1 #if 0 /* Moved to malloc.h */
2 /* ---------- To make a malloc.h, start cutting here ------------ */
5 A version of malloc/free/realloc written by Doug Lea and released to the
6 public domain. Send questions/comments/complaints/performance data
9 * VERSION 2.6.6 Sun Mar 5 19:10:03 2000 Doug Lea (dl at gee)
11 Note: There may be an updated version of this malloc obtainable at
12 ftp://g.oswego.edu/pub/misc/malloc.c
13 Check before installing!
15 * Why use this malloc?
17 This is not the fastest, most space-conserving, most portable, or
18 most tunable malloc ever written. However it is among the fastest
19 while also being among the most space-conserving, portable and tunable.
20 Consistent balance across these factors results in a good general-purpose
21 allocator. For a high-level description, see
22 http://g.oswego.edu/dl/html/malloc.html
24 * Synopsis of public routines
26 (Much fuller descriptions are contained in the program documentation below.)
29 Return a pointer to a newly allocated chunk of at least n bytes, or null
30 if no space is available.
32 Release the chunk of memory pointed to by p, or no effect if p is null.
33 realloc(Void_t* p, size_t n);
34 Return a pointer to a chunk of size n that contains the same data
35 as does chunk p up to the minimum of (n, p's size) bytes, or null
36 if no space is available. The returned pointer may or may not be
37 the same as p. If p is null, equivalent to malloc. Unless the
38 #define REALLOC_ZERO_BYTES_FREES below is set, realloc with a
39 size argument of zero (re)allocates a minimum-sized chunk.
40 memalign(size_t alignment, size_t n);
41 Return a pointer to a newly allocated chunk of n bytes, aligned
42 in accord with the alignment argument, which must be a power of
45 Equivalent to memalign(pagesize, n), where pagesize is the page
46 size of the system (or as near to this as can be figured out from
47 all the includes/defines below.)
49 Equivalent to valloc(minimum-page-that-holds(n)), that is,
50 round up n to nearest pagesize.
51 calloc(size_t unit, size_t quantity);
52 Returns a pointer to quantity * unit bytes, with all locations
55 Equivalent to free(p).
56 malloc_trim(size_t pad);
57 Release all but pad bytes of freed top-most memory back
58 to the system. Return 1 if successful, else 0.
59 malloc_usable_size(Void_t* p);
60 Report the number usable allocated bytes associated with allocated
61 chunk p. This may or may not report more bytes than were requested,
62 due to alignment and minimum size constraints.
64 Prints brief summary statistics.
66 Returns (by copy) a struct containing various summary statistics.
67 mallopt(int parameter_number, int parameter_value)
68 Changes one of the tunable parameters described below. Returns
69 1 if successful in changing the parameter, else 0.
74 8 byte alignment is currently hardwired into the design. This
75 seems to suffice for all current machines and C compilers.
77 Assumed pointer representation: 4 or 8 bytes
78 Code for 8-byte pointers is untested by me but has worked
79 reliably by Wolfram Gloger, who contributed most of the
80 changes supporting this.
82 Assumed size_t representation: 4 or 8 bytes
83 Note that size_t is allowed to be 4 bytes even if pointers are 8.
85 Minimum overhead per allocated chunk: 4 or 8 bytes
86 Each malloced chunk has a hidden overhead of 4 bytes holding size
87 and status information.
89 Minimum allocated size: 4-byte ptrs: 16 bytes (including 4 overhead)
90 8-byte ptrs: 24/32 bytes (including, 4/8 overhead)
92 When a chunk is freed, 12 (for 4byte ptrs) or 20 (for 8 byte
93 ptrs but 4 byte size) or 24 (for 8/8) additional bytes are
94 needed; 4 (8) for a trailing size field
95 and 8 (16) bytes for free list pointers. Thus, the minimum
96 allocatable size is 16/24/32 bytes.
98 Even a request for zero bytes (i.e., malloc(0)) returns a
99 pointer to something of the minimum allocatable size.
101 Maximum allocated size: 4-byte size_t: 2^31 - 8 bytes
102 8-byte size_t: 2^63 - 16 bytes
104 It is assumed that (possibly signed) size_t bit values suffice to
105 represent chunk sizes. `Possibly signed' is due to the fact
106 that `size_t' may be defined on a system as either a signed or
107 an unsigned type. To be conservative, values that would appear
108 as negative numbers are avoided.
109 Requests for sizes with a negative sign bit when the request
110 size is treaded as a long will return null.
112 Maximum overhead wastage per allocated chunk: normally 15 bytes
114 Alignnment demands, plus the minimum allocatable size restriction
115 make the normal worst-case wastage 15 bytes (i.e., up to 15
116 more bytes will be allocated than were requested in malloc), with
118 1. Because requests for zero bytes allocate non-zero space,
119 the worst case wastage for a request of zero bytes is 24 bytes.
120 2. For requests >= mmap_threshold that are serviced via
121 mmap(), the worst case wastage is 8 bytes plus the remainder
122 from a system page (the minimal mmap unit); typically 4096 bytes.
126 Here are some features that are NOT currently supported
128 * No user-definable hooks for callbacks and the like.
129 * No automated mechanism for fully checking that all accesses
130 to malloced memory stay within their bounds.
131 * No support for compaction.
133 * Synopsis of compile-time options:
135 People have reported using previous versions of this malloc on all
136 versions of Unix, sometimes by tweaking some of the defines
137 below. It has been tested most extensively on Solaris and
138 Linux. It is also reported to work on WIN32 platforms.
139 People have also reported adapting this malloc for use in
140 stand-alone embedded systems.
142 The implementation is in straight, hand-tuned ANSI C. Among other
143 consequences, it uses a lot of macros. Because of this, to be at
144 all usable, this code should be compiled using an optimizing compiler
145 (for example gcc -O2) that can simplify expressions and control
148 __STD_C (default: derived from C compiler defines)
149 Nonzero if using ANSI-standard C compiler, a C++ compiler, or
150 a C compiler sufficiently close to ANSI to get away with it.
151 DEBUG (default: NOT defined)
152 Define to enable debugging. Adds fairly extensive assertion-based
153 checking to help track down memory errors, but noticeably slows down
155 REALLOC_ZERO_BYTES_FREES (default: NOT defined)
156 Define this if you think that realloc(p, 0) should be equivalent
157 to free(p). Otherwise, since malloc returns a unique pointer for
158 malloc(0), so does realloc(p, 0).
159 HAVE_MEMCPY (default: defined)
160 Define if you are not otherwise using ANSI STD C, but still
161 have memcpy and memset in your C library and want to use them.
162 Otherwise, simple internal versions are supplied.
163 USE_MEMCPY (default: 1 if HAVE_MEMCPY is defined, 0 otherwise)
164 Define as 1 if you want the C library versions of memset and
165 memcpy called in realloc and calloc (otherwise macro versions are used).
166 At least on some platforms, the simple macro versions usually
167 outperform libc versions.
168 HAVE_MMAP (default: defined as 1)
169 Define to non-zero to optionally make malloc() use mmap() to
170 allocate very large blocks.
171 HAVE_MREMAP (default: defined as 0 unless Linux libc set)
172 Define to non-zero to optionally make realloc() use mremap() to
173 reallocate very large blocks.
174 malloc_getpagesize (default: derived from system #includes)
175 Either a constant or routine call returning the system page size.
176 HAVE_USR_INCLUDE_MALLOC_H (default: NOT defined)
177 Optionally define if you are on a system with a /usr/include/malloc.h
178 that declares struct mallinfo. It is not at all necessary to
179 define this even if you do, but will ensure consistency.
180 INTERNAL_SIZE_T (default: size_t)
181 Define to a 32-bit type (probably `unsigned int') if you are on a
182 64-bit machine, yet do not want or need to allow malloc requests of
183 greater than 2^31 to be handled. This saves space, especially for
185 INTERNAL_LINUX_C_LIB (default: NOT defined)
186 Defined only when compiled as part of Linux libc.
187 Also note that there is some odd internal name-mangling via defines
188 (for example, internally, `malloc' is named `mALLOc') needed
189 when compiling in this case. These look funny but don't otherwise
191 WIN32 (default: undefined)
192 Define this on MS win (95, nt) platforms to compile in sbrk emulation.
193 LACKS_UNISTD_H (default: undefined if not WIN32)
194 Define this if your system does not have a <unistd.h>.
195 LACKS_SYS_PARAM_H (default: undefined if not WIN32)
196 Define this if your system does not have a <sys/param.h>.
197 MORECORE (default: sbrk)
198 The name of the routine to call to obtain more memory from the system.
199 MORECORE_FAILURE (default: -1)
200 The value returned upon failure of MORECORE.
201 MORECORE_CLEARS (default 1)
202 True (1) if the routine mapped to MORECORE zeroes out memory (which
204 DEFAULT_TRIM_THRESHOLD
206 DEFAULT_MMAP_THRESHOLD
208 Default values of tunable parameters (described in detail below)
209 controlling interaction with host system routines (sbrk, mmap, etc).
210 These values may also be changed dynamically via mallopt(). The
211 preset defaults are those that give best performance for typical
213 USE_DL_PREFIX (default: undefined)
214 Prefix all public routines with the string 'dl'. Useful to
215 quickly avoid procedure declaration conflicts and linker symbol
216 conflicts with existing memory allocation routines.
234 #endif /*__cplusplus*/
239 #if (__STD_C || defined(WIN32))
247 #include <stddef.h> /* for size_t */
249 #include <sys/types.h>
256 #include <stdio.h> /* needed for malloc_stats */
267 Because freed chunks may be overwritten with link fields, this
268 malloc will often die when freed memory is overwritten by user
269 programs. This can be very effective (albeit in an annoying way)
270 in helping track down dangling pointers.
272 If you compile with -DDEBUG, a number of assertion checks are
273 enabled that will catch more memory errors. You probably won't be
274 able to make much sense of the actual assertion errors, but they
275 should help you locate incorrectly overwritten memory. The
276 checking is fairly extensive, and will slow down execution
277 noticeably. Calling malloc_stats or mallinfo with DEBUG set will
278 attempt to check every non-mmapped allocated and free chunk in the
279 course of computing the summmaries. (By nature, mmapped regions
280 cannot be checked very much automatically.)
282 Setting DEBUG may also be helpful if you are trying to modify
283 this code. The assertions in the check routines spell out in more
284 detail the assumptions and invariants underlying the algorithms.
291 #define assert(x) ((void)0)
296 INTERNAL_SIZE_T is the word-size used for internal bookkeeping
297 of chunk sizes. On a 64-bit machine, you can reduce malloc
298 overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int'
299 at the expense of not being able to handle requests greater than
300 2^31. This limitation is hardly ever a concern; you are encouraged
301 to set this. However, the default version is the same as size_t.
304 #ifndef INTERNAL_SIZE_T
305 #define INTERNAL_SIZE_T size_t
309 REALLOC_ZERO_BYTES_FREES should be set if a call to
310 realloc with zero bytes should be the same as a call to free.
311 Some people think it should. Otherwise, since this malloc
312 returns a unique pointer for malloc(0), so does realloc(p, 0).
316 /* #define REALLOC_ZERO_BYTES_FREES */
320 WIN32 causes an emulation of sbrk to be compiled in
321 mmap-based options are not currently supported in WIN32.
326 #define MORECORE wsbrk
329 #define LACKS_UNISTD_H
330 #define LACKS_SYS_PARAM_H
333 Include 'windows.h' to get the necessary declarations for the
334 Microsoft Visual C++ data structures and routines used in the 'sbrk'
337 Define WIN32_LEAN_AND_MEAN so that only the essential Microsoft
338 Visual C++ header files are included.
340 #define WIN32_LEAN_AND_MEAN
346 HAVE_MEMCPY should be defined if you are not otherwise using
347 ANSI STD C, but still have memcpy and memset in your C library
348 and want to use them in calloc and realloc. Otherwise simple
349 macro versions are defined here.
351 USE_MEMCPY should be defined as 1 if you actually want to
352 have memset and memcpy called. People report that the macro
353 versions are often enough faster than libc versions on many
354 systems that it is better to use them.
368 #if (__STD_C || defined(HAVE_MEMCPY))
371 void* memset(void*, int, size_t);
372 void* memcpy(void*, const void*, size_t);
375 /* On Win32 platforms, 'memset()' and 'memcpy()' are already declared in */
386 /* The following macros are only invoked with (2n+1)-multiples of
387 INTERNAL_SIZE_T units, with a positive integer n. This is exploited
388 for fast inline execution when n is small. */
390 #define MALLOC_ZERO(charp, nbytes) \
392 INTERNAL_SIZE_T mzsz = (nbytes); \
393 if(mzsz <= 9*sizeof(mzsz)) { \
394 INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp); \
395 if(mzsz >= 5*sizeof(mzsz)) { *mz++ = 0; \
397 if(mzsz >= 7*sizeof(mzsz)) { *mz++ = 0; \
399 if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0; \
404 } else memset((charp), 0, mzsz); \
407 #define MALLOC_COPY(dest,src,nbytes) \
409 INTERNAL_SIZE_T mcsz = (nbytes); \
410 if(mcsz <= 9*sizeof(mcsz)) { \
411 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src); \
412 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest); \
413 if(mcsz >= 5*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
414 *mcdst++ = *mcsrc++; \
415 if(mcsz >= 7*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
416 *mcdst++ = *mcsrc++; \
417 if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \
418 *mcdst++ = *mcsrc++; }}} \
419 *mcdst++ = *mcsrc++; \
420 *mcdst++ = *mcsrc++; \
422 } else memcpy(dest, src, mcsz); \
425 #else /* !USE_MEMCPY */
427 /* Use Duff's device for good zeroing/copying performance. */
429 #define MALLOC_ZERO(charp, nbytes) \
431 INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp); \
432 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
433 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
435 case 0: for(;;) { *mzp++ = 0; \
436 case 7: *mzp++ = 0; \
437 case 6: *mzp++ = 0; \
438 case 5: *mzp++ = 0; \
439 case 4: *mzp++ = 0; \
440 case 3: *mzp++ = 0; \
441 case 2: *mzp++ = 0; \
442 case 1: *mzp++ = 0; if(mcn <= 0) break; mcn--; } \
446 #define MALLOC_COPY(dest,src,nbytes) \
448 INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src; \
449 INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest; \
450 long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \
451 if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \
453 case 0: for(;;) { *mcdst++ = *mcsrc++; \
454 case 7: *mcdst++ = *mcsrc++; \
455 case 6: *mcdst++ = *mcsrc++; \
456 case 5: *mcdst++ = *mcsrc++; \
457 case 4: *mcdst++ = *mcsrc++; \
458 case 3: *mcdst++ = *mcsrc++; \
459 case 2: *mcdst++ = *mcsrc++; \
460 case 1: *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; } \
468 Define HAVE_MMAP to optionally make malloc() use mmap() to
469 allocate very large blocks. These will be returned to the
470 operating system immediately after a free().
478 Define HAVE_MREMAP to make realloc() use mremap() to re-allocate
479 large blocks. This is currently only possible on Linux with
480 kernel versions newer than 1.3.77.
484 #ifdef INTERNAL_LINUX_C_LIB
485 #define HAVE_MREMAP 1
487 #define HAVE_MREMAP 0
495 #include <sys/mman.h>
497 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
498 #define MAP_ANONYMOUS MAP_ANON
501 #endif /* HAVE_MMAP */
504 Access to system page size. To the extent possible, this malloc
505 manages memory from the system in page-size units.
507 The following mechanics for getpagesize were adapted from
508 bsd/gnu getpagesize.h
511 #ifndef LACKS_UNISTD_H
515 #ifndef malloc_getpagesize
516 # ifdef _SC_PAGESIZE /* some SVR4 systems omit an underscore */
517 # ifndef _SC_PAGE_SIZE
518 # define _SC_PAGE_SIZE _SC_PAGESIZE
521 # ifdef _SC_PAGE_SIZE
522 # define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
524 # if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
525 extern size_t getpagesize();
526 # define malloc_getpagesize getpagesize()
529 # define malloc_getpagesize (4096) /* TBD: Use 'GetSystemInfo' instead */
531 # ifndef LACKS_SYS_PARAM_H
532 # include <sys/param.h>
534 # ifdef EXEC_PAGESIZE
535 # define malloc_getpagesize EXEC_PAGESIZE
539 # define malloc_getpagesize NBPG
541 # define malloc_getpagesize (NBPG * CLSIZE)
545 # define malloc_getpagesize NBPC
548 # define malloc_getpagesize PAGESIZE
550 # define malloc_getpagesize (4096) /* just guess */
563 This version of malloc supports the standard SVID/XPG mallinfo
564 routine that returns a struct containing the same kind of
565 information you can get from malloc_stats. It should work on
566 any SVID/XPG compliant system that has a /usr/include/malloc.h
567 defining struct mallinfo. (If you'd like to install such a thing
568 yourself, cut out the preliminary declarations as described above
569 and below and save them in a malloc.h file. But there's no
570 compelling reason to bother to do this.)
572 The main declaration needed is the mallinfo struct that is returned
573 (by-copy) by mallinfo(). The SVID/XPG malloinfo struct contains a
574 bunch of fields, most of which are not even meaningful in this
575 version of malloc. Some of these fields are are instead filled by
576 mallinfo() with other numbers that might possibly be of interest.
578 HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
579 /usr/include/malloc.h file that includes a declaration of struct
580 mallinfo. If so, it is included; else an SVID2/XPG2 compliant
581 version is declared below. These must be precisely the same for
586 /* #define HAVE_USR_INCLUDE_MALLOC_H */
588 #if HAVE_USR_INCLUDE_MALLOC_H
589 #include "/usr/include/malloc.h"
592 /* SVID2/XPG mallinfo structure */
595 int arena; /* total space allocated from system */
596 int ordblks; /* number of non-inuse chunks */
597 int smblks; /* unused -- always zero */
598 int hblks; /* number of mmapped regions */
599 int hblkhd; /* total space in mmapped regions */
600 int usmblks; /* unused -- always zero */
601 int fsmblks; /* unused -- always zero */
602 int uordblks; /* total allocated space */
603 int fordblks; /* total non-inuse space */
604 int keepcost; /* top-most, releasable (via malloc_trim) space */
607 /* SVID2/XPG mallopt options */
609 #define M_MXFAST 1 /* UNUSED in this malloc */
610 #define M_NLBLKS 2 /* UNUSED in this malloc */
611 #define M_GRAIN 3 /* UNUSED in this malloc */
612 #define M_KEEP 4 /* UNUSED in this malloc */
616 /* mallopt options that actually do something */
618 #define M_TRIM_THRESHOLD -1
620 #define M_MMAP_THRESHOLD -3
621 #define M_MMAP_MAX -4
624 #ifndef DEFAULT_TRIM_THRESHOLD
625 #define DEFAULT_TRIM_THRESHOLD (128 * 1024)
629 M_TRIM_THRESHOLD is the maximum amount of unused top-most memory
630 to keep before releasing via malloc_trim in free().
632 Automatic trimming is mainly useful in long-lived programs.
633 Because trimming via sbrk can be slow on some systems, and can
634 sometimes be wasteful (in cases where programs immediately
635 afterward allocate more large chunks) the value should be high
636 enough so that your overall system performance would improve by
639 The trim threshold and the mmap control parameters (see below)
640 can be traded off with one another. Trimming and mmapping are
641 two different ways of releasing unused memory back to the
642 system. Between these two, it is often possible to keep
643 system-level demands of a long-lived program down to a bare
644 minimum. For example, in one test suite of sessions measuring
645 the XF86 X server on Linux, using a trim threshold of 128K and a
646 mmap threshold of 192K led to near-minimal long term resource
649 If you are using this malloc in a long-lived program, it should
650 pay to experiment with these values. As a rough guide, you
651 might set to a value close to the average size of a process
652 (program) running on your system. Releasing this much memory
653 would allow such a process to run in memory. Generally, it's
654 worth it to tune for trimming rather tham memory mapping when a
655 program undergoes phases where several large chunks are
656 allocated and released in ways that can reuse each other's
657 storage, perhaps mixed with phases where there are no such
658 chunks at all. And in well-behaved long-lived programs,
659 controlling release of large blocks via trimming versus mapping
662 However, in most programs, these parameters serve mainly as
663 protection against the system-level effects of carrying around
664 massive amounts of unneeded memory. Since frequent calls to
665 sbrk, mmap, and munmap otherwise degrade performance, the default
666 parameters are set to relatively high values that serve only as
669 The default trim value is high enough to cause trimming only in
670 fairly extreme (by current memory consumption standards) cases.
671 It must be greater than page size to have any useful effect. To
672 disable trimming completely, you can set to (unsigned long)(-1);
678 #ifndef DEFAULT_TOP_PAD
679 #define DEFAULT_TOP_PAD (0)
683 M_TOP_PAD is the amount of extra `padding' space to allocate or
684 retain whenever sbrk is called. It is used in two ways internally:
686 * When sbrk is called to extend the top of the arena to satisfy
687 a new malloc request, this much padding is added to the sbrk
690 * When malloc_trim is called automatically from free(),
691 it is used as the `pad' argument.
693 In both cases, the actual amount of padding is rounded
694 so that the end of the arena is always a system page boundary.
696 The main reason for using padding is to avoid calling sbrk so
697 often. Having even a small pad greatly reduces the likelihood
698 that nearly every malloc request during program start-up (or
699 after trimming) will invoke sbrk, which needlessly wastes
702 Automatic rounding-up to page-size units is normally sufficient
703 to avoid measurable overhead, so the default is 0. However, in
704 systems where sbrk is relatively slow, it can pay to increase
705 this value, at the expense of carrying around more memory than
711 #ifndef DEFAULT_MMAP_THRESHOLD
712 #define DEFAULT_MMAP_THRESHOLD (128 * 1024)
717 M_MMAP_THRESHOLD is the request size threshold for using mmap()
718 to service a request. Requests of at least this size that cannot
719 be allocated using already-existing space will be serviced via mmap.
720 (If enough normal freed space already exists it is used instead.)
722 Using mmap segregates relatively large chunks of memory so that
723 they can be individually obtained and released from the host
724 system. A request serviced through mmap is never reused by any
725 other request (at least not directly; the system may just so
726 happen to remap successive requests to the same locations).
728 Segregating space in this way has the benefit that mmapped space
729 can ALWAYS be individually released back to the system, which
730 helps keep the system level memory demands of a long-lived
731 program low. Mapped memory can never become `locked' between
732 other chunks, as can happen with normally allocated chunks, which
733 menas that even trimming via malloc_trim would not release them.
735 However, it has the disadvantages that:
737 1. The space cannot be reclaimed, consolidated, and then
738 used to service later requests, as happens with normal chunks.
739 2. It can lead to more wastage because of mmap page alignment
741 3. It causes malloc performance to be more dependent on host
742 system memory management support routines which may vary in
743 implementation quality and may impose arbitrary
744 limitations. Generally, servicing a request via normal
745 malloc steps is faster than going through a system's mmap.
747 All together, these considerations should lead you to use mmap
748 only for relatively large requests.
754 #ifndef DEFAULT_MMAP_MAX
756 #define DEFAULT_MMAP_MAX (64)
758 #define DEFAULT_MMAP_MAX (0)
763 M_MMAP_MAX is the maximum number of requests to simultaneously
764 service using mmap. This parameter exists because:
766 1. Some systems have a limited number of internal tables for
768 2. In most systems, overreliance on mmap can degrade overall
770 3. If a program allocates many large regions, it is probably
771 better off using normal sbrk-based allocation routines that
772 can reclaim and reallocate normal heap memory. Using a
773 small value allows transition into this mode after the
774 first few allocations.
776 Setting to 0 disables all use of mmap. If HAVE_MMAP is not set,
777 the default value is 0, and attempts to set it to non-zero values
778 in mallopt will fail.
783 USE_DL_PREFIX will prefix all public routines with the string 'dl'.
784 Useful to quickly avoid procedure declaration conflicts and linker
785 symbol conflicts with existing memory allocation routines.
789 /* #define USE_DL_PREFIX */
794 Special defines for linux libc
796 Except when compiled using these special defines for Linux libc
797 using weak aliases, this malloc is NOT designed to work in
798 multithreaded applications. No semaphores or other concurrency
799 control are provided to ensure that multiple malloc or free calls
800 don't run at the same time, which could be disasterous. A single
801 semaphore could be used across malloc, realloc, and free (which is
802 essentially the effect of the linux weak alias approach). It would
803 be hard to obtain finer granularity.
808 #ifdef INTERNAL_LINUX_C_LIB
812 Void_t * __default_morecore_init (ptrdiff_t);
813 Void_t *(*__morecore)(ptrdiff_t) = __default_morecore_init;
817 Void_t * __default_morecore_init ();
818 Void_t *(*__morecore)() = __default_morecore_init;
822 #define MORECORE (*__morecore)
823 #define MORECORE_FAILURE 0
824 #define MORECORE_CLEARS 1
826 #else /* INTERNAL_LINUX_C_LIB */
829 extern Void_t* sbrk(ptrdiff_t);
831 extern Void_t* sbrk();
835 #define MORECORE sbrk
838 #ifndef MORECORE_FAILURE
839 #define MORECORE_FAILURE -1
842 #ifndef MORECORE_CLEARS
843 #define MORECORE_CLEARS 1
846 #endif /* INTERNAL_LINUX_C_LIB */
848 #if defined(INTERNAL_LINUX_C_LIB) && defined(__ELF__)
850 #define cALLOc __libc_calloc
851 #define fREe __libc_free
852 #define mALLOc __libc_malloc
853 #define mEMALIGn __libc_memalign
854 #define rEALLOc __libc_realloc
855 #define vALLOc __libc_valloc
856 #define pvALLOc __libc_pvalloc
857 #define mALLINFo __libc_mallinfo
858 #define mALLOPt __libc_mallopt
860 #pragma weak calloc = __libc_calloc
861 #pragma weak free = __libc_free
862 #pragma weak cfree = __libc_free
863 #pragma weak malloc = __libc_malloc
864 #pragma weak memalign = __libc_memalign
865 #pragma weak realloc = __libc_realloc
866 #pragma weak valloc = __libc_valloc
867 #pragma weak pvalloc = __libc_pvalloc
868 #pragma weak mallinfo = __libc_mallinfo
869 #pragma weak mallopt = __libc_mallopt
874 #define cALLOc dlcalloc
876 #define mALLOc dlmalloc
877 #define mEMALIGn dlmemalign
878 #define rEALLOc dlrealloc
879 #define vALLOc dlvalloc
880 #define pvALLOc dlpvalloc
881 #define mALLINFo dlmallinfo
882 #define mALLOPt dlmallopt
883 #else /* USE_DL_PREFIX */
884 #define cALLOc calloc
886 #define mALLOc malloc
887 #define mEMALIGn memalign
888 #define rEALLOc realloc
889 #define vALLOc valloc
890 #define pvALLOc pvalloc
891 #define mALLINFo mallinfo
892 #define mALLOPt mallopt
893 #endif /* USE_DL_PREFIX */
897 /* Public routines */
901 Void_t* mALLOc(size_t);
903 Void_t* rEALLOc(Void_t*, size_t);
904 Void_t* mEMALIGn(size_t, size_t);
905 Void_t* vALLOc(size_t);
906 Void_t* pvALLOc(size_t);
907 Void_t* cALLOc(size_t, size_t);
909 int malloc_trim(size_t);
910 size_t malloc_usable_size(Void_t*);
912 int mALLOPt(int, int);
913 struct mallinfo mALLINFo(void);
924 size_t malloc_usable_size();
927 struct mallinfo mALLINFo();
932 }; /* end of extern "C" */
935 /* ---------- To make a malloc.h, end cutting here ------------ */
936 #else /* Moved to malloc.h */
941 static void malloc_update_mallinfo (void);
942 void malloc_stats (void);
944 static void malloc_update_mallinfo ();
949 #endif /* 0 */ /* Moved to malloc.h */
952 DECLARE_GLOBAL_DATA_PTR;
955 Emulation of sbrk for WIN32
956 All code within the ifdef WIN32 is untested by me.
958 Thanks to Martin Fong and others for supplying this.
964 #define AlignPage(add) (((add) + (malloc_getpagesize-1)) & \
965 ~(malloc_getpagesize-1))
966 #define AlignPage64K(add) (((add) + (0x10000 - 1)) & ~(0x10000 - 1))
968 /* resrve 64MB to insure large contiguous space */
969 #define RESERVED_SIZE (1024*1024*64)
970 #define NEXT_SIZE (2048*1024)
971 #define TOP_MEMORY ((unsigned long)2*1024*1024*1024)
973 struct GmListElement;
974 typedef struct GmListElement GmListElement;
982 static GmListElement* head = 0;
983 static unsigned int gNextAddress = 0;
984 static unsigned int gAddressBase = 0;
985 static unsigned int gAllocatedSize = 0;
988 GmListElement* makeGmListElement (void* bas)
991 this = (GmListElement*)(void*)LocalAlloc (0, sizeof (GmListElement));
1005 assert ( (head == NULL) || (head->base == (void*)gAddressBase));
1006 if (gAddressBase && (gNextAddress - gAddressBase))
1008 rval = VirtualFree ((void*)gAddressBase,
1009 gNextAddress - gAddressBase,
1015 GmListElement* next = head->next;
1016 rval = VirtualFree (head->base, 0, MEM_RELEASE);
1024 void* findRegion (void* start_address, unsigned long size)
1026 MEMORY_BASIC_INFORMATION info;
1027 if (size >= TOP_MEMORY) return NULL;
1029 while ((unsigned long)start_address + size < TOP_MEMORY)
1031 VirtualQuery (start_address, &info, sizeof (info));
1032 if ((info.State == MEM_FREE) && (info.RegionSize >= size))
1033 return start_address;
1036 /* Requested region is not available so see if the */
1037 /* next region is available. Set 'start_address' */
1038 /* to the next region and call 'VirtualQuery()' */
1041 start_address = (char*)info.BaseAddress + info.RegionSize;
1043 /* Make sure we start looking for the next region */
1044 /* on the *next* 64K boundary. Otherwise, even if */
1045 /* the new region is free according to */
1046 /* 'VirtualQuery()', the subsequent call to */
1047 /* 'VirtualAlloc()' (which follows the call to */
1048 /* this routine in 'wsbrk()') will round *down* */
1049 /* the requested address to a 64K boundary which */
1050 /* we already know is an address in the */
1051 /* unavailable region. Thus, the subsequent call */
1052 /* to 'VirtualAlloc()' will fail and bring us back */
1053 /* here, causing us to go into an infinite loop. */
1056 (void *) AlignPage64K((unsigned long) start_address);
1064 void* wsbrk (long size)
1069 if (gAddressBase == 0)
1071 gAllocatedSize = max (RESERVED_SIZE, AlignPage (size));
1072 gNextAddress = gAddressBase =
1073 (unsigned int)VirtualAlloc (NULL, gAllocatedSize,
1074 MEM_RESERVE, PAGE_NOACCESS);
1075 } else if (AlignPage (gNextAddress + size) > (gAddressBase +
1078 long new_size = max (NEXT_SIZE, AlignPage (size));
1079 void* new_address = (void*)(gAddressBase+gAllocatedSize);
1082 new_address = findRegion (new_address, new_size);
1084 if (new_address == 0)
1087 gAddressBase = gNextAddress =
1088 (unsigned int)VirtualAlloc (new_address, new_size,
1089 MEM_RESERVE, PAGE_NOACCESS);
1090 /* repeat in case of race condition */
1091 /* The region that we found has been snagged */
1092 /* by another thread */
1094 while (gAddressBase == 0);
1096 assert (new_address == (void*)gAddressBase);
1098 gAllocatedSize = new_size;
1100 if (!makeGmListElement ((void*)gAddressBase))
1103 if ((size + gNextAddress) > AlignPage (gNextAddress))
1106 res = VirtualAlloc ((void*)AlignPage (gNextAddress),
1107 (size + gNextAddress -
1108 AlignPage (gNextAddress)),
1109 MEM_COMMIT, PAGE_READWRITE);
1113 tmp = (void*)gNextAddress;
1114 gNextAddress = (unsigned int)tmp + size;
1119 unsigned int alignedGoal = AlignPage (gNextAddress + size);
1120 /* Trim by releasing the virtual memory */
1121 if (alignedGoal >= gAddressBase)
1123 VirtualFree ((void*)alignedGoal, gNextAddress - alignedGoal,
1125 gNextAddress = gNextAddress + size;
1126 return (void*)gNextAddress;
1130 VirtualFree ((void*)gAddressBase, gNextAddress - gAddressBase,
1132 gNextAddress = gAddressBase;
1138 return (void*)gNextAddress;
1153 INTERNAL_SIZE_T prev_size; /* Size of previous chunk (if free). */
1154 INTERNAL_SIZE_T size; /* Size in bytes, including overhead. */
1155 struct malloc_chunk* fd; /* double links -- used only if free. */
1156 struct malloc_chunk* bk;
1159 typedef struct malloc_chunk* mchunkptr;
1163 malloc_chunk details:
1165 (The following includes lightly edited explanations by Colin Plumb.)
1167 Chunks of memory are maintained using a `boundary tag' method as
1168 described in e.g., Knuth or Standish. (See the paper by Paul
1169 Wilson ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a
1170 survey of such techniques.) Sizes of free chunks are stored both
1171 in the front of each chunk and at the end. This makes
1172 consolidating fragmented chunks into bigger chunks very fast. The
1173 size fields also hold bits representing whether chunks are free or
1176 An allocated chunk looks like this:
1179 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1180 | Size of previous chunk, if allocated | |
1181 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1182 | Size of chunk, in bytes |P|
1183 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1184 | User data starts here... .
1186 . (malloc_usable_space() bytes) .
1188 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1190 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1193 Where "chunk" is the front of the chunk for the purpose of most of
1194 the malloc code, but "mem" is the pointer that is returned to the
1195 user. "Nextchunk" is the beginning of the next contiguous chunk.
1197 Chunks always begin on even word boundries, so the mem portion
1198 (which is returned to the user) is also on an even word boundary, and
1199 thus double-word aligned.
1201 Free chunks are stored in circular doubly-linked lists, and look like this:
1203 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1204 | Size of previous chunk |
1205 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1206 `head:' | Size of chunk, in bytes |P|
1207 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1208 | Forward pointer to next chunk in list |
1209 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1210 | Back pointer to previous chunk in list |
1211 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1212 | Unused space (may be 0 bytes long) .
1215 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1216 `foot:' | Size of chunk, in bytes |
1217 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1219 The P (PREV_INUSE) bit, stored in the unused low-order bit of the
1220 chunk size (which is always a multiple of two words), is an in-use
1221 bit for the *previous* chunk. If that bit is *clear*, then the
1222 word before the current chunk size contains the previous chunk
1223 size, and can be used to find the front of the previous chunk.
1224 (The very first chunk allocated always has this bit set,
1225 preventing access to non-existent (or non-owned) memory.)
1227 Note that the `foot' of the current chunk is actually represented
1228 as the prev_size of the NEXT chunk. (This makes it easier to
1229 deal with alignments etc).
1231 The two exceptions to all this are
1233 1. The special chunk `top', which doesn't bother using the
1234 trailing size field since there is no
1235 next contiguous chunk that would have to index off it. (After
1236 initialization, `top' is forced to always exist. If it would
1237 become less than MINSIZE bytes long, it is replenished via
1240 2. Chunks allocated via mmap, which have the second-lowest-order
1241 bit (IS_MMAPPED) set in their size fields. Because they are
1242 never merged or traversed from any other chunk, they have no
1243 foot size or inuse information.
1245 Available chunks are kept in any of several places (all declared below):
1247 * `av': An array of chunks serving as bin headers for consolidated
1248 chunks. Each bin is doubly linked. The bins are approximately
1249 proportionally (log) spaced. There are a lot of these bins
1250 (128). This may look excessive, but works very well in
1251 practice. All procedures maintain the invariant that no
1252 consolidated chunk physically borders another one. Chunks in
1253 bins are kept in size order, with ties going to the
1254 approximately least recently used chunk.
1256 The chunks in each bin are maintained in decreasing sorted order by
1257 size. This is irrelevant for the small bins, which all contain
1258 the same-sized chunks, but facilitates best-fit allocation for
1259 larger chunks. (These lists are just sequential. Keeping them in
1260 order almost never requires enough traversal to warrant using
1261 fancier ordered data structures.) Chunks of the same size are
1262 linked with the most recently freed at the front, and allocations
1263 are taken from the back. This results in LRU or FIFO allocation
1264 order, which tends to give each chunk an equal opportunity to be
1265 consolidated with adjacent freed chunks, resulting in larger free
1266 chunks and less fragmentation.
1268 * `top': The top-most available chunk (i.e., the one bordering the
1269 end of available memory) is treated specially. It is never
1270 included in any bin, is used only if no other chunk is
1271 available, and is released back to the system if it is very
1272 large (see M_TRIM_THRESHOLD).
1274 * `last_remainder': A bin holding only the remainder of the
1275 most recently split (non-top) chunk. This bin is checked
1276 before other non-fitting chunks, so as to provide better
1277 locality for runs of sequentially allocated chunks.
1279 * Implicitly, through the host system's memory mapping tables.
1280 If supported, requests greater than a threshold are usually
1281 serviced via calls to mmap, and then later released via munmap.
1285 /* sizes, alignments */
1287 #define SIZE_SZ (sizeof(INTERNAL_SIZE_T))
1288 #define MALLOC_ALIGNMENT (SIZE_SZ + SIZE_SZ)
1289 #define MALLOC_ALIGN_MASK (MALLOC_ALIGNMENT - 1)
1290 #define MINSIZE (sizeof(struct malloc_chunk))
1292 /* conversion from malloc headers to user pointers, and back */
1294 #define chunk2mem(p) ((Void_t*)((char*)(p) + 2*SIZE_SZ))
1295 #define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ))
1297 /* pad request bytes into a usable size */
1299 #define request2size(req) \
1300 (((long)((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) < \
1301 (long)(MINSIZE + MALLOC_ALIGN_MASK)) ? MINSIZE : \
1302 (((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) & ~(MALLOC_ALIGN_MASK)))
1304 /* Check if m has acceptable alignment */
1306 #define aligned_OK(m) (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
1312 Physical chunk operations
1316 /* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
1318 #define PREV_INUSE 0x1
1320 /* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
1322 #define IS_MMAPPED 0x2
1324 /* Bits to mask off when extracting size */
1326 #define SIZE_BITS (PREV_INUSE|IS_MMAPPED)
1329 /* Ptr to next physical malloc_chunk. */
1331 #define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~PREV_INUSE) ))
1333 /* Ptr to previous physical malloc_chunk */
1335 #define prev_chunk(p)\
1336 ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) ))
1339 /* Treat space at ptr + offset as a chunk */
1341 #define chunk_at_offset(p, s) ((mchunkptr)(((char*)(p)) + (s)))
1347 Dealing with use bits
1350 /* extract p's inuse bit */
1353 ((((mchunkptr)(((char*)(p))+((p)->size & ~PREV_INUSE)))->size) & PREV_INUSE)
1355 /* extract inuse bit of previous chunk */
1357 #define prev_inuse(p) ((p)->size & PREV_INUSE)
1359 /* check for mmap()'ed chunk */
1361 #define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
1363 /* set/clear chunk as in use without otherwise disturbing */
1365 #define set_inuse(p)\
1366 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size |= PREV_INUSE
1368 #define clear_inuse(p)\
1369 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size &= ~(PREV_INUSE)
1371 /* check/set/clear inuse bits in known places */
1373 #define inuse_bit_at_offset(p, s)\
1374 (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE)
1376 #define set_inuse_bit_at_offset(p, s)\
1377 (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE)
1379 #define clear_inuse_bit_at_offset(p, s)\
1380 (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE))
1386 Dealing with size fields
1389 /* Get size, ignoring use bits */
1391 #define chunksize(p) ((p)->size & ~(SIZE_BITS))
1393 /* Set size at head, without disturbing its use bit */
1395 #define set_head_size(p, s) ((p)->size = (((p)->size & PREV_INUSE) | (s)))
1397 /* Set size/use ignoring previous bits in header */
1399 #define set_head(p, s) ((p)->size = (s))
1401 /* Set size at footer (only when chunk is not in use) */
1403 #define set_foot(p, s) (((mchunkptr)((char*)(p) + (s)))->prev_size = (s))
1412 The bins, `av_' are an array of pairs of pointers serving as the
1413 heads of (initially empty) doubly-linked lists of chunks, laid out
1414 in a way so that each pair can be treated as if it were in a
1415 malloc_chunk. (This way, the fd/bk offsets for linking bin heads
1416 and chunks are the same).
1418 Bins for sizes < 512 bytes contain chunks of all the same size, spaced
1419 8 bytes apart. Larger bins are approximately logarithmically
1420 spaced. (See the table below.) The `av_' array is never mentioned
1421 directly in the code, but instead via bin access macros.
1429 4 bins of size 32768
1430 2 bins of size 262144
1431 1 bin of size what's left
1433 There is actually a little bit of slop in the numbers in bin_index
1434 for the sake of speed. This makes no difference elsewhere.
1436 The special chunks `top' and `last_remainder' get their own bins,
1437 (this is implemented via yet more trickery with the av_ array),
1438 although `top' is never properly linked to its bin since it is
1439 always handled specially.
1443 #define NAV 128 /* number of bins */
1445 typedef struct malloc_chunk* mbinptr;
1449 #define bin_at(i) ((mbinptr)((char*)&(av_[2*(i) + 2]) - 2*SIZE_SZ))
1450 #define next_bin(b) ((mbinptr)((char*)(b) + 2 * sizeof(mbinptr)))
1451 #define prev_bin(b) ((mbinptr)((char*)(b) - 2 * sizeof(mbinptr)))
1454 The first 2 bins are never indexed. The corresponding av_ cells are instead
1455 used for bookkeeping. This is not to save space, but to simplify
1456 indexing, maintain locality, and avoid some initialization tests.
1459 #define top (bin_at(0)->fd) /* The topmost chunk */
1460 #define last_remainder (bin_at(1)) /* remainder from last split */
1464 Because top initially points to its own bin with initial
1465 zero size, thus forcing extension on the first malloc request,
1466 we avoid having any special code in malloc to check whether
1467 it even exists yet. But we still need to in malloc_extend_top.
1470 #define initial_top ((mchunkptr)(bin_at(0)))
1472 /* Helper macro to initialize bins */
1474 #define IAV(i) bin_at(i), bin_at(i)
1476 static mbinptr av_[NAV * 2 + 2] = {
1478 IAV(0), IAV(1), IAV(2), IAV(3), IAV(4), IAV(5), IAV(6), IAV(7),
1479 IAV(8), IAV(9), IAV(10), IAV(11), IAV(12), IAV(13), IAV(14), IAV(15),
1480 IAV(16), IAV(17), IAV(18), IAV(19), IAV(20), IAV(21), IAV(22), IAV(23),
1481 IAV(24), IAV(25), IAV(26), IAV(27), IAV(28), IAV(29), IAV(30), IAV(31),
1482 IAV(32), IAV(33), IAV(34), IAV(35), IAV(36), IAV(37), IAV(38), IAV(39),
1483 IAV(40), IAV(41), IAV(42), IAV(43), IAV(44), IAV(45), IAV(46), IAV(47),
1484 IAV(48), IAV(49), IAV(50), IAV(51), IAV(52), IAV(53), IAV(54), IAV(55),
1485 IAV(56), IAV(57), IAV(58), IAV(59), IAV(60), IAV(61), IAV(62), IAV(63),
1486 IAV(64), IAV(65), IAV(66), IAV(67), IAV(68), IAV(69), IAV(70), IAV(71),
1487 IAV(72), IAV(73), IAV(74), IAV(75), IAV(76), IAV(77), IAV(78), IAV(79),
1488 IAV(80), IAV(81), IAV(82), IAV(83), IAV(84), IAV(85), IAV(86), IAV(87),
1489 IAV(88), IAV(89), IAV(90), IAV(91), IAV(92), IAV(93), IAV(94), IAV(95),
1490 IAV(96), IAV(97), IAV(98), IAV(99), IAV(100), IAV(101), IAV(102), IAV(103),
1491 IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111),
1492 IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119),
1493 IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127)
1496 void malloc_bin_reloc (void)
1498 unsigned long *p = (unsigned long *)(&av_[2]);
1500 for (i=2; i<(sizeof(av_)/sizeof(mbinptr)); ++i) {
1501 *p++ += gd->reloc_off;
1506 /* field-extraction macros */
1508 #define first(b) ((b)->fd)
1509 #define last(b) ((b)->bk)
1515 #define bin_index(sz) \
1516 (((((unsigned long)(sz)) >> 9) == 0) ? (((unsigned long)(sz)) >> 3): \
1517 ((((unsigned long)(sz)) >> 9) <= 4) ? 56 + (((unsigned long)(sz)) >> 6): \
1518 ((((unsigned long)(sz)) >> 9) <= 20) ? 91 + (((unsigned long)(sz)) >> 9): \
1519 ((((unsigned long)(sz)) >> 9) <= 84) ? 110 + (((unsigned long)(sz)) >> 12): \
1520 ((((unsigned long)(sz)) >> 9) <= 340) ? 119 + (((unsigned long)(sz)) >> 15): \
1521 ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18): \
1524 bins for chunks < 512 are all spaced 8 bytes apart, and hold
1525 identically sized chunks. This is exploited in malloc.
1528 #define MAX_SMALLBIN 63
1529 #define MAX_SMALLBIN_SIZE 512
1530 #define SMALLBIN_WIDTH 8
1532 #define smallbin_index(sz) (((unsigned long)(sz)) >> 3)
1535 Requests are `small' if both the corresponding and the next bin are small
1538 #define is_small_request(nb) (nb < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH)
1543 To help compensate for the large number of bins, a one-level index
1544 structure is used for bin-by-bin searching. `binblocks' is a
1545 one-word bitvector recording whether groups of BINBLOCKWIDTH bins
1546 have any (possibly) non-empty bins, so they can be skipped over
1547 all at once during during traversals. The bits are NOT always
1548 cleared as soon as all bins in a block are empty, but instead only
1549 when all are noticed to be empty during traversal in malloc.
1552 #define BINBLOCKWIDTH 4 /* bins per block */
1554 #define binblocks (bin_at(0)->size) /* bitvector of nonempty blocks */
1556 /* bin<->block macros */
1558 #define idx2binblock(ix) ((unsigned)1 << (ix / BINBLOCKWIDTH))
1559 #define mark_binblock(ii) (binblocks |= idx2binblock(ii))
1560 #define clear_binblock(ii) (binblocks &= ~(idx2binblock(ii)))
1566 /* Other static bookkeeping data */
1568 /* variables holding tunable values */
1570 static unsigned long trim_threshold = DEFAULT_TRIM_THRESHOLD;
1571 static unsigned long top_pad = DEFAULT_TOP_PAD;
1572 static unsigned int n_mmaps_max = DEFAULT_MMAP_MAX;
1573 static unsigned long mmap_threshold = DEFAULT_MMAP_THRESHOLD;
1575 /* The first value returned from sbrk */
1576 static char* sbrk_base = (char*)(-1);
1578 /* The maximum memory obtained from system via sbrk */
1579 static unsigned long max_sbrked_mem = 0;
1581 /* The maximum via either sbrk or mmap */
1582 static unsigned long max_total_mem = 0;
1584 /* internal working copy of mallinfo */
1585 static struct mallinfo current_mallinfo = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
1587 /* The total memory obtained from system via sbrk */
1588 #define sbrked_mem (current_mallinfo.arena)
1590 /* Tracking mmaps */
1593 static unsigned int n_mmaps = 0;
1595 static unsigned long mmapped_mem = 0;
1597 static unsigned int max_n_mmaps = 0;
1598 static unsigned long max_mmapped_mem = 0;
1611 These routines make a number of assertions about the states
1612 of data structures that should be true at all times. If any
1613 are not true, it's very likely that a user program has somehow
1614 trashed memory. (It's also possible that there is a coding error
1615 in malloc. In which case, please report it!)
1619 static void do_check_chunk(mchunkptr p)
1621 static void do_check_chunk(p) mchunkptr p;
1624 #if 0 /* causes warnings because assert() is off */
1625 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1628 /* No checkable chunk is mmapped */
1629 assert(!chunk_is_mmapped(p));
1631 /* Check for legal address ... */
1632 assert((char*)p >= sbrk_base);
1634 assert((char*)p + sz <= (char*)top);
1636 assert((char*)p + sz <= sbrk_base + sbrked_mem);
1642 static void do_check_free_chunk(mchunkptr p)
1644 static void do_check_free_chunk(p) mchunkptr p;
1647 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1648 #if 0 /* causes warnings because assert() is off */
1649 mchunkptr next = chunk_at_offset(p, sz);
1654 /* Check whether it claims to be free ... */
1657 /* Unless a special marker, must have OK fields */
1658 if ((long)sz >= (long)MINSIZE)
1660 assert((sz & MALLOC_ALIGN_MASK) == 0);
1661 assert(aligned_OK(chunk2mem(p)));
1662 /* ... matching footer field */
1663 assert(next->prev_size == sz);
1664 /* ... and is fully consolidated */
1665 assert(prev_inuse(p));
1666 assert (next == top || inuse(next));
1668 /* ... and has minimally sane links */
1669 assert(p->fd->bk == p);
1670 assert(p->bk->fd == p);
1672 else /* markers are always of size SIZE_SZ */
1673 assert(sz == SIZE_SZ);
1677 static void do_check_inuse_chunk(mchunkptr p)
1679 static void do_check_inuse_chunk(p) mchunkptr p;
1682 mchunkptr next = next_chunk(p);
1685 /* Check whether it claims to be in use ... */
1688 /* ... and is surrounded by OK chunks.
1689 Since more things can be checked with free chunks than inuse ones,
1690 if an inuse chunk borders them and debug is on, it's worth doing them.
1694 mchunkptr prv = prev_chunk(p);
1695 assert(next_chunk(prv) == p);
1696 do_check_free_chunk(prv);
1700 assert(prev_inuse(next));
1701 assert(chunksize(next) >= MINSIZE);
1703 else if (!inuse(next))
1704 do_check_free_chunk(next);
1709 static void do_check_malloced_chunk(mchunkptr p, INTERNAL_SIZE_T s)
1711 static void do_check_malloced_chunk(p, s) mchunkptr p; INTERNAL_SIZE_T s;
1714 #if 0 /* causes warnings because assert() is off */
1715 INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1719 do_check_inuse_chunk(p);
1721 /* Legal size ... */
1722 assert((long)sz >= (long)MINSIZE);
1723 assert((sz & MALLOC_ALIGN_MASK) == 0);
1725 assert(room < (long)MINSIZE);
1727 /* ... and alignment */
1728 assert(aligned_OK(chunk2mem(p)));
1731 /* ... and was allocated at front of an available chunk */
1732 assert(prev_inuse(p));
1737 #define check_free_chunk(P) do_check_free_chunk(P)
1738 #define check_inuse_chunk(P) do_check_inuse_chunk(P)
1739 #define check_chunk(P) do_check_chunk(P)
1740 #define check_malloced_chunk(P,N) do_check_malloced_chunk(P,N)
1742 #define check_free_chunk(P)
1743 #define check_inuse_chunk(P)
1744 #define check_chunk(P)
1745 #define check_malloced_chunk(P,N)
1751 Macro-based internal utilities
1756 Linking chunks in bin lists.
1757 Call these only with variables, not arbitrary expressions, as arguments.
1761 Place chunk p of size s in its bin, in size order,
1762 putting it ahead of others of same size.
1766 #define frontlink(P, S, IDX, BK, FD) \
1768 if (S < MAX_SMALLBIN_SIZE) \
1770 IDX = smallbin_index(S); \
1771 mark_binblock(IDX); \
1776 FD->bk = BK->fd = P; \
1780 IDX = bin_index(S); \
1783 if (FD == BK) mark_binblock(IDX); \
1786 while (FD != BK && S < chunksize(FD)) FD = FD->fd; \
1791 FD->bk = BK->fd = P; \
1796 /* take a chunk off a list */
1798 #define unlink(P, BK, FD) \
1806 /* Place p as the last remainder */
1808 #define link_last_remainder(P) \
1810 last_remainder->fd = last_remainder->bk = P; \
1811 P->fd = P->bk = last_remainder; \
1814 /* Clear the last_remainder bin */
1816 #define clear_last_remainder \
1817 (last_remainder->fd = last_remainder->bk = last_remainder)
1823 /* Routines dealing with mmap(). */
1828 static mchunkptr mmap_chunk(size_t size)
1830 static mchunkptr mmap_chunk(size) size_t size;
1833 size_t page_mask = malloc_getpagesize - 1;
1836 #ifndef MAP_ANONYMOUS
1840 if(n_mmaps >= n_mmaps_max) return 0; /* too many regions */
1842 /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because
1843 * there is no following chunk whose prev_size field could be used.
1845 size = (size + SIZE_SZ + page_mask) & ~page_mask;
1847 #ifdef MAP_ANONYMOUS
1848 p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE,
1849 MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
1850 #else /* !MAP_ANONYMOUS */
1853 fd = open("/dev/zero", O_RDWR);
1854 if(fd < 0) return 0;
1856 p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0);
1859 if(p == (mchunkptr)-1) return 0;
1862 if (n_mmaps > max_n_mmaps) max_n_mmaps = n_mmaps;
1864 /* We demand that eight bytes into a page must be 8-byte aligned. */
1865 assert(aligned_OK(chunk2mem(p)));
1867 /* The offset to the start of the mmapped region is stored
1868 * in the prev_size field of the chunk; normally it is zero,
1869 * but that can be changed in memalign().
1872 set_head(p, size|IS_MMAPPED);
1874 mmapped_mem += size;
1875 if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1876 max_mmapped_mem = mmapped_mem;
1877 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1878 max_total_mem = mmapped_mem + sbrked_mem;
1883 static void munmap_chunk(mchunkptr p)
1885 static void munmap_chunk(p) mchunkptr p;
1888 INTERNAL_SIZE_T size = chunksize(p);
1891 assert (chunk_is_mmapped(p));
1892 assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1893 assert((n_mmaps > 0));
1894 assert(((p->prev_size + size) & (malloc_getpagesize-1)) == 0);
1897 mmapped_mem -= (size + p->prev_size);
1899 ret = munmap((char *)p - p->prev_size, size + p->prev_size);
1901 /* munmap returns non-zero on failure */
1908 static mchunkptr mremap_chunk(mchunkptr p, size_t new_size)
1910 static mchunkptr mremap_chunk(p, new_size) mchunkptr p; size_t new_size;
1913 size_t page_mask = malloc_getpagesize - 1;
1914 INTERNAL_SIZE_T offset = p->prev_size;
1915 INTERNAL_SIZE_T size = chunksize(p);
1918 assert (chunk_is_mmapped(p));
1919 assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1920 assert((n_mmaps > 0));
1921 assert(((size + offset) & (malloc_getpagesize-1)) == 0);
1923 /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
1924 new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
1926 cp = (char *)mremap((char *)p - offset, size + offset, new_size, 1);
1928 if (cp == (char *)-1) return 0;
1930 p = (mchunkptr)(cp + offset);
1932 assert(aligned_OK(chunk2mem(p)));
1934 assert((p->prev_size == offset));
1935 set_head(p, (new_size - offset)|IS_MMAPPED);
1937 mmapped_mem -= size + offset;
1938 mmapped_mem += new_size;
1939 if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1940 max_mmapped_mem = mmapped_mem;
1941 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1942 max_total_mem = mmapped_mem + sbrked_mem;
1946 #endif /* HAVE_MREMAP */
1948 #endif /* HAVE_MMAP */
1954 Extend the top-most chunk by obtaining memory from system.
1955 Main interface to sbrk (but see also malloc_trim).
1959 static void malloc_extend_top(INTERNAL_SIZE_T nb)
1961 static void malloc_extend_top(nb) INTERNAL_SIZE_T nb;
1964 char* brk; /* return value from sbrk */
1965 INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of sbrked space */
1966 INTERNAL_SIZE_T correction; /* bytes for 2nd sbrk call */
1967 char* new_brk; /* return of 2nd sbrk call */
1968 INTERNAL_SIZE_T top_size; /* new size of top chunk */
1970 mchunkptr old_top = top; /* Record state of old top */
1971 INTERNAL_SIZE_T old_top_size = chunksize(old_top);
1972 char* old_end = (char*)(chunk_at_offset(old_top, old_top_size));
1974 /* Pad request with top_pad plus minimal overhead */
1976 INTERNAL_SIZE_T sbrk_size = nb + top_pad + MINSIZE;
1977 unsigned long pagesz = malloc_getpagesize;
1979 /* If not the first time through, round to preserve page boundary */
1980 /* Otherwise, we need to correct to a page size below anyway. */
1981 /* (We also correct below if an intervening foreign sbrk call.) */
1983 if (sbrk_base != (char*)(-1))
1984 sbrk_size = (sbrk_size + (pagesz - 1)) & ~(pagesz - 1);
1986 brk = (char*)(MORECORE (sbrk_size));
1988 /* Fail if sbrk failed or if a foreign sbrk call killed our space */
1989 if (brk == (char*)(MORECORE_FAILURE) ||
1990 (brk < old_end && old_top != initial_top))
1993 sbrked_mem += sbrk_size;
1995 if (brk == old_end) /* can just add bytes to current top */
1997 top_size = sbrk_size + old_top_size;
1998 set_head(top, top_size | PREV_INUSE);
2002 if (sbrk_base == (char*)(-1)) /* First time through. Record base */
2004 else /* Someone else called sbrk(). Count those bytes as sbrked_mem. */
2005 sbrked_mem += brk - (char*)old_end;
2007 /* Guarantee alignment of first new chunk made from this space */
2008 front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
2009 if (front_misalign > 0)
2011 correction = (MALLOC_ALIGNMENT) - front_misalign;
2017 /* Guarantee the next brk will be at a page boundary */
2019 correction += ((((unsigned long)(brk + sbrk_size))+(pagesz-1)) &
2020 ~(pagesz - 1)) - ((unsigned long)(brk + sbrk_size));
2022 /* Allocate correction */
2023 new_brk = (char*)(MORECORE (correction));
2024 if (new_brk == (char*)(MORECORE_FAILURE)) return;
2026 sbrked_mem += correction;
2028 top = (mchunkptr)brk;
2029 top_size = new_brk - brk + correction;
2030 set_head(top, top_size | PREV_INUSE);
2032 if (old_top != initial_top)
2035 /* There must have been an intervening foreign sbrk call. */
2036 /* A double fencepost is necessary to prevent consolidation */
2038 /* If not enough space to do this, then user did something very wrong */
2039 if (old_top_size < MINSIZE)
2041 set_head(top, PREV_INUSE); /* will force null return from malloc */
2045 /* Also keep size a multiple of MALLOC_ALIGNMENT */
2046 old_top_size = (old_top_size - 3*SIZE_SZ) & ~MALLOC_ALIGN_MASK;
2047 set_head_size(old_top, old_top_size);
2048 chunk_at_offset(old_top, old_top_size )->size =
2050 chunk_at_offset(old_top, old_top_size + SIZE_SZ)->size =
2052 /* If possible, release the rest. */
2053 if (old_top_size >= MINSIZE)
2054 fREe(chunk2mem(old_top));
2058 if ((unsigned long)sbrked_mem > (unsigned long)max_sbrked_mem)
2059 max_sbrked_mem = sbrked_mem;
2060 if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
2061 max_total_mem = mmapped_mem + sbrked_mem;
2063 /* We always land on a page boundary */
2064 assert(((unsigned long)((char*)top + top_size) & (pagesz - 1)) == 0);
2070 /* Main public routines */
2076 The requested size is first converted into a usable form, `nb'.
2077 This currently means to add 4 bytes overhead plus possibly more to
2078 obtain 8-byte alignment and/or to obtain a size of at least
2079 MINSIZE (currently 16 bytes), the smallest allocatable size.
2080 (All fits are considered `exact' if they are within MINSIZE bytes.)
2082 From there, the first successful of the following steps is taken:
2084 1. The bin corresponding to the request size is scanned, and if
2085 a chunk of exactly the right size is found, it is taken.
2087 2. The most recently remaindered chunk is used if it is big
2088 enough. This is a form of (roving) first fit, used only in
2089 the absence of exact fits. Runs of consecutive requests use
2090 the remainder of the chunk used for the previous such request
2091 whenever possible. This limited use of a first-fit style
2092 allocation strategy tends to give contiguous chunks
2093 coextensive lifetimes, which improves locality and can reduce
2094 fragmentation in the long run.
2096 3. Other bins are scanned in increasing size order, using a
2097 chunk big enough to fulfill the request, and splitting off
2098 any remainder. This search is strictly by best-fit; i.e.,
2099 the smallest (with ties going to approximately the least
2100 recently used) chunk that fits is selected.
2102 4. If large enough, the chunk bordering the end of memory
2103 (`top') is split off. (This use of `top' is in accord with
2104 the best-fit search rule. In effect, `top' is treated as
2105 larger (and thus less well fitting) than any other available
2106 chunk since it can be extended to be as large as necessary
2107 (up to system limitations).
2109 5. If the request size meets the mmap threshold and the
2110 system supports mmap, and there are few enough currently
2111 allocated mmapped regions, and a call to mmap succeeds,
2112 the request is allocated via direct memory mapping.
2114 6. Otherwise, the top of memory is extended by
2115 obtaining more space from the system (normally using sbrk,
2116 but definable to anything else via the MORECORE macro).
2117 Memory is gathered from the system (in system page-sized
2118 units) in a way that allows chunks obtained across different
2119 sbrk calls to be consolidated, but does not require
2120 contiguous memory. Thus, it should be safe to intersperse
2121 mallocs with other sbrk calls.
2124 All allocations are made from the the `lowest' part of any found
2125 chunk. (The implementation invariant is that prev_inuse is
2126 always true of any allocated chunk; i.e., that each allocated
2127 chunk borders either a previously allocated and still in-use chunk,
2128 or the base of its memory arena.)
2133 Void_t* mALLOc(size_t bytes)
2135 Void_t* mALLOc(bytes) size_t bytes;
2138 mchunkptr victim; /* inspected/selected chunk */
2139 INTERNAL_SIZE_T victim_size; /* its size */
2140 int idx; /* index for bin traversal */
2141 mbinptr bin; /* associated bin */
2142 mchunkptr remainder; /* remainder from a split */
2143 long remainder_size; /* its size */
2144 int remainder_index; /* its bin index */
2145 unsigned long block; /* block traverser bit */
2146 int startidx; /* first bin of a traversed block */
2147 mchunkptr fwd; /* misc temp for linking */
2148 mchunkptr bck; /* misc temp for linking */
2149 mbinptr q; /* misc temp */
2153 if ((long)bytes < 0) return 0;
2155 nb = request2size(bytes); /* padded request size; */
2157 /* Check for exact match in a bin */
2159 if (is_small_request(nb)) /* Faster version for small requests */
2161 idx = smallbin_index(nb);
2163 /* No traversal or size check necessary for small bins. */
2168 /* Also scan the next one, since it would have a remainder < MINSIZE */
2176 victim_size = chunksize(victim);
2177 unlink(victim, bck, fwd);
2178 set_inuse_bit_at_offset(victim, victim_size);
2179 check_malloced_chunk(victim, nb);
2180 return chunk2mem(victim);
2183 idx += 2; /* Set for bin scan below. We've already scanned 2 bins. */
2188 idx = bin_index(nb);
2191 for (victim = last(bin); victim != bin; victim = victim->bk)
2193 victim_size = chunksize(victim);
2194 remainder_size = victim_size - nb;
2196 if (remainder_size >= (long)MINSIZE) /* too big */
2198 --idx; /* adjust to rescan below after checking last remainder */
2202 else if (remainder_size >= 0) /* exact fit */
2204 unlink(victim, bck, fwd);
2205 set_inuse_bit_at_offset(victim, victim_size);
2206 check_malloced_chunk(victim, nb);
2207 return chunk2mem(victim);
2215 /* Try to use the last split-off remainder */
2217 if ( (victim = last_remainder->fd) != last_remainder)
2219 victim_size = chunksize(victim);
2220 remainder_size = victim_size - nb;
2222 if (remainder_size >= (long)MINSIZE) /* re-split */
2224 remainder = chunk_at_offset(victim, nb);
2225 set_head(victim, nb | PREV_INUSE);
2226 link_last_remainder(remainder);
2227 set_head(remainder, remainder_size | PREV_INUSE);
2228 set_foot(remainder, remainder_size);
2229 check_malloced_chunk(victim, nb);
2230 return chunk2mem(victim);
2233 clear_last_remainder;
2235 if (remainder_size >= 0) /* exhaust */
2237 set_inuse_bit_at_offset(victim, victim_size);
2238 check_malloced_chunk(victim, nb);
2239 return chunk2mem(victim);
2242 /* Else place in bin */
2244 frontlink(victim, victim_size, remainder_index, bck, fwd);
2248 If there are any possibly nonempty big-enough blocks,
2249 search for best fitting chunk by scanning bins in blockwidth units.
2252 if ( (block = idx2binblock(idx)) <= binblocks)
2255 /* Get to the first marked block */
2257 if ( (block & binblocks) == 0)
2259 /* force to an even block boundary */
2260 idx = (idx & ~(BINBLOCKWIDTH - 1)) + BINBLOCKWIDTH;
2262 while ((block & binblocks) == 0)
2264 idx += BINBLOCKWIDTH;
2269 /* For each possibly nonempty block ... */
2272 startidx = idx; /* (track incomplete blocks) */
2273 q = bin = bin_at(idx);
2275 /* For each bin in this block ... */
2278 /* Find and use first big enough chunk ... */
2280 for (victim = last(bin); victim != bin; victim = victim->bk)
2282 victim_size = chunksize(victim);
2283 remainder_size = victim_size - nb;
2285 if (remainder_size >= (long)MINSIZE) /* split */
2287 remainder = chunk_at_offset(victim, nb);
2288 set_head(victim, nb | PREV_INUSE);
2289 unlink(victim, bck, fwd);
2290 link_last_remainder(remainder);
2291 set_head(remainder, remainder_size | PREV_INUSE);
2292 set_foot(remainder, remainder_size);
2293 check_malloced_chunk(victim, nb);
2294 return chunk2mem(victim);
2297 else if (remainder_size >= 0) /* take */
2299 set_inuse_bit_at_offset(victim, victim_size);
2300 unlink(victim, bck, fwd);
2301 check_malloced_chunk(victim, nb);
2302 return chunk2mem(victim);
2307 bin = next_bin(bin);
2309 } while ((++idx & (BINBLOCKWIDTH - 1)) != 0);
2311 /* Clear out the block bit. */
2313 do /* Possibly backtrack to try to clear a partial block */
2315 if ((startidx & (BINBLOCKWIDTH - 1)) == 0)
2317 binblocks &= ~block;
2322 } while (first(q) == q);
2324 /* Get to the next possibly nonempty block */
2326 if ( (block <<= 1) <= binblocks && (block != 0) )
2328 while ((block & binblocks) == 0)
2330 idx += BINBLOCKWIDTH;
2340 /* Try to use top chunk */
2342 /* Require that there be a remainder, ensuring top always exists */
2343 if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2347 /* If big and would otherwise need to extend, try to use mmap instead */
2348 if ((unsigned long)nb >= (unsigned long)mmap_threshold &&
2349 (victim = mmap_chunk(nb)) != 0)
2350 return chunk2mem(victim);
2354 malloc_extend_top(nb);
2355 if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2356 return 0; /* propagate failure */
2360 set_head(victim, nb | PREV_INUSE);
2361 top = chunk_at_offset(victim, nb);
2362 set_head(top, remainder_size | PREV_INUSE);
2363 check_malloced_chunk(victim, nb);
2364 return chunk2mem(victim);
2377 1. free(0) has no effect.
2379 2. If the chunk was allocated via mmap, it is release via munmap().
2381 3. If a returned chunk borders the current high end of memory,
2382 it is consolidated into the top, and if the total unused
2383 topmost memory exceeds the trim threshold, malloc_trim is
2386 4. Other chunks are consolidated as they arrive, and
2387 placed in corresponding bins. (This includes the case of
2388 consolidating with the current `last_remainder').
2394 void fREe(Void_t* mem)
2396 void fREe(mem) Void_t* mem;
2399 mchunkptr p; /* chunk corresponding to mem */
2400 INTERNAL_SIZE_T hd; /* its head field */
2401 INTERNAL_SIZE_T sz; /* its size */
2402 int idx; /* its bin index */
2403 mchunkptr next; /* next contiguous chunk */
2404 INTERNAL_SIZE_T nextsz; /* its size */
2405 INTERNAL_SIZE_T prevsz; /* size of previous contiguous chunk */
2406 mchunkptr bck; /* misc temp for linking */
2407 mchunkptr fwd; /* misc temp for linking */
2408 int islr; /* track whether merging with last_remainder */
2410 if (mem == 0) /* free(0) has no effect */
2417 if (hd & IS_MMAPPED) /* release mmapped memory. */
2424 check_inuse_chunk(p);
2426 sz = hd & ~PREV_INUSE;
2427 next = chunk_at_offset(p, sz);
2428 nextsz = chunksize(next);
2430 if (next == top) /* merge with top */
2434 if (!(hd & PREV_INUSE)) /* consolidate backward */
2436 prevsz = p->prev_size;
2437 p = chunk_at_offset(p, -((long) prevsz));
2439 unlink(p, bck, fwd);
2442 set_head(p, sz | PREV_INUSE);
2444 if ((unsigned long)(sz) >= (unsigned long)trim_threshold)
2445 malloc_trim(top_pad);
2449 set_head(next, nextsz); /* clear inuse bit */
2453 if (!(hd & PREV_INUSE)) /* consolidate backward */
2455 prevsz = p->prev_size;
2456 p = chunk_at_offset(p, -((long) prevsz));
2459 if (p->fd == last_remainder) /* keep as last_remainder */
2462 unlink(p, bck, fwd);
2465 if (!(inuse_bit_at_offset(next, nextsz))) /* consolidate forward */
2469 if (!islr && next->fd == last_remainder) /* re-insert last_remainder */
2472 link_last_remainder(p);
2475 unlink(next, bck, fwd);
2479 set_head(p, sz | PREV_INUSE);
2482 frontlink(p, sz, idx, bck, fwd);
2493 Chunks that were obtained via mmap cannot be extended or shrunk
2494 unless HAVE_MREMAP is defined, in which case mremap is used.
2495 Otherwise, if their reallocation is for additional space, they are
2496 copied. If for less, they are just left alone.
2498 Otherwise, if the reallocation is for additional space, and the
2499 chunk can be extended, it is, else a malloc-copy-free sequence is
2500 taken. There are several different ways that a chunk could be
2501 extended. All are tried:
2503 * Extending forward into following adjacent free chunk.
2504 * Shifting backwards, joining preceding adjacent space
2505 * Both shifting backwards and extending forward.
2506 * Extending into newly sbrked space
2508 Unless the #define REALLOC_ZERO_BYTES_FREES is set, realloc with a
2509 size argument of zero (re)allocates a minimum-sized chunk.
2511 If the reallocation is for less space, and the new request is for
2512 a `small' (<512 bytes) size, then the newly unused space is lopped
2515 The old unix realloc convention of allowing the last-free'd chunk
2516 to be used as an argument to realloc is no longer supported.
2517 I don't know of any programs still relying on this feature,
2518 and allowing it would also allow too many other incorrect
2519 usages of realloc to be sensible.
2526 Void_t* rEALLOc(Void_t* oldmem, size_t bytes)
2528 Void_t* rEALLOc(oldmem, bytes) Void_t* oldmem; size_t bytes;
2531 INTERNAL_SIZE_T nb; /* padded request size */
2533 mchunkptr oldp; /* chunk corresponding to oldmem */
2534 INTERNAL_SIZE_T oldsize; /* its size */
2536 mchunkptr newp; /* chunk to return */
2537 INTERNAL_SIZE_T newsize; /* its size */
2538 Void_t* newmem; /* corresponding user mem */
2540 mchunkptr next; /* next contiguous chunk after oldp */
2541 INTERNAL_SIZE_T nextsize; /* its size */
2543 mchunkptr prev; /* previous contiguous chunk before oldp */
2544 INTERNAL_SIZE_T prevsize; /* its size */
2546 mchunkptr remainder; /* holds split off extra space from newp */
2547 INTERNAL_SIZE_T remainder_size; /* its size */
2549 mchunkptr bck; /* misc temp for linking */
2550 mchunkptr fwd; /* misc temp for linking */
2552 #ifdef REALLOC_ZERO_BYTES_FREES
2553 if (bytes == 0) { fREe(oldmem); return 0; }
2556 if ((long)bytes < 0) return 0;
2558 /* realloc of null is supposed to be same as malloc */
2559 if (oldmem == 0) return mALLOc(bytes);
2561 newp = oldp = mem2chunk(oldmem);
2562 newsize = oldsize = chunksize(oldp);
2565 nb = request2size(bytes);
2568 if (chunk_is_mmapped(oldp))
2571 newp = mremap_chunk(oldp, nb);
2572 if(newp) return chunk2mem(newp);
2574 /* Note the extra SIZE_SZ overhead. */
2575 if(oldsize - SIZE_SZ >= nb) return oldmem; /* do nothing */
2576 /* Must alloc, copy, free. */
2577 newmem = mALLOc(bytes);
2578 if (newmem == 0) return 0; /* propagate failure */
2579 MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ);
2585 check_inuse_chunk(oldp);
2587 if ((long)(oldsize) < (long)(nb))
2590 /* Try expanding forward */
2592 next = chunk_at_offset(oldp, oldsize);
2593 if (next == top || !inuse(next))
2595 nextsize = chunksize(next);
2597 /* Forward into top only if a remainder */
2600 if ((long)(nextsize + newsize) >= (long)(nb + MINSIZE))
2602 newsize += nextsize;
2603 top = chunk_at_offset(oldp, nb);
2604 set_head(top, (newsize - nb) | PREV_INUSE);
2605 set_head_size(oldp, nb);
2606 return chunk2mem(oldp);
2610 /* Forward into next chunk */
2611 else if (((long)(nextsize + newsize) >= (long)(nb)))
2613 unlink(next, bck, fwd);
2614 newsize += nextsize;
2624 /* Try shifting backwards. */
2626 if (!prev_inuse(oldp))
2628 prev = prev_chunk(oldp);
2629 prevsize = chunksize(prev);
2631 /* try forward + backward first to save a later consolidation */
2638 if ((long)(nextsize + prevsize + newsize) >= (long)(nb + MINSIZE))
2640 unlink(prev, bck, fwd);
2642 newsize += prevsize + nextsize;
2643 newmem = chunk2mem(newp);
2644 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2645 top = chunk_at_offset(newp, nb);
2646 set_head(top, (newsize - nb) | PREV_INUSE);
2647 set_head_size(newp, nb);
2652 /* into next chunk */
2653 else if (((long)(nextsize + prevsize + newsize) >= (long)(nb)))
2655 unlink(next, bck, fwd);
2656 unlink(prev, bck, fwd);
2658 newsize += nextsize + prevsize;
2659 newmem = chunk2mem(newp);
2660 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2666 if (prev != 0 && (long)(prevsize + newsize) >= (long)nb)
2668 unlink(prev, bck, fwd);
2670 newsize += prevsize;
2671 newmem = chunk2mem(newp);
2672 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2679 newmem = mALLOc (bytes);
2681 if (newmem == 0) /* propagate failure */
2684 /* Avoid copy if newp is next chunk after oldp. */
2685 /* (This can only happen when new chunk is sbrk'ed.) */
2687 if ( (newp = mem2chunk(newmem)) == next_chunk(oldp))
2689 newsize += chunksize(newp);
2694 /* Otherwise copy, free, and exit */
2695 MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2701 split: /* split off extra room in old or expanded chunk */
2703 if (newsize - nb >= MINSIZE) /* split off remainder */
2705 remainder = chunk_at_offset(newp, nb);
2706 remainder_size = newsize - nb;
2707 set_head_size(newp, nb);
2708 set_head(remainder, remainder_size | PREV_INUSE);
2709 set_inuse_bit_at_offset(remainder, remainder_size);
2710 fREe(chunk2mem(remainder)); /* let free() deal with it */
2714 set_head_size(newp, newsize);
2715 set_inuse_bit_at_offset(newp, newsize);
2718 check_inuse_chunk(newp);
2719 return chunk2mem(newp);
2729 memalign requests more than enough space from malloc, finds a spot
2730 within that chunk that meets the alignment request, and then
2731 possibly frees the leading and trailing space.
2733 The alignment argument must be a power of two. This property is not
2734 checked by memalign, so misuse may result in random runtime errors.
2736 8-byte alignment is guaranteed by normal malloc calls, so don't
2737 bother calling memalign with an argument of 8 or less.
2739 Overreliance on memalign is a sure way to fragment space.
2745 Void_t* mEMALIGn(size_t alignment, size_t bytes)
2747 Void_t* mEMALIGn(alignment, bytes) size_t alignment; size_t bytes;
2750 INTERNAL_SIZE_T nb; /* padded request size */
2751 char* m; /* memory returned by malloc call */
2752 mchunkptr p; /* corresponding chunk */
2753 char* brk; /* alignment point within p */
2754 mchunkptr newp; /* chunk to return */
2755 INTERNAL_SIZE_T newsize; /* its size */
2756 INTERNAL_SIZE_T leadsize; /* leading space befor alignment point */
2757 mchunkptr remainder; /* spare room at end to split off */
2758 long remainder_size; /* its size */
2760 if ((long)bytes < 0) return 0;
2762 /* If need less alignment than we give anyway, just relay to malloc */
2764 if (alignment <= MALLOC_ALIGNMENT) return mALLOc(bytes);
2766 /* Otherwise, ensure that it is at least a minimum chunk size */
2768 if (alignment < MINSIZE) alignment = MINSIZE;
2770 /* Call malloc with worst case padding to hit alignment. */
2772 nb = request2size(bytes);
2773 m = (char*)(mALLOc(nb + alignment + MINSIZE));
2775 if (m == 0) return 0; /* propagate failure */
2779 if ((((unsigned long)(m)) % alignment) == 0) /* aligned */
2782 if(chunk_is_mmapped(p))
2783 return chunk2mem(p); /* nothing more to do */
2786 else /* misaligned */
2789 Find an aligned spot inside chunk.
2790 Since we need to give back leading space in a chunk of at
2791 least MINSIZE, if the first calculation places us at
2792 a spot with less than MINSIZE leader, we can move to the
2793 next aligned spot -- we've allocated enough total room so that
2794 this is always possible.
2797 brk = (char*)mem2chunk(((unsigned long)(m + alignment - 1)) & -((signed) alignment));
2798 if ((long)(brk - (char*)(p)) < MINSIZE) brk = brk + alignment;
2800 newp = (mchunkptr)brk;
2801 leadsize = brk - (char*)(p);
2802 newsize = chunksize(p) - leadsize;
2805 if(chunk_is_mmapped(p))
2807 newp->prev_size = p->prev_size + leadsize;
2808 set_head(newp, newsize|IS_MMAPPED);
2809 return chunk2mem(newp);
2813 /* give back leader, use the rest */
2815 set_head(newp, newsize | PREV_INUSE);
2816 set_inuse_bit_at_offset(newp, newsize);
2817 set_head_size(p, leadsize);
2821 assert (newsize >= nb && (((unsigned long)(chunk2mem(p))) % alignment) == 0);
2824 /* Also give back spare room at the end */
2826 remainder_size = chunksize(p) - nb;
2828 if (remainder_size >= (long)MINSIZE)
2830 remainder = chunk_at_offset(p, nb);
2831 set_head(remainder, remainder_size | PREV_INUSE);
2832 set_head_size(p, nb);
2833 fREe(chunk2mem(remainder));
2836 check_inuse_chunk(p);
2837 return chunk2mem(p);
2845 valloc just invokes memalign with alignment argument equal
2846 to the page size of the system (or as near to this as can
2847 be figured out from all the includes/defines above.)
2851 Void_t* vALLOc(size_t bytes)
2853 Void_t* vALLOc(bytes) size_t bytes;
2856 return mEMALIGn (malloc_getpagesize, bytes);
2860 pvalloc just invokes valloc for the nearest pagesize
2861 that will accommodate request
2866 Void_t* pvALLOc(size_t bytes)
2868 Void_t* pvALLOc(bytes) size_t bytes;
2871 size_t pagesize = malloc_getpagesize;
2872 return mEMALIGn (pagesize, (bytes + pagesize - 1) & ~(pagesize - 1));
2877 calloc calls malloc, then zeroes out the allocated chunk.
2882 Void_t* cALLOc(size_t n, size_t elem_size)
2884 Void_t* cALLOc(n, elem_size) size_t n; size_t elem_size;
2888 INTERNAL_SIZE_T csz;
2890 INTERNAL_SIZE_T sz = n * elem_size;
2893 /* check if expand_top called, in which case don't need to clear */
2895 mchunkptr oldtop = top;
2896 INTERNAL_SIZE_T oldtopsize = chunksize(top);
2898 Void_t* mem = mALLOc (sz);
2900 if ((long)n < 0) return 0;
2908 /* Two optional cases in which clearing not necessary */
2912 if (chunk_is_mmapped(p)) return mem;
2918 if (p == oldtop && csz > oldtopsize)
2920 /* clear only the bytes from non-freshly-sbrked memory */
2925 MALLOC_ZERO(mem, csz - SIZE_SZ);
2932 cfree just calls free. It is needed/defined on some systems
2933 that pair it with calloc, presumably for odd historical reasons.
2937 #if !defined(INTERNAL_LINUX_C_LIB) || !defined(__ELF__)
2939 void cfree(Void_t *mem)
2941 void cfree(mem) Void_t *mem;
2952 Malloc_trim gives memory back to the system (via negative
2953 arguments to sbrk) if there is unused memory at the `high' end of
2954 the malloc pool. You can call this after freeing large blocks of
2955 memory to potentially reduce the system-level memory requirements
2956 of a program. However, it cannot guarantee to reduce memory. Under
2957 some allocation patterns, some large free blocks of memory will be
2958 locked between two used chunks, so they cannot be given back to
2961 The `pad' argument to malloc_trim represents the amount of free
2962 trailing space to leave untrimmed. If this argument is zero,
2963 only the minimum amount of memory to maintain internal data
2964 structures will be left (one page or less). Non-zero arguments
2965 can be supplied to maintain enough trailing space to service
2966 future expected allocations without having to re-obtain memory
2969 Malloc_trim returns 1 if it actually released any memory, else 0.
2974 int malloc_trim(size_t pad)
2976 int malloc_trim(pad) size_t pad;
2979 long top_size; /* Amount of top-most memory */
2980 long extra; /* Amount to release */
2981 char* current_brk; /* address returned by pre-check sbrk call */
2982 char* new_brk; /* address returned by negative sbrk call */
2984 unsigned long pagesz = malloc_getpagesize;
2986 top_size = chunksize(top);
2987 extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
2989 if (extra < (long)pagesz) /* Not enough memory to release */
2994 /* Test to make sure no one else called sbrk */
2995 current_brk = (char*)(MORECORE (0));
2996 if (current_brk != (char*)(top) + top_size)
2997 return 0; /* Apparently we don't own memory; must fail */
3001 new_brk = (char*)(MORECORE (-extra));
3003 if (new_brk == (char*)(MORECORE_FAILURE)) /* sbrk failed? */
3005 /* Try to figure out what we have */
3006 current_brk = (char*)(MORECORE (0));
3007 top_size = current_brk - (char*)top;
3008 if (top_size >= (long)MINSIZE) /* if not, we are very very dead! */
3010 sbrked_mem = current_brk - sbrk_base;
3011 set_head(top, top_size | PREV_INUSE);
3019 /* Success. Adjust top accordingly. */
3020 set_head(top, (top_size - extra) | PREV_INUSE);
3021 sbrked_mem -= extra;
3034 This routine tells you how many bytes you can actually use in an
3035 allocated chunk, which may be more than you requested (although
3036 often not). You can use this many bytes without worrying about
3037 overwriting other allocated objects. Not a particularly great
3038 programming practice, but still sometimes useful.
3043 size_t malloc_usable_size(Void_t* mem)
3045 size_t malloc_usable_size(mem) Void_t* mem;
3054 if(!chunk_is_mmapped(p))
3056 if (!inuse(p)) return 0;
3057 check_inuse_chunk(p);
3058 return chunksize(p) - SIZE_SZ;
3060 return chunksize(p) - 2*SIZE_SZ;
3067 /* Utility to update current_mallinfo for malloc_stats and mallinfo() */
3070 static void malloc_update_mallinfo()
3079 INTERNAL_SIZE_T avail = chunksize(top);
3080 int navail = ((long)(avail) >= (long)MINSIZE)? 1 : 0;
3082 for (i = 1; i < NAV; ++i)
3085 for (p = last(b); p != b; p = p->bk)
3088 check_free_chunk(p);
3089 for (q = next_chunk(p);
3090 q < top && inuse(q) && (long)(chunksize(q)) >= (long)MINSIZE;
3092 check_inuse_chunk(q);
3094 avail += chunksize(p);
3099 current_mallinfo.ordblks = navail;
3100 current_mallinfo.uordblks = sbrked_mem - avail;
3101 current_mallinfo.fordblks = avail;
3102 current_mallinfo.hblks = n_mmaps;
3103 current_mallinfo.hblkhd = mmapped_mem;
3104 current_mallinfo.keepcost = chunksize(top);
3115 Prints on the amount of space obtain from the system (both
3116 via sbrk and mmap), the maximum amount (which may be more than
3117 current if malloc_trim and/or munmap got called), the maximum
3118 number of simultaneous mmap regions used, and the current number
3119 of bytes allocated via malloc (or realloc, etc) but not yet
3120 freed. (Note that this is the number of bytes allocated, not the
3121 number requested. It will be larger than the number requested
3122 because of alignment and bookkeeping overhead.)
3129 malloc_update_mallinfo();
3130 printf("max system bytes = %10u\n",
3131 (unsigned int)(max_total_mem));
3132 printf("system bytes = %10u\n",
3133 (unsigned int)(sbrked_mem + mmapped_mem));
3134 printf("in use bytes = %10u\n",
3135 (unsigned int)(current_mallinfo.uordblks + mmapped_mem));
3137 printf("max mmap regions = %10u\n",
3138 (unsigned int)max_n_mmaps);
3144 mallinfo returns a copy of updated current mallinfo.
3148 struct mallinfo mALLINFo()
3150 malloc_update_mallinfo();
3151 return current_mallinfo;
3161 mallopt is the general SVID/XPG interface to tunable parameters.
3162 The format is to provide a (parameter-number, parameter-value) pair.
3163 mallopt then sets the corresponding parameter to the argument
3164 value if it can (i.e., so long as the value is meaningful),
3165 and returns 1 if successful else 0.
3167 See descriptions of tunable parameters above.
3172 int mALLOPt(int param_number, int value)
3174 int mALLOPt(param_number, value) int param_number; int value;
3177 switch(param_number)
3179 case M_TRIM_THRESHOLD:
3180 trim_threshold = value; return 1;
3182 top_pad = value; return 1;
3183 case M_MMAP_THRESHOLD:
3184 mmap_threshold = value; return 1;
3187 n_mmaps_max = value; return 1;
3189 if (value != 0) return 0; else n_mmaps_max = value; return 1;
3201 V2.6.6 Sun Dec 5 07:42:19 1999 Doug Lea (dl at gee)
3202 * return null for negative arguments
3203 * Added Several WIN32 cleanups from Martin C. Fong <mcfong@yahoo.com>
3204 * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'
3205 (e.g. WIN32 platforms)
3206 * Cleanup up header file inclusion for WIN32 platforms
3207 * Cleanup code to avoid Microsoft Visual C++ compiler complaints
3208 * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing
3209 memory allocation routines
3210 * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)
3211 * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to
3212 usage of 'assert' in non-WIN32 code
3213 * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to
3215 * Always call 'fREe()' rather than 'free()'
3217 V2.6.5 Wed Jun 17 15:57:31 1998 Doug Lea (dl at gee)
3218 * Fixed ordering problem with boundary-stamping
3220 V2.6.3 Sun May 19 08:17:58 1996 Doug Lea (dl at gee)
3221 * Added pvalloc, as recommended by H.J. Liu
3222 * Added 64bit pointer support mainly from Wolfram Gloger
3223 * Added anonymously donated WIN32 sbrk emulation
3224 * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
3225 * malloc_extend_top: fix mask error that caused wastage after
3227 * Add linux mremap support code from HJ Liu
3229 V2.6.2 Tue Dec 5 06:52:55 1995 Doug Lea (dl at gee)
3230 * Integrated most documentation with the code.
3231 * Add support for mmap, with help from
3232 Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3233 * Use last_remainder in more cases.
3234 * Pack bins using idea from colin@nyx10.cs.du.edu
3235 * Use ordered bins instead of best-fit threshhold
3236 * Eliminate block-local decls to simplify tracing and debugging.
3237 * Support another case of realloc via move into top
3238 * Fix error occuring when initial sbrk_base not word-aligned.
3239 * Rely on page size for units instead of SBRK_UNIT to
3240 avoid surprises about sbrk alignment conventions.
3241 * Add mallinfo, mallopt. Thanks to Raymond Nijssen
3242 (raymond@es.ele.tue.nl) for the suggestion.
3243 * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
3244 * More precautions for cases where other routines call sbrk,
3245 courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3246 * Added macros etc., allowing use in linux libc from
3247 H.J. Lu (hjl@gnu.ai.mit.edu)
3248 * Inverted this history list
3250 V2.6.1 Sat Dec 2 14:10:57 1995 Doug Lea (dl at gee)
3251 * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
3252 * Removed all preallocation code since under current scheme
3253 the work required to undo bad preallocations exceeds
3254 the work saved in good cases for most test programs.
3255 * No longer use return list or unconsolidated bins since
3256 no scheme using them consistently outperforms those that don't
3257 given above changes.
3258 * Use best fit for very large chunks to prevent some worst-cases.
3259 * Added some support for debugging
3261 V2.6.0 Sat Nov 4 07:05:23 1995 Doug Lea (dl at gee)
3262 * Removed footers when chunks are in use. Thanks to
3263 Paul Wilson (wilson@cs.texas.edu) for the suggestion.
3265 V2.5.4 Wed Nov 1 07:54:51 1995 Doug Lea (dl at gee)
3266 * Added malloc_trim, with help from Wolfram Gloger
3267 (wmglo@Dent.MED.Uni-Muenchen.DE).
3269 V2.5.3 Tue Apr 26 10:16:01 1994 Doug Lea (dl at g)
3271 V2.5.2 Tue Apr 5 16:20:40 1994 Doug Lea (dl at g)
3272 * realloc: try to expand in both directions
3273 * malloc: swap order of clean-bin strategy;
3274 * realloc: only conditionally expand backwards
3275 * Try not to scavenge used bins
3276 * Use bin counts as a guide to preallocation
3277 * Occasionally bin return list chunks in first scan
3278 * Add a few optimizations from colin@nyx10.cs.du.edu
3280 V2.5.1 Sat Aug 14 15:40:43 1993 Doug Lea (dl at g)
3281 * faster bin computation & slightly different binning
3282 * merged all consolidations to one part of malloc proper
3283 (eliminating old malloc_find_space & malloc_clean_bin)
3284 * Scan 2 returns chunks (not just 1)
3285 * Propagate failure in realloc if malloc returns 0
3286 * Add stuff to allow compilation on non-ANSI compilers
3287 from kpv@research.att.com
3289 V2.5 Sat Aug 7 07:41:59 1993 Doug Lea (dl at g.oswego.edu)
3290 * removed potential for odd address access in prev_chunk
3291 * removed dependency on getpagesize.h
3292 * misc cosmetics and a bit more internal documentation
3293 * anticosmetics: mangled names in macros to evade debugger strangeness
3294 * tested on sparc, hp-700, dec-mips, rs6000
3295 with gcc & native cc (hp, dec only) allowing
3296 Detlefs & Zorn comparison study (in SIGPLAN Notices.)
3298 Trial version Fri Aug 28 13:14:29 1992 Doug Lea (dl at g.oswego.edu)
3299 * Based loosely on libg++-1.2X malloc. (It retains some of the overall
3300 structure of old version, but most details differ.)
projects / u-boot / blob