3 #if 0 /* Moved to malloc.h */
4 /* ---------- To make a malloc.h, start cutting here ------------ */
7 A version of malloc/free/realloc written by Doug Lea and released to the
8 public domain. Send questions/comments/complaints/performance data
11 * VERSION 2.6.6 Sun Mar 5 19:10:03 2000 Doug Lea (dl at gee)
13 Note: There may be an updated version of this malloc obtainable at
14 ftp://g.oswego.edu/pub/misc/malloc.c
15 Check before installing!
17 * Why use this malloc?
19 This is not the fastest, most space-conserving, most portable, or
20 most tunable malloc ever written. However it is among the fastest
21 while also being among the most space-conserving, portable and tunable.
22 Consistent balance across these factors results in a good general-purpose
23 allocator. For a high-level description, see
24 http://g.oswego.edu/dl/html/malloc.html
26 * Synopsis of public routines
28 (Much fuller descriptions are contained in the program documentation below.)
31 Return a pointer to a newly allocated chunk of at least n bytes, or null
32 if no space is available.
34 Release the chunk of memory pointed to by p, or no effect if p is null.
35 realloc(Void_t* p, size_t n);
36 Return a pointer to a chunk of size n that contains the same data
37 as does chunk p up to the minimum of (n, p's size) bytes, or null
38 if no space is available. The returned pointer may or may not be
39 the same as p. If p is null, equivalent to malloc. Unless the
40 #define REALLOC_ZERO_BYTES_FREES below is set, realloc with a
41 size argument of zero (re)allocates a minimum-sized chunk.
42 memalign(size_t alignment, size_t n);
43 Return a pointer to a newly allocated chunk of n bytes, aligned
44 in accord with the alignment argument, which must be a power of
47 Equivalent to memalign(pagesize, n), where pagesize is the page
48 size of the system (or as near to this as can be figured out from
49 all the includes/defines below.)
51 Equivalent to valloc(minimum-page-that-holds(n)), that is,
52 round up n to nearest pagesize.
53 calloc(size_t unit, size_t quantity);
54 Returns a pointer to quantity * unit bytes, with all locations
57 Equivalent to free(p).
58 malloc_trim(size_t pad);
59 Release all but pad bytes of freed top-most memory back
60 to the system. Return 1 if successful, else 0.
61 malloc_usable_size(Void_t* p);
62 Report the number usable allocated bytes associated with allocated
63 chunk p. This may or may not report more bytes than were requested,
64 due to alignment and minimum size constraints.
66 Prints brief summary statistics.
68 Returns (by copy) a struct containing various summary statistics.
69 mallopt(int parameter_number, int parameter_value)
70 Changes one of the tunable parameters described below. Returns
71 1 if successful in changing the parameter, else 0.
76 8 byte alignment is currently hardwired into the design. This
77 seems to suffice for all current machines and C compilers.
79 Assumed pointer representation: 4 or 8 bytes
80 Code for 8-byte pointers is untested by me but has worked
81 reliably by Wolfram Gloger, who contributed most of the
82 changes supporting this.
84 Assumed size_t representation: 4 or 8 bytes
85 Note that size_t is allowed to be 4 bytes even if pointers are 8.
87 Minimum overhead per allocated chunk: 4 or 8 bytes
88 Each malloced chunk has a hidden overhead of 4 bytes holding size
89 and status information.
91 Minimum allocated size: 4-byte ptrs: 16 bytes (including 4 overhead)
92 8-byte ptrs: 24/32 bytes (including, 4/8 overhead)
94 When a chunk is freed, 12 (for 4byte ptrs) or 20 (for 8 byte
95 ptrs but 4 byte size) or 24 (for 8/8) additional bytes are
96 needed; 4 (8) for a trailing size field
97 and 8 (16) bytes for free list pointers. Thus, the minimum
98 allocatable size is 16/24/32 bytes.
100 Even a request for zero bytes (i.e., malloc(0)) returns a
101 pointer to something of the minimum allocatable size.
103 Maximum allocated size: 4-byte size_t: 2^31 - 8 bytes
104 8-byte size_t: 2^63 - 16 bytes
106 It is assumed that (possibly signed) size_t bit values suffice to
107 represent chunk sizes. `Possibly signed' is due to the fact
108 that `size_t' may be defined on a system as either a signed or
109 an unsigned type. To be conservative, values that would appear
110 as negative numbers are avoided.
111 Requests for sizes with a negative sign bit when the request
112 size is treaded as a long will return null.
114 Maximum overhead wastage per allocated chunk: normally 15 bytes
116 Alignnment demands, plus the minimum allocatable size restriction
117 make the normal worst-case wastage 15 bytes (i.e., up to 15
118 more bytes will be allocated than were requested in malloc), with
120 1. Because requests for zero bytes allocate non-zero space,
121 the worst case wastage for a request of zero bytes is 24 bytes.
122 2. For requests >= mmap_threshold that are serviced via
123 mmap(), the worst case wastage is 8 bytes plus the remainder
124 from a system page (the minimal mmap unit); typically 4096 bytes.
128 Here are some features that are NOT currently supported
130 * No user-definable hooks for callbacks and the like.
131 * No automated mechanism for fully checking that all accesses
132 to malloced memory stay within their bounds.
133 * No support for compaction.
135 * Synopsis of compile-time options:
137 People have reported using previous versions of this malloc on all
138 versions of Unix, sometimes by tweaking some of the defines
139 below. It has been tested most extensively on Solaris and
140 Linux. It is also reported to work on WIN32 platforms.
141 People have also reported adapting this malloc for use in
142 stand-alone embedded systems.
144 The implementation is in straight, hand-tuned ANSI C. Among other
145 consequences, it uses a lot of macros. Because of this, to be at
146 all usable, this code should be compiled using an optimizing compiler
147 (for example gcc -O2) that can simplify expressions and control
150 __STD_C (default: derived from C compiler defines)
151 Nonzero if using ANSI-standard C compiler, a C++ compiler, or
152 a C compiler sufficiently close to ANSI to get away with it.
153 DEBUG (default: NOT defined)
154 Define to enable debugging. Adds fairly extensive assertion-based
155 checking to help track down memory errors, but noticeably slows down
157 REALLOC_ZERO_BYTES_FREES (default: NOT defined)
158 Define this if you think that realloc(p, 0) should be equivalent
159 to free(p). Otherwise, since malloc returns a unique pointer for
160 malloc(0), so does realloc(p, 0).
161 HAVE_MEMCPY (default: defined)
162 Define if you are not otherwise using ANSI STD C, but still
163 have memcpy and memset in your C library and want to use them.
164 Otherwise, simple internal versions are supplied.
165 USE_MEMCPY (default: 1 if HAVE_MEMCPY is defined, 0 otherwise)
166 Define as 1 if you want the C library versions of memset and
167 memcpy called in realloc and calloc (otherwise macro versions are used).
168 At least on some platforms, the simple macro versions usually
169 outperform libc versions.
170 HAVE_MMAP (default: defined as 1)
171 Define to non-zero to optionally make malloc() use mmap() to
172 allocate very large blocks.
173 HAVE_MREMAP (default: defined as 0 unless Linux libc set)
174 Define to non-zero to optionally make realloc() use mremap() to
175 reallocate very large blocks.
176 malloc_getpagesize (default: derived from system #includes)
177 Either a constant or routine call returning the system page size.
178 HAVE_USR_INCLUDE_MALLOC_H (default: NOT defined)
179 Optionally define if you are on a system with a /usr/include/malloc.h
180 that declares struct mallinfo. It is not at all necessary to
181 define this even if you do, but will ensure consistency.
182 INTERNAL_SIZE_T (default: size_t)
183 Define to a 32-bit type (probably `unsigned int') if you are on a
184 64-bit machine, yet do not want or need to allow malloc requests of
185 greater than 2^31 to be handled. This saves space, especially for
187 INTERNAL_LINUX_C_LIB (default: NOT defined)
188 Defined only when compiled as part of Linux libc.
189 Also note that there is some odd internal name-mangling via defines
190 (for example, internally, `malloc' is named `mALLOc') needed
191 when compiling in this case. These look funny but don't otherwise
193 WIN32 (default: undefined)
194 Define this on MS win (95, nt) platforms to compile in sbrk emulation.
195 LACKS_UNISTD_H (default: undefined if not WIN32)
196 Define this if your system does not have a <unistd.h>.
197 LACKS_SYS_PARAM_H (default: undefined if not WIN32)
198 Define this if your system does not have a <sys/param.h>.
199 MORECORE (default: sbrk)
200 The name of the routine to call to obtain more memory from the system.
201 MORECORE_FAILURE (default: -1)
202 The value returned upon failure of MORECORE.
203 MORECORE_CLEARS (default 1)
204 True (1) if the routine mapped to MORECORE zeroes out memory (which
206 DEFAULT_TRIM_THRESHOLD
208 DEFAULT_MMAP_THRESHOLD
210 Default values of tunable parameters (described in detail below)
211 controlling interaction with host system routines (sbrk, mmap, etc).
212 These values may also be changed dynamically via mallopt(). The
213 preset defaults are those that give best performance for typical
215 USE_DL_PREFIX (default: undefined)
216 Prefix all public routines with the string 'dl'. Useful to
217 quickly avoid procedure declaration conflicts and linker symbol
218 conflicts with existing memory allocation routines.
235 #endif /*__cplusplus*/
240 #if (__STD_C || defined(WIN32))
248 #include <stddef.h> /* for size_t */
250 #include <sys/types.h>
257 #include <stdio.h> /* needed for malloc_stats */
268 Because freed chunks may be overwritten with link fields, this
269 malloc will often die when freed memory is overwritten by user
270 programs. This can be very effective (albeit in an annoying way)
271 in helping track down dangling pointers.
273 If you compile with -DDEBUG, a number of assertion checks are
274 enabled that will catch more memory errors. You probably won't be
275 able to make much sense of the actual assertion errors, but they
276 should help you locate incorrectly overwritten memory. The
277 checking is fairly extensive, and will slow down execution
278 noticeably. Calling malloc_stats or mallinfo with DEBUG set will
279 attempt to check every non-mmapped allocated and free chunk in the
280 course of computing the summmaries. (By nature, mmapped regions
281 cannot be checked very much automatically.)
283 Setting DEBUG may also be helpful if you are trying to modify
284 this code. The assertions in the check routines spell out in more
285 detail the assumptions and invariants underlying the algorithms.
292 #define assert(x) ((void)0)
297 INTERNAL_SIZE_T is the word-size used for internal bookkeeping
298 of chunk sizes. On a 64-bit machine, you can reduce malloc
299 overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int'
300 at the expense of not being able to handle requests greater than
301 2^31. This limitation is hardly ever a concern; you are encouraged
302 to set this. However, the default version is the same as size_t.
305 #ifndef INTERNAL_SIZE_T
306 #define INTERNAL_SIZE_T size_t
310 REALLOC_ZERO_BYTES_FREES should be set if a call to
311 realloc with zero bytes should be the same as a call to free.
312 Some people think it should. Otherwise, since this malloc
313 returns a unique pointer for malloc(0), so does realloc(p, 0).
317 /* #define REALLOC_ZERO_BYTES_FREES */
321 WIN32 causes an emulation of sbrk to be compiled in
322 mmap-based options are not currently supported in WIN32.
327 #define MORECORE wsbrk
330 #define LACKS_UNISTD_H
331 #define LACKS_SYS_PARAM_H
334 Include 'windows.h' to get the necessary declarations for the
335 Microsoft Visual C++ data structures and routines used in the 'sbrk'
338 Define WIN32_LEAN_AND_MEAN so that only the essential Microsoft
339 Visual C++ header files are included.
341 #define WIN32_LEAN_AND_MEAN
347 HAVE_MEMCPY should be defined if you are not otherwise using
348 ANSI STD C, but still have memcpy and memset in your C library
349 and want to use them in calloc and realloc. Otherwise simple
350 macro versions are defined here.
352 USE_MEMCPY should be defined as 1 if you actually want to
353 have memset and memcpy called. People report that the macro
354 versions are often enough faster than libc versions on many
355 systems that it is better to use them.
369 #if (__STD_C || defined(HAVE_MEMCPY))
372 void* memset(void*, int, size_t);
373 void* memcpy(void*, const void*, size_t);
376 /* On Win32 platforms, 'memset()' and 'memcpy()' are already declared in */
387 /* The following macros are only invoked with (2n+1)-multiples of
388 INTERNAL_SIZE_T units, with a positive integer n. This is exploited
389 for fast inline execution when n is small. */
391 #define MALLOC_ZERO(charp, nbytes) \
393 INTERNAL_SIZE_T mzsz = (nbytes); \
394 if(mzsz <= 9*sizeof(mzsz)) { \
395 INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp); \
396 if(mzsz >= 5*sizeof(mzsz)) { *mz++ = 0; \
398 if(mzsz >= 7*sizeof(mzsz)) { *mz++ = 0; \
400 if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0; \
405 } else memset((charp), 0, mzsz); \
408 #define MALLOC_COPY(dest,src,nbytes) \
410 INTERNAL_SIZE_T mcsz = (nbytes); \
411 if(mcsz <= 9*sizeof(mcsz)) { \
412 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src); \
413 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest); \
414 if(mcsz >= 5*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
415 *mcdst++ = *mcsrc++; \
416 if(mcsz >= 7*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
417 *mcdst++ = *mcsrc++; \
418 if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
419 *mcdst++ = *mcsrc++; }}} \
420 *mcdst++ = *mcsrc++; \
421 *mcdst++ = *mcsrc++; \
423 } else memcpy(dest, src, mcsz); \
426 #else /* !USE_MEMCPY */
428 /* Use Duff's device for good zeroing/copying performance. */
430 #define MALLOC_ZERO(charp, nbytes) \
432 INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp); \
433 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
434 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
436 case 0: for(;;) { *mzp++ = 0; \
437 case 7: *mzp++ = 0; \
438 case 6: *mzp++ = 0; \
439 case 5: *mzp++ = 0; \
440 case 4: *mzp++ = 0; \
441 case 3: *mzp++ = 0; \
442 case 2: *mzp++ = 0; \
443 case 1: *mzp++ = 0; if(mcn <= 0) break; mcn--; } \
447 #define MALLOC_COPY(dest,src,nbytes) \
449 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src; \
450 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest; \
451 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
452 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
454 case 0: for(;;) { *mcdst++ = *mcsrc++; \
455 case 7: *mcdst++ = *mcsrc++; \
456 case 6: *mcdst++ = *mcsrc++; \
457 case 5: *mcdst++ = *mcsrc++; \
458 case 4: *mcdst++ = *mcsrc++; \
459 case 3: *mcdst++ = *mcsrc++; \
460 case 2: *mcdst++ = *mcsrc++; \
461 case 1: *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; } \
469 Define HAVE_MMAP to optionally make malloc() use mmap() to
470 allocate very large blocks. These will be returned to the
471 operating system immediately after a free().
479 Define HAVE_MREMAP to make realloc() use mremap() to re-allocate
480 large blocks. This is currently only possible on Linux with
481 kernel versions newer than 1.3.77.
485 #ifdef INTERNAL_LINUX_C_LIB
486 #define HAVE_MREMAP 1
488 #define HAVE_MREMAP 0
496 #include <sys/mman.h>
498 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
499 #define MAP_ANONYMOUS MAP_ANON
502 #endif /* HAVE_MMAP */
505 Access to system page size. To the extent possible, this malloc
506 manages memory from the system in page-size units.
508 The following mechanics for getpagesize were adapted from
509 bsd/gnu getpagesize.h
512 #ifndef LACKS_UNISTD_H
516 #ifndef malloc_getpagesize
517 # ifdef _SC_PAGESIZE /* some SVR4 systems omit an underscore */
518 # ifndef _SC_PAGE_SIZE
519 # define _SC_PAGE_SIZE _SC_PAGESIZE
522 # ifdef _SC_PAGE_SIZE
523 # define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
525 # if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
526 extern size_t getpagesize();
527 # define malloc_getpagesize getpagesize()
530 # define malloc_getpagesize (4096) /* TBD: Use 'GetSystemInfo' instead */
532 # ifndef LACKS_SYS_PARAM_H
533 # include <sys/param.h>
535 # ifdef EXEC_PAGESIZE
536 # define malloc_getpagesize EXEC_PAGESIZE
540 # define malloc_getpagesize NBPG
542 # define malloc_getpagesize (NBPG * CLSIZE)
546 # define malloc_getpagesize NBPC
549 # define malloc_getpagesize PAGESIZE
551 # define malloc_getpagesize (4096) /* just guess */
564 This version of malloc supports the standard SVID/XPG mallinfo
565 routine that returns a struct containing the same kind of
566 information you can get from malloc_stats. It should work on
567 any SVID/XPG compliant system that has a /usr/include/malloc.h
568 defining struct mallinfo. (If you'd like to install such a thing
569 yourself, cut out the preliminary declarations as described above
570 and below and save them in a malloc.h file. But there's no
571 compelling reason to bother to do this.)
573 The main declaration needed is the mallinfo struct that is returned
574 (by-copy) by mallinfo(). The SVID/XPG malloinfo struct contains a
575 bunch of fields, most of which are not even meaningful in this
576 version of malloc. Some of these fields are are instead filled by
577 mallinfo() with other numbers that might possibly be of interest.
579 HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
580 /usr/include/malloc.h file that includes a declaration of struct
581 mallinfo. If so, it is included; else an SVID2/XPG2 compliant
582 version is declared below. These must be precisely the same for
587 /* #define HAVE_USR_INCLUDE_MALLOC_H */
589 #if HAVE_USR_INCLUDE_MALLOC_H
590 #include "/usr/include/malloc.h"
593 /* SVID2/XPG mallinfo structure */
596 int arena; /* total space allocated from system */
597 int ordblks; /* number of non-inuse chunks */
598 int smblks; /* unused -- always zero */
599 int hblks; /* number of mmapped regions */
600 int hblkhd; /* total space in mmapped regions */
601 int usmblks; /* unused -- always zero */
602 int fsmblks; /* unused -- always zero */
603 int uordblks; /* total allocated space */
604 int fordblks; /* total non-inuse space */
605 int keepcost; /* top-most, releasable (via malloc_trim) space */
608 /* SVID2/XPG mallopt options */
610 #define M_MXFAST 1 /* UNUSED in this malloc */
611 #define M_NLBLKS 2 /* UNUSED in this malloc */
612 #define M_GRAIN 3 /* UNUSED in this malloc */
613 #define M_KEEP 4 /* UNUSED in this malloc */
617 /* mallopt options that actually do something */
619 #define M_TRIM_THRESHOLD -1
621 #define M_MMAP_THRESHOLD -3
622 #define M_MMAP_MAX -4
625 #ifndef DEFAULT_TRIM_THRESHOLD
626 #define DEFAULT_TRIM_THRESHOLD (128 * 1024)
630 M_TRIM_THRESHOLD is the maximum amount of unused top-most memory
631 to keep before releasing via malloc_trim in free().
633 Automatic trimming is mainly useful in long-lived programs.
634 Because trimming via sbrk can be slow on some systems, and can
635 sometimes be wasteful (in cases where programs immediately
636 afterward allocate more large chunks) the value should be high
637 enough so that your overall system performance would improve by
640 The trim threshold and the mmap control parameters (see below)
641 can be traded off with one another. Trimming and mmapping are
642 two different ways of releasing unused memory back to the
643 system. Between these two, it is often possible to keep
644 system-level demands of a long-lived program down to a bare
645 minimum. For example, in one test suite of sessions measuring
646 the XF86 X server on Linux, using a trim threshold of 128K and a
647 mmap threshold of 192K led to near-minimal long term resource
650 If you are using this malloc in a long-lived program, it should
651 pay to experiment with these values. As a rough guide, you
652 might set to a value close to the average size of a process
653 (program) running on your system. Releasing this much memory
654 would allow such a process to run in memory. Generally, it's
655 worth it to tune for trimming rather tham memory mapping when a
656 program undergoes phases where several large chunks are
657 allocated and released in ways that can reuse each other's
658 storage, perhaps mixed with phases where there are no such
659 chunks at all. And in well-behaved long-lived programs,
660 controlling release of large blocks via trimming versus mapping
663 However, in most programs, these parameters serve mainly as
664 protection against the system-level effects of carrying around
665 massive amounts of unneeded memory. Since frequent calls to
666 sbrk, mmap, and munmap otherwise degrade performance, the default
667 parameters are set to relatively high values that serve only as
670 The default trim value is high enough to cause trimming only in
671 fairly extreme (by current memory consumption standards) cases.
672 It must be greater than page size to have any useful effect. To
673 disable trimming completely, you can set to (unsigned long)(-1);
679 #ifndef DEFAULT_TOP_PAD
680 #define DEFAULT_TOP_PAD (0)
684 M_TOP_PAD is the amount of extra `padding' space to allocate or
685 retain whenever sbrk is called. It is used in two ways internally:
687 * When sbrk is called to extend the top of the arena to satisfy
688 a new malloc request, this much padding is added to the sbrk
691 * When malloc_trim is called automatically from free(),
692 it is used as the `pad' argument.
694 In both cases, the actual amount of padding is rounded
695 so that the end of the arena is always a system page boundary.
697 The main reason for using padding is to avoid calling sbrk so
698 often. Having even a small pad greatly reduces the likelihood
699 that nearly every malloc request during program start-up (or
700 after trimming) will invoke sbrk, which needlessly wastes
703 Automatic rounding-up to page-size units is normally sufficient
704 to avoid measurable overhead, so the default is 0. However, in
705 systems where sbrk is relatively slow, it can pay to increase
706 this value, at the expense of carrying around more memory than
712 #ifndef DEFAULT_MMAP_THRESHOLD
713 #define DEFAULT_MMAP_THRESHOLD (128 * 1024)
718 M_MMAP_THRESHOLD is the request size threshold for using mmap()
719 to service a request. Requests of at least this size that cannot
720 be allocated using already-existing space will be serviced via mmap.
721 (If enough normal freed space already exists it is used instead.)
723 Using mmap segregates relatively large chunks of memory so that
724 they can be individually obtained and released from the host
725 system. A request serviced through mmap is never reused by any
726 other request (at least not directly; the system may just so
727 happen to remap successive requests to the same locations).
729 Segregating space in this way has the benefit that mmapped space
730 can ALWAYS be individually released back to the system, which
731 helps keep the system level memory demands of a long-lived
732 program low. Mapped memory can never become `locked' between
733 other chunks, as can happen with normally allocated chunks, which
734 menas that even trimming via malloc_trim would not release them.
736 However, it has the disadvantages that:
738 1. The space cannot be reclaimed, consolidated, and then
739 used to service later requests, as happens with normal chunks.
740 2. It can lead to more wastage because of mmap page alignment
742 3. It causes malloc performance to be more dependent on host
743 system memory management support routines which may vary in
744 implementation quality and may impose arbitrary
745 limitations. Generally, servicing a request via normal
746 malloc steps is faster than going through a system's mmap.
748 All together, these considerations should lead you to use mmap
749 only for relatively large requests.
755 #ifndef DEFAULT_MMAP_MAX
757 #define DEFAULT_MMAP_MAX (64)
759 #define DEFAULT_MMAP_MAX (0)
764 M_MMAP_MAX is the maximum number of requests to simultaneously
765 service using mmap. This parameter exists because:
767 1. Some systems have a limited number of internal tables for
769 2. In most systems, overreliance on mmap can degrade overall
771 3. If a program allocates many large regions, it is probably
772 better off using normal sbrk-based allocation routines that
773 can reclaim and reallocate normal heap memory. Using a
774 small value allows transition into this mode after the
775 first few allocations.
777 Setting to 0 disables all use of mmap. If HAVE_MMAP is not set,
778 the default value is 0, and attempts to set it to non-zero values
779 in mallopt will fail.
784 USE_DL_PREFIX will prefix all public routines with the string 'dl'.
785 Useful to quickly avoid procedure declaration conflicts and linker
786 symbol conflicts with existing memory allocation routines.
790 /* #define USE_DL_PREFIX */
795 Special defines for linux libc
797 Except when compiled using these special defines for Linux libc
798 using weak aliases, this malloc is NOT designed to work in
799 multithreaded applications. No semaphores or other concurrency
800 control are provided to ensure that multiple malloc or free calls
801 don't run at the same time, which could be disasterous. A single
802 semaphore could be used across malloc, realloc, and free (which is
803 essentially the effect of the linux weak alias approach). It would
804 be hard to obtain finer granularity.
809 #ifdef INTERNAL_LINUX_C_LIB
813 Void_t * __default_morecore_init (ptrdiff_t);
814 Void_t *(*__morecore)(ptrdiff_t) = __default_morecore_init;
818 Void_t * __default_morecore_init ();
819 Void_t *(*__morecore)() = __default_morecore_init;
823 #define MORECORE (*__morecore)
824 #define MORECORE_FAILURE 0
825 #define MORECORE_CLEARS 1
827 #else /* INTERNAL_LINUX_C_LIB */
830 extern Void_t* sbrk(ptrdiff_t);
832 extern Void_t* sbrk();
836 #define MORECORE sbrk
839 #ifndef MORECORE_FAILURE
840 #define MORECORE_FAILURE -1
843 #ifndef MORECORE_CLEARS
844 #define MORECORE_CLEARS 1
847 #endif /* INTERNAL_LINUX_C_LIB */
849 #if defined(INTERNAL_LINUX_C_LIB) && defined(__ELF__)
851 #define cALLOc __libc_calloc
852 #define fREe __libc_free
853 #define mALLOc __libc_malloc
854 #define mEMALIGn __libc_memalign
855 #define rEALLOc __libc_realloc
856 #define vALLOc __libc_valloc
857 #define pvALLOc __libc_pvalloc
858 #define mALLINFo __libc_mallinfo
859 #define mALLOPt __libc_mallopt
861 #pragma weak calloc = __libc_calloc
862 #pragma weak free = __libc_free
863 #pragma weak cfree = __libc_free
864 #pragma weak malloc = __libc_malloc
865 #pragma weak memalign = __libc_memalign
866 #pragma weak realloc = __libc_realloc
867 #pragma weak valloc = __libc_valloc
868 #pragma weak pvalloc = __libc_pvalloc
869 #pragma weak mallinfo = __libc_mallinfo
870 #pragma weak mallopt = __libc_mallopt
875 #define cALLOc dlcalloc
877 #define mALLOc dlmalloc
878 #define mEMALIGn dlmemalign
879 #define rEALLOc dlrealloc
880 #define vALLOc dlvalloc
881 #define pvALLOc dlpvalloc
882 #define mALLINFo dlmallinfo
883 #define mALLOPt dlmallopt
884 #else /* USE_DL_PREFIX */
885 #define cALLOc calloc
887 #define mALLOc malloc
888 #define mEMALIGn memalign
889 #define rEALLOc realloc
890 #define vALLOc valloc
891 #define pvALLOc pvalloc
892 #define mALLINFo mallinfo
893 #define mALLOPt mallopt
894 #endif /* USE_DL_PREFIX */
898 /* Public routines */
902 Void_t* mALLOc(size_t);
904 Void_t* rEALLOc(Void_t*, size_t);
905 Void_t* mEMALIGn(size_t, size_t);
906 Void_t* vALLOc(size_t);
907 Void_t* pvALLOc(size_t);
908 Void_t* cALLOc(size_t, size_t);
910 int malloc_trim(size_t);
911 size_t malloc_usable_size(Void_t*);
913 int mALLOPt(int, int);
914 struct mallinfo mALLINFo(void);
925 size_t malloc_usable_size();
928 struct mallinfo mALLINFo();
933 }; /* end of extern "C" */
936 /* ---------- To make a malloc.h, end cutting here ------------ */
937 #endif /* 0 */ /* Moved to malloc.h */
942 static void malloc_update_mallinfo (void);
943 void malloc_stats (void);
945 static void malloc_update_mallinfo ();
950 DECLARE_GLOBAL_DATA_PTR;
953 Emulation of sbrk for WIN32
954 All code within the ifdef WIN32 is untested by me.
956 Thanks to Martin Fong and others for supplying this.
962 #define AlignPage(add) (((add) + (malloc_getpagesize-1)) & \
963 ~(malloc_getpagesize-1))
964 #define AlignPage64K(add) (((add) + (0x10000 - 1)) & ~(0x10000 - 1))
966 /* resrve 64MB to insure large contiguous space */
967 #define RESERVED_SIZE (1024*1024*64)
968 #define NEXT_SIZE (2048*1024)
969 #define TOP_MEMORY ((unsigned long)2*1024*1024*1024)
971 struct GmListElement;
972 typedef struct GmListElement GmListElement;
980 static GmListElement* head = 0;
981 static unsigned int gNextAddress = 0;
982 static unsigned int gAddressBase = 0;
983 static unsigned int gAllocatedSize = 0;
986 GmListElement* makeGmListElement (void* bas)
989 this = (GmListElement*)(void*)LocalAlloc (0, sizeof (GmListElement));
1003 assert ( (head == NULL) || (head->base == (void*)gAddressBase));
1004 if (gAddressBase && (gNextAddress - gAddressBase))
1006 rval = VirtualFree ((void*)gAddressBase,
1007 gNextAddress - gAddressBase,
1013 GmListElement* next = head->next;
1014 rval = VirtualFree (head->base, 0, MEM_RELEASE);
1022 void* findRegion (void* start_address, unsigned long size)
1024 MEMORY_BASIC_INFORMATION info;
1025 if (size >= TOP_MEMORY) return NULL;
1027 while ((unsigned long)start_address + size < TOP_MEMORY)
1029 VirtualQuery (start_address, &info, sizeof (info));
1030 if ((info.State == MEM_FREE) && (info.RegionSize >= size))
1031 return start_address;
1034 /* Requested region is not available so see if the */
1035 /* next region is available. Set 'start_address' */
1036 /* to the next region and call 'VirtualQuery()' */
1039 start_address = (char*)info.BaseAddress + info.RegionSize;
1041 /* Make sure we start looking for the next region */
1042 /* on the *next* 64K boundary. Otherwise, even if */
1043 /* the new region is free according to */
1044 /* 'VirtualQuery()', the subsequent call to */
1045 /* 'VirtualAlloc()' (which follows the call to */
1046 /* this routine in 'wsbrk()') will round *down* */
1047 /* the requested address to a 64K boundary which */
1048 /* we already know is an address in the */
1049 /* unavailable region. Thus, the subsequent call */
1050 /* to 'VirtualAlloc()' will fail and bring us back */
1051 /* here, causing us to go into an infinite loop. */
1054 (void *) AlignPage64K((unsigned long) start_address);
1062 void* wsbrk (long size)
1067 if (gAddressBase == 0)
1069 gAllocatedSize = max (RESERVED_SIZE, AlignPage (size));
1070 gNextAddress = gAddressBase =
1071 (unsigned int)VirtualAlloc (NULL, gAllocatedSize,
1072 MEM_RESERVE, PAGE_NOACCESS);
1073 } else if (AlignPage (gNextAddress + size) > (gAddressBase +
1076 long new_size = max (NEXT_SIZE, AlignPage (size));
1077 void* new_address = (void*)(gAddressBase+gAllocatedSize);
1080 new_address = findRegion (new_address, new_size);
1082 if (new_address == 0)
1085 gAddressBase = gNextAddress =
1086 (unsigned int)VirtualAlloc (new_address, new_size,
1087 MEM_RESERVE, PAGE_NOACCESS);
1088 /* repeat in case of race condition */
1089 /* The region that we found has been snagged */
1090 /* by another thread */
1092 while (gAddressBase == 0);
1094 assert (new_address == (void*)gAddressBase);
1096 gAllocatedSize = new_size;
1098 if (!makeGmListElement ((void*)gAddressBase))
1101 if ((size + gNextAddress) > AlignPage (gNextAddress))
1104 res = VirtualAlloc ((void*)AlignPage (gNextAddress),
1105 (size + gNextAddress -
1106 AlignPage (gNextAddress)),
1107 MEM_COMMIT, PAGE_READWRITE);
1111 tmp = (void*)gNextAddress;
1112 gNextAddress = (unsigned int)tmp + size;
1117 unsigned int alignedGoal = AlignPage (gNextAddress + size);
1118 /* Trim by releasing the virtual memory */
1119 if (alignedGoal >= gAddressBase)
1121 VirtualFree ((void*)alignedGoal, gNextAddress - alignedGoal,
1123 gNextAddress = gNextAddress + size;
1124 return (void*)gNextAddress;
1128 VirtualFree ((void*)gAddressBase, gNextAddress - gAddressBase,
1130 gNextAddress = gAddressBase;
1136 return (void*)gNextAddress;
1151 INTERNAL_SIZE_T prev_size; /* Size of previous chunk (if free). */
1152 INTERNAL_SIZE_T size; /* Size in bytes, including overhead. */
1153 struct malloc_chunk* fd; /* double links -- used only if free. */
1154 struct malloc_chunk* bk;
1155 } __attribute__((__may_alias__)) ;
1157 typedef struct malloc_chunk* mchunkptr;
1161 malloc_chunk details:
1163 (The following includes lightly edited explanations by Colin Plumb.)
1165 Chunks of memory are maintained using a `boundary tag' method as
1166 described in e.g., Knuth or Standish. (See the paper by Paul
1167 Wilson ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a
1168 survey of such techniques.) Sizes of free chunks are stored both
1169 in the front of each chunk and at the end. This makes
1170 consolidating fragmented chunks into bigger chunks very fast. The
1171 size fields also hold bits representing whether chunks are free or
1174 An allocated chunk looks like this:
1177 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1178 | Size of previous chunk, if allocated | |
1179 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1180 | Size of chunk, in bytes |P|
1181 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1182 | User data starts here... .
1184 . (malloc_usable_space() bytes) .
1186 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1188 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1191 Where "chunk" is the front of the chunk for the purpose of most of
1192 the malloc code, but "mem" is the pointer that is returned to the
1193 user. "Nextchunk" is the beginning of the next contiguous chunk.
1195 Chunks always begin on even word boundries, so the mem portion
1196 (which is returned to the user) is also on an even word boundary, and
1197 thus double-word aligned.
1199 Free chunks are stored in circular doubly-linked lists, and look like this:
1201 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1202 | Size of previous chunk |
1203 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1204 `head:' | Size of chunk, in bytes |P|
1205 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1206 | Forward pointer to next chunk in list |
1207 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1208 | Back pointer to previous chunk in list |
1209 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1210 | Unused space (may be 0 bytes long) .
1213 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1214 `foot:' | Size of chunk, in bytes |
1215 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1217 The P (PREV_INUSE) bit, stored in the unused low-order bit of the
1218 chunk size (which is always a multiple of two words), is an in-use
1219 bit for the *previous* chunk. If that bit is *clear*, then the
1220 word before the current chunk size contains the previous chunk
1221 size, and can be used to find the front of the previous chunk.
1222 (The very first chunk allocated always has this bit set,
1223 preventing access to non-existent (or non-owned) memory.)
1225 Note that the `foot' of the current chunk is actually represented
1226 as the prev_size of the NEXT chunk. (This makes it easier to
1227 deal with alignments etc).
1229 The two exceptions to all this are
1231 1. The special chunk `top', which doesn't bother using the
1232 trailing size field since there is no
1233 next contiguous chunk that would have to index off it. (After
1234 initialization, `top' is forced to always exist. If it would
1235 become less than MINSIZE bytes long, it is replenished via
1238 2. Chunks allocated via mmap, which have the second-lowest-order
1239 bit (IS_MMAPPED) set in their size fields. Because they are
1240 never merged or traversed from any other chunk, they have no
1241 foot size or inuse information.
1243 Available chunks are kept in any of several places (all declared below):
1245 * `av': An array of chunks serving as bin headers for consolidated
1246 chunks. Each bin is doubly linked. The bins are approximately
1247 proportionally (log) spaced. There are a lot of these bins
1248 (128). This may look excessive, but works very well in
1249 practice. All procedures maintain the invariant that no
1250 consolidated chunk physically borders another one. Chunks in
1251 bins are kept in size order, with ties going to the
1252 approximately least recently used chunk.
1254 The chunks in each bin are maintained in decreasing sorted order by
1255 size. This is irrelevant for the small bins, which all contain
1256 the same-sized chunks, but facilitates best-fit allocation for
1257 larger chunks. (These lists are just sequential. Keeping them in
1258 order almost never requires enough traversal to warrant using
1259 fancier ordered data structures.) Chunks of the same size are
1260 linked with the most recently freed at the front, and allocations
1261 are taken from the back. This results in LRU or FIFO allocation
1262 order, which tends to give each chunk an equal opportunity to be
1263 consolidated with adjacent freed chunks, resulting in larger free
1264 chunks and less fragmentation.
1266 * `top': The top-most available chunk (i.e., the one bordering the
1267 end of available memory) is treated specially. It is never
1268 included in any bin, is used only if no other chunk is
1269 available, and is released back to the system if it is very
1270 large (see M_TRIM_THRESHOLD).
1272 * `last_remainder': A bin holding only the remainder of the
1273 most recently split (non-top) chunk. This bin is checked
1274 before other non-fitting chunks, so as to provide better
1275 locality for runs of sequentially allocated chunks.
1277 * Implicitly, through the host system's memory mapping tables.
1278 If supported, requests greater than a threshold are usually
1279 serviced via calls to mmap, and then later released via munmap.
1283 /* sizes, alignments */
1285 #define SIZE_SZ (sizeof(INTERNAL_SIZE_T))
1286 #define MALLOC_ALIGNMENT (SIZE_SZ + SIZE_SZ)
1287 #define MALLOC_ALIGN_MASK (MALLOC_ALIGNMENT - 1)
1288 #define MINSIZE (sizeof(struct malloc_chunk))
1290 /* conversion from malloc headers to user pointers, and back */
1292 #define chunk2mem(p) ((Void_t*)((char*)(p) + 2*SIZE_SZ))
1293 #define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ))
1295 /* pad request bytes into a usable size */
1297 #define request2size(req) \
1298 (((long)((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) < \
1299 (long)(MINSIZE + MALLOC_ALIGN_MASK)) ? MINSIZE : \
1300 (((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) & ~(MALLOC_ALIGN_MASK)))
1302 /* Check if m has acceptable alignment */
1304 #define aligned_OK(m) (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
1310 Physical chunk operations
1314 /* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
1316 #define PREV_INUSE 0x1
1318 /* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
1320 #define IS_MMAPPED 0x2
1322 /* Bits to mask off when extracting size */
1324 #define SIZE_BITS (PREV_INUSE|IS_MMAPPED)
1327 /* Ptr to next physical malloc_chunk. */
1329 #define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~PREV_INUSE) ))
1331 /* Ptr to previous physical malloc_chunk */
1333 #define prev_chunk(p)\
1334 ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) ))
1337 /* Treat space at ptr + offset as a chunk */
1339 #define chunk_at_offset(p, s) ((mchunkptr)(((char*)(p)) + (s)))
1345 Dealing with use bits
1348 /* extract p's inuse bit */
1351 ((((mchunkptr)(((char*)(p))+((p)->size & ~PREV_INUSE)))->size) & PREV_INUSE)
1353 /* extract inuse bit of previous chunk */
1355 #define prev_inuse(p) ((p)->size & PREV_INUSE)
1357 /* check for mmap()'ed chunk */
1359 #define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
1361 /* set/clear chunk as in use without otherwise disturbing */
1363 #define set_inuse(p)\
1364 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size |= PREV_INUSE
1366 #define clear_inuse(p)\
1367 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size &= ~(PREV_INUSE)
1369 /* check/set/clear inuse bits in known places */
1371 #define inuse_bit_at_offset(p, s)\
1372 (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE)
1374 #define set_inuse_bit_at_offset(p, s)\
1375 (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE)
1377 #define clear_inuse_bit_at_offset(p, s)\
1378 (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE))
1384 Dealing with size fields
1387 /* Get size, ignoring use bits */
1389 #define chunksize(p) ((p)->size & ~(SIZE_BITS))
1391 /* Set size at head, without disturbing its use bit */
1393 #define set_head_size(p, s) ((p)->size = (((p)->size & PREV_INUSE) | (s)))
1395 /* Set size/use ignoring previous bits in header */
1397 #define set_head(p, s) ((p)->size = (s))
1399 /* Set size at footer (only when chunk is not in use) */
1401 #define set_foot(p, s) (((mchunkptr)((char*)(p) + (s)))->prev_size = (s))
1410 The bins, `av_' are an array of pairs of pointers serving as the
1411 heads of (initially empty) doubly-linked lists of chunks, laid out
1412 in a way so that each pair can be treated as if it were in a
1413 malloc_chunk. (This way, the fd/bk offsets for linking bin heads
1414 and chunks are the same).
1416 Bins for sizes < 512 bytes contain chunks of all the same size, spaced
1417 8 bytes apart. Larger bins are approximately logarithmically
1418 spaced. (See the table below.) The `av_' array is never mentioned
1419 directly in the code, but instead via bin access macros.
1427 4 bins of size 32768
1428 2 bins of size 262144
1429 1 bin of size what's left
1431 There is actually a little bit of slop in the numbers in bin_index
1432 for the sake of speed. This makes no difference elsewhere.
1434 The special chunks `top' and `last_remainder' get their own bins,
1435 (this is implemented via yet more trickery with the av_ array),
1436 although `top' is never properly linked to its bin since it is
1437 always handled specially.
1441 #define NAV 128 /* number of bins */
1443 typedef struct malloc_chunk* mbinptr;
1447 #define bin_at(i) ((mbinptr)((char*)&(av_[2*(i) + 2]) - 2*SIZE_SZ))
1448 #define next_bin(b) ((mbinptr)((char*)(b) + 2 * sizeof(mbinptr)))
1449 #define prev_bin(b) ((mbinptr)((char*)(b) - 2 * sizeof(mbinptr)))
1452 The first 2 bins are never indexed. The corresponding av_ cells are instead
1453 used for bookkeeping. This is not to save space, but to simplify
1454 indexing, maintain locality, and avoid some initialization tests.
1457 #define top (av_[2]) /* The topmost chunk */
1458 #define last_remainder (bin_at(1)) /* remainder from last split */
1462 Because top initially points to its own bin with initial
1463 zero size, thus forcing extension on the first malloc request,
1464 we avoid having any special code in malloc to check whether
1465 it even exists yet. But we still need to in malloc_extend_top.
1468 #define initial_top ((mchunkptr)(bin_at(0)))
1470 /* Helper macro to initialize bins */
1472 #define IAV(i) bin_at(i), bin_at(i)
1474 static mbinptr av_[NAV * 2 + 2] = {
1476 IAV(0), IAV(1), IAV(2), IAV(3), IAV(4), IAV(5), IAV(6), IAV(7),
1477 IAV(8), IAV(9), IAV(10), IAV(11), IAV(12), IAV(13), IAV(14), IAV(15),
1478 IAV(16), IAV(17), IAV(18), IAV(19), IAV(20), IAV(21), IAV(22), IAV(23),
1479 IAV(24), IAV(25), IAV(26), IAV(27), IAV(28), IAV(29), IAV(30), IAV(31),
1480 IAV(32), IAV(33), IAV(34), IAV(35), IAV(36), IAV(37), IAV(38), IAV(39),
1481 IAV(40), IAV(41), IAV(42), IAV(43), IAV(44), IAV(45), IAV(46), IAV(47),
1482 IAV(48), IAV(49), IAV(50), IAV(51), IAV(52), IAV(53), IAV(54), IAV(55),
1483 IAV(56), IAV(57), IAV(58), IAV(59), IAV(60), IAV(61), IAV(62), IAV(63),
1484 IAV(64), IAV(65), IAV(66), IAV(67), IAV(68), IAV(69), IAV(70), IAV(71),
1485 IAV(72), IAV(73), IAV(74), IAV(75), IAV(76), IAV(77), IAV(78), IAV(79),
1486 IAV(80), IAV(81), IAV(82), IAV(83), IAV(84), IAV(85), IAV(86), IAV(87),
1487 IAV(88), IAV(89), IAV(90), IAV(91), IAV(92), IAV(93), IAV(94), IAV(95),
1488 IAV(96), IAV(97), IAV(98), IAV(99), IAV(100), IAV(101), IAV(102), IAV(103),
1489 IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111),
1490 IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119),
1491 IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127)
1494 #ifndef CONFIG_RELOC_FIXUP_WORKS
1495 void malloc_bin_reloc (void)
1497 unsigned long *p = (unsigned long *)(&av_[2]);
1499 for (i=2; i<(sizeof(av_)/sizeof(mbinptr)); ++i) {
1500 *p++ += gd->reloc_off;
1505 ulong mem_malloc_start = 0;
1506 ulong mem_malloc_end = 0;
1507 ulong mem_malloc_brk = 0;
1509 void *sbrk(ptrdiff_t increment)
1511 ulong old = mem_malloc_brk;
1512 ulong new = old + increment;
1514 if ((new < mem_malloc_start) || (new > mem_malloc_end))
1515 return (void *)MORECORE_FAILURE;
1517 mem_malloc_brk = new;
1522 void mem_malloc_init(ulong start, ulong size)
1524 mem_malloc_start = start;
1525 mem_malloc_end = start + size;
1526 mem_malloc_brk = start;
1528 memset((void *)mem_malloc_start, 0, size);
1531 /* field-extraction macros */
1533 #define first(b) ((b)->fd)
1534 #define last(b) ((b)->bk)
1540 #define bin_index(sz) \
1541 (((((unsigned long)(sz)) >> 9) == 0) ? (((unsigned long)(sz)) >> 3): \
1542 ((((unsigned long)(sz)) >> 9) <= 4) ? 56 + (((unsigned long)(sz)) >> 6): \
1543 ((((unsigned long)(sz)) >> 9) <= 20) ? 91 + (((unsigned long)(sz)) >> 9): \
1544 ((((unsigned long)(sz)) >> 9) <= 84) ? 110 + (((unsigned long)(sz)) >> 12): \
1545 ((((unsigned long)(sz)) >> 9) <= 340) ? 119 + (((unsigned long)(sz)) >> 15): \
1546 ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18): \
1549 bins for chunks < 512 are all spaced 8 bytes apart, and hold
1550 identically sized chunks. This is exploited in malloc.
1553 #define MAX_SMALLBIN 63
1554 #define MAX_SMALLBIN_SIZE 512
1555 #define SMALLBIN_WIDTH 8
1557 #define smallbin_index(sz) (((unsigned long)(sz)) >> 3)
1560 Requests are `small' if both the corresponding and the next bin are small
1563 #define is_small_request(nb) (nb < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH)
1568 To help compensate for the large number of bins, a one-level index
1569 structure is used for bin-by-bin searching. `binblocks' is a
1570 one-word bitvector recording whether groups of BINBLOCKWIDTH bins
1571 have any (possibly) non-empty bins, so they can be skipped over
1572 all at once during during traversals. The bits are NOT always
1573 cleared as soon as all bins in a block are empty, but instead only
1574 when all are noticed to be empty during traversal in malloc.
1577 #define BINBLOCKWIDTH 4 /* bins per block */
1579 #define binblocks_r ((INTERNAL_SIZE_T)av_[1]) /* bitvector of nonempty blocks */
1580 #define binblocks_w (av_[1])
1582 /* bin<->block macros */
1584 #define idx2binblock(ix) ((unsigned)1 << (ix / BINBLOCKWIDTH))
1585 #define mark_binblock(ii) (binblocks_w = (mbinptr)(binblocks_r | idx2binblock(ii)))
1586 #define clear_binblock(ii) (binblocks_w = (mbinptr)(binblocks_r & ~(idx2binblock(ii))))
1592 /* Other static bookkeeping data */
1594 /* variables holding tunable values */
1596 static unsigned long trim_threshold = DEFAULT_TRIM_THRESHOLD;
1597 static unsigned long top_pad = DEFAULT_TOP_PAD;
1598 static unsigned int n_mmaps_max = DEFAULT_MMAP_MAX;
1599 static unsigned long mmap_threshold = DEFAULT_MMAP_THRESHOLD;
1601 /* The first value returned from sbrk */
1602 static char* sbrk_base = (char*)(-1);
1604 /* The maximum memory obtained from system via sbrk */
1605 static unsigned long max_sbrked_mem = 0;
1607 /* The maximum via either sbrk or mmap */
1608 static unsigned long max_total_mem = 0;
1610 /* internal working copy of mallinfo */
1611 static struct mallinfo current_mallinfo = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
1613 /* The total memory obtained from system via sbrk */
1614 #define sbrked_mem (current_mallinfo.arena)
1616 /* Tracking mmaps */
1619 static unsigned int n_mmaps = 0;
1621 static unsigned long mmapped_mem = 0;
1623 static unsigned int max_n_mmaps = 0;
1624 static unsigned long max_mmapped_mem = 0;
1637 These routines make a number of assertions about the states
1638 of data structures that should be true at all times. If any
1639 are not true, it's very likely that a user program has somehow
1640 trashed memory. (It's also possible that there is a coding error
1641 in malloc. In which case, please report it!)
1645 static void do_check_chunk(mchunkptr p)
1647 static void do_check_chunk(p) mchunkptr p;
1650 #if 0 /* causes warnings because assert() is off */
1651 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1654 /* No checkable chunk is mmapped */
1655 assert(!chunk_is_mmapped(p));
1657 /* Check for legal address ... */
1658 assert((char*)p >= sbrk_base);
1660 assert((char*)p + sz <= (char*)top);
1662 assert((char*)p + sz <= sbrk_base + sbrked_mem);
1668 static void do_check_free_chunk(mchunkptr p)
1670 static void do_check_free_chunk(p) mchunkptr p;
1673 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1674 #if 0 /* causes warnings because assert() is off */
1675 mchunkptr next = chunk_at_offset(p, sz);
1680 /* Check whether it claims to be free ... */
1683 /* Unless a special marker, must have OK fields */
1684 if ((long)sz >= (long)MINSIZE)
1686 assert((sz & MALLOC_ALIGN_MASK) == 0);
1687 assert(aligned_OK(chunk2mem(p)));
1688 /* ... matching footer field */
1689 assert(next->prev_size == sz);
1690 /* ... and is fully consolidated */
1691 assert(prev_inuse(p));
1692 assert (next == top || inuse(next));
1694 /* ... and has minimally sane links */
1695 assert(p->fd->bk == p);
1696 assert(p->bk->fd == p);
1698 else /* markers are always of size SIZE_SZ */
1699 assert(sz == SIZE_SZ);
1703 static void do_check_inuse_chunk(mchunkptr p)
1705 static void do_check_inuse_chunk(p) mchunkptr p;
1708 mchunkptr next = next_chunk(p);
1711 /* Check whether it claims to be in use ... */
1714 /* ... and is surrounded by OK chunks.
1715 Since more things can be checked with free chunks than inuse ones,
1716 if an inuse chunk borders them and debug is on, it's worth doing them.
1720 mchunkptr prv = prev_chunk(p);
1721 assert(next_chunk(prv) == p);
1722 do_check_free_chunk(prv);
1726 assert(prev_inuse(next));
1727 assert(chunksize(next) >= MINSIZE);
1729 else if (!inuse(next))
1730 do_check_free_chunk(next);
1735 static void do_check_malloced_chunk(mchunkptr p, INTERNAL_SIZE_T s)
1737 static void do_check_malloced_chunk(p, s) mchunkptr p; INTERNAL_SIZE_T s;
1740 #if 0 /* causes warnings because assert() is off */
1741 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1745 do_check_inuse_chunk(p);
1747 /* Legal size ... */
1748 assert((long)sz >= (long)MINSIZE);
1749 assert((sz & MALLOC_ALIGN_MASK) == 0);
1751 assert(room < (long)MINSIZE);
1753 /* ... and alignment */
1754 assert(aligned_OK(chunk2mem(p)));
1757 /* ... and was allocated at front of an available chunk */
1758 assert(prev_inuse(p));
1763 #define check_free_chunk(P) do_check_free_chunk(P)
1764 #define check_inuse_chunk(P) do_check_inuse_chunk(P)
1765 #define check_chunk(P) do_check_chunk(P)
1766 #define check_malloced_chunk(P,N) do_check_malloced_chunk(P,N)
1768 #define check_free_chunk(P)
1769 #define check_inuse_chunk(P)
1770 #define check_chunk(P)
1771 #define check_malloced_chunk(P,N)
1777 Macro-based internal utilities
1782 Linking chunks in bin lists.
1783 Call these only with variables, not arbitrary expressions, as arguments.
1787 Place chunk p of size s in its bin, in size order,
1788 putting it ahead of others of same size.
1792 #define frontlink(P, S, IDX, BK, FD) \
1794 if (S < MAX_SMALLBIN_SIZE) \
1796 IDX = smallbin_index(S); \
1797 mark_binblock(IDX); \
1802 FD->bk = BK->fd = P; \
1806 IDX = bin_index(S); \
1809 if (FD == BK) mark_binblock(IDX); \
1812 while (FD != BK && S < chunksize(FD)) FD = FD->fd; \
1817 FD->bk = BK->fd = P; \
1822 /* take a chunk off a list */
1824 #define unlink(P, BK, FD) \
1832 /* Place p as the last remainder */
1834 #define link_last_remainder(P) \
1836 last_remainder->fd = last_remainder->bk = P; \
1837 P->fd = P->bk = last_remainder; \
1840 /* Clear the last_remainder bin */
1842 #define clear_last_remainder \
1843 (last_remainder->fd = last_remainder->bk = last_remainder)
1849 /* Routines dealing with mmap(). */
1854 static mchunkptr mmap_chunk(size_t size)
1856 static mchunkptr mmap_chunk(size) size_t size;
1859 size_t page_mask = malloc_getpagesize - 1;
1862 #ifndef MAP_ANONYMOUS
1866 if(n_mmaps >= n_mmaps_max) return 0; /* too many regions */
1868 /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because
1869 * there is no following chunk whose prev_size field could be used.
1871 size = (size + SIZE_SZ + page_mask) & ~page_mask;
1873 #ifdef MAP_ANONYMOUS
1874 p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE,
1875 MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
1876 #else /* !MAP_ANONYMOUS */
1879 fd = open("/dev/zero", O_RDWR);
1880 if(fd < 0) return 0;
1882 p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0);
1885 if(p == (mchunkptr)-1) return 0;
1888 if (n_mmaps > max_n_mmaps) max_n_mmaps = n_mmaps;
1890 /* We demand that eight bytes into a page must be 8-byte aligned. */
1891 assert(aligned_OK(chunk2mem(p)));
1893 /* The offset to the start of the mmapped region is stored
1894 * in the prev_size field of the chunk; normally it is zero,
1895 * but that can be changed in memalign().
1898 set_head(p, size|IS_MMAPPED);
1900 mmapped_mem += size;
1901 if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1902 max_mmapped_mem = mmapped_mem;
1903 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1904 max_total_mem = mmapped_mem + sbrked_mem;
1909 static void munmap_chunk(mchunkptr p)
1911 static void munmap_chunk(p) mchunkptr p;
1914 INTERNAL_SIZE_T size = chunksize(p);
1917 assert (chunk_is_mmapped(p));
1918 assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1919 assert((n_mmaps > 0));
1920 assert(((p->prev_size + size) & (malloc_getpagesize-1)) == 0);
1923 mmapped_mem -= (size + p->prev_size);
1925 ret = munmap((char *)p - p->prev_size, size + p->prev_size);
1927 /* munmap returns non-zero on failure */
1934 static mchunkptr mremap_chunk(mchunkptr p, size_t new_size)
1936 static mchunkptr mremap_chunk(p, new_size) mchunkptr p; size_t new_size;
1939 size_t page_mask = malloc_getpagesize - 1;
1940 INTERNAL_SIZE_T offset = p->prev_size;
1941 INTERNAL_SIZE_T size = chunksize(p);
1944 assert (chunk_is_mmapped(p));
1945 assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1946 assert((n_mmaps > 0));
1947 assert(((size + offset) & (malloc_getpagesize-1)) == 0);
1949 /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
1950 new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
1952 cp = (char *)mremap((char *)p - offset, size + offset, new_size, 1);
1954 if (cp == (char *)-1) return 0;
1956 p = (mchunkptr)(cp + offset);
1958 assert(aligned_OK(chunk2mem(p)));
1960 assert((p->prev_size == offset));
1961 set_head(p, (new_size - offset)|IS_MMAPPED);
1963 mmapped_mem -= size + offset;
1964 mmapped_mem += new_size;
1965 if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1966 max_mmapped_mem = mmapped_mem;
1967 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1968 max_total_mem = mmapped_mem + sbrked_mem;
1972 #endif /* HAVE_MREMAP */
1974 #endif /* HAVE_MMAP */
1980 Extend the top-most chunk by obtaining memory from system.
1981 Main interface to sbrk (but see also malloc_trim).
1985 static void malloc_extend_top(INTERNAL_SIZE_T nb)
1987 static void malloc_extend_top(nb) INTERNAL_SIZE_T nb;
1990 char* brk; /* return value from sbrk */
1991 INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of sbrked space */
1992 INTERNAL_SIZE_T correction; /* bytes for 2nd sbrk call */
1993 char* new_brk; /* return of 2nd sbrk call */
1994 INTERNAL_SIZE_T top_size; /* new size of top chunk */
1996 mchunkptr old_top = top; /* Record state of old top */
1997 INTERNAL_SIZE_T old_top_size = chunksize(old_top);
1998 char* old_end = (char*)(chunk_at_offset(old_top, old_top_size));
2000 /* Pad request with top_pad plus minimal overhead */
2002 INTERNAL_SIZE_T sbrk_size = nb + top_pad + MINSIZE;
2003 unsigned long pagesz = malloc_getpagesize;
2005 /* If not the first time through, round to preserve page boundary */
2006 /* Otherwise, we need to correct to a page size below anyway. */
2007 /* (We also correct below if an intervening foreign sbrk call.) */
2009 if (sbrk_base != (char*)(-1))
2010 sbrk_size = (sbrk_size + (pagesz - 1)) & ~(pagesz - 1);
2012 brk = (char*)(MORECORE (sbrk_size));
2014 /* Fail if sbrk failed or if a foreign sbrk call killed our space */
2015 if (brk == (char*)(MORECORE_FAILURE) ||
2016 (brk < old_end && old_top != initial_top))
2019 sbrked_mem += sbrk_size;
2021 if (brk == old_end) /* can just add bytes to current top */
2023 top_size = sbrk_size + old_top_size;
2024 set_head(top, top_size | PREV_INUSE);
2028 if (sbrk_base == (char*)(-1)) /* First time through. Record base */
2030 else /* Someone else called sbrk(). Count those bytes as sbrked_mem. */
2031 sbrked_mem += brk - (char*)old_end;
2033 /* Guarantee alignment of first new chunk made from this space */
2034 front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
2035 if (front_misalign > 0)
2037 correction = (MALLOC_ALIGNMENT) - front_misalign;
2043 /* Guarantee the next brk will be at a page boundary */
2045 correction += ((((unsigned long)(brk + sbrk_size))+(pagesz-1)) &
2046 ~(pagesz - 1)) - ((unsigned long)(brk + sbrk_size));
2048 /* Allocate correction */
2049 new_brk = (char*)(MORECORE (correction));
2050 if (new_brk == (char*)(MORECORE_FAILURE)) return;
2052 sbrked_mem += correction;
2054 top = (mchunkptr)brk;
2055 top_size = new_brk - brk + correction;
2056 set_head(top, top_size | PREV_INUSE);
2058 if (old_top != initial_top)
2061 /* There must have been an intervening foreign sbrk call. */
2062 /* A double fencepost is necessary to prevent consolidation */
2064 /* If not enough space to do this, then user did something very wrong */
2065 if (old_top_size < MINSIZE)
2067 set_head(top, PREV_INUSE); /* will force null return from malloc */
2071 /* Also keep size a multiple of MALLOC_ALIGNMENT */
2072 old_top_size = (old_top_size - 3*SIZE_SZ) & ~MALLOC_ALIGN_MASK;
2073 set_head_size(old_top, old_top_size);
2074 chunk_at_offset(old_top, old_top_size )->size =
2076 chunk_at_offset(old_top, old_top_size + SIZE_SZ)->size =
2078 /* If possible, release the rest. */
2079 if (old_top_size >= MINSIZE)
2080 fREe(chunk2mem(old_top));
2084 if ((unsigned long)sbrked_mem > (unsigned long)max_sbrked_mem)
2085 max_sbrked_mem = sbrked_mem;
2086 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
2087 max_total_mem = mmapped_mem + sbrked_mem;
2089 /* We always land on a page boundary */
2090 assert(((unsigned long)((char*)top + top_size) & (pagesz - 1)) == 0);
2096 /* Main public routines */
2102 The requested size is first converted into a usable form, `nb'.
2103 This currently means to add 4 bytes overhead plus possibly more to
2104 obtain 8-byte alignment and/or to obtain a size of at least
2105 MINSIZE (currently 16 bytes), the smallest allocatable size.
2106 (All fits are considered `exact' if they are within MINSIZE bytes.)
2108 From there, the first successful of the following steps is taken:
2110 1. The bin corresponding to the request size is scanned, and if
2111 a chunk of exactly the right size is found, it is taken.
2113 2. The most recently remaindered chunk is used if it is big
2114 enough. This is a form of (roving) first fit, used only in
2115 the absence of exact fits. Runs of consecutive requests use
2116 the remainder of the chunk used for the previous such request
2117 whenever possible. This limited use of a first-fit style
2118 allocation strategy tends to give contiguous chunks
2119 coextensive lifetimes, which improves locality and can reduce
2120 fragmentation in the long run.
2122 3. Other bins are scanned in increasing size order, using a
2123 chunk big enough to fulfill the request, and splitting off
2124 any remainder. This search is strictly by best-fit; i.e.,
2125 the smallest (with ties going to approximately the least
2126 recently used) chunk that fits is selected.
2128 4. If large enough, the chunk bordering the end of memory
2129 (`top') is split off. (This use of `top' is in accord with
2130 the best-fit search rule. In effect, `top' is treated as
2131 larger (and thus less well fitting) than any other available
2132 chunk since it can be extended to be as large as necessary
2133 (up to system limitations).
2135 5. If the request size meets the mmap threshold and the
2136 system supports mmap, and there are few enough currently
2137 allocated mmapped regions, and a call to mmap succeeds,
2138 the request is allocated via direct memory mapping.
2140 6. Otherwise, the top of memory is extended by
2141 obtaining more space from the system (normally using sbrk,
2142 but definable to anything else via the MORECORE macro).
2143 Memory is gathered from the system (in system page-sized
2144 units) in a way that allows chunks obtained across different
2145 sbrk calls to be consolidated, but does not require
2146 contiguous memory. Thus, it should be safe to intersperse
2147 mallocs with other sbrk calls.
2150 All allocations are made from the the `lowest' part of any found
2151 chunk. (The implementation invariant is that prev_inuse is
2152 always true of any allocated chunk; i.e., that each allocated
2153 chunk borders either a previously allocated and still in-use chunk,
2154 or the base of its memory arena.)
2159 Void_t* mALLOc(size_t bytes)
2161 Void_t* mALLOc(bytes) size_t bytes;
2164 mchunkptr victim; /* inspected/selected chunk */
2165 INTERNAL_SIZE_T victim_size; /* its size */
2166 int idx; /* index for bin traversal */
2167 mbinptr bin; /* associated bin */
2168 mchunkptr remainder; /* remainder from a split */
2169 long remainder_size; /* its size */
2170 int remainder_index; /* its bin index */
2171 unsigned long block; /* block traverser bit */
2172 int startidx; /* first bin of a traversed block */
2173 mchunkptr fwd; /* misc temp for linking */
2174 mchunkptr bck; /* misc temp for linking */
2175 mbinptr q; /* misc temp */
2179 /* check if mem_malloc_init() was run */
2180 if ((mem_malloc_start == 0) && (mem_malloc_end == 0)) {
2181 /* not initialized yet */
2185 if ((long)bytes < 0) return 0;
2187 nb = request2size(bytes); /* padded request size; */
2189 /* Check for exact match in a bin */
2191 if (is_small_request(nb)) /* Faster version for small requests */
2193 idx = smallbin_index(nb);
2195 /* No traversal or size check necessary for small bins. */
2200 /* Also scan the next one, since it would have a remainder < MINSIZE */
2208 victim_size = chunksize(victim);
2209 unlink(victim, bck, fwd);
2210 set_inuse_bit_at_offset(victim, victim_size);
2211 check_malloced_chunk(victim, nb);
2212 return chunk2mem(victim);
2215 idx += 2; /* Set for bin scan below. We've already scanned 2 bins. */
2220 idx = bin_index(nb);
2223 for (victim = last(bin); victim != bin; victim = victim->bk)
2225 victim_size = chunksize(victim);
2226 remainder_size = victim_size - nb;
2228 if (remainder_size >= (long)MINSIZE) /* too big */
2230 --idx; /* adjust to rescan below after checking last remainder */
2234 else if (remainder_size >= 0) /* exact fit */
2236 unlink(victim, bck, fwd);
2237 set_inuse_bit_at_offset(victim, victim_size);
2238 check_malloced_chunk(victim, nb);
2239 return chunk2mem(victim);
2247 /* Try to use the last split-off remainder */
2249 if ( (victim = last_remainder->fd) != last_remainder)
2251 victim_size = chunksize(victim);
2252 remainder_size = victim_size - nb;
2254 if (remainder_size >= (long)MINSIZE) /* re-split */
2256 remainder = chunk_at_offset(victim, nb);
2257 set_head(victim, nb | PREV_INUSE);
2258 link_last_remainder(remainder);
2259 set_head(remainder, remainder_size | PREV_INUSE);
2260 set_foot(remainder, remainder_size);
2261 check_malloced_chunk(victim, nb);
2262 return chunk2mem(victim);
2265 clear_last_remainder;
2267 if (remainder_size >= 0) /* exhaust */
2269 set_inuse_bit_at_offset(victim, victim_size);
2270 check_malloced_chunk(victim, nb);
2271 return chunk2mem(victim);
2274 /* Else place in bin */
2276 frontlink(victim, victim_size, remainder_index, bck, fwd);
2280 If there are any possibly nonempty big-enough blocks,
2281 search for best fitting chunk by scanning bins in blockwidth units.
2284 if ( (block = idx2binblock(idx)) <= binblocks_r)
2287 /* Get to the first marked block */
2289 if ( (block & binblocks_r) == 0)
2291 /* force to an even block boundary */
2292 idx = (idx & ~(BINBLOCKWIDTH - 1)) + BINBLOCKWIDTH;
2294 while ((block & binblocks_r) == 0)
2296 idx += BINBLOCKWIDTH;
2301 /* For each possibly nonempty block ... */
2304 startidx = idx; /* (track incomplete blocks) */
2305 q = bin = bin_at(idx);
2307 /* For each bin in this block ... */
2310 /* Find and use first big enough chunk ... */
2312 for (victim = last(bin); victim != bin; victim = victim->bk)
2314 victim_size = chunksize(victim);
2315 remainder_size = victim_size - nb;
2317 if (remainder_size >= (long)MINSIZE) /* split */
2319 remainder = chunk_at_offset(victim, nb);
2320 set_head(victim, nb | PREV_INUSE);
2321 unlink(victim, bck, fwd);
2322 link_last_remainder(remainder);
2323 set_head(remainder, remainder_size | PREV_INUSE);
2324 set_foot(remainder, remainder_size);
2325 check_malloced_chunk(victim, nb);
2326 return chunk2mem(victim);
2329 else if (remainder_size >= 0) /* take */
2331 set_inuse_bit_at_offset(victim, victim_size);
2332 unlink(victim, bck, fwd);
2333 check_malloced_chunk(victim, nb);
2334 return chunk2mem(victim);
2339 bin = next_bin(bin);
2341 } while ((++idx & (BINBLOCKWIDTH - 1)) != 0);
2343 /* Clear out the block bit. */
2345 do /* Possibly backtrack to try to clear a partial block */
2347 if ((startidx & (BINBLOCKWIDTH - 1)) == 0)
2349 av_[1] = (mbinptr)(binblocks_r & ~block);
2354 } while (first(q) == q);
2356 /* Get to the next possibly nonempty block */
2358 if ( (block <<= 1) <= binblocks_r && (block != 0) )
2360 while ((block & binblocks_r) == 0)
2362 idx += BINBLOCKWIDTH;
2372 /* Try to use top chunk */
2374 /* Require that there be a remainder, ensuring top always exists */
2375 if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2379 /* If big and would otherwise need to extend, try to use mmap instead */
2380 if ((unsigned long)nb >= (unsigned long)mmap_threshold &&
2381 (victim = mmap_chunk(nb)) != 0)
2382 return chunk2mem(victim);
2386 malloc_extend_top(nb);
2387 if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2388 return 0; /* propagate failure */
2392 set_head(victim, nb | PREV_INUSE);
2393 top = chunk_at_offset(victim, nb);
2394 set_head(top, remainder_size | PREV_INUSE);
2395 check_malloced_chunk(victim, nb);
2396 return chunk2mem(victim);
2409 1. free(0) has no effect.
2411 2. If the chunk was allocated via mmap, it is release via munmap().
2413 3. If a returned chunk borders the current high end of memory,
2414 it is consolidated into the top, and if the total unused
2415 topmost memory exceeds the trim threshold, malloc_trim is
2418 4. Other chunks are consolidated as they arrive, and
2419 placed in corresponding bins. (This includes the case of
2420 consolidating with the current `last_remainder').
2426 void fREe(Void_t* mem)
2428 void fREe(mem) Void_t* mem;
2431 mchunkptr p; /* chunk corresponding to mem */
2432 INTERNAL_SIZE_T hd; /* its head field */
2433 INTERNAL_SIZE_T sz; /* its size */
2434 int idx; /* its bin index */
2435 mchunkptr next; /* next contiguous chunk */
2436 INTERNAL_SIZE_T nextsz; /* its size */
2437 INTERNAL_SIZE_T prevsz; /* size of previous contiguous chunk */
2438 mchunkptr bck; /* misc temp for linking */
2439 mchunkptr fwd; /* misc temp for linking */
2440 int islr; /* track whether merging with last_remainder */
2442 if (mem == 0) /* free(0) has no effect */
2449 if (hd & IS_MMAPPED) /* release mmapped memory. */
2456 check_inuse_chunk(p);
2458 sz = hd & ~PREV_INUSE;
2459 next = chunk_at_offset(p, sz);
2460 nextsz = chunksize(next);
2462 if (next == top) /* merge with top */
2466 if (!(hd & PREV_INUSE)) /* consolidate backward */
2468 prevsz = p->prev_size;
2469 p = chunk_at_offset(p, -((long) prevsz));
2471 unlink(p, bck, fwd);
2474 set_head(p, sz | PREV_INUSE);
2476 if ((unsigned long)(sz) >= (unsigned long)trim_threshold)
2477 malloc_trim(top_pad);
2481 set_head(next, nextsz); /* clear inuse bit */
2485 if (!(hd & PREV_INUSE)) /* consolidate backward */
2487 prevsz = p->prev_size;
2488 p = chunk_at_offset(p, -((long) prevsz));
2491 if (p->fd == last_remainder) /* keep as last_remainder */
2494 unlink(p, bck, fwd);
2497 if (!(inuse_bit_at_offset(next, nextsz))) /* consolidate forward */
2501 if (!islr && next->fd == last_remainder) /* re-insert last_remainder */
2504 link_last_remainder(p);
2507 unlink(next, bck, fwd);
2511 set_head(p, sz | PREV_INUSE);
2514 frontlink(p, sz, idx, bck, fwd);
2525 Chunks that were obtained via mmap cannot be extended or shrunk
2526 unless HAVE_MREMAP is defined, in which case mremap is used.
2527 Otherwise, if their reallocation is for additional space, they are
2528 copied. If for less, they are just left alone.
2530 Otherwise, if the reallocation is for additional space, and the
2531 chunk can be extended, it is, else a malloc-copy-free sequence is
2532 taken. There are several different ways that a chunk could be
2533 extended. All are tried:
2535 * Extending forward into following adjacent free chunk.
2536 * Shifting backwards, joining preceding adjacent space
2537 * Both shifting backwards and extending forward.
2538 * Extending into newly sbrked space
2540 Unless the #define REALLOC_ZERO_BYTES_FREES is set, realloc with a
2541 size argument of zero (re)allocates a minimum-sized chunk.
2543 If the reallocation is for less space, and the new request is for
2544 a `small' (<512 bytes) size, then the newly unused space is lopped
2547 The old unix realloc convention of allowing the last-free'd chunk
2548 to be used as an argument to realloc is no longer supported.
2549 I don't know of any programs still relying on this feature,
2550 and allowing it would also allow too many other incorrect
2551 usages of realloc to be sensible.
2558 Void_t* rEALLOc(Void_t* oldmem, size_t bytes)
2560 Void_t* rEALLOc(oldmem, bytes) Void_t* oldmem; size_t bytes;
2563 INTERNAL_SIZE_T nb; /* padded request size */
2565 mchunkptr oldp; /* chunk corresponding to oldmem */
2566 INTERNAL_SIZE_T oldsize; /* its size */
2568 mchunkptr newp; /* chunk to return */
2569 INTERNAL_SIZE_T newsize; /* its size */
2570 Void_t* newmem; /* corresponding user mem */
2572 mchunkptr next; /* next contiguous chunk after oldp */
2573 INTERNAL_SIZE_T nextsize; /* its size */
2575 mchunkptr prev; /* previous contiguous chunk before oldp */
2576 INTERNAL_SIZE_T prevsize; /* its size */
2578 mchunkptr remainder; /* holds split off extra space from newp */
2579 INTERNAL_SIZE_T remainder_size; /* its size */
2581 mchunkptr bck; /* misc temp for linking */
2582 mchunkptr fwd; /* misc temp for linking */
2584 #ifdef REALLOC_ZERO_BYTES_FREES
2585 if (bytes == 0) { fREe(oldmem); return 0; }
2588 if ((long)bytes < 0) return 0;
2590 /* realloc of null is supposed to be same as malloc */
2591 if (oldmem == 0) return mALLOc(bytes);
2593 newp = oldp = mem2chunk(oldmem);
2594 newsize = oldsize = chunksize(oldp);
2597 nb = request2size(bytes);
2600 if (chunk_is_mmapped(oldp))
2603 newp = mremap_chunk(oldp, nb);
2604 if(newp) return chunk2mem(newp);
2606 /* Note the extra SIZE_SZ overhead. */
2607 if(oldsize - SIZE_SZ >= nb) return oldmem; /* do nothing */
2608 /* Must alloc, copy, free. */
2609 newmem = mALLOc(bytes);
2610 if (newmem == 0) return 0; /* propagate failure */
2611 MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ);
2617 check_inuse_chunk(oldp);
2619 if ((long)(oldsize) < (long)(nb))
2622 /* Try expanding forward */
2624 next = chunk_at_offset(oldp, oldsize);
2625 if (next == top || !inuse(next))
2627 nextsize = chunksize(next);
2629 /* Forward into top only if a remainder */
2632 if ((long)(nextsize + newsize) >= (long)(nb + MINSIZE))
2634 newsize += nextsize;
2635 top = chunk_at_offset(oldp, nb);
2636 set_head(top, (newsize - nb) | PREV_INUSE);
2637 set_head_size(oldp, nb);
2638 return chunk2mem(oldp);
2642 /* Forward into next chunk */
2643 else if (((long)(nextsize + newsize) >= (long)(nb)))
2645 unlink(next, bck, fwd);
2646 newsize += nextsize;
2656 /* Try shifting backwards. */
2658 if (!prev_inuse(oldp))
2660 prev = prev_chunk(oldp);
2661 prevsize = chunksize(prev);
2663 /* try forward + backward first to save a later consolidation */
2670 if ((long)(nextsize + prevsize + newsize) >= (long)(nb + MINSIZE))
2672 unlink(prev, bck, fwd);
2674 newsize += prevsize + nextsize;
2675 newmem = chunk2mem(newp);
2676 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2677 top = chunk_at_offset(newp, nb);
2678 set_head(top, (newsize - nb) | PREV_INUSE);
2679 set_head_size(newp, nb);
2684 /* into next chunk */
2685 else if (((long)(nextsize + prevsize + newsize) >= (long)(nb)))
2687 unlink(next, bck, fwd);
2688 unlink(prev, bck, fwd);
2690 newsize += nextsize + prevsize;
2691 newmem = chunk2mem(newp);
2692 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2698 if (prev != 0 && (long)(prevsize + newsize) >= (long)nb)
2700 unlink(prev, bck, fwd);
2702 newsize += prevsize;
2703 newmem = chunk2mem(newp);
2704 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2711 newmem = mALLOc (bytes);
2713 if (newmem == 0) /* propagate failure */
2716 /* Avoid copy if newp is next chunk after oldp. */
2717 /* (This can only happen when new chunk is sbrk'ed.) */
2719 if ( (newp = mem2chunk(newmem)) == next_chunk(oldp))
2721 newsize += chunksize(newp);
2726 /* Otherwise copy, free, and exit */
2727 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2733 split: /* split off extra room in old or expanded chunk */
2735 if (newsize - nb >= MINSIZE) /* split off remainder */
2737 remainder = chunk_at_offset(newp, nb);
2738 remainder_size = newsize - nb;
2739 set_head_size(newp, nb);
2740 set_head(remainder, remainder_size | PREV_INUSE);
2741 set_inuse_bit_at_offset(remainder, remainder_size);
2742 fREe(chunk2mem(remainder)); /* let free() deal with it */
2746 set_head_size(newp, newsize);
2747 set_inuse_bit_at_offset(newp, newsize);
2750 check_inuse_chunk(newp);
2751 return chunk2mem(newp);
2761 memalign requests more than enough space from malloc, finds a spot
2762 within that chunk that meets the alignment request, and then
2763 possibly frees the leading and trailing space.
2765 The alignment argument must be a power of two. This property is not
2766 checked by memalign, so misuse may result in random runtime errors.
2768 8-byte alignment is guaranteed by normal malloc calls, so don't
2769 bother calling memalign with an argument of 8 or less.
2771 Overreliance on memalign is a sure way to fragment space.
2777 Void_t* mEMALIGn(size_t alignment, size_t bytes)
2779 Void_t* mEMALIGn(alignment, bytes) size_t alignment; size_t bytes;
2782 INTERNAL_SIZE_T nb; /* padded request size */
2783 char* m; /* memory returned by malloc call */
2784 mchunkptr p; /* corresponding chunk */
2785 char* brk; /* alignment point within p */
2786 mchunkptr newp; /* chunk to return */
2787 INTERNAL_SIZE_T newsize; /* its size */
2788 INTERNAL_SIZE_T leadsize; /* leading space befor alignment point */
2789 mchunkptr remainder; /* spare room at end to split off */
2790 long remainder_size; /* its size */
2792 if ((long)bytes < 0) return 0;
2794 /* If need less alignment than we give anyway, just relay to malloc */
2796 if (alignment <= MALLOC_ALIGNMENT) return mALLOc(bytes);
2798 /* Otherwise, ensure that it is at least a minimum chunk size */
2800 if (alignment < MINSIZE) alignment = MINSIZE;
2802 /* Call malloc with worst case padding to hit alignment. */
2804 nb = request2size(bytes);
2805 m = (char*)(mALLOc(nb + alignment + MINSIZE));
2807 if (m == 0) return 0; /* propagate failure */
2811 if ((((unsigned long)(m)) % alignment) == 0) /* aligned */
2814 if(chunk_is_mmapped(p))
2815 return chunk2mem(p); /* nothing more to do */
2818 else /* misaligned */
2821 Find an aligned spot inside chunk.
2822 Since we need to give back leading space in a chunk of at
2823 least MINSIZE, if the first calculation places us at
2824 a spot with less than MINSIZE leader, we can move to the
2825 next aligned spot -- we've allocated enough total room so that
2826 this is always possible.
2829 brk = (char*)mem2chunk(((unsigned long)(m + alignment - 1)) & -((signed) alignment));
2830 if ((long)(brk - (char*)(p)) < MINSIZE) brk = brk + alignment;
2832 newp = (mchunkptr)brk;
2833 leadsize = brk - (char*)(p);
2834 newsize = chunksize(p) - leadsize;
2837 if(chunk_is_mmapped(p))
2839 newp->prev_size = p->prev_size + leadsize;
2840 set_head(newp, newsize|IS_MMAPPED);
2841 return chunk2mem(newp);
2845 /* give back leader, use the rest */
2847 set_head(newp, newsize | PREV_INUSE);
2848 set_inuse_bit_at_offset(newp, newsize);
2849 set_head_size(p, leadsize);
2853 assert (newsize >= nb && (((unsigned long)(chunk2mem(p))) % alignment) == 0);
2856 /* Also give back spare room at the end */
2858 remainder_size = chunksize(p) - nb;
2860 if (remainder_size >= (long)MINSIZE)
2862 remainder = chunk_at_offset(p, nb);
2863 set_head(remainder, remainder_size | PREV_INUSE);
2864 set_head_size(p, nb);
2865 fREe(chunk2mem(remainder));
2868 check_inuse_chunk(p);
2869 return chunk2mem(p);
2877 valloc just invokes memalign with alignment argument equal
2878 to the page size of the system (or as near to this as can
2879 be figured out from all the includes/defines above.)
2883 Void_t* vALLOc(size_t bytes)
2885 Void_t* vALLOc(bytes) size_t bytes;
2888 return mEMALIGn (malloc_getpagesize, bytes);
2892 pvalloc just invokes valloc for the nearest pagesize
2893 that will accommodate request
2898 Void_t* pvALLOc(size_t bytes)
2900 Void_t* pvALLOc(bytes) size_t bytes;
2903 size_t pagesize = malloc_getpagesize;
2904 return mEMALIGn (pagesize, (bytes + pagesize - 1) & ~(pagesize - 1));
2909 calloc calls malloc, then zeroes out the allocated chunk.
2914 Void_t* cALLOc(size_t n, size_t elem_size)
2916 Void_t* cALLOc(n, elem_size) size_t n; size_t elem_size;
2920 INTERNAL_SIZE_T csz;
2922 INTERNAL_SIZE_T sz = n * elem_size;
2925 /* check if expand_top called, in which case don't need to clear */
2927 mchunkptr oldtop = top;
2928 INTERNAL_SIZE_T oldtopsize = chunksize(top);
2930 Void_t* mem = mALLOc (sz);
2932 if ((long)n < 0) return 0;
2940 /* Two optional cases in which clearing not necessary */
2944 if (chunk_is_mmapped(p)) return mem;
2950 if (p == oldtop && csz > oldtopsize)
2952 /* clear only the bytes from non-freshly-sbrked memory */
2957 MALLOC_ZERO(mem, csz - SIZE_SZ);
2964 cfree just calls free. It is needed/defined on some systems
2965 that pair it with calloc, presumably for odd historical reasons.
2969 #if !defined(INTERNAL_LINUX_C_LIB) || !defined(__ELF__)
2971 void cfree(Void_t *mem)
2973 void cfree(mem) Void_t *mem;
2984 Malloc_trim gives memory back to the system (via negative
2985 arguments to sbrk) if there is unused memory at the `high' end of
2986 the malloc pool. You can call this after freeing large blocks of
2987 memory to potentially reduce the system-level memory requirements
2988 of a program. However, it cannot guarantee to reduce memory. Under
2989 some allocation patterns, some large free blocks of memory will be
2990 locked between two used chunks, so they cannot be given back to
2993 The `pad' argument to malloc_trim represents the amount of free
2994 trailing space to leave untrimmed. If this argument is zero,
2995 only the minimum amount of memory to maintain internal data
2996 structures will be left (one page or less). Non-zero arguments
2997 can be supplied to maintain enough trailing space to service
2998 future expected allocations without having to re-obtain memory
3001 Malloc_trim returns 1 if it actually released any memory, else 0.
3006 int malloc_trim(size_t pad)
3008 int malloc_trim(pad) size_t pad;
3011 long top_size; /* Amount of top-most memory */
3012 long extra; /* Amount to release */
3013 char* current_brk; /* address returned by pre-check sbrk call */
3014 char* new_brk; /* address returned by negative sbrk call */
3016 unsigned long pagesz = malloc_getpagesize;
3018 top_size = chunksize(top);
3019 extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
3021 if (extra < (long)pagesz) /* Not enough memory to release */
3026 /* Test to make sure no one else called sbrk */
3027 current_brk = (char*)(MORECORE (0));
3028 if (current_brk != (char*)(top) + top_size)
3029 return 0; /* Apparently we don't own memory; must fail */
3033 new_brk = (char*)(MORECORE (-extra));
3035 if (new_brk == (char*)(MORECORE_FAILURE)) /* sbrk failed? */
3037 /* Try to figure out what we have */
3038 current_brk = (char*)(MORECORE (0));
3039 top_size = current_brk - (char*)top;
3040 if (top_size >= (long)MINSIZE) /* if not, we are very very dead! */
3042 sbrked_mem = current_brk - sbrk_base;
3043 set_head(top, top_size | PREV_INUSE);
3051 /* Success. Adjust top accordingly. */
3052 set_head(top, (top_size - extra) | PREV_INUSE);
3053 sbrked_mem -= extra;
3066 This routine tells you how many bytes you can actually use in an
3067 allocated chunk, which may be more than you requested (although
3068 often not). You can use this many bytes without worrying about
3069 overwriting other allocated objects. Not a particularly great
3070 programming practice, but still sometimes useful.
3075 size_t malloc_usable_size(Void_t* mem)
3077 size_t malloc_usable_size(mem) Void_t* mem;
3086 if(!chunk_is_mmapped(p))
3088 if (!inuse(p)) return 0;
3089 check_inuse_chunk(p);
3090 return chunksize(p) - SIZE_SZ;
3092 return chunksize(p) - 2*SIZE_SZ;
3099 /* Utility to update current_mallinfo for malloc_stats and mallinfo() */
3102 static void malloc_update_mallinfo()
3111 INTERNAL_SIZE_T avail = chunksize(top);
3112 int navail = ((long)(avail) >= (long)MINSIZE)? 1 : 0;
3114 for (i = 1; i < NAV; ++i)
3117 for (p = last(b); p != b; p = p->bk)
3120 check_free_chunk(p);
3121 for (q = next_chunk(p);
3122 q < top && inuse(q) && (long)(chunksize(q)) >= (long)MINSIZE;
3124 check_inuse_chunk(q);
3126 avail += chunksize(p);
3131 current_mallinfo.ordblks = navail;
3132 current_mallinfo.uordblks = sbrked_mem - avail;
3133 current_mallinfo.fordblks = avail;
3134 current_mallinfo.hblks = n_mmaps;
3135 current_mallinfo.hblkhd = mmapped_mem;
3136 current_mallinfo.keepcost = chunksize(top);
3147 Prints on the amount of space obtain from the system (both
3148 via sbrk and mmap), the maximum amount (which may be more than
3149 current if malloc_trim and/or munmap got called), the maximum
3150 number of simultaneous mmap regions used, and the current number
3151 of bytes allocated via malloc (or realloc, etc) but not yet
3152 freed. (Note that this is the number of bytes allocated, not the
3153 number requested. It will be larger than the number requested
3154 because of alignment and bookkeeping overhead.)
3161 malloc_update_mallinfo();
3162 printf("max system bytes = %10u\n",
3163 (unsigned int)(max_total_mem));
3164 printf("system bytes = %10u\n",
3165 (unsigned int)(sbrked_mem + mmapped_mem));
3166 printf("in use bytes = %10u\n",
3167 (unsigned int)(current_mallinfo.uordblks + mmapped_mem));
3169 printf("max mmap regions = %10u\n",
3170 (unsigned int)max_n_mmaps);
3176 mallinfo returns a copy of updated current mallinfo.
3180 struct mallinfo mALLINFo()
3182 malloc_update_mallinfo();
3183 return current_mallinfo;
3193 mallopt is the general SVID/XPG interface to tunable parameters.
3194 The format is to provide a (parameter-number, parameter-value) pair.
3195 mallopt then sets the corresponding parameter to the argument
3196 value if it can (i.e., so long as the value is meaningful),
3197 and returns 1 if successful else 0.
3199 See descriptions of tunable parameters above.
3204 int mALLOPt(int param_number, int value)
3206 int mALLOPt(param_number, value) int param_number; int value;
3209 switch(param_number)
3211 case M_TRIM_THRESHOLD:
3212 trim_threshold = value; return 1;
3214 top_pad = value; return 1;
3215 case M_MMAP_THRESHOLD:
3216 mmap_threshold = value; return 1;
3219 n_mmaps_max = value; return 1;
3221 if (value != 0) return 0; else n_mmaps_max = value; return 1;
3233 V2.6.6 Sun Dec 5 07:42:19 1999 Doug Lea (dl at gee)
3234 * return null for negative arguments
3235 * Added Several WIN32 cleanups from Martin C. Fong <mcfong@yahoo.com>
3236 * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'
3237 (e.g. WIN32 platforms)
3238 * Cleanup up header file inclusion for WIN32 platforms
3239 * Cleanup code to avoid Microsoft Visual C++ compiler complaints
3240 * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing
3241 memory allocation routines
3242 * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)
3243 * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to
3244 usage of 'assert' in non-WIN32 code
3245 * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to
3247 * Always call 'fREe()' rather than 'free()'
3249 V2.6.5 Wed Jun 17 15:57:31 1998 Doug Lea (dl at gee)
3250 * Fixed ordering problem with boundary-stamping
3252 V2.6.3 Sun May 19 08:17:58 1996 Doug Lea (dl at gee)
3253 * Added pvalloc, as recommended by H.J. Liu
3254 * Added 64bit pointer support mainly from Wolfram Gloger
3255 * Added anonymously donated WIN32 sbrk emulation
3256 * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
3257 * malloc_extend_top: fix mask error that caused wastage after
3259 * Add linux mremap support code from HJ Liu
3261 V2.6.2 Tue Dec 5 06:52:55 1995 Doug Lea (dl at gee)
3262 * Integrated most documentation with the code.
3263 * Add support for mmap, with help from
3264 Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3265 * Use last_remainder in more cases.
3266 * Pack bins using idea from colin@nyx10.cs.du.edu
3267 * Use ordered bins instead of best-fit threshhold
3268 * Eliminate block-local decls to simplify tracing and debugging.
3269 * Support another case of realloc via move into top
3270 * Fix error occuring when initial sbrk_base not word-aligned.
3271 * Rely on page size for units instead of SBRK_UNIT to
3272 avoid surprises about sbrk alignment conventions.
3273 * Add mallinfo, mallopt. Thanks to Raymond Nijssen
3274 (raymond@es.ele.tue.nl) for the suggestion.
3275 * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
3276 * More precautions for cases where other routines call sbrk,
3277 courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3278 * Added macros etc., allowing use in linux libc from
3279 H.J. Lu (hjl@gnu.ai.mit.edu)
3280 * Inverted this history list
3282 V2.6.1 Sat Dec 2 14:10:57 1995 Doug Lea (dl at gee)
3283 * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
3284 * Removed all preallocation code since under current scheme
3285 the work required to undo bad preallocations exceeds
3286 the work saved in good cases for most test programs.
3287 * No longer use return list or unconsolidated bins since
3288 no scheme using them consistently outperforms those that don't
3289 given above changes.
3290 * Use best fit for very large chunks to prevent some worst-cases.
3291 * Added some support for debugging
3293 V2.6.0 Sat Nov 4 07:05:23 1995 Doug Lea (dl at gee)
3294 * Removed footers when chunks are in use. Thanks to
3295 Paul Wilson (wilson@cs.texas.edu) for the suggestion.
3297 V2.5.4 Wed Nov 1 07:54:51 1995 Doug Lea (dl at gee)
3298 * Added malloc_trim, with help from Wolfram Gloger
3299 (wmglo@Dent.MED.Uni-Muenchen.DE).
3301 V2.5.3 Tue Apr 26 10:16:01 1994 Doug Lea (dl at g)
3303 V2.5.2 Tue Apr 5 16:20:40 1994 Doug Lea (dl at g)
3304 * realloc: try to expand in both directions
3305 * malloc: swap order of clean-bin strategy;
3306 * realloc: only conditionally expand backwards
3307 * Try not to scavenge used bins
3308 * Use bin counts as a guide to preallocation
3309 * Occasionally bin return list chunks in first scan
3310 * Add a few optimizations from colin@nyx10.cs.du.edu
3312 V2.5.1 Sat Aug 14 15:40:43 1993 Doug Lea (dl at g)
3313 * faster bin computation & slightly different binning
3314 * merged all consolidations to one part of malloc proper
3315 (eliminating old malloc_find_space & malloc_clean_bin)
3316 * Scan 2 returns chunks (not just 1)
3317 * Propagate failure in realloc if malloc returns 0
3318 * Add stuff to allow compilation on non-ANSI compilers
3319 from kpv@research.att.com
3321 V2.5 Sat Aug 7 07:41:59 1993 Doug Lea (dl at g.oswego.edu)
3322 * removed potential for odd address access in prev_chunk
3323 * removed dependency on getpagesize.h
3324 * misc cosmetics and a bit more internal documentation
3325 * anticosmetics: mangled names in macros to evade debugger strangeness
3326 * tested on sparc, hp-700, dec-mips, rs6000
3327 with gcc & native cc (hp, dec only) allowing
3328 Detlefs & Zorn comparison study (in SIGPLAN Notices.)
3330 Trial version Fri Aug 28 13:14:29 1992 Doug Lea (dl at g.oswego.edu)
3331 * Based loosely on libg++-1.2X malloc. (It retains some of the overall
3332 structure of old version, but most details differ.)