3661 freed. (Note that this is the number of bytes allocated, not the
3662 number requested. It will be larger than the number requested
3663 because of alignment and bookkeeping overhead.) When not compiled
3664 for multiple threads, the maximum amount of allocated memory
3665 (which may be more than current if malloc_trim and/or munmap got
3666 called) is also reported. When using mmap(), prints the maximum
3667 number of simultaneous mmap regions used, too.
3676 unsigned int in_use_b
= mmapped_mem
, system_b
= in_use_b
;
3678 long stat_lock_direct
= 0, stat_lock_loop
= 0, stat_lock_wait
= 0;
3681 for(i
=0, ar_ptr
= &main_arena
;; i
++) {
3682 malloc_update_mallinfo(ar_ptr
, &mi
);
3683 fprintf(stderr
, "Arena %d:\n", i
);
3684 fprintf(stderr
, "system bytes = %10u\n", (unsigned int)mi
.arena
);
3685 fprintf(stderr
, "in use bytes = %10u\n", (unsigned int)mi
.uordblks
);
3686 system_b
+= mi
.arena
;
3687 in_use_b
+= mi
.uordblks
;
3689 stat_lock_direct
+= ar_ptr
->stat_lock_direct
;
3690 stat_lock_loop
+= ar_ptr
->stat_lock_loop
;
3691 stat_lock_wait
+= ar_ptr
->stat_lock_wait
;
3693 #if !defined(NO_THREADS) && MALLOC_DEBUG > 1
3694 if(ar_ptr
!= &main_arena
) {
3696 (void)mutex_lock(&ar_ptr
->mutex
);
3697 heap
= heap_for_ptr(top(ar_ptr
));
3698 while(heap
) { dump_heap(heap
); heap
= heap
->prev
; }
3699 (void)mutex_unlock(&ar_ptr
->mutex
);
3702 ar_ptr
= ar_ptr
->next
;
3703 if(ar_ptr
== &main_arena
) break;
3705 fprintf(stderr
, "Total (incl. mmap):\n");
3706 fprintf(stderr
, "system bytes = %10u\n", system_b
);
3707 fprintf(stderr
, "in use bytes = %10u\n", in_use_b
);
3709 fprintf(stderr
, "max system bytes = %10u\n", (unsigned int)max_total_mem
);
3712 fprintf(stderr
, "max mmap regions = %10u\n", (unsigned int)max_n_mmaps
);
3715 fprintf(stderr
, "heaps created = %10d\n", stat_n_heaps
);
3716 fprintf(stderr
, "locked directly = %10ld\n", stat_lock_direct
);
3717 fprintf(stderr
, "locked in loop = %10ld\n", stat_lock_loop
);
3718 fprintf(stderr
, "locked waiting = %10ld\n", stat_lock_wait
);
3719 fprintf(stderr
, "locked total = %10ld\n",
3720 stat_lock_direct
+ stat_lock_loop
+ stat_lock_wait
);
3725 mallinfo returns a copy of updated current mallinfo.
3726 The information reported is for the arena last used by the thread.
3729 struct mallinfo
mALLINFo()
3732 Void_t
*vptr
= NULL
;
3735 tsd_getspecific(arena_key
, vptr
);
3737 malloc_update_mallinfo((vptr
? (arena
*)vptr
: &main_arena
), &mi
);
3747 mallopt is the general SVID/XPG interface to tunable parameters.
3748 The format is to provide a (parameter-number, parameter-value) pair.
3749 mallopt then sets the corresponding parameter to the argument
3750 value if it can (i.e., so long as the value is meaningful),
3751 and returns 1 if successful else 0.
3753 See descriptions of tunable parameters above.
3758 int mALLOPt(int param_number
, int value
)
3760 int mALLOPt(param_number
, value
) int param_number
; int value
;
3763 switch(param_number
)
3765 case M_TRIM_THRESHOLD
:
3766 trim_threshold
= value
; return 1;
3768 top_pad
= value
; return 1;
3769 case M_MMAP_THRESHOLD
:
3771 /* Forbid setting the threshold too high. */
3772 if((unsigned long)value
> HEAP_MAX_SIZE
/2) return 0;
3774 mmap_threshold
= value
; return 1;
3777 n_mmaps_max
= value
; return 1;
3779 if (value
!= 0) return 0; else n_mmaps_max
= value
; return 1;
3781 case M_CHECK_ACTION
:
3782 check_action
= value
; return 1;
3791 /* Get/set state: malloc_get_state() records the current state of all
3792 malloc variables (_except_ for the actual heap contents and `hook'
3793 function pointers) in a system dependent, opaque data structure.
3794 This data structure is dynamically allocated and can be free()d
3795 after use. malloc_set_state() restores the state of all malloc
3796 variables to the previously obtained state. This is especially
3797 useful when using this malloc as part of a shared library, and when
3798 the heap contents are saved/restored via some other method. The
3799 primary example for this is GNU Emacs with its `dumping' procedure.
3800 `Hook' function pointers are never saved or restored by these
3803 #define MALLOC_STATE_MAGIC 0x444c4541l
3804 #define MALLOC_STATE_VERSION (0*0x100l + 0l) /* major*0x100 + minor */
3806 struct malloc_state
{
3809 mbinptr av
[NAV
* 2 + 2];
3811 int sbrked_mem_bytes
;
3812 unsigned long trim_threshold
;
3813 unsigned long top_pad
;
3814 unsigned int n_mmaps_max
;
3815 unsigned long mmap_threshold
;
3817 unsigned long max_sbrked_mem
;
3818 unsigned long max_total_mem
;
3819 unsigned int n_mmaps
;
3820 unsigned int max_n_mmaps
;
3821 unsigned long mmapped_mem
;
3822 unsigned long max_mmapped_mem
;
3829 struct malloc_state
* ms
;
3834 (void)mutex_lock(&main_arena
.mutex
);
3835 victim
= chunk_alloc(&main_arena
, request2size(sizeof(*ms
)));
3837 (void)mutex_unlock(&main_arena
.mutex
);
3840 ms
= (struct malloc_state
*)chunk2mem(victim
);
3841 ms
->magic
= MALLOC_STATE_MAGIC
;
3842 ms
->version
= MALLOC_STATE_VERSION
;
3843 ms
->av
[0] = main_arena
.av
[0];
3844 ms
->av
[1] = main_arena
.av
[1];
3845 for(i
=0; i
<NAV
; i
++) {
3846 b
= bin_at(&main_arena
, i
);
3848 ms
->av
[2*i
+2] = ms
->av
[2*i
+3] = 0; /* empty bin (or initial top) */
3850 ms
->av
[2*i
+2] = first(b
);
3851 ms
->av
[2*i
+3] = last(b
);
3854 ms
->sbrk_base
= sbrk_base
;
3855 ms
->sbrked_mem_bytes
= sbrked_mem
;
3856 ms
->trim_threshold
= trim_threshold
;
3857 ms
->top_pad
= top_pad
;
3858 ms
->n_mmaps_max
= n_mmaps_max
;
3859 ms
->mmap_threshold
= mmap_threshold
;
3860 ms
->check_action
= check_action
;
3861 ms
->max_sbrked_mem
= max_sbrked_mem
;
3863 ms
->max_total_mem
= max_total_mem
;
3865 ms
->max_total_mem
= 0;
3867 ms
->n_mmaps
= n_mmaps
;
3868 ms
->max_n_mmaps
= max_n_mmaps
;
3869 ms
->mmapped_mem
= mmapped_mem
;
3870 ms
->max_mmapped_mem
= max_mmapped_mem
;
3871 (void)mutex_unlock(&main_arena
.mutex
);
3877 mALLOC_SET_STATe(Void_t
* msptr
)
3879 mALLOC_SET_STATe(msptr
) Void_t
* msptr
;
3882 struct malloc_state
* ms
= (struct malloc_state
*)msptr
;
3887 if(ms
->magic
!= MALLOC_STATE_MAGIC
) return -1;
3888 /* Must fail if the major version is too high. */
3889 if((ms
->version
& ~0xffl
) > (MALLOC_STATE_VERSION
& ~0xffl
)) return -2;
3890 (void)mutex_lock(&main_arena
.mutex
);
3891 main_arena
.av
[0] = ms
->av
[0];
3892 main_arena
.av
[1] = ms
->av
[1];
3893 for(i
=0; i
<NAV
; i
++) {
3894 b
= bin_at(&main_arena
, i
);
3895 if(ms
->av
[2*i
+2] == 0)
3896 first(b
) = last(b
) = b
;
3898 first(b
) = ms
->av
[2*i
+2];
3899 last(b
) = ms
->av
[2*i
+3];
3901 /* Make sure the links to the `av'-bins in the heap are correct. */
3907 sbrk_base
= ms
->sbrk_base
;
3908 sbrked_mem
= ms
->sbrked_mem_bytes
;
3909 trim_threshold
= ms
->trim_threshold
;
3910 top_pad
= ms
->top_pad
;
3911 n_mmaps_max
= ms
->n_mmaps_max
;
3912 mmap_threshold
= ms
->mmap_threshold
;
3913 check_action
= ms
->check_action
;
3914 max_sbrked_mem
= ms
->max_sbrked_mem
;
3916 max_total_mem
= ms
->max_total_mem
;
3918 n_mmaps
= ms
->n_mmaps
;
3919 max_n_mmaps
= ms
->max_n_mmaps
;
3920 mmapped_mem
= ms
->mmapped_mem
;
3921 max_mmapped_mem
= ms
->max_mmapped_mem
;
3922 /* add version-dependent code here */
3923 (void)mutex_unlock(&main_arena
.mutex
);
3929 #if defined(_LIBC) || defined(MALLOC_HOOKS)
3931 /* A simple, standard set of debugging hooks. Overhead is `only' one
3932 byte per chunk; still this will catch most cases of double frees or
3935 #define MAGICBYTE(p) ( ( ((size_t)p >> 3) ^ ((size_t)p >> 11)) & 0xFF )
3937 /* Convert a pointer to be free()d or realloc()ed to a valid chunk
3938 pointer. If the provided pointer is not valid, return NULL. The
3939 goal here is to avoid crashes, unlike in the MALLOC_DEBUG code. */
3943 mem2chunk_check(Void_t
* mem
)
3945 mem2chunk_check(mem
) Void_t
* mem
;
3952 if(!aligned_OK(p
)) return NULL
;
3953 if( (char*)p
>=sbrk_base
&& (char*)p
<(sbrk_base
+sbrked_mem
) ) {
3954 /* Must be a chunk in conventional heap memory. */
3955 if(chunk_is_mmapped(p
) ||
3956 ( (sz
= chunksize(p
)), ((char*)p
+ sz
)>=(sbrk_base
+sbrked_mem
) ) ||
3957 sz
<MINSIZE
|| sz
&MALLOC_ALIGN_MASK
|| !inuse(p
) ||
3958 ( !prev_inuse(p
) && (p
->prev_size
&MALLOC_ALIGN_MASK
||
3959 (long)prev_chunk(p
)<(long)sbrk_base
||
3960 next_chunk(prev_chunk(p
))!=p
) ))
3962 if(*((unsigned char*)p
+ sz
+ (SIZE_SZ
-1)) != MAGICBYTE(p
))
3964 *((unsigned char*)p
+ sz
+ (SIZE_SZ
-1)) ^= 0xFF;
3966 unsigned long offset
, page_mask
= malloc_getpagesize
-1;
3968 /* mmap()ed chunks have MALLOC_ALIGNMENT or higher power-of-two
3969 alignment relative to the beginning of a page. Check this
3971 offset
= (unsigned long)mem
& page_mask
;
3972 if((offset
!=MALLOC_ALIGNMENT
&& offset
!=0 && offset
!=0x10 &&
3973 offset
!=0x20 && offset
!=0x40 && offset
!=0x80 && offset
!=0x100 &&
3974 offset
!=0x200 && offset
!=0x400 && offset
!=0x800 && offset
!=0x1000 &&
3976 !chunk_is_mmapped(p
) || (p
->size
& PREV_INUSE
) ||
3977 ( (((unsigned long)p
- p
->prev_size
) & page_mask
) != 0 ) ||
3978 ( (sz
= chunksize(p
)), ((p
->prev_size
+ sz
) & page_mask
) != 0 ) )
3980 if(*((unsigned char*)p
+ sz
- 1) != MAGICBYTE(p
))
3982 *((unsigned char*)p
+ sz
- 1) ^= 0xFF;
3989 malloc_check(size_t sz
)
3991 malloc_check(sz
) size_t sz
;
3995 INTERNAL_SIZE_T nb
= request2size(sz
+ 1);
3997 (void)mutex_lock(&main_arena
.mutex
);
3998 victim
= chunk_alloc(&main_arena
, nb
);
3999 (void)mutex_unlock(&main_arena
.mutex
);
4000 if(!victim
) return NULL
;
4001 nb
= chunksize(victim
);
4002 if(chunk_is_mmapped(victim
))
4006 *((unsigned char*)victim
+ nb
) = MAGICBYTE(victim
);
4007 return chunk2mem(victim
);
4012 free_check(Void_t
* mem
)
4014 free_check(mem
) Void_t
* mem
;
4020 (void)mutex_lock(&main_arena
.mutex
);
4021 p
= mem2chunk_check(mem
);
4023 (void)mutex_unlock(&main_arena
.mutex
);
4024 switch(check_action
) {
4026 fprintf(stderr
, "free(): invalid pointer %lx!\n", (long)(mem
));
4034 if (chunk_is_mmapped(p
)) {
4035 (void)mutex_unlock(&main_arena
.mutex
);
4040 #if 0 /* Erase freed memory. */
4041 memset(mem
, 0, chunksize(p
) - (SIZE_SZ
+1));
4043 chunk_free(&main_arena
, p
);
4044 (void)mutex_unlock(&main_arena
.mutex
);
4049 realloc_check(Void_t
* oldmem
, size_t bytes
)
4051 realloc_check(oldmem
, bytes
) Void_t
* oldmem
; size_t bytes
;
4054 mchunkptr oldp
, newp
;
4055 INTERNAL_SIZE_T nb
, oldsize
;
4057 if (oldmem
== 0) return malloc_check(bytes
);
4058 (void)mutex_lock(&main_arena
.mutex
);
4059 oldp
= mem2chunk_check(oldmem
);
4061 (void)mutex_unlock(&main_arena
.mutex
);
4062 switch(check_action
) {
4064 fprintf(stderr
, "realloc(): invalid pointer %lx!\n", (long)(oldmem
));
4069 return malloc_check(bytes
);
4071 oldsize
= chunksize(oldp
);
4073 nb
= request2size(bytes
+1);
4076 if (chunk_is_mmapped(oldp
)) {
4078 newp
= mremap_chunk(oldp
, nb
);
4081 /* Note the extra SIZE_SZ overhead. */
4082 if(oldsize
- SIZE_SZ
>= nb
) newp
= oldp
; /* do nothing */
4084 /* Must alloc, copy, free. */
4085 newp
= chunk_alloc(&main_arena
, nb
);
4087 MALLOC_COPY(chunk2mem(newp
), oldmem
, oldsize
- 2*SIZE_SZ
);
4095 #endif /* HAVE_MMAP */
4096 newp
= chunk_realloc(&main_arena
, oldp
, oldsize
, nb
);
4097 #if 0 /* Erase freed memory. */
4098 nb
= chunksize(newp
);
4099 if(oldp
<newp
|| oldp
>=chunk_at_offset(newp
, nb
)) {
4100 memset((char*)oldmem
+ 2*sizeof(mbinptr
), 0,
4101 oldsize
- (2*sizeof(mbinptr
)+2*SIZE_SZ
+1));
4102 } else if(nb
> oldsize
+SIZE_SZ
) {
4103 memset((char*)chunk2mem(newp
) + oldsize
, 0, nb
- (oldsize
+SIZE_SZ
));
4109 (void)mutex_unlock(&main_arena
.mutex
);
4111 if(!newp
) return NULL
;
4112 nb
= chunksize(newp
);
4113 if(chunk_is_mmapped(newp
))
4117 *((unsigned char*)newp
+ nb
) = MAGICBYTE(newp
);
4118 return chunk2mem(newp
);
4123 memalign_check(size_t alignment
, size_t bytes
)
4125 memalign_check(alignment
, bytes
) size_t alignment
; size_t bytes
;
4131 if (alignment
<= MALLOC_ALIGNMENT
) return malloc_check(bytes
);
4132 if (alignment
< MINSIZE
) alignment
= MINSIZE
;
4134 nb
= request2size(bytes
+1);
4135 (void)mutex_lock(&main_arena
.mutex
);
4136 p
= chunk_align(&main_arena
, nb
, alignment
);
4137 (void)mutex_unlock(&main_arena
.mutex
);
4140 if(chunk_is_mmapped(p
))
4144 *((unsigned char*)p
+ nb
) = MAGICBYTE(p
);
4145 return chunk2mem(p
);
4148 /* The following hooks are used when the global initialization in
4149 ptmalloc_init() hasn't completed yet. */
4153 malloc_starter(size_t sz
)
4155 malloc_starter(sz
) size_t sz
;
4158 mchunkptr victim
= chunk_alloc(&main_arena
, request2size(sz
));
4160 return victim
? chunk2mem(victim
) : 0;
4165 free_starter(Void_t
* mem
)
4167 free_starter(mem
) Void_t
* mem
;
4175 if (chunk_is_mmapped(p
)) {
4180 chunk_free(&main_arena
, p
);
4183 #endif /* defined(_LIBC) || defined(MALLOC_HOOKS) */
4188 weak_alias (__libc_calloc
, __calloc
) weak_alias (__libc_calloc
, calloc
)
4189 weak_alias (__libc_free
, __cfree
) weak_alias (__libc_free
, cfree
)
4190 weak_alias (__libc_free
, __free
) weak_alias (__libc_free
, free
)
4191 weak_alias (__libc_malloc
, __malloc
) weak_alias (__libc_malloc
, malloc
)
4192 weak_alias (__libc_memalign
, __memalign
) weak_alias (__libc_memalign
, memalign
)
4193 weak_alias (__libc_realloc
, __realloc
) weak_alias (__libc_realloc
, realloc
)
4194 weak_alias (__libc_valloc
, __valloc
) weak_alias (__libc_valloc
, valloc
)
4195 weak_alias (__libc_pvalloc
, __pvalloc
) weak_alias (__libc_pvalloc
, pvalloc
)
4196 weak_alias (__libc_mallinfo
, __mallinfo
) weak_alias (__libc_mallinfo
, mallinfo
)
4197 weak_alias (__libc_mallopt
, __mallopt
) weak_alias (__libc_mallopt
, mallopt
)
4199 weak_alias (__malloc_stats
, malloc_stats
)
4200 weak_alias (__malloc_usable_size
, malloc_usable_size
)
4201 weak_alias (__malloc_trim
, malloc_trim
)
4202 weak_alias (__malloc_get_state
, malloc_get_state
)
4203 weak_alias (__malloc_set_state
, malloc_set_state
)
4210 V2.6.4-pt3 Thu Feb 20 1997 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
4211 * Added malloc_get/set_state() (mainly for use in GNU emacs),
4212 using interface from Marcus Daniels
4213 * All parameters are now adjustable via environment variables
4215 V2.6.4-pt2 Sat Dec 14 1996 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
4216 * Added debugging hooks
4217 * Fixed possible deadlock in realloc() when out of memory
4218 * Don't pollute namespace in glibc: use __getpagesize, __mmap, etc.
4220 V2.6.4-pt Wed Dec 4 1996 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
4221 * Very minor updates from the released 2.6.4 version.
4222 * Trimmed include file down to exported data structures.
4223 * Changes from H.J. Lu for glibc-2.0.
4225 V2.6.3i-pt Sep 16 1996 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
4226 * Many changes for multiple threads
4227 * Introduced arenas and heaps
4229 V2.6.3 Sun May 19 08:17:58 1996 Doug Lea (dl at gee)
4230 * Added pvalloc, as recommended by H.J. Liu
4231 * Added 64bit pointer support mainly from Wolfram Gloger
4232 * Added anonymously donated WIN32 sbrk emulation
4233 * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
4234 * malloc_extend_top: fix mask error that caused wastage after
4236 * Add linux mremap support code from HJ Liu
4238 V2.6.2 Tue Dec 5 06:52:55 1995 Doug Lea (dl at gee)
4239 * Integrated most documentation with the code.
4240 * Add support for mmap, with help from
4241 Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
4242 * Use last_remainder in more cases.
4243 * Pack bins using idea from colin@nyx10.cs.du.edu
4244 * Use ordered bins instead of best-fit threshold
4245 * Eliminate block-local decls to simplify tracing and debugging.
4246 * Support another case of realloc via move into top
4247 * Fix error occurring when initial sbrk_base not word-aligned.
4248 * Rely on page size for units instead of SBRK_UNIT to
4249 avoid surprises about sbrk alignment conventions.
4250 * Add mallinfo, mallopt. Thanks to Raymond Nijssen
4251 (raymond@es.ele.tue.nl) for the suggestion.
4252 * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
4253 * More precautions for cases where other routines call sbrk,
4254 courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
4255 * Added macros etc., allowing use in linux libc from
4256 H.J. Lu (hjl@gnu.ai.mit.edu)
4257 * Inverted this history list
4259 V2.6.1 Sat Dec 2 14:10:57 1995 Doug Lea (dl at gee)
4260 * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
4261 * Removed all preallocation code since under current scheme
4262 the work required to undo bad preallocations exceeds
4263 the work saved in good cases for most test programs.
4264 * No longer use return list or unconsolidated bins since
4265 no scheme using them consistently outperforms those that don't
4266 given above changes.
4267 * Use best fit for very large chunks to prevent some worst-cases.
4268 * Added some support for debugging
4270 V2.6.0 Sat Nov 4 07:05:23 1995 Doug Lea (dl at gee)
4271 * Removed footers when chunks are in use. Thanks to
4272 Paul Wilson (wilson@cs.texas.edu) for the suggestion.
4274 V2.5.4 Wed Nov 1 07:54:51 1995 Doug Lea (dl at gee)
4275 * Added malloc_trim, with help from Wolfram Gloger
4276 (wmglo@Dent.MED.Uni-Muenchen.DE).
4278 V2.5.3 Tue Apr 26 10:16:01 1994 Doug Lea (dl at g)
4280 V2.5.2 Tue Apr 5 16:20:40 1994 Doug Lea (dl at g)
4281 * realloc: try to expand in both directions
4282 * malloc: swap order of clean-bin strategy;
4283 * realloc: only conditionally expand backwards
4284 * Try not to scavenge used bins
4285 * Use bin counts as a guide to preallocation
4286 * Occasionally bin return list chunks in first scan
4287 * Add a few optimizations from colin@nyx10.cs.du.edu
4289 V2.5.1 Sat Aug 14 15:40:43 1993 Doug Lea (dl at g)
4290 * faster bin computation & slightly different binning
4291 * merged all consolidations to one part of malloc proper
4292 (eliminating old malloc_find_space & malloc_clean_bin)
4293 * Scan 2 returns chunks (not just 1)
4294 * Propagate failure in realloc if malloc returns 0
4295 * Add stuff to allow compilation on non-ANSI compilers
4296 from kpv@research.att.com
4298 V2.5 Sat Aug 7 07:41:59 1993 Doug Lea (dl at g.oswego.edu)
4299 * removed potential for odd address access in prev_chunk
4300 * removed dependency on getpagesize.h
4301 * misc cosmetics and a bit more internal documentation
4302 * anticosmetics: mangled names in macros to evade debugger strangeness
4303 * tested on sparc, hp-700, dec-mips, rs6000
4304 with gcc & native cc (hp, dec only) allowing
4305 Detlefs & Zorn comparison study (in SIGPLAN Notices.)
4307 Trial version Fri Aug 28 13:14:29 1992 Doug Lea (dl at g.oswego.edu)
4308 * Based loosely on libg++-1.2X malloc. (It retains some of the overall
4309 structure of old version, but most details differ.)