xref: /openbmc/u-boot/common/dlmalloc.c (revision 7d9b5bae)
1 #include <common.h>
2 
3 #if 0	/* Moved to malloc.h */
4 /* ---------- To make a malloc.h, start cutting here ------------ */
5 
6 /*
7   A version of malloc/free/realloc written by Doug Lea and released to the
8   public domain.  Send questions/comments/complaints/performance data
9   to dl@cs.oswego.edu
10 
11 * VERSION 2.6.6  Sun Mar  5 19:10:03 2000  Doug Lea  (dl at gee)
12 
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!
16 
17 * Why use this malloc?
18 
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
25 
26 * Synopsis of public routines
27 
28   (Much fuller descriptions are contained in the program documentation below.)
29 
30   malloc(size_t n);
31      Return a pointer to a newly allocated chunk of at least n bytes, or null
32      if no space is available.
33   free(Void_t* p);
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
45      two.
46   valloc(size_t n);
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.)
50   pvalloc(size_t n);
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
55      set to zero.
56   cfree(Void_t* p);
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.
65   malloc_stats();
66      Prints brief summary statistics.
67   mallinfo()
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.
72 
73 * Vital statistics:
74 
75   Alignment:                            8-byte
76        8 byte alignment is currently hardwired into the design.  This
77        seems to suffice for all current machines and C compilers.
78 
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.
83 
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.
86 
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.
90 
91   Minimum allocated size: 4-byte ptrs:  16 bytes    (including 4 overhead)
92 			  8-byte ptrs:  24/32 bytes (including, 4/8 overhead)
93 
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.
99 
100        Even a request for zero bytes (i.e., malloc(0)) returns a
101        pointer to something of the minimum allocatable size.
102 
103   Maximum allocated size: 4-byte size_t: 2^31 -  8 bytes
104 			  8-byte size_t: 2^63 - 16 bytes
105 
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.
113 
114   Maximum overhead wastage per allocated chunk: normally 15 bytes
115 
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
119        two exceptions:
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.
125 
126 * Limitations
127 
128     Here are some features that are NOT currently supported
129 
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.
134 
135 * Synopsis of compile-time options:
136 
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.
143 
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
148     paths.
149 
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
156      execution.
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
186      very small chunks.
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
192      affect anything.
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
205      holds for sbrk).
206   DEFAULT_TRIM_THRESHOLD
207   DEFAULT_TOP_PAD
208   DEFAULT_MMAP_THRESHOLD
209   DEFAULT_MMAP_MAX
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
214      programs/systems.
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.
219 
220 
221 */
222 
223 
224 
225 
226 /* Preliminaries */
227 
228 #ifndef __STD_C
229 #ifdef __STDC__
230 #define __STD_C     1
231 #else
232 #if __cplusplus
233 #define __STD_C     1
234 #else
235 #define __STD_C     0
236 #endif /*__cplusplus*/
237 #endif /*__STDC__*/
238 #endif /*__STD_C*/
239 
240 #ifndef Void_t
241 #if (__STD_C || defined(WIN32))
242 #define Void_t      void
243 #else
244 #define Void_t      char
245 #endif
246 #endif /*Void_t*/
247 
248 #if __STD_C
249 #include <stddef.h>   /* for size_t */
250 #else
251 #include <sys/types.h>
252 #endif
253 
254 #ifdef __cplusplus
255 extern "C" {
256 #endif
257 
258 #include <stdio.h>    /* needed for malloc_stats */
259 
260 
261 /*
262   Compile-time options
263 */
264 
265 
266 /*
267     Debugging:
268 
269     Because freed chunks may be overwritten with link fields, this
270     malloc will often die when freed memory is overwritten by user
271     programs.  This can be very effective (albeit in an annoying way)
272     in helping track down dangling pointers.
273 
274     If you compile with -DDEBUG, a number of assertion checks are
275     enabled that will catch more memory errors. You probably won't be
276     able to make much sense of the actual assertion errors, but they
277     should help you locate incorrectly overwritten memory.  The
278     checking is fairly extensive, and will slow down execution
279     noticeably. Calling malloc_stats or mallinfo with DEBUG set will
280     attempt to check every non-mmapped allocated and free chunk in the
281     course of computing the summmaries. (By nature, mmapped regions
282     cannot be checked very much automatically.)
283 
284     Setting DEBUG may also be helpful if you are trying to modify
285     this code. The assertions in the check routines spell out in more
286     detail the assumptions and invariants underlying the algorithms.
287 
288 */
289 
290 #ifdef DEBUG
291 #include <assert.h>
292 #else
293 #define assert(x) ((void)0)
294 #endif
295 
296 
297 /*
298   INTERNAL_SIZE_T is the word-size used for internal bookkeeping
299   of chunk sizes. On a 64-bit machine, you can reduce malloc
300   overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int'
301   at the expense of not being able to handle requests greater than
302   2^31. This limitation is hardly ever a concern; you are encouraged
303   to set this. However, the default version is the same as size_t.
304 */
305 
306 #ifndef INTERNAL_SIZE_T
307 #define INTERNAL_SIZE_T size_t
308 #endif
309 
310 /*
311   REALLOC_ZERO_BYTES_FREES should be set if a call to
312   realloc with zero bytes should be the same as a call to free.
313   Some people think it should. Otherwise, since this malloc
314   returns a unique pointer for malloc(0), so does realloc(p, 0).
315 */
316 
317 
318 /*   #define REALLOC_ZERO_BYTES_FREES */
319 
320 
321 /*
322   WIN32 causes an emulation of sbrk to be compiled in
323   mmap-based options are not currently supported in WIN32.
324 */
325 
326 /* #define WIN32 */
327 #ifdef WIN32
328 #define MORECORE wsbrk
329 #define HAVE_MMAP 0
330 
331 #define LACKS_UNISTD_H
332 #define LACKS_SYS_PARAM_H
333 
334 /*
335   Include 'windows.h' to get the necessary declarations for the
336   Microsoft Visual C++ data structures and routines used in the 'sbrk'
337   emulation.
338 
339   Define WIN32_LEAN_AND_MEAN so that only the essential Microsoft
340   Visual C++ header files are included.
341 */
342 #define WIN32_LEAN_AND_MEAN
343 #include <windows.h>
344 #endif
345 
346 
347 /*
348   HAVE_MEMCPY should be defined if you are not otherwise using
349   ANSI STD C, but still have memcpy and memset in your C library
350   and want to use them in calloc and realloc. Otherwise simple
351   macro versions are defined here.
352 
353   USE_MEMCPY should be defined as 1 if you actually want to
354   have memset and memcpy called. People report that the macro
355   versions are often enough faster than libc versions on many
356   systems that it is better to use them.
357 
358 */
359 
360 #define HAVE_MEMCPY
361 
362 #ifndef USE_MEMCPY
363 #ifdef HAVE_MEMCPY
364 #define USE_MEMCPY 1
365 #else
366 #define USE_MEMCPY 0
367 #endif
368 #endif
369 
370 #if (__STD_C || defined(HAVE_MEMCPY))
371 
372 #if __STD_C
373 void* memset(void*, int, size_t);
374 void* memcpy(void*, const void*, size_t);
375 #else
376 #ifdef WIN32
377 /* On Win32 platforms, 'memset()' and 'memcpy()' are already declared in */
378 /* 'windows.h' */
379 #else
380 Void_t* memset();
381 Void_t* memcpy();
382 #endif
383 #endif
384 #endif
385 
386 #if USE_MEMCPY
387 
388 /* The following macros are only invoked with (2n+1)-multiples of
389    INTERNAL_SIZE_T units, with a positive integer n. This is exploited
390    for fast inline execution when n is small. */
391 
392 #define MALLOC_ZERO(charp, nbytes)                                            \
393 do {                                                                          \
394   INTERNAL_SIZE_T mzsz = (nbytes);                                            \
395   if(mzsz <= 9*sizeof(mzsz)) {                                                \
396     INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp);                         \
397     if(mzsz >= 5*sizeof(mzsz)) {     *mz++ = 0;                               \
398 				     *mz++ = 0;                               \
399       if(mzsz >= 7*sizeof(mzsz)) {   *mz++ = 0;                               \
400 				     *mz++ = 0;                               \
401 	if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0;                               \
402 				     *mz++ = 0; }}}                           \
403 				     *mz++ = 0;                               \
404 				     *mz++ = 0;                               \
405 				     *mz   = 0;                               \
406   } else memset((charp), 0, mzsz);                                            \
407 } while(0)
408 
409 #define MALLOC_COPY(dest,src,nbytes)                                          \
410 do {                                                                          \
411   INTERNAL_SIZE_T mcsz = (nbytes);                                            \
412   if(mcsz <= 9*sizeof(mcsz)) {                                                \
413     INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src);                        \
414     INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest);                       \
415     if(mcsz >= 5*sizeof(mcsz)) {     *mcdst++ = *mcsrc++;                     \
416 				     *mcdst++ = *mcsrc++;                     \
417       if(mcsz >= 7*sizeof(mcsz)) {   *mcdst++ = *mcsrc++;                     \
418 				     *mcdst++ = *mcsrc++;                     \
419 	if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++;                     \
420 				     *mcdst++ = *mcsrc++; }}}                 \
421 				     *mcdst++ = *mcsrc++;                     \
422 				     *mcdst++ = *mcsrc++;                     \
423 				     *mcdst   = *mcsrc  ;                     \
424   } else memcpy(dest, src, mcsz);                                             \
425 } while(0)
426 
427 #else /* !USE_MEMCPY */
428 
429 /* Use Duff's device for good zeroing/copying performance. */
430 
431 #define MALLOC_ZERO(charp, nbytes)                                            \
432 do {                                                                          \
433   INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp);                           \
434   long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn;                         \
435   if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
436   switch (mctmp) {                                                            \
437     case 0: for(;;) { *mzp++ = 0;                                             \
438     case 7:           *mzp++ = 0;                                             \
439     case 6:           *mzp++ = 0;                                             \
440     case 5:           *mzp++ = 0;                                             \
441     case 4:           *mzp++ = 0;                                             \
442     case 3:           *mzp++ = 0;                                             \
443     case 2:           *mzp++ = 0;                                             \
444     case 1:           *mzp++ = 0; if(mcn <= 0) break; mcn--; }                \
445   }                                                                           \
446 } while(0)
447 
448 #define MALLOC_COPY(dest,src,nbytes)                                          \
449 do {                                                                          \
450   INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src;                            \
451   INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest;                           \
452   long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn;                         \
453   if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
454   switch (mctmp) {                                                            \
455     case 0: for(;;) { *mcdst++ = *mcsrc++;                                    \
456     case 7:           *mcdst++ = *mcsrc++;                                    \
457     case 6:           *mcdst++ = *mcsrc++;                                    \
458     case 5:           *mcdst++ = *mcsrc++;                                    \
459     case 4:           *mcdst++ = *mcsrc++;                                    \
460     case 3:           *mcdst++ = *mcsrc++;                                    \
461     case 2:           *mcdst++ = *mcsrc++;                                    \
462     case 1:           *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; }       \
463   }                                                                           \
464 } while(0)
465 
466 #endif
467 
468 
469 /*
470   Define HAVE_MMAP to optionally make malloc() use mmap() to
471   allocate very large blocks.  These will be returned to the
472   operating system immediately after a free().
473 */
474 
475 #ifndef HAVE_MMAP
476 #define HAVE_MMAP 1
477 #endif
478 
479 /*
480   Define HAVE_MREMAP to make realloc() use mremap() to re-allocate
481   large blocks.  This is currently only possible on Linux with
482   kernel versions newer than 1.3.77.
483 */
484 
485 #ifndef HAVE_MREMAP
486 #ifdef INTERNAL_LINUX_C_LIB
487 #define HAVE_MREMAP 1
488 #else
489 #define HAVE_MREMAP 0
490 #endif
491 #endif
492 
493 #if HAVE_MMAP
494 
495 #include <unistd.h>
496 #include <fcntl.h>
497 #include <sys/mman.h>
498 
499 #if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
500 #define MAP_ANONYMOUS MAP_ANON
501 #endif
502 
503 #endif /* HAVE_MMAP */
504 
505 /*
506   Access to system page size. To the extent possible, this malloc
507   manages memory from the system in page-size units.
508 
509   The following mechanics for getpagesize were adapted from
510   bsd/gnu getpagesize.h
511 */
512 
513 #ifndef LACKS_UNISTD_H
514 #  include <unistd.h>
515 #endif
516 
517 #ifndef malloc_getpagesize
518 #  ifdef _SC_PAGESIZE         /* some SVR4 systems omit an underscore */
519 #    ifndef _SC_PAGE_SIZE
520 #      define _SC_PAGE_SIZE _SC_PAGESIZE
521 #    endif
522 #  endif
523 #  ifdef _SC_PAGE_SIZE
524 #    define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
525 #  else
526 #    if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
527        extern size_t getpagesize();
528 #      define malloc_getpagesize getpagesize()
529 #    else
530 #      ifdef WIN32
531 #        define malloc_getpagesize (4096) /* TBD: Use 'GetSystemInfo' instead */
532 #      else
533 #        ifndef LACKS_SYS_PARAM_H
534 #          include <sys/param.h>
535 #        endif
536 #        ifdef EXEC_PAGESIZE
537 #          define malloc_getpagesize EXEC_PAGESIZE
538 #        else
539 #          ifdef NBPG
540 #            ifndef CLSIZE
541 #              define malloc_getpagesize NBPG
542 #            else
543 #              define malloc_getpagesize (NBPG * CLSIZE)
544 #            endif
545 #          else
546 #            ifdef NBPC
547 #              define malloc_getpagesize NBPC
548 #            else
549 #              ifdef PAGESIZE
550 #                define malloc_getpagesize PAGESIZE
551 #              else
552 #                define malloc_getpagesize (4096) /* just guess */
553 #              endif
554 #            endif
555 #          endif
556 #        endif
557 #      endif
558 #    endif
559 #  endif
560 #endif
561 
562 
563 /*
564 
565   This version of malloc supports the standard SVID/XPG mallinfo
566   routine that returns a struct containing the same kind of
567   information you can get from malloc_stats. It should work on
568   any SVID/XPG compliant system that has a /usr/include/malloc.h
569   defining struct mallinfo. (If you'd like to install such a thing
570   yourself, cut out the preliminary declarations as described above
571   and below and save them in a malloc.h file. But there's no
572   compelling reason to bother to do this.)
573 
574   The main declaration needed is the mallinfo struct that is returned
575   (by-copy) by mallinfo().  The SVID/XPG malloinfo struct contains a
576   bunch of fields, most of which are not even meaningful in this
577   version of malloc. Some of these fields are are instead filled by
578   mallinfo() with other numbers that might possibly be of interest.
579 
580   HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
581   /usr/include/malloc.h file that includes a declaration of struct
582   mallinfo.  If so, it is included; else an SVID2/XPG2 compliant
583   version is declared below.  These must be precisely the same for
584   mallinfo() to work.
585 
586 */
587 
588 /* #define HAVE_USR_INCLUDE_MALLOC_H */
589 
590 #if HAVE_USR_INCLUDE_MALLOC_H
591 #include "/usr/include/malloc.h"
592 #else
593 
594 /* SVID2/XPG mallinfo structure */
595 
596 struct mallinfo {
597   int arena;    /* total space allocated from system */
598   int ordblks;  /* number of non-inuse chunks */
599   int smblks;   /* unused -- always zero */
600   int hblks;    /* number of mmapped regions */
601   int hblkhd;   /* total space in mmapped regions */
602   int usmblks;  /* unused -- always zero */
603   int fsmblks;  /* unused -- always zero */
604   int uordblks; /* total allocated space */
605   int fordblks; /* total non-inuse space */
606   int keepcost; /* top-most, releasable (via malloc_trim) space */
607 };
608 
609 /* SVID2/XPG mallopt options */
610 
611 #define M_MXFAST  1    /* UNUSED in this malloc */
612 #define M_NLBLKS  2    /* UNUSED in this malloc */
613 #define M_GRAIN   3    /* UNUSED in this malloc */
614 #define M_KEEP    4    /* UNUSED in this malloc */
615 
616 #endif
617 
618 /* mallopt options that actually do something */
619 
620 #define M_TRIM_THRESHOLD    -1
621 #define M_TOP_PAD           -2
622 #define M_MMAP_THRESHOLD    -3
623 #define M_MMAP_MAX          -4
624 
625 
626 #ifndef DEFAULT_TRIM_THRESHOLD
627 #define DEFAULT_TRIM_THRESHOLD (128 * 1024)
628 #endif
629 
630 /*
631     M_TRIM_THRESHOLD is the maximum amount of unused top-most memory
632       to keep before releasing via malloc_trim in free().
633 
634       Automatic trimming is mainly useful in long-lived programs.
635       Because trimming via sbrk can be slow on some systems, and can
636       sometimes be wasteful (in cases where programs immediately
637       afterward allocate more large chunks) the value should be high
638       enough so that your overall system performance would improve by
639       releasing.
640 
641       The trim threshold and the mmap control parameters (see below)
642       can be traded off with one another. Trimming and mmapping are
643       two different ways of releasing unused memory back to the
644       system. Between these two, it is often possible to keep
645       system-level demands of a long-lived program down to a bare
646       minimum. For example, in one test suite of sessions measuring
647       the XF86 X server on Linux, using a trim threshold of 128K and a
648       mmap threshold of 192K led to near-minimal long term resource
649       consumption.
650 
651       If you are using this malloc in a long-lived program, it should
652       pay to experiment with these values.  As a rough guide, you
653       might set to a value close to the average size of a process
654       (program) running on your system.  Releasing this much memory
655       would allow such a process to run in memory.  Generally, it's
656       worth it to tune for trimming rather tham memory mapping when a
657       program undergoes phases where several large chunks are
658       allocated and released in ways that can reuse each other's
659       storage, perhaps mixed with phases where there are no such
660       chunks at all.  And in well-behaved long-lived programs,
661       controlling release of large blocks via trimming versus mapping
662       is usually faster.
663 
664       However, in most programs, these parameters serve mainly as
665       protection against the system-level effects of carrying around
666       massive amounts of unneeded memory. Since frequent calls to
667       sbrk, mmap, and munmap otherwise degrade performance, the default
668       parameters are set to relatively high values that serve only as
669       safeguards.
670 
671       The default trim value is high enough to cause trimming only in
672       fairly extreme (by current memory consumption standards) cases.
673       It must be greater than page size to have any useful effect.  To
674       disable trimming completely, you can set to (unsigned long)(-1);
675 
676 
677 */
678 
679 
680 #ifndef DEFAULT_TOP_PAD
681 #define DEFAULT_TOP_PAD        (0)
682 #endif
683 
684 /*
685     M_TOP_PAD is the amount of extra `padding' space to allocate or
686       retain whenever sbrk is called. It is used in two ways internally:
687 
688       * When sbrk is called to extend the top of the arena to satisfy
689 	a new malloc request, this much padding is added to the sbrk
690 	request.
691 
692       * When malloc_trim is called automatically from free(),
693 	it is used as the `pad' argument.
694 
695       In both cases, the actual amount of padding is rounded
696       so that the end of the arena is always a system page boundary.
697 
698       The main reason for using padding is to avoid calling sbrk so
699       often. Having even a small pad greatly reduces the likelihood
700       that nearly every malloc request during program start-up (or
701       after trimming) will invoke sbrk, which needlessly wastes
702       time.
703 
704       Automatic rounding-up to page-size units is normally sufficient
705       to avoid measurable overhead, so the default is 0.  However, in
706       systems where sbrk is relatively slow, it can pay to increase
707       this value, at the expense of carrying around more memory than
708       the program needs.
709 
710 */
711 
712 
713 #ifndef DEFAULT_MMAP_THRESHOLD
714 #define DEFAULT_MMAP_THRESHOLD (128 * 1024)
715 #endif
716 
717 /*
718 
719     M_MMAP_THRESHOLD is the request size threshold for using mmap()
720       to service a request. Requests of at least this size that cannot
721       be allocated using already-existing space will be serviced via mmap.
722       (If enough normal freed space already exists it is used instead.)
723 
724       Using mmap segregates relatively large chunks of memory so that
725       they can be individually obtained and released from the host
726       system. A request serviced through mmap is never reused by any
727       other request (at least not directly; the system may just so
728       happen to remap successive requests to the same locations).
729 
730       Segregating space in this way has the benefit that mmapped space
731       can ALWAYS be individually released back to the system, which
732       helps keep the system level memory demands of a long-lived
733       program low. Mapped memory can never become `locked' between
734       other chunks, as can happen with normally allocated chunks, which
735       menas that even trimming via malloc_trim would not release them.
736 
737       However, it has the disadvantages that:
738 
739 	 1. The space cannot be reclaimed, consolidated, and then
740 	    used to service later requests, as happens with normal chunks.
741 	 2. It can lead to more wastage because of mmap page alignment
742 	    requirements
743 	 3. It causes malloc performance to be more dependent on host
744 	    system memory management support routines which may vary in
745 	    implementation quality and may impose arbitrary
746 	    limitations. Generally, servicing a request via normal
747 	    malloc steps is faster than going through a system's mmap.
748 
749       All together, these considerations should lead you to use mmap
750       only for relatively large requests.
751 
752 
753 */
754 
755 
756 #ifndef DEFAULT_MMAP_MAX
757 #if HAVE_MMAP
758 #define DEFAULT_MMAP_MAX       (64)
759 #else
760 #define DEFAULT_MMAP_MAX       (0)
761 #endif
762 #endif
763 
764 /*
765     M_MMAP_MAX is the maximum number of requests to simultaneously
766       service using mmap. This parameter exists because:
767 
768 	 1. Some systems have a limited number of internal tables for
769 	    use by mmap.
770 	 2. In most systems, overreliance on mmap can degrade overall
771 	    performance.
772 	 3. If a program allocates many large regions, it is probably
773 	    better off using normal sbrk-based allocation routines that
774 	    can reclaim and reallocate normal heap memory. Using a
775 	    small value allows transition into this mode after the
776 	    first few allocations.
777 
778       Setting to 0 disables all use of mmap.  If HAVE_MMAP is not set,
779       the default value is 0, and attempts to set it to non-zero values
780       in mallopt will fail.
781 */
782 
783 
784 /*
785     USE_DL_PREFIX will prefix all public routines with the string 'dl'.
786       Useful to quickly avoid procedure declaration conflicts and linker
787       symbol conflicts with existing memory allocation routines.
788 
789 */
790 
791 /* #define USE_DL_PREFIX */
792 
793 
794 /*
795 
796   Special defines for linux libc
797 
798   Except when compiled using these special defines for Linux libc
799   using weak aliases, this malloc is NOT designed to work in
800   multithreaded applications.  No semaphores or other concurrency
801   control are provided to ensure that multiple malloc or free calls
802   don't run at the same time, which could be disasterous. A single
803   semaphore could be used across malloc, realloc, and free (which is
804   essentially the effect of the linux weak alias approach). It would
805   be hard to obtain finer granularity.
806 
807 */
808 
809 
810 #ifdef INTERNAL_LINUX_C_LIB
811 
812 #if __STD_C
813 
814 Void_t * __default_morecore_init (ptrdiff_t);
815 Void_t *(*__morecore)(ptrdiff_t) = __default_morecore_init;
816 
817 #else
818 
819 Void_t * __default_morecore_init ();
820 Void_t *(*__morecore)() = __default_morecore_init;
821 
822 #endif
823 
824 #define MORECORE (*__morecore)
825 #define MORECORE_FAILURE 0
826 #define MORECORE_CLEARS 1
827 
828 #else /* INTERNAL_LINUX_C_LIB */
829 
830 #if __STD_C
831 extern Void_t*     sbrk(ptrdiff_t);
832 #else
833 extern Void_t*     sbrk();
834 #endif
835 
836 #ifndef MORECORE
837 #define MORECORE sbrk
838 #endif
839 
840 #ifndef MORECORE_FAILURE
841 #define MORECORE_FAILURE -1
842 #endif
843 
844 #ifndef MORECORE_CLEARS
845 #define MORECORE_CLEARS 1
846 #endif
847 
848 #endif /* INTERNAL_LINUX_C_LIB */
849 
850 #if defined(INTERNAL_LINUX_C_LIB) && defined(__ELF__)
851 
852 #define cALLOc		__libc_calloc
853 #define fREe		__libc_free
854 #define mALLOc		__libc_malloc
855 #define mEMALIGn	__libc_memalign
856 #define rEALLOc		__libc_realloc
857 #define vALLOc		__libc_valloc
858 #define pvALLOc		__libc_pvalloc
859 #define mALLINFo	__libc_mallinfo
860 #define mALLOPt		__libc_mallopt
861 
862 #pragma weak calloc = __libc_calloc
863 #pragma weak free = __libc_free
864 #pragma weak cfree = __libc_free
865 #pragma weak malloc = __libc_malloc
866 #pragma weak memalign = __libc_memalign
867 #pragma weak realloc = __libc_realloc
868 #pragma weak valloc = __libc_valloc
869 #pragma weak pvalloc = __libc_pvalloc
870 #pragma weak mallinfo = __libc_mallinfo
871 #pragma weak mallopt = __libc_mallopt
872 
873 #else
874 
875 #ifdef USE_DL_PREFIX
876 #define cALLOc		dlcalloc
877 #define fREe		dlfree
878 #define mALLOc		dlmalloc
879 #define mEMALIGn	dlmemalign
880 #define rEALLOc		dlrealloc
881 #define vALLOc		dlvalloc
882 #define pvALLOc		dlpvalloc
883 #define mALLINFo	dlmallinfo
884 #define mALLOPt		dlmallopt
885 #else /* USE_DL_PREFIX */
886 #define cALLOc		calloc
887 #define fREe		free
888 #define mALLOc		malloc
889 #define mEMALIGn	memalign
890 #define rEALLOc		realloc
891 #define vALLOc		valloc
892 #define pvALLOc		pvalloc
893 #define mALLINFo	mallinfo
894 #define mALLOPt		mallopt
895 #endif /* USE_DL_PREFIX */
896 
897 #endif
898 
899 /* Public routines */
900 
901 #if __STD_C
902 
903 Void_t* mALLOc(size_t);
904 void    fREe(Void_t*);
905 Void_t* rEALLOc(Void_t*, size_t);
906 Void_t* mEMALIGn(size_t, size_t);
907 Void_t* vALLOc(size_t);
908 Void_t* pvALLOc(size_t);
909 Void_t* cALLOc(size_t, size_t);
910 void    cfree(Void_t*);
911 int     malloc_trim(size_t);
912 size_t  malloc_usable_size(Void_t*);
913 void    malloc_stats();
914 int     mALLOPt(int, int);
915 struct mallinfo mALLINFo(void);
916 #else
917 Void_t* mALLOc();
918 void    fREe();
919 Void_t* rEALLOc();
920 Void_t* mEMALIGn();
921 Void_t* vALLOc();
922 Void_t* pvALLOc();
923 Void_t* cALLOc();
924 void    cfree();
925 int     malloc_trim();
926 size_t  malloc_usable_size();
927 void    malloc_stats();
928 int     mALLOPt();
929 struct mallinfo mALLINFo();
930 #endif
931 
932 
933 #ifdef __cplusplus
934 };  /* end of extern "C" */
935 #endif
936 
937 /* ---------- To make a malloc.h, end cutting here ------------ */
938 #else				/* Moved to malloc.h */
939 
940 #include <malloc.h>
941 #if 0
942 #if __STD_C
943 static void malloc_update_mallinfo (void);
944 void malloc_stats (void);
945 #else
946 static void malloc_update_mallinfo ();
947 void malloc_stats();
948 #endif
949 #endif	/* 0 */
950 
951 #endif	/* 0 */			/* Moved to malloc.h */
952 
953 DECLARE_GLOBAL_DATA_PTR;
954 
955 /*
956   Emulation of sbrk for WIN32
957   All code within the ifdef WIN32 is untested by me.
958 
959   Thanks to Martin Fong and others for supplying this.
960 */
961 
962 
963 #ifdef WIN32
964 
965 #define AlignPage(add) (((add) + (malloc_getpagesize-1)) & \
966 ~(malloc_getpagesize-1))
967 #define AlignPage64K(add) (((add) + (0x10000 - 1)) & ~(0x10000 - 1))
968 
969 /* resrve 64MB to insure large contiguous space */
970 #define RESERVED_SIZE (1024*1024*64)
971 #define NEXT_SIZE (2048*1024)
972 #define TOP_MEMORY ((unsigned long)2*1024*1024*1024)
973 
974 struct GmListElement;
975 typedef struct GmListElement GmListElement;
976 
977 struct GmListElement
978 {
979 	GmListElement* next;
980 	void* base;
981 };
982 
983 static GmListElement* head = 0;
984 static unsigned int gNextAddress = 0;
985 static unsigned int gAddressBase = 0;
986 static unsigned int gAllocatedSize = 0;
987 
988 static
989 GmListElement* makeGmListElement (void* bas)
990 {
991 	GmListElement* this;
992 	this = (GmListElement*)(void*)LocalAlloc (0, sizeof (GmListElement));
993 	assert (this);
994 	if (this)
995 	{
996 		this->base = bas;
997 		this->next = head;
998 		head = this;
999 	}
1000 	return this;
1001 }
1002 
1003 void gcleanup ()
1004 {
1005 	BOOL rval;
1006 	assert ( (head == NULL) || (head->base == (void*)gAddressBase));
1007 	if (gAddressBase && (gNextAddress - gAddressBase))
1008 	{
1009 		rval = VirtualFree ((void*)gAddressBase,
1010 							gNextAddress - gAddressBase,
1011 							MEM_DECOMMIT);
1012 	assert (rval);
1013 	}
1014 	while (head)
1015 	{
1016 		GmListElement* next = head->next;
1017 		rval = VirtualFree (head->base, 0, MEM_RELEASE);
1018 		assert (rval);
1019 		LocalFree (head);
1020 		head = next;
1021 	}
1022 }
1023 
1024 static
1025 void* findRegion (void* start_address, unsigned long size)
1026 {
1027 	MEMORY_BASIC_INFORMATION info;
1028 	if (size >= TOP_MEMORY) return NULL;
1029 
1030 	while ((unsigned long)start_address + size < TOP_MEMORY)
1031 	{
1032 		VirtualQuery (start_address, &info, sizeof (info));
1033 		if ((info.State == MEM_FREE) && (info.RegionSize >= size))
1034 			return start_address;
1035 		else
1036 		{
1037 			/* Requested region is not available so see if the */
1038 			/* next region is available.  Set 'start_address' */
1039 			/* to the next region and call 'VirtualQuery()' */
1040 			/* again. */
1041 
1042 			start_address = (char*)info.BaseAddress + info.RegionSize;
1043 
1044 			/* Make sure we start looking for the next region */
1045 			/* on the *next* 64K boundary.  Otherwise, even if */
1046 			/* the new region is free according to */
1047 			/* 'VirtualQuery()', the subsequent call to */
1048 			/* 'VirtualAlloc()' (which follows the call to */
1049 			/* this routine in 'wsbrk()') will round *down* */
1050 			/* the requested address to a 64K boundary which */
1051 			/* we already know is an address in the */
1052 			/* unavailable region.  Thus, the subsequent call */
1053 			/* to 'VirtualAlloc()' will fail and bring us back */
1054 			/* here, causing us to go into an infinite loop. */
1055 
1056 			start_address =
1057 				(void *) AlignPage64K((unsigned long) start_address);
1058 		}
1059 	}
1060 	return NULL;
1061 
1062 }
1063 
1064 
1065 void* wsbrk (long size)
1066 {
1067 	void* tmp;
1068 	if (size > 0)
1069 	{
1070 		if (gAddressBase == 0)
1071 		{
1072 			gAllocatedSize = max (RESERVED_SIZE, AlignPage (size));
1073 			gNextAddress = gAddressBase =
1074 				(unsigned int)VirtualAlloc (NULL, gAllocatedSize,
1075 											MEM_RESERVE, PAGE_NOACCESS);
1076 		} else if (AlignPage (gNextAddress + size) > (gAddressBase +
1077 gAllocatedSize))
1078 		{
1079 			long new_size = max (NEXT_SIZE, AlignPage (size));
1080 			void* new_address = (void*)(gAddressBase+gAllocatedSize);
1081 			do
1082 			{
1083 				new_address = findRegion (new_address, new_size);
1084 
1085 				if (new_address == 0)
1086 					return (void*)-1;
1087 
1088 				gAddressBase = gNextAddress =
1089 					(unsigned int)VirtualAlloc (new_address, new_size,
1090 												MEM_RESERVE, PAGE_NOACCESS);
1091 				/* repeat in case of race condition */
1092 				/* The region that we found has been snagged */
1093 				/* by another thread */
1094 			}
1095 			while (gAddressBase == 0);
1096 
1097 			assert (new_address == (void*)gAddressBase);
1098 
1099 			gAllocatedSize = new_size;
1100 
1101 			if (!makeGmListElement ((void*)gAddressBase))
1102 				return (void*)-1;
1103 		}
1104 		if ((size + gNextAddress) > AlignPage (gNextAddress))
1105 		{
1106 			void* res;
1107 			res = VirtualAlloc ((void*)AlignPage (gNextAddress),
1108 								(size + gNextAddress -
1109 								 AlignPage (gNextAddress)),
1110 								MEM_COMMIT, PAGE_READWRITE);
1111 			if (res == 0)
1112 				return (void*)-1;
1113 		}
1114 		tmp = (void*)gNextAddress;
1115 		gNextAddress = (unsigned int)tmp + size;
1116 		return tmp;
1117 	}
1118 	else if (size < 0)
1119 	{
1120 		unsigned int alignedGoal = AlignPage (gNextAddress + size);
1121 		/* Trim by releasing the virtual memory */
1122 		if (alignedGoal >= gAddressBase)
1123 		{
1124 			VirtualFree ((void*)alignedGoal, gNextAddress - alignedGoal,
1125 						 MEM_DECOMMIT);
1126 			gNextAddress = gNextAddress + size;
1127 			return (void*)gNextAddress;
1128 		}
1129 		else
1130 		{
1131 			VirtualFree ((void*)gAddressBase, gNextAddress - gAddressBase,
1132 						 MEM_DECOMMIT);
1133 			gNextAddress = gAddressBase;
1134 			return (void*)-1;
1135 		}
1136 	}
1137 	else
1138 	{
1139 		return (void*)gNextAddress;
1140 	}
1141 }
1142 
1143 #endif
1144 
1145 
1146 
1147 /*
1148   Type declarations
1149 */
1150 
1151 
1152 struct malloc_chunk
1153 {
1154   INTERNAL_SIZE_T prev_size; /* Size of previous chunk (if free). */
1155   INTERNAL_SIZE_T size;      /* Size in bytes, including overhead. */
1156   struct malloc_chunk* fd;   /* double links -- used only if free. */
1157   struct malloc_chunk* bk;
1158 };
1159 
1160 typedef struct malloc_chunk* mchunkptr;
1161 
1162 /*
1163 
1164    malloc_chunk details:
1165 
1166     (The following includes lightly edited explanations by Colin Plumb.)
1167 
1168     Chunks of memory are maintained using a `boundary tag' method as
1169     described in e.g., Knuth or Standish.  (See the paper by Paul
1170     Wilson ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a
1171     survey of such techniques.)  Sizes of free chunks are stored both
1172     in the front of each chunk and at the end.  This makes
1173     consolidating fragmented chunks into bigger chunks very fast.  The
1174     size fields also hold bits representing whether chunks are free or
1175     in use.
1176 
1177     An allocated chunk looks like this:
1178 
1179 
1180     chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1181 	    |             Size of previous chunk, if allocated            | |
1182 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1183 	    |             Size of chunk, in bytes                         |P|
1184       mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1185 	    |             User data starts here...                          .
1186 	    .                                                               .
1187 	    .             (malloc_usable_space() bytes)                     .
1188 	    .                                                               |
1189 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1190 	    |             Size of chunk                                     |
1191 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1192 
1193 
1194     Where "chunk" is the front of the chunk for the purpose of most of
1195     the malloc code, but "mem" is the pointer that is returned to the
1196     user.  "Nextchunk" is the beginning of the next contiguous chunk.
1197 
1198     Chunks always begin on even word boundries, so the mem portion
1199     (which is returned to the user) is also on an even word boundary, and
1200     thus double-word aligned.
1201 
1202     Free chunks are stored in circular doubly-linked lists, and look like this:
1203 
1204     chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1205 	    |             Size of previous chunk                            |
1206 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1207     `head:' |             Size of chunk, in bytes                         |P|
1208       mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1209 	    |             Forward pointer to next chunk in list             |
1210 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1211 	    |             Back pointer to previous chunk in list            |
1212 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1213 	    |             Unused space (may be 0 bytes long)                .
1214 	    .                                                               .
1215 	    .                                                               |
1216 nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1217     `foot:' |             Size of chunk, in bytes                           |
1218 	    +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
1219 
1220     The P (PREV_INUSE) bit, stored in the unused low-order bit of the
1221     chunk size (which is always a multiple of two words), is an in-use
1222     bit for the *previous* chunk.  If that bit is *clear*, then the
1223     word before the current chunk size contains the previous chunk
1224     size, and can be used to find the front of the previous chunk.
1225     (The very first chunk allocated always has this bit set,
1226     preventing access to non-existent (or non-owned) memory.)
1227 
1228     Note that the `foot' of the current chunk is actually represented
1229     as the prev_size of the NEXT chunk. (This makes it easier to
1230     deal with alignments etc).
1231 
1232     The two exceptions to all this are
1233 
1234      1. The special chunk `top', which doesn't bother using the
1235 	trailing size field since there is no
1236 	next contiguous chunk that would have to index off it. (After
1237 	initialization, `top' is forced to always exist.  If it would
1238 	become less than MINSIZE bytes long, it is replenished via
1239 	malloc_extend_top.)
1240 
1241      2. Chunks allocated via mmap, which have the second-lowest-order
1242 	bit (IS_MMAPPED) set in their size fields.  Because they are
1243 	never merged or traversed from any other chunk, they have no
1244 	foot size or inuse information.
1245 
1246     Available chunks are kept in any of several places (all declared below):
1247 
1248     * `av': An array of chunks serving as bin headers for consolidated
1249        chunks. Each bin is doubly linked.  The bins are approximately
1250        proportionally (log) spaced.  There are a lot of these bins
1251        (128). This may look excessive, but works very well in
1252        practice.  All procedures maintain the invariant that no
1253        consolidated chunk physically borders another one. Chunks in
1254        bins are kept in size order, with ties going to the
1255        approximately least recently used chunk.
1256 
1257        The chunks in each bin are maintained in decreasing sorted order by
1258        size.  This is irrelevant for the small bins, which all contain
1259        the same-sized chunks, but facilitates best-fit allocation for
1260        larger chunks. (These lists are just sequential. Keeping them in
1261        order almost never requires enough traversal to warrant using
1262        fancier ordered data structures.)  Chunks of the same size are
1263        linked with the most recently freed at the front, and allocations
1264        are taken from the back.  This results in LRU or FIFO allocation
1265        order, which tends to give each chunk an equal opportunity to be
1266        consolidated with adjacent freed chunks, resulting in larger free
1267        chunks and less fragmentation.
1268 
1269     * `top': The top-most available chunk (i.e., the one bordering the
1270        end of available memory) is treated specially. It is never
1271        included in any bin, is used only if no other chunk is
1272        available, and is released back to the system if it is very
1273        large (see M_TRIM_THRESHOLD).
1274 
1275     * `last_remainder': A bin holding only the remainder of the
1276        most recently split (non-top) chunk. This bin is checked
1277        before other non-fitting chunks, so as to provide better
1278        locality for runs of sequentially allocated chunks.
1279 
1280     *  Implicitly, through the host system's memory mapping tables.
1281        If supported, requests greater than a threshold are usually
1282        serviced via calls to mmap, and then later released via munmap.
1283 
1284 */
1285 
1286 /*  sizes, alignments */
1287 
1288 #define SIZE_SZ                (sizeof(INTERNAL_SIZE_T))
1289 #define MALLOC_ALIGNMENT       (SIZE_SZ + SIZE_SZ)
1290 #define MALLOC_ALIGN_MASK      (MALLOC_ALIGNMENT - 1)
1291 #define MINSIZE                (sizeof(struct malloc_chunk))
1292 
1293 /* conversion from malloc headers to user pointers, and back */
1294 
1295 #define chunk2mem(p)   ((Void_t*)((char*)(p) + 2*SIZE_SZ))
1296 #define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ))
1297 
1298 /* pad request bytes into a usable size */
1299 
1300 #define request2size(req) \
1301  (((long)((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) < \
1302   (long)(MINSIZE + MALLOC_ALIGN_MASK)) ? MINSIZE : \
1303    (((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) & ~(MALLOC_ALIGN_MASK)))
1304 
1305 /* Check if m has acceptable alignment */
1306 
1307 #define aligned_OK(m)    (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
1308 
1309 
1310 
1311 
1312 /*
1313   Physical chunk operations
1314 */
1315 
1316 
1317 /* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
1318 
1319 #define PREV_INUSE 0x1
1320 
1321 /* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
1322 
1323 #define IS_MMAPPED 0x2
1324 
1325 /* Bits to mask off when extracting size */
1326 
1327 #define SIZE_BITS (PREV_INUSE|IS_MMAPPED)
1328 
1329 
1330 /* Ptr to next physical malloc_chunk. */
1331 
1332 #define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~PREV_INUSE) ))
1333 
1334 /* Ptr to previous physical malloc_chunk */
1335 
1336 #define prev_chunk(p)\
1337    ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) ))
1338 
1339 
1340 /* Treat space at ptr + offset as a chunk */
1341 
1342 #define chunk_at_offset(p, s)  ((mchunkptr)(((char*)(p)) + (s)))
1343 
1344 
1345 
1346 
1347 /*
1348   Dealing with use bits
1349 */
1350 
1351 /* extract p's inuse bit */
1352 
1353 #define inuse(p)\
1354 ((((mchunkptr)(((char*)(p))+((p)->size & ~PREV_INUSE)))->size) & PREV_INUSE)
1355 
1356 /* extract inuse bit of previous chunk */
1357 
1358 #define prev_inuse(p)  ((p)->size & PREV_INUSE)
1359 
1360 /* check for mmap()'ed chunk */
1361 
1362 #define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
1363 
1364 /* set/clear chunk as in use without otherwise disturbing */
1365 
1366 #define set_inuse(p)\
1367 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size |= PREV_INUSE
1368 
1369 #define clear_inuse(p)\
1370 ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size &= ~(PREV_INUSE)
1371 
1372 /* check/set/clear inuse bits in known places */
1373 
1374 #define inuse_bit_at_offset(p, s)\
1375  (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE)
1376 
1377 #define set_inuse_bit_at_offset(p, s)\
1378  (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE)
1379 
1380 #define clear_inuse_bit_at_offset(p, s)\
1381  (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE))
1382 
1383 
1384 
1385 
1386 /*
1387   Dealing with size fields
1388 */
1389 
1390 /* Get size, ignoring use bits */
1391 
1392 #define chunksize(p)          ((p)->size & ~(SIZE_BITS))
1393 
1394 /* Set size at head, without disturbing its use bit */
1395 
1396 #define set_head_size(p, s)   ((p)->size = (((p)->size & PREV_INUSE) | (s)))
1397 
1398 /* Set size/use ignoring previous bits in header */
1399 
1400 #define set_head(p, s)        ((p)->size = (s))
1401 
1402 /* Set size at footer (only when chunk is not in use) */
1403 
1404 #define set_foot(p, s)   (((mchunkptr)((char*)(p) + (s)))->prev_size = (s))
1405 
1406 
1407 
1408 
1409 
1410 /*
1411    Bins
1412 
1413     The bins, `av_' are an array of pairs of pointers serving as the
1414     heads of (initially empty) doubly-linked lists of chunks, laid out
1415     in a way so that each pair can be treated as if it were in a
1416     malloc_chunk. (This way, the fd/bk offsets for linking bin heads
1417     and chunks are the same).
1418 
1419     Bins for sizes < 512 bytes contain chunks of all the same size, spaced
1420     8 bytes apart. Larger bins are approximately logarithmically
1421     spaced. (See the table below.) The `av_' array is never mentioned
1422     directly in the code, but instead via bin access macros.
1423 
1424     Bin layout:
1425 
1426     64 bins of size       8
1427     32 bins of size      64
1428     16 bins of size     512
1429      8 bins of size    4096
1430      4 bins of size   32768
1431      2 bins of size  262144
1432      1 bin  of size what's left
1433 
1434     There is actually a little bit of slop in the numbers in bin_index
1435     for the sake of speed. This makes no difference elsewhere.
1436 
1437     The special chunks `top' and `last_remainder' get their own bins,
1438     (this is implemented via yet more trickery with the av_ array),
1439     although `top' is never properly linked to its bin since it is
1440     always handled specially.
1441 
1442 */
1443 
1444 #define NAV             128   /* number of bins */
1445 
1446 typedef struct malloc_chunk* mbinptr;
1447 
1448 /* access macros */
1449 
1450 #define bin_at(i)      ((mbinptr)((char*)&(av_[2*(i) + 2]) - 2*SIZE_SZ))
1451 #define next_bin(b)    ((mbinptr)((char*)(b) + 2 * sizeof(mbinptr)))
1452 #define prev_bin(b)    ((mbinptr)((char*)(b) - 2 * sizeof(mbinptr)))
1453 
1454 /*
1455    The first 2 bins are never indexed. The corresponding av_ cells are instead
1456    used for bookkeeping. This is not to save space, but to simplify
1457    indexing, maintain locality, and avoid some initialization tests.
1458 */
1459 
1460 #define top            (av_[2])          /* The topmost chunk */
1461 #define last_remainder (bin_at(1))       /* remainder from last split */
1462 
1463 
1464 /*
1465    Because top initially points to its own bin with initial
1466    zero size, thus forcing extension on the first malloc request,
1467    we avoid having any special code in malloc to check whether
1468    it even exists yet. But we still need to in malloc_extend_top.
1469 */
1470 
1471 #define initial_top    ((mchunkptr)(bin_at(0)))
1472 
1473 /* Helper macro to initialize bins */
1474 
1475 #define IAV(i)  bin_at(i), bin_at(i)
1476 
1477 static mbinptr av_[NAV * 2 + 2] = {
1478  0, 0,
1479  IAV(0),   IAV(1),   IAV(2),   IAV(3),   IAV(4),   IAV(5),   IAV(6),   IAV(7),
1480  IAV(8),   IAV(9),   IAV(10),  IAV(11),  IAV(12),  IAV(13),  IAV(14),  IAV(15),
1481  IAV(16),  IAV(17),  IAV(18),  IAV(19),  IAV(20),  IAV(21),  IAV(22),  IAV(23),
1482  IAV(24),  IAV(25),  IAV(26),  IAV(27),  IAV(28),  IAV(29),  IAV(30),  IAV(31),
1483  IAV(32),  IAV(33),  IAV(34),  IAV(35),  IAV(36),  IAV(37),  IAV(38),  IAV(39),
1484  IAV(40),  IAV(41),  IAV(42),  IAV(43),  IAV(44),  IAV(45),  IAV(46),  IAV(47),
1485  IAV(48),  IAV(49),  IAV(50),  IAV(51),  IAV(52),  IAV(53),  IAV(54),  IAV(55),
1486  IAV(56),  IAV(57),  IAV(58),  IAV(59),  IAV(60),  IAV(61),  IAV(62),  IAV(63),
1487  IAV(64),  IAV(65),  IAV(66),  IAV(67),  IAV(68),  IAV(69),  IAV(70),  IAV(71),
1488  IAV(72),  IAV(73),  IAV(74),  IAV(75),  IAV(76),  IAV(77),  IAV(78),  IAV(79),
1489  IAV(80),  IAV(81),  IAV(82),  IAV(83),  IAV(84),  IAV(85),  IAV(86),  IAV(87),
1490  IAV(88),  IAV(89),  IAV(90),  IAV(91),  IAV(92),  IAV(93),  IAV(94),  IAV(95),
1491  IAV(96),  IAV(97),  IAV(98),  IAV(99),  IAV(100), IAV(101), IAV(102), IAV(103),
1492  IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111),
1493  IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119),
1494  IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127)
1495 };
1496 
1497 void malloc_bin_reloc (void)
1498 {
1499 	unsigned long *p = (unsigned long *)(&av_[2]);
1500 	int i;
1501 	for (i=2; i<(sizeof(av_)/sizeof(mbinptr)); ++i) {
1502 		*p++ += gd->reloc_off;
1503 	}
1504 }
1505 
1506 
1507 /* field-extraction macros */
1508 
1509 #define first(b) ((b)->fd)
1510 #define last(b)  ((b)->bk)
1511 
1512 /*
1513   Indexing into bins
1514 */
1515 
1516 #define bin_index(sz)                                                          \
1517 (((((unsigned long)(sz)) >> 9) ==    0) ?       (((unsigned long)(sz)) >>  3): \
1518  ((((unsigned long)(sz)) >> 9) <=    4) ?  56 + (((unsigned long)(sz)) >>  6): \
1519  ((((unsigned long)(sz)) >> 9) <=   20) ?  91 + (((unsigned long)(sz)) >>  9): \
1520  ((((unsigned long)(sz)) >> 9) <=   84) ? 110 + (((unsigned long)(sz)) >> 12): \
1521  ((((unsigned long)(sz)) >> 9) <=  340) ? 119 + (((unsigned long)(sz)) >> 15): \
1522  ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18): \
1523 					  126)
1524 /*
1525   bins for chunks < 512 are all spaced 8 bytes apart, and hold
1526   identically sized chunks. This is exploited in malloc.
1527 */
1528 
1529 #define MAX_SMALLBIN         63
1530 #define MAX_SMALLBIN_SIZE   512
1531 #define SMALLBIN_WIDTH        8
1532 
1533 #define smallbin_index(sz)  (((unsigned long)(sz)) >> 3)
1534 
1535 /*
1536    Requests are `small' if both the corresponding and the next bin are small
1537 */
1538 
1539 #define is_small_request(nb) (nb < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH)
1540 
1541 
1542 
1543 /*
1544     To help compensate for the large number of bins, a one-level index
1545     structure is used for bin-by-bin searching.  `binblocks' is a
1546     one-word bitvector recording whether groups of BINBLOCKWIDTH bins
1547     have any (possibly) non-empty bins, so they can be skipped over
1548     all at once during during traversals. The bits are NOT always
1549     cleared as soon as all bins in a block are empty, but instead only
1550     when all are noticed to be empty during traversal in malloc.
1551 */
1552 
1553 #define BINBLOCKWIDTH     4   /* bins per block */
1554 
1555 #define binblocks_r     ((INTERNAL_SIZE_T)av_[1]) /* bitvector of nonempty blocks */
1556 #define binblocks_w     (av_[1])
1557 
1558 /* bin<->block macros */
1559 
1560 #define idx2binblock(ix)    ((unsigned)1 << (ix / BINBLOCKWIDTH))
1561 #define mark_binblock(ii)   (binblocks_w = (mbinptr)(binblocks_r | idx2binblock(ii)))
1562 #define clear_binblock(ii)  (binblocks_w = (mbinptr)(binblocks_r & ~(idx2binblock(ii))))
1563 
1564 
1565 
1566 
1567 
1568 /*  Other static bookkeeping data */
1569 
1570 /* variables holding tunable values */
1571 
1572 static unsigned long trim_threshold   = DEFAULT_TRIM_THRESHOLD;
1573 static unsigned long top_pad          = DEFAULT_TOP_PAD;
1574 static unsigned int  n_mmaps_max      = DEFAULT_MMAP_MAX;
1575 static unsigned long mmap_threshold   = DEFAULT_MMAP_THRESHOLD;
1576 
1577 /* The first value returned from sbrk */
1578 static char* sbrk_base = (char*)(-1);
1579 
1580 /* The maximum memory obtained from system via sbrk */
1581 static unsigned long max_sbrked_mem = 0;
1582 
1583 /* The maximum via either sbrk or mmap */
1584 static unsigned long max_total_mem = 0;
1585 
1586 /* internal working copy of mallinfo */
1587 static struct mallinfo current_mallinfo = {  0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
1588 
1589 /* The total memory obtained from system via sbrk */
1590 #define sbrked_mem  (current_mallinfo.arena)
1591 
1592 /* Tracking mmaps */
1593 
1594 #if 0
1595 static unsigned int n_mmaps = 0;
1596 #endif	/* 0 */
1597 static unsigned long mmapped_mem = 0;
1598 #if HAVE_MMAP
1599 static unsigned int max_n_mmaps = 0;
1600 static unsigned long max_mmapped_mem = 0;
1601 #endif
1602 
1603 
1604 
1605 /*
1606   Debugging support
1607 */
1608 
1609 #ifdef DEBUG
1610 
1611 
1612 /*
1613   These routines make a number of assertions about the states
1614   of data structures that should be true at all times. If any
1615   are not true, it's very likely that a user program has somehow
1616   trashed memory. (It's also possible that there is a coding error
1617   in malloc. In which case, please report it!)
1618 */
1619 
1620 #if __STD_C
1621 static void do_check_chunk(mchunkptr p)
1622 #else
1623 static void do_check_chunk(p) mchunkptr p;
1624 #endif
1625 {
1626 #if 0	/* causes warnings because assert() is off */
1627   INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1628 #endif	/* 0 */
1629 
1630   /* No checkable chunk is mmapped */
1631   assert(!chunk_is_mmapped(p));
1632 
1633   /* Check for legal address ... */
1634   assert((char*)p >= sbrk_base);
1635   if (p != top)
1636     assert((char*)p + sz <= (char*)top);
1637   else
1638     assert((char*)p + sz <= sbrk_base + sbrked_mem);
1639 
1640 }
1641 
1642 
1643 #if __STD_C
1644 static void do_check_free_chunk(mchunkptr p)
1645 #else
1646 static void do_check_free_chunk(p) mchunkptr p;
1647 #endif
1648 {
1649   INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1650 #if 0	/* causes warnings because assert() is off */
1651   mchunkptr next = chunk_at_offset(p, sz);
1652 #endif	/* 0 */
1653 
1654   do_check_chunk(p);
1655 
1656   /* Check whether it claims to be free ... */
1657   assert(!inuse(p));
1658 
1659   /* Unless a special marker, must have OK fields */
1660   if ((long)sz >= (long)MINSIZE)
1661   {
1662     assert((sz & MALLOC_ALIGN_MASK) == 0);
1663     assert(aligned_OK(chunk2mem(p)));
1664     /* ... matching footer field */
1665     assert(next->prev_size == sz);
1666     /* ... and is fully consolidated */
1667     assert(prev_inuse(p));
1668     assert (next == top || inuse(next));
1669 
1670     /* ... and has minimally sane links */
1671     assert(p->fd->bk == p);
1672     assert(p->bk->fd == p);
1673   }
1674   else /* markers are always of size SIZE_SZ */
1675     assert(sz == SIZE_SZ);
1676 }
1677 
1678 #if __STD_C
1679 static void do_check_inuse_chunk(mchunkptr p)
1680 #else
1681 static void do_check_inuse_chunk(p) mchunkptr p;
1682 #endif
1683 {
1684   mchunkptr next = next_chunk(p);
1685   do_check_chunk(p);
1686 
1687   /* Check whether it claims to be in use ... */
1688   assert(inuse(p));
1689 
1690   /* ... and is surrounded by OK chunks.
1691     Since more things can be checked with free chunks than inuse ones,
1692     if an inuse chunk borders them and debug is on, it's worth doing them.
1693   */
1694   if (!prev_inuse(p))
1695   {
1696     mchunkptr prv = prev_chunk(p);
1697     assert(next_chunk(prv) == p);
1698     do_check_free_chunk(prv);
1699   }
1700   if (next == top)
1701   {
1702     assert(prev_inuse(next));
1703     assert(chunksize(next) >= MINSIZE);
1704   }
1705   else if (!inuse(next))
1706     do_check_free_chunk(next);
1707 
1708 }
1709 
1710 #if __STD_C
1711 static void do_check_malloced_chunk(mchunkptr p, INTERNAL_SIZE_T s)
1712 #else
1713 static void do_check_malloced_chunk(p, s) mchunkptr p; INTERNAL_SIZE_T s;
1714 #endif
1715 {
1716 #if 0	/* causes warnings because assert() is off */
1717   INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
1718   long room = sz - s;
1719 #endif	/* 0 */
1720 
1721   do_check_inuse_chunk(p);
1722 
1723   /* Legal size ... */
1724   assert((long)sz >= (long)MINSIZE);
1725   assert((sz & MALLOC_ALIGN_MASK) == 0);
1726   assert(room >= 0);
1727   assert(room < (long)MINSIZE);
1728 
1729   /* ... and alignment */
1730   assert(aligned_OK(chunk2mem(p)));
1731 
1732 
1733   /* ... and was allocated at front of an available chunk */
1734   assert(prev_inuse(p));
1735 
1736 }
1737 
1738 
1739 #define check_free_chunk(P)  do_check_free_chunk(P)
1740 #define check_inuse_chunk(P) do_check_inuse_chunk(P)
1741 #define check_chunk(P) do_check_chunk(P)
1742 #define check_malloced_chunk(P,N) do_check_malloced_chunk(P,N)
1743 #else
1744 #define check_free_chunk(P)
1745 #define check_inuse_chunk(P)
1746 #define check_chunk(P)
1747 #define check_malloced_chunk(P,N)
1748 #endif
1749 
1750 
1751 
1752 /*
1753   Macro-based internal utilities
1754 */
1755 
1756 
1757 /*
1758   Linking chunks in bin lists.
1759   Call these only with variables, not arbitrary expressions, as arguments.
1760 */
1761 
1762 /*
1763   Place chunk p of size s in its bin, in size order,
1764   putting it ahead of others of same size.
1765 */
1766 
1767 
1768 #define frontlink(P, S, IDX, BK, FD)                                          \
1769 {                                                                             \
1770   if (S < MAX_SMALLBIN_SIZE)                                                  \
1771   {                                                                           \
1772     IDX = smallbin_index(S);                                                  \
1773     mark_binblock(IDX);                                                       \
1774     BK = bin_at(IDX);                                                         \
1775     FD = BK->fd;                                                              \
1776     P->bk = BK;                                                               \
1777     P->fd = FD;                                                               \
1778     FD->bk = BK->fd = P;                                                      \
1779   }                                                                           \
1780   else                                                                        \
1781   {                                                                           \
1782     IDX = bin_index(S);                                                       \
1783     BK = bin_at(IDX);                                                         \
1784     FD = BK->fd;                                                              \
1785     if (FD == BK) mark_binblock(IDX);                                         \
1786     else                                                                      \
1787     {                                                                         \
1788       while (FD != BK && S < chunksize(FD)) FD = FD->fd;                      \
1789       BK = FD->bk;                                                            \
1790     }                                                                         \
1791     P->bk = BK;                                                               \
1792     P->fd = FD;                                                               \
1793     FD->bk = BK->fd = P;                                                      \
1794   }                                                                           \
1795 }
1796 
1797 
1798 /* take a chunk off a list */
1799 
1800 #define unlink(P, BK, FD)                                                     \
1801 {                                                                             \
1802   BK = P->bk;                                                                 \
1803   FD = P->fd;                                                                 \
1804   FD->bk = BK;                                                                \
1805   BK->fd = FD;                                                                \
1806 }                                                                             \
1807 
1808 /* Place p as the last remainder */
1809 
1810 #define link_last_remainder(P)                                                \
1811 {                                                                             \
1812   last_remainder->fd = last_remainder->bk =  P;                               \
1813   P->fd = P->bk = last_remainder;                                             \
1814 }
1815 
1816 /* Clear the last_remainder bin */
1817 
1818 #define clear_last_remainder \
1819   (last_remainder->fd = last_remainder->bk = last_remainder)
1820 
1821 
1822 
1823 
1824 
1825 /* Routines dealing with mmap(). */
1826 
1827 #if HAVE_MMAP
1828 
1829 #if __STD_C
1830 static mchunkptr mmap_chunk(size_t size)
1831 #else
1832 static mchunkptr mmap_chunk(size) size_t size;
1833 #endif
1834 {
1835   size_t page_mask = malloc_getpagesize - 1;
1836   mchunkptr p;
1837 
1838 #ifndef MAP_ANONYMOUS
1839   static int fd = -1;
1840 #endif
1841 
1842   if(n_mmaps >= n_mmaps_max) return 0; /* too many regions */
1843 
1844   /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because
1845    * there is no following chunk whose prev_size field could be used.
1846    */
1847   size = (size + SIZE_SZ + page_mask) & ~page_mask;
1848 
1849 #ifdef MAP_ANONYMOUS
1850   p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE,
1851 		      MAP_PRIVATE|MAP_ANONYMOUS, -1, 0);
1852 #else /* !MAP_ANONYMOUS */
1853   if (fd < 0)
1854   {
1855     fd = open("/dev/zero", O_RDWR);
1856     if(fd < 0) return 0;
1857   }
1858   p = (mchunkptr)mmap(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE, fd, 0);
1859 #endif
1860 
1861   if(p == (mchunkptr)-1) return 0;
1862 
1863   n_mmaps++;
1864   if (n_mmaps > max_n_mmaps) max_n_mmaps = n_mmaps;
1865 
1866   /* We demand that eight bytes into a page must be 8-byte aligned. */
1867   assert(aligned_OK(chunk2mem(p)));
1868 
1869   /* The offset to the start of the mmapped region is stored
1870    * in the prev_size field of the chunk; normally it is zero,
1871    * but that can be changed in memalign().
1872    */
1873   p->prev_size = 0;
1874   set_head(p, size|IS_MMAPPED);
1875 
1876   mmapped_mem += size;
1877   if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1878     max_mmapped_mem = mmapped_mem;
1879   if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1880     max_total_mem = mmapped_mem + sbrked_mem;
1881   return p;
1882 }
1883 
1884 #if __STD_C
1885 static void munmap_chunk(mchunkptr p)
1886 #else
1887 static void munmap_chunk(p) mchunkptr p;
1888 #endif
1889 {
1890   INTERNAL_SIZE_T size = chunksize(p);
1891   int ret;
1892 
1893   assert (chunk_is_mmapped(p));
1894   assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1895   assert((n_mmaps > 0));
1896   assert(((p->prev_size + size) & (malloc_getpagesize-1)) == 0);
1897 
1898   n_mmaps--;
1899   mmapped_mem -= (size + p->prev_size);
1900 
1901   ret = munmap((char *)p - p->prev_size, size + p->prev_size);
1902 
1903   /* munmap returns non-zero on failure */
1904   assert(ret == 0);
1905 }
1906 
1907 #if HAVE_MREMAP
1908 
1909 #if __STD_C
1910 static mchunkptr mremap_chunk(mchunkptr p, size_t new_size)
1911 #else
1912 static mchunkptr mremap_chunk(p, new_size) mchunkptr p; size_t new_size;
1913 #endif
1914 {
1915   size_t page_mask = malloc_getpagesize - 1;
1916   INTERNAL_SIZE_T offset = p->prev_size;
1917   INTERNAL_SIZE_T size = chunksize(p);
1918   char *cp;
1919 
1920   assert (chunk_is_mmapped(p));
1921   assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
1922   assert((n_mmaps > 0));
1923   assert(((size + offset) & (malloc_getpagesize-1)) == 0);
1924 
1925   /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
1926   new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
1927 
1928   cp = (char *)mremap((char *)p - offset, size + offset, new_size, 1);
1929 
1930   if (cp == (char *)-1) return 0;
1931 
1932   p = (mchunkptr)(cp + offset);
1933 
1934   assert(aligned_OK(chunk2mem(p)));
1935 
1936   assert((p->prev_size == offset));
1937   set_head(p, (new_size - offset)|IS_MMAPPED);
1938 
1939   mmapped_mem -= size + offset;
1940   mmapped_mem += new_size;
1941   if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
1942     max_mmapped_mem = mmapped_mem;
1943   if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
1944     max_total_mem = mmapped_mem + sbrked_mem;
1945   return p;
1946 }
1947 
1948 #endif /* HAVE_MREMAP */
1949 
1950 #endif /* HAVE_MMAP */
1951 
1952 
1953 
1954 
1955 /*
1956   Extend the top-most chunk by obtaining memory from system.
1957   Main interface to sbrk (but see also malloc_trim).
1958 */
1959 
1960 #if __STD_C
1961 static void malloc_extend_top(INTERNAL_SIZE_T nb)
1962 #else
1963 static void malloc_extend_top(nb) INTERNAL_SIZE_T nb;
1964 #endif
1965 {
1966   char*     brk;                  /* return value from sbrk */
1967   INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of sbrked space */
1968   INTERNAL_SIZE_T correction;     /* bytes for 2nd sbrk call */
1969   char*     new_brk;              /* return of 2nd sbrk call */
1970   INTERNAL_SIZE_T top_size;       /* new size of top chunk */
1971 
1972   mchunkptr old_top     = top;  /* Record state of old top */
1973   INTERNAL_SIZE_T old_top_size = chunksize(old_top);
1974   char*     old_end      = (char*)(chunk_at_offset(old_top, old_top_size));
1975 
1976   /* Pad request with top_pad plus minimal overhead */
1977 
1978   INTERNAL_SIZE_T    sbrk_size     = nb + top_pad + MINSIZE;
1979   unsigned long pagesz    = malloc_getpagesize;
1980 
1981   /* If not the first time through, round to preserve page boundary */
1982   /* Otherwise, we need to correct to a page size below anyway. */
1983   /* (We also correct below if an intervening foreign sbrk call.) */
1984 
1985   if (sbrk_base != (char*)(-1))
1986     sbrk_size = (sbrk_size + (pagesz - 1)) & ~(pagesz - 1);
1987 
1988   brk = (char*)(MORECORE (sbrk_size));
1989 
1990   /* Fail if sbrk failed or if a foreign sbrk call killed our space */
1991   if (brk == (char*)(MORECORE_FAILURE) ||
1992       (brk < old_end && old_top != initial_top))
1993     return;
1994 
1995   sbrked_mem += sbrk_size;
1996 
1997   if (brk == old_end) /* can just add bytes to current top */
1998   {
1999     top_size = sbrk_size + old_top_size;
2000     set_head(top, top_size | PREV_INUSE);
2001   }
2002   else
2003   {
2004     if (sbrk_base == (char*)(-1))  /* First time through. Record base */
2005       sbrk_base = brk;
2006     else  /* Someone else called sbrk().  Count those bytes as sbrked_mem. */
2007       sbrked_mem += brk - (char*)old_end;
2008 
2009     /* Guarantee alignment of first new chunk made from this space */
2010     front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
2011     if (front_misalign > 0)
2012     {
2013       correction = (MALLOC_ALIGNMENT) - front_misalign;
2014       brk += correction;
2015     }
2016     else
2017       correction = 0;
2018 
2019     /* Guarantee the next brk will be at a page boundary */
2020 
2021     correction += ((((unsigned long)(brk + sbrk_size))+(pagesz-1)) &
2022 		   ~(pagesz - 1)) - ((unsigned long)(brk + sbrk_size));
2023 
2024     /* Allocate correction */
2025     new_brk = (char*)(MORECORE (correction));
2026     if (new_brk == (char*)(MORECORE_FAILURE)) return;
2027 
2028     sbrked_mem += correction;
2029 
2030     top = (mchunkptr)brk;
2031     top_size = new_brk - brk + correction;
2032     set_head(top, top_size | PREV_INUSE);
2033 
2034     if (old_top != initial_top)
2035     {
2036 
2037       /* There must have been an intervening foreign sbrk call. */
2038       /* A double fencepost is necessary to prevent consolidation */
2039 
2040       /* If not enough space to do this, then user did something very wrong */
2041       if (old_top_size < MINSIZE)
2042       {
2043 	set_head(top, PREV_INUSE); /* will force null return from malloc */
2044 	return;
2045       }
2046 
2047       /* Also keep size a multiple of MALLOC_ALIGNMENT */
2048       old_top_size = (old_top_size - 3*SIZE_SZ) & ~MALLOC_ALIGN_MASK;
2049       set_head_size(old_top, old_top_size);
2050       chunk_at_offset(old_top, old_top_size          )->size =
2051 	SIZE_SZ|PREV_INUSE;
2052       chunk_at_offset(old_top, old_top_size + SIZE_SZ)->size =
2053 	SIZE_SZ|PREV_INUSE;
2054       /* If possible, release the rest. */
2055       if (old_top_size >= MINSIZE)
2056 	fREe(chunk2mem(old_top));
2057     }
2058   }
2059 
2060   if ((unsigned long)sbrked_mem > (unsigned long)max_sbrked_mem)
2061     max_sbrked_mem = sbrked_mem;
2062   if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem)
2063     max_total_mem = mmapped_mem + sbrked_mem;
2064 
2065   /* We always land on a page boundary */
2066   assert(((unsigned long)((char*)top + top_size) & (pagesz - 1)) == 0);
2067 }
2068 
2069 
2070 
2071 
2072 /* Main public routines */
2073 
2074 
2075 /*
2076   Malloc Algorthim:
2077 
2078     The requested size is first converted into a usable form, `nb'.
2079     This currently means to add 4 bytes overhead plus possibly more to
2080     obtain 8-byte alignment and/or to obtain a size of at least
2081     MINSIZE (currently 16 bytes), the smallest allocatable size.
2082     (All fits are considered `exact' if they are within MINSIZE bytes.)
2083 
2084     From there, the first successful of the following steps is taken:
2085 
2086       1. The bin corresponding to the request size is scanned, and if
2087 	 a chunk of exactly the right size is found, it is taken.
2088 
2089       2. The most recently remaindered chunk is used if it is big
2090 	 enough.  This is a form of (roving) first fit, used only in
2091 	 the absence of exact fits. Runs of consecutive requests use
2092 	 the remainder of the chunk used for the previous such request
2093 	 whenever possible. This limited use of a first-fit style
2094 	 allocation strategy tends to give contiguous chunks
2095 	 coextensive lifetimes, which improves locality and can reduce
2096 	 fragmentation in the long run.
2097 
2098       3. Other bins are scanned in increasing size order, using a
2099 	 chunk big enough to fulfill the request, and splitting off
2100 	 any remainder.  This search is strictly by best-fit; i.e.,
2101 	 the smallest (with ties going to approximately the least
2102 	 recently used) chunk that fits is selected.
2103 
2104       4. If large enough, the chunk bordering the end of memory
2105 	 (`top') is split off. (This use of `top' is in accord with
2106 	 the best-fit search rule.  In effect, `top' is treated as
2107 	 larger (and thus less well fitting) than any other available
2108 	 chunk since it can be extended to be as large as necessary
2109 	 (up to system limitations).
2110 
2111       5. If the request size meets the mmap threshold and the
2112 	 system supports mmap, and there are few enough currently
2113 	 allocated mmapped regions, and a call to mmap succeeds,
2114 	 the request is allocated via direct memory mapping.
2115 
2116       6. Otherwise, the top of memory is extended by
2117 	 obtaining more space from the system (normally using sbrk,
2118 	 but definable to anything else via the MORECORE macro).
2119 	 Memory is gathered from the system (in system page-sized
2120 	 units) in a way that allows chunks obtained across different
2121 	 sbrk calls to be consolidated, but does not require
2122 	 contiguous memory. Thus, it should be safe to intersperse
2123 	 mallocs with other sbrk calls.
2124 
2125 
2126       All allocations are made from the the `lowest' part of any found
2127       chunk. (The implementation invariant is that prev_inuse is
2128       always true of any allocated chunk; i.e., that each allocated
2129       chunk borders either a previously allocated and still in-use chunk,
2130       or the base of its memory arena.)
2131 
2132 */
2133 
2134 #if __STD_C
2135 Void_t* mALLOc(size_t bytes)
2136 #else
2137 Void_t* mALLOc(bytes) size_t bytes;
2138 #endif
2139 {
2140   mchunkptr victim;                  /* inspected/selected chunk */
2141   INTERNAL_SIZE_T victim_size;       /* its size */
2142   int       idx;                     /* index for bin traversal */
2143   mbinptr   bin;                     /* associated bin */
2144   mchunkptr remainder;               /* remainder from a split */
2145   long      remainder_size;          /* its size */
2146   int       remainder_index;         /* its bin index */
2147   unsigned long block;               /* block traverser bit */
2148   int       startidx;                /* first bin of a traversed block */
2149   mchunkptr fwd;                     /* misc temp for linking */
2150   mchunkptr bck;                     /* misc temp for linking */
2151   mbinptr q;                         /* misc temp */
2152 
2153   INTERNAL_SIZE_T nb;
2154 
2155   if ((long)bytes < 0) return 0;
2156 
2157   nb = request2size(bytes);  /* padded request size; */
2158 
2159   /* Check for exact match in a bin */
2160 
2161   if (is_small_request(nb))  /* Faster version for small requests */
2162   {
2163     idx = smallbin_index(nb);
2164 
2165     /* No traversal or size check necessary for small bins.  */
2166 
2167     q = bin_at(idx);
2168     victim = last(q);
2169 
2170     /* Also scan the next one, since it would have a remainder < MINSIZE */
2171     if (victim == q)
2172     {
2173       q = next_bin(q);
2174       victim = last(q);
2175     }
2176     if (victim != q)
2177     {
2178       victim_size = chunksize(victim);
2179       unlink(victim, bck, fwd);
2180       set_inuse_bit_at_offset(victim, victim_size);
2181       check_malloced_chunk(victim, nb);
2182       return chunk2mem(victim);
2183     }
2184 
2185     idx += 2; /* Set for bin scan below. We've already scanned 2 bins. */
2186 
2187   }
2188   else
2189   {
2190     idx = bin_index(nb);
2191     bin = bin_at(idx);
2192 
2193     for (victim = last(bin); victim != bin; victim = victim->bk)
2194     {
2195       victim_size = chunksize(victim);
2196       remainder_size = victim_size - nb;
2197 
2198       if (remainder_size >= (long)MINSIZE) /* too big */
2199       {
2200 	--idx; /* adjust to rescan below after checking last remainder */
2201 	break;
2202       }
2203 
2204       else if (remainder_size >= 0) /* exact fit */
2205       {
2206 	unlink(victim, bck, fwd);
2207 	set_inuse_bit_at_offset(victim, victim_size);
2208 	check_malloced_chunk(victim, nb);
2209 	return chunk2mem(victim);
2210       }
2211     }
2212 
2213     ++idx;
2214 
2215   }
2216 
2217   /* Try to use the last split-off remainder */
2218 
2219   if ( (victim = last_remainder->fd) != last_remainder)
2220   {
2221     victim_size = chunksize(victim);
2222     remainder_size = victim_size - nb;
2223 
2224     if (remainder_size >= (long)MINSIZE) /* re-split */
2225     {
2226       remainder = chunk_at_offset(victim, nb);
2227       set_head(victim, nb | PREV_INUSE);
2228       link_last_remainder(remainder);
2229       set_head(remainder, remainder_size | PREV_INUSE);
2230       set_foot(remainder, remainder_size);
2231       check_malloced_chunk(victim, nb);
2232       return chunk2mem(victim);
2233     }
2234 
2235     clear_last_remainder;
2236 
2237     if (remainder_size >= 0)  /* exhaust */
2238     {
2239       set_inuse_bit_at_offset(victim, victim_size);
2240       check_malloced_chunk(victim, nb);
2241       return chunk2mem(victim);
2242     }
2243 
2244     /* Else place in bin */
2245 
2246     frontlink(victim, victim_size, remainder_index, bck, fwd);
2247   }
2248 
2249   /*
2250      If there are any possibly nonempty big-enough blocks,
2251      search for best fitting chunk by scanning bins in blockwidth units.
2252   */
2253 
2254   if ( (block = idx2binblock(idx)) <= binblocks_r)
2255   {
2256 
2257     /* Get to the first marked block */
2258 
2259     if ( (block & binblocks_r) == 0)
2260     {
2261       /* force to an even block boundary */
2262       idx = (idx & ~(BINBLOCKWIDTH - 1)) + BINBLOCKWIDTH;
2263       block <<= 1;
2264       while ((block & binblocks_r) == 0)
2265       {
2266 	idx += BINBLOCKWIDTH;
2267 	block <<= 1;
2268       }
2269     }
2270 
2271     /* For each possibly nonempty block ... */
2272     for (;;)
2273     {
2274       startidx = idx;          /* (track incomplete blocks) */
2275       q = bin = bin_at(idx);
2276 
2277       /* For each bin in this block ... */
2278       do
2279       {
2280 	/* Find and use first big enough chunk ... */
2281 
2282 	for (victim = last(bin); victim != bin; victim = victim->bk)
2283 	{
2284 	  victim_size = chunksize(victim);
2285 	  remainder_size = victim_size - nb;
2286 
2287 	  if (remainder_size >= (long)MINSIZE) /* split */
2288 	  {
2289 	    remainder = chunk_at_offset(victim, nb);
2290 	    set_head(victim, nb | PREV_INUSE);
2291 	    unlink(victim, bck, fwd);
2292 	    link_last_remainder(remainder);
2293 	    set_head(remainder, remainder_size | PREV_INUSE);
2294 	    set_foot(remainder, remainder_size);
2295 	    check_malloced_chunk(victim, nb);
2296 	    return chunk2mem(victim);
2297 	  }
2298 
2299 	  else if (remainder_size >= 0)  /* take */
2300 	  {
2301 	    set_inuse_bit_at_offset(victim, victim_size);
2302 	    unlink(victim, bck, fwd);
2303 	    check_malloced_chunk(victim, nb);
2304 	    return chunk2mem(victim);
2305 	  }
2306 
2307 	}
2308 
2309        bin = next_bin(bin);
2310 
2311       } while ((++idx & (BINBLOCKWIDTH - 1)) != 0);
2312 
2313       /* Clear out the block bit. */
2314 
2315       do   /* Possibly backtrack to try to clear a partial block */
2316       {
2317 	if ((startidx & (BINBLOCKWIDTH - 1)) == 0)
2318 	{
2319 	  av_[1] = (mbinptr)(binblocks_r & ~block);
2320 	  break;
2321 	}
2322 	--startidx;
2323        q = prev_bin(q);
2324       } while (first(q) == q);
2325 
2326       /* Get to the next possibly nonempty block */
2327 
2328       if ( (block <<= 1) <= binblocks_r && (block != 0) )
2329       {
2330 	while ((block & binblocks_r) == 0)
2331 	{
2332 	  idx += BINBLOCKWIDTH;
2333 	  block <<= 1;
2334 	}
2335       }
2336       else
2337 	break;
2338     }
2339   }
2340 
2341 
2342   /* Try to use top chunk */
2343 
2344   /* Require that there be a remainder, ensuring top always exists  */
2345   if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2346   {
2347 
2348 #if HAVE_MMAP
2349     /* If big and would otherwise need to extend, try to use mmap instead */
2350     if ((unsigned long)nb >= (unsigned long)mmap_threshold &&
2351 	(victim = mmap_chunk(nb)) != 0)
2352       return chunk2mem(victim);
2353 #endif
2354 
2355     /* Try to extend */
2356     malloc_extend_top(nb);
2357     if ( (remainder_size = chunksize(top) - nb) < (long)MINSIZE)
2358       return 0; /* propagate failure */
2359   }
2360 
2361   victim = top;
2362   set_head(victim, nb | PREV_INUSE);
2363   top = chunk_at_offset(victim, nb);
2364   set_head(top, remainder_size | PREV_INUSE);
2365   check_malloced_chunk(victim, nb);
2366   return chunk2mem(victim);
2367 
2368 }
2369 
2370 
2371 
2372 
2373 /*
2374 
2375   free() algorithm :
2376 
2377     cases:
2378 
2379        1. free(0) has no effect.
2380 
2381        2. If the chunk was allocated via mmap, it is release via munmap().
2382 
2383        3. If a returned chunk borders the current high end of memory,
2384 	  it is consolidated into the top, and if the total unused
2385 	  topmost memory exceeds the trim threshold, malloc_trim is
2386 	  called.
2387 
2388        4. Other chunks are consolidated as they arrive, and
2389 	  placed in corresponding bins. (This includes the case of
2390 	  consolidating with the current `last_remainder').
2391 
2392 */
2393 
2394 
2395 #if __STD_C
2396 void fREe(Void_t* mem)
2397 #else
2398 void fREe(mem) Void_t* mem;
2399 #endif
2400 {
2401   mchunkptr p;         /* chunk corresponding to mem */
2402   INTERNAL_SIZE_T hd;  /* its head field */
2403   INTERNAL_SIZE_T sz;  /* its size */
2404   int       idx;       /* its bin index */
2405   mchunkptr next;      /* next contiguous chunk */
2406   INTERNAL_SIZE_T nextsz; /* its size */
2407   INTERNAL_SIZE_T prevsz; /* size of previous contiguous chunk */
2408   mchunkptr bck;       /* misc temp for linking */
2409   mchunkptr fwd;       /* misc temp for linking */
2410   int       islr;      /* track whether merging with last_remainder */
2411 
2412   if (mem == 0)                              /* free(0) has no effect */
2413     return;
2414 
2415   p = mem2chunk(mem);
2416   hd = p->size;
2417 
2418 #if HAVE_MMAP
2419   if (hd & IS_MMAPPED)                       /* release mmapped memory. */
2420   {
2421     munmap_chunk(p);
2422     return;
2423   }
2424 #endif
2425 
2426   check_inuse_chunk(p);
2427 
2428   sz = hd & ~PREV_INUSE;
2429   next = chunk_at_offset(p, sz);
2430   nextsz = chunksize(next);
2431 
2432   if (next == top)                            /* merge with top */
2433   {
2434     sz += nextsz;
2435 
2436     if (!(hd & PREV_INUSE))                    /* consolidate backward */
2437     {
2438       prevsz = p->prev_size;
2439       p = chunk_at_offset(p, -((long) prevsz));
2440       sz += prevsz;
2441       unlink(p, bck, fwd);
2442     }
2443 
2444     set_head(p, sz | PREV_INUSE);
2445     top = p;
2446     if ((unsigned long)(sz) >= (unsigned long)trim_threshold)
2447       malloc_trim(top_pad);
2448     return;
2449   }
2450 
2451   set_head(next, nextsz);                    /* clear inuse bit */
2452 
2453   islr = 0;
2454 
2455   if (!(hd & PREV_INUSE))                    /* consolidate backward */
2456   {
2457     prevsz = p->prev_size;
2458     p = chunk_at_offset(p, -((long) prevsz));
2459     sz += prevsz;
2460 
2461     if (p->fd == last_remainder)             /* keep as last_remainder */
2462       islr = 1;
2463     else
2464       unlink(p, bck, fwd);
2465   }
2466 
2467   if (!(inuse_bit_at_offset(next, nextsz)))   /* consolidate forward */
2468   {
2469     sz += nextsz;
2470 
2471     if (!islr && next->fd == last_remainder)  /* re-insert last_remainder */
2472     {
2473       islr = 1;
2474       link_last_remainder(p);
2475     }
2476     else
2477       unlink(next, bck, fwd);
2478   }
2479 
2480 
2481   set_head(p, sz | PREV_INUSE);
2482   set_foot(p, sz);
2483   if (!islr)
2484     frontlink(p, sz, idx, bck, fwd);
2485 }
2486 
2487 
2488 
2489 
2490 
2491 /*
2492 
2493   Realloc algorithm:
2494 
2495     Chunks that were obtained via mmap cannot be extended or shrunk
2496     unless HAVE_MREMAP is defined, in which case mremap is used.
2497     Otherwise, if their reallocation is for additional space, they are
2498     copied.  If for less, they are just left alone.
2499 
2500     Otherwise, if the reallocation is for additional space, and the
2501     chunk can be extended, it is, else a malloc-copy-free sequence is
2502     taken.  There are several different ways that a chunk could be
2503     extended. All are tried:
2504 
2505        * Extending forward into following adjacent free chunk.
2506        * Shifting backwards, joining preceding adjacent space
2507        * Both shifting backwards and extending forward.
2508        * Extending into newly sbrked space
2509 
2510     Unless the #define REALLOC_ZERO_BYTES_FREES is set, realloc with a
2511     size argument of zero (re)allocates a minimum-sized chunk.
2512 
2513     If the reallocation is for less space, and the new request is for
2514     a `small' (<512 bytes) size, then the newly unused space is lopped
2515     off and freed.
2516 
2517     The old unix realloc convention of allowing the last-free'd chunk
2518     to be used as an argument to realloc is no longer supported.
2519     I don't know of any programs still relying on this feature,
2520     and allowing it would also allow too many other incorrect
2521     usages of realloc to be sensible.
2522 
2523 
2524 */
2525 
2526 
2527 #if __STD_C
2528 Void_t* rEALLOc(Void_t* oldmem, size_t bytes)
2529 #else
2530 Void_t* rEALLOc(oldmem, bytes) Void_t* oldmem; size_t bytes;
2531 #endif
2532 {
2533   INTERNAL_SIZE_T    nb;      /* padded request size */
2534 
2535   mchunkptr oldp;             /* chunk corresponding to oldmem */
2536   INTERNAL_SIZE_T    oldsize; /* its size */
2537 
2538   mchunkptr newp;             /* chunk to return */
2539   INTERNAL_SIZE_T    newsize; /* its size */
2540   Void_t*   newmem;           /* corresponding user mem */
2541 
2542   mchunkptr next;             /* next contiguous chunk after oldp */
2543   INTERNAL_SIZE_T  nextsize;  /* its size */
2544 
2545   mchunkptr prev;             /* previous contiguous chunk before oldp */
2546   INTERNAL_SIZE_T  prevsize;  /* its size */
2547 
2548   mchunkptr remainder;        /* holds split off extra space from newp */
2549   INTERNAL_SIZE_T  remainder_size;   /* its size */
2550 
2551   mchunkptr bck;              /* misc temp for linking */
2552   mchunkptr fwd;              /* misc temp for linking */
2553 
2554 #ifdef REALLOC_ZERO_BYTES_FREES
2555   if (bytes == 0) { fREe(oldmem); return 0; }
2556 #endif
2557 
2558   if ((long)bytes < 0) return 0;
2559 
2560   /* realloc of null is supposed to be same as malloc */
2561   if (oldmem == 0) return mALLOc(bytes);
2562 
2563   newp    = oldp    = mem2chunk(oldmem);
2564   newsize = oldsize = chunksize(oldp);
2565 
2566 
2567   nb = request2size(bytes);
2568 
2569 #if HAVE_MMAP
2570   if (chunk_is_mmapped(oldp))
2571   {
2572 #if HAVE_MREMAP
2573     newp = mremap_chunk(oldp, nb);
2574     if(newp) return chunk2mem(newp);
2575 #endif
2576     /* Note the extra SIZE_SZ overhead. */
2577     if(oldsize - SIZE_SZ >= nb) return oldmem; /* do nothing */
2578     /* Must alloc, copy, free. */
2579     newmem = mALLOc(bytes);
2580     if (newmem == 0) return 0; /* propagate failure */
2581     MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ);
2582     munmap_chunk(oldp);
2583     return newmem;
2584   }
2585 #endif
2586 
2587   check_inuse_chunk(oldp);
2588 
2589   if ((long)(oldsize) < (long)(nb))
2590   {
2591 
2592     /* Try expanding forward */
2593 
2594     next = chunk_at_offset(oldp, oldsize);
2595     if (next == top || !inuse(next))
2596     {
2597       nextsize = chunksize(next);
2598 
2599       /* Forward into top only if a remainder */
2600       if (next == top)
2601       {
2602 	if ((long)(nextsize + newsize) >= (long)(nb + MINSIZE))
2603 	{
2604 	  newsize += nextsize;
2605 	  top = chunk_at_offset(oldp, nb);
2606 	  set_head(top, (newsize - nb) | PREV_INUSE);
2607 	  set_head_size(oldp, nb);
2608 	  return chunk2mem(oldp);
2609 	}
2610       }
2611 
2612       /* Forward into next chunk */
2613       else if (((long)(nextsize + newsize) >= (long)(nb)))
2614       {
2615 	unlink(next, bck, fwd);
2616 	newsize  += nextsize;
2617 	goto split;
2618       }
2619     }
2620     else
2621     {
2622       next = 0;
2623       nextsize = 0;
2624     }
2625 
2626     /* Try shifting backwards. */
2627 
2628     if (!prev_inuse(oldp))
2629     {
2630       prev = prev_chunk(oldp);
2631       prevsize = chunksize(prev);
2632 
2633       /* try forward + backward first to save a later consolidation */
2634 
2635       if (next != 0)
2636       {
2637 	/* into top */
2638 	if (next == top)
2639 	{
2640 	  if ((long)(nextsize + prevsize + newsize) >= (long)(nb + MINSIZE))
2641 	  {
2642 	    unlink(prev, bck, fwd);
2643 	    newp = prev;
2644 	    newsize += prevsize + nextsize;
2645 	    newmem = chunk2mem(newp);
2646 	    MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2647 	    top = chunk_at_offset(newp, nb);
2648 	    set_head(top, (newsize - nb) | PREV_INUSE);
2649 	    set_head_size(newp, nb);
2650 	    return newmem;
2651 	  }
2652 	}
2653 
2654 	/* into next chunk */
2655 	else if (((long)(nextsize + prevsize + newsize) >= (long)(nb)))
2656 	{
2657 	  unlink(next, bck, fwd);
2658 	  unlink(prev, bck, fwd);
2659 	  newp = prev;
2660 	  newsize += nextsize + prevsize;
2661 	  newmem = chunk2mem(newp);
2662 	  MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2663 	  goto split;
2664 	}
2665       }
2666 
2667       /* backward only */
2668       if (prev != 0 && (long)(prevsize + newsize) >= (long)nb)
2669       {
2670 	unlink(prev, bck, fwd);
2671 	newp = prev;
2672 	newsize += prevsize;
2673 	newmem = chunk2mem(newp);
2674 	MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2675 	goto split;
2676       }
2677     }
2678 
2679     /* Must allocate */
2680 
2681     newmem = mALLOc (bytes);
2682 
2683     if (newmem == 0)  /* propagate failure */
2684       return 0;
2685 
2686     /* Avoid copy if newp is next chunk after oldp. */
2687     /* (This can only happen when new chunk is sbrk'ed.) */
2688 
2689     if ( (newp = mem2chunk(newmem)) == next_chunk(oldp))
2690     {
2691       newsize += chunksize(newp);
2692       newp = oldp;
2693       goto split;
2694     }
2695 
2696     /* Otherwise copy, free, and exit */
2697     MALLOC_COPY(newmem, oldmem, oldsize - SIZE_SZ);
2698     fREe(oldmem);
2699     return newmem;
2700   }
2701 
2702 
2703  split:  /* split off extra room in old or expanded chunk */
2704 
2705   if (newsize - nb >= MINSIZE) /* split off remainder */
2706   {
2707     remainder = chunk_at_offset(newp, nb);
2708     remainder_size = newsize - nb;
2709     set_head_size(newp, nb);
2710     set_head(remainder, remainder_size | PREV_INUSE);
2711     set_inuse_bit_at_offset(remainder, remainder_size);
2712     fREe(chunk2mem(remainder)); /* let free() deal with it */
2713   }
2714   else
2715   {
2716     set_head_size(newp, newsize);
2717     set_inuse_bit_at_offset(newp, newsize);
2718   }
2719 
2720   check_inuse_chunk(newp);
2721   return chunk2mem(newp);
2722 }
2723 
2724 
2725 
2726 
2727 /*
2728 
2729   memalign algorithm:
2730 
2731     memalign requests more than enough space from malloc, finds a spot
2732     within that chunk that meets the alignment request, and then
2733     possibly frees the leading and trailing space.
2734 
2735     The alignment argument must be a power of two. This property is not
2736     checked by memalign, so misuse may result in random runtime errors.
2737 
2738     8-byte alignment is guaranteed by normal malloc calls, so don't
2739     bother calling memalign with an argument of 8 or less.
2740 
2741     Overreliance on memalign is a sure way to fragment space.
2742 
2743 */
2744 
2745 
2746 #if __STD_C
2747 Void_t* mEMALIGn(size_t alignment, size_t bytes)
2748 #else
2749 Void_t* mEMALIGn(alignment, bytes) size_t alignment; size_t bytes;
2750 #endif
2751 {
2752   INTERNAL_SIZE_T    nb;      /* padded  request size */
2753   char*     m;                /* memory returned by malloc call */
2754   mchunkptr p;                /* corresponding chunk */
2755   char*     brk;              /* alignment point within p */
2756   mchunkptr newp;             /* chunk to return */
2757   INTERNAL_SIZE_T  newsize;   /* its size */
2758   INTERNAL_SIZE_T  leadsize;  /* leading space befor alignment point */
2759   mchunkptr remainder;        /* spare room at end to split off */
2760   long      remainder_size;   /* its size */
2761 
2762   if ((long)bytes < 0) return 0;
2763 
2764   /* If need less alignment than we give anyway, just relay to malloc */
2765 
2766   if (alignment <= MALLOC_ALIGNMENT) return mALLOc(bytes);
2767 
2768   /* Otherwise, ensure that it is at least a minimum chunk size */
2769 
2770   if (alignment <  MINSIZE) alignment = MINSIZE;
2771 
2772   /* Call malloc with worst case padding to hit alignment. */
2773 
2774   nb = request2size(bytes);
2775   m  = (char*)(mALLOc(nb + alignment + MINSIZE));
2776 
2777   if (m == 0) return 0; /* propagate failure */
2778 
2779   p = mem2chunk(m);
2780 
2781   if ((((unsigned long)(m)) % alignment) == 0) /* aligned */
2782   {
2783 #if HAVE_MMAP
2784     if(chunk_is_mmapped(p))
2785       return chunk2mem(p); /* nothing more to do */
2786 #endif
2787   }
2788   else /* misaligned */
2789   {
2790     /*
2791       Find an aligned spot inside chunk.
2792       Since we need to give back leading space in a chunk of at
2793       least MINSIZE, if the first calculation places us at
2794       a spot with less than MINSIZE leader, we can move to the
2795       next aligned spot -- we've allocated enough total room so that
2796       this is always possible.
2797     */
2798 
2799     brk = (char*)mem2chunk(((unsigned long)(m + alignment - 1)) & -((signed) alignment));
2800     if ((long)(brk - (char*)(p)) < MINSIZE) brk = brk + alignment;
2801 
2802     newp = (mchunkptr)brk;
2803     leadsize = brk - (char*)(p);
2804     newsize = chunksize(p) - leadsize;
2805 
2806 #if HAVE_MMAP
2807     if(chunk_is_mmapped(p))
2808     {
2809       newp->prev_size = p->prev_size + leadsize;
2810       set_head(newp, newsize|IS_MMAPPED);
2811       return chunk2mem(newp);
2812     }
2813 #endif
2814 
2815     /* give back leader, use the rest */
2816 
2817     set_head(newp, newsize | PREV_INUSE);
2818     set_inuse_bit_at_offset(newp, newsize);
2819     set_head_size(p, leadsize);
2820     fREe(chunk2mem(p));
2821     p = newp;
2822 
2823     assert (newsize >= nb && (((unsigned long)(chunk2mem(p))) % alignment) == 0);
2824   }
2825 
2826   /* Also give back spare room at the end */
2827 
2828   remainder_size = chunksize(p) - nb;
2829 
2830   if (remainder_size >= (long)MINSIZE)
2831   {
2832     remainder = chunk_at_offset(p, nb);
2833     set_head(remainder, remainder_size | PREV_INUSE);
2834     set_head_size(p, nb);
2835     fREe(chunk2mem(remainder));
2836   }
2837 
2838   check_inuse_chunk(p);
2839   return chunk2mem(p);
2840 
2841 }
2842 
2843 
2844 
2845 
2846 /*
2847     valloc just invokes memalign with alignment argument equal
2848     to the page size of the system (or as near to this as can
2849     be figured out from all the includes/defines above.)
2850 */
2851 
2852 #if __STD_C
2853 Void_t* vALLOc(size_t bytes)
2854 #else
2855 Void_t* vALLOc(bytes) size_t bytes;
2856 #endif
2857 {
2858   return mEMALIGn (malloc_getpagesize, bytes);
2859 }
2860 
2861 /*
2862   pvalloc just invokes valloc for the nearest pagesize
2863   that will accommodate request
2864 */
2865 
2866 
2867 #if __STD_C
2868 Void_t* pvALLOc(size_t bytes)
2869 #else
2870 Void_t* pvALLOc(bytes) size_t bytes;
2871 #endif
2872 {
2873   size_t pagesize = malloc_getpagesize;
2874   return mEMALIGn (pagesize, (bytes + pagesize - 1) & ~(pagesize - 1));
2875 }
2876 
2877 /*
2878 
2879   calloc calls malloc, then zeroes out the allocated chunk.
2880 
2881 */
2882 
2883 #if __STD_C
2884 Void_t* cALLOc(size_t n, size_t elem_size)
2885 #else
2886 Void_t* cALLOc(n, elem_size) size_t n; size_t elem_size;
2887 #endif
2888 {
2889   mchunkptr p;
2890   INTERNAL_SIZE_T csz;
2891 
2892   INTERNAL_SIZE_T sz = n * elem_size;
2893 
2894 
2895   /* check if expand_top called, in which case don't need to clear */
2896 #if MORECORE_CLEARS
2897   mchunkptr oldtop = top;
2898   INTERNAL_SIZE_T oldtopsize = chunksize(top);
2899 #endif
2900   Void_t* mem = mALLOc (sz);
2901 
2902   if ((long)n < 0) return 0;
2903 
2904   if (mem == 0)
2905     return 0;
2906   else
2907   {
2908     p = mem2chunk(mem);
2909 
2910     /* Two optional cases in which clearing not necessary */
2911 
2912 
2913 #if HAVE_MMAP
2914     if (chunk_is_mmapped(p)) return mem;
2915 #endif
2916 
2917     csz = chunksize(p);
2918 
2919 #if MORECORE_CLEARS
2920     if (p == oldtop && csz > oldtopsize)
2921     {
2922       /* clear only the bytes from non-freshly-sbrked memory */
2923       csz = oldtopsize;
2924     }
2925 #endif
2926 
2927     MALLOC_ZERO(mem, csz - SIZE_SZ);
2928     return mem;
2929   }
2930 }
2931 
2932 /*
2933 
2934   cfree just calls free. It is needed/defined on some systems
2935   that pair it with calloc, presumably for odd historical reasons.
2936 
2937 */
2938 
2939 #if !defined(INTERNAL_LINUX_C_LIB) || !defined(__ELF__)
2940 #if __STD_C
2941 void cfree(Void_t *mem)
2942 #else
2943 void cfree(mem) Void_t *mem;
2944 #endif
2945 {
2946   fREe(mem);
2947 }
2948 #endif
2949 
2950 
2951 
2952 /*
2953 
2954     Malloc_trim gives memory back to the system (via negative
2955     arguments to sbrk) if there is unused memory at the `high' end of
2956     the malloc pool. You can call this after freeing large blocks of
2957     memory to potentially reduce the system-level memory requirements
2958     of a program. However, it cannot guarantee to reduce memory. Under
2959     some allocation patterns, some large free blocks of memory will be
2960     locked between two used chunks, so they cannot be given back to
2961     the system.
2962 
2963     The `pad' argument to malloc_trim represents the amount of free
2964     trailing space to leave untrimmed. If this argument is zero,
2965     only the minimum amount of memory to maintain internal data
2966     structures will be left (one page or less). Non-zero arguments
2967     can be supplied to maintain enough trailing space to service
2968     future expected allocations without having to re-obtain memory
2969     from the system.
2970 
2971     Malloc_trim returns 1 if it actually released any memory, else 0.
2972 
2973 */
2974 
2975 #if __STD_C
2976 int malloc_trim(size_t pad)
2977 #else
2978 int malloc_trim(pad) size_t pad;
2979 #endif
2980 {
2981   long  top_size;        /* Amount of top-most memory */
2982   long  extra;           /* Amount to release */
2983   char* current_brk;     /* address returned by pre-check sbrk call */
2984   char* new_brk;         /* address returned by negative sbrk call */
2985 
2986   unsigned long pagesz = malloc_getpagesize;
2987 
2988   top_size = chunksize(top);
2989   extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
2990 
2991   if (extra < (long)pagesz)  /* Not enough memory to release */
2992     return 0;
2993 
2994   else
2995   {
2996     /* Test to make sure no one else called sbrk */
2997     current_brk = (char*)(MORECORE (0));
2998     if (current_brk != (char*)(top) + top_size)
2999       return 0;     /* Apparently we don't own memory; must fail */
3000 
3001     else
3002     {
3003       new_brk = (char*)(MORECORE (-extra));
3004 
3005       if (new_brk == (char*)(MORECORE_FAILURE)) /* sbrk failed? */
3006       {
3007 	/* Try to figure out what we have */
3008 	current_brk = (char*)(MORECORE (0));
3009 	top_size = current_brk - (char*)top;
3010 	if (top_size >= (long)MINSIZE) /* if not, we are very very dead! */
3011 	{
3012 	  sbrked_mem = current_brk - sbrk_base;
3013 	  set_head(top, top_size | PREV_INUSE);
3014 	}
3015 	check_chunk(top);
3016 	return 0;
3017       }
3018 
3019       else
3020       {
3021 	/* Success. Adjust top accordingly. */
3022 	set_head(top, (top_size - extra) | PREV_INUSE);
3023 	sbrked_mem -= extra;
3024 	check_chunk(top);
3025 	return 1;
3026       }
3027     }
3028   }
3029 }
3030 
3031 
3032 
3033 /*
3034   malloc_usable_size:
3035 
3036     This routine tells you how many bytes you can actually use in an
3037     allocated chunk, which may be more than you requested (although
3038     often not). You can use this many bytes without worrying about
3039     overwriting other allocated objects. Not a particularly great
3040     programming practice, but still sometimes useful.
3041 
3042 */
3043 
3044 #if __STD_C
3045 size_t malloc_usable_size(Void_t* mem)
3046 #else
3047 size_t malloc_usable_size(mem) Void_t* mem;
3048 #endif
3049 {
3050   mchunkptr p;
3051   if (mem == 0)
3052     return 0;
3053   else
3054   {
3055     p = mem2chunk(mem);
3056     if(!chunk_is_mmapped(p))
3057     {
3058       if (!inuse(p)) return 0;
3059       check_inuse_chunk(p);
3060       return chunksize(p) - SIZE_SZ;
3061     }
3062     return chunksize(p) - 2*SIZE_SZ;
3063   }
3064 }
3065 
3066 
3067 
3068 
3069 /* Utility to update current_mallinfo for malloc_stats and mallinfo() */
3070 
3071 #if 0
3072 static void malloc_update_mallinfo()
3073 {
3074   int i;
3075   mbinptr b;
3076   mchunkptr p;
3077 #ifdef DEBUG
3078   mchunkptr q;
3079 #endif
3080 
3081   INTERNAL_SIZE_T avail = chunksize(top);
3082   int   navail = ((long)(avail) >= (long)MINSIZE)? 1 : 0;
3083 
3084   for (i = 1; i < NAV; ++i)
3085   {
3086     b = bin_at(i);
3087     for (p = last(b); p != b; p = p->bk)
3088     {
3089 #ifdef DEBUG
3090       check_free_chunk(p);
3091       for (q = next_chunk(p);
3092 	   q < top && inuse(q) && (long)(chunksize(q)) >= (long)MINSIZE;
3093 	   q = next_chunk(q))
3094 	check_inuse_chunk(q);
3095 #endif
3096       avail += chunksize(p);
3097       navail++;
3098     }
3099   }
3100 
3101   current_mallinfo.ordblks = navail;
3102   current_mallinfo.uordblks = sbrked_mem - avail;
3103   current_mallinfo.fordblks = avail;
3104   current_mallinfo.hblks = n_mmaps;
3105   current_mallinfo.hblkhd = mmapped_mem;
3106   current_mallinfo.keepcost = chunksize(top);
3107 
3108 }
3109 #endif	/* 0 */
3110 
3111 
3112 
3113 /*
3114 
3115   malloc_stats:
3116 
3117     Prints on the amount of space obtain from the system (both
3118     via sbrk and mmap), the maximum amount (which may be more than
3119     current if malloc_trim and/or munmap got called), the maximum
3120     number of simultaneous mmap regions used, and the current number
3121     of bytes allocated via malloc (or realloc, etc) but not yet
3122     freed. (Note that this is the number of bytes allocated, not the
3123     number requested. It will be larger than the number requested
3124     because of alignment and bookkeeping overhead.)
3125 
3126 */
3127 
3128 #if 0
3129 void malloc_stats()
3130 {
3131   malloc_update_mallinfo();
3132   printf("max system bytes = %10u\n",
3133 	  (unsigned int)(max_total_mem));
3134   printf("system bytes     = %10u\n",
3135 	  (unsigned int)(sbrked_mem + mmapped_mem));
3136   printf("in use bytes     = %10u\n",
3137 	  (unsigned int)(current_mallinfo.uordblks + mmapped_mem));
3138 #if HAVE_MMAP
3139   printf("max mmap regions = %10u\n",
3140 	  (unsigned int)max_n_mmaps);
3141 #endif
3142 }
3143 #endif	/* 0 */
3144 
3145 /*
3146   mallinfo returns a copy of updated current mallinfo.
3147 */
3148 
3149 #if 0
3150 struct mallinfo mALLINFo()
3151 {
3152   malloc_update_mallinfo();
3153   return current_mallinfo;
3154 }
3155 #endif	/* 0 */
3156 
3157 
3158 
3159 
3160 /*
3161   mallopt:
3162 
3163     mallopt is the general SVID/XPG interface to tunable parameters.
3164     The format is to provide a (parameter-number, parameter-value) pair.
3165     mallopt then sets the corresponding parameter to the argument
3166     value if it can (i.e., so long as the value is meaningful),
3167     and returns 1 if successful else 0.
3168 
3169     See descriptions of tunable parameters above.
3170 
3171 */
3172 
3173 #if __STD_C
3174 int mALLOPt(int param_number, int value)
3175 #else
3176 int mALLOPt(param_number, value) int param_number; int value;
3177 #endif
3178 {
3179   switch(param_number)
3180   {
3181     case M_TRIM_THRESHOLD:
3182       trim_threshold = value; return 1;
3183     case M_TOP_PAD:
3184       top_pad = value; return 1;
3185     case M_MMAP_THRESHOLD:
3186       mmap_threshold = value; return 1;
3187     case M_MMAP_MAX:
3188 #if HAVE_MMAP
3189       n_mmaps_max = value; return 1;
3190 #else
3191       if (value != 0) return 0; else  n_mmaps_max = value; return 1;
3192 #endif
3193 
3194     default:
3195       return 0;
3196   }
3197 }
3198 
3199 /*
3200 
3201 History:
3202 
3203     V2.6.6 Sun Dec  5 07:42:19 1999  Doug Lea  (dl at gee)
3204       * return null for negative arguments
3205       * Added Several WIN32 cleanups from Martin C. Fong <mcfong@yahoo.com>
3206 	 * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'
3207 	  (e.g. WIN32 platforms)
3208 	 * Cleanup up header file inclusion for WIN32 platforms
3209 	 * Cleanup code to avoid Microsoft Visual C++ compiler complaints
3210 	 * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing
3211 	   memory allocation routines
3212 	 * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)
3213 	 * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to
3214 	   usage of 'assert' in non-WIN32 code
3215 	 * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to
3216 	   avoid infinite loop
3217       * Always call 'fREe()' rather than 'free()'
3218 
3219     V2.6.5 Wed Jun 17 15:57:31 1998  Doug Lea  (dl at gee)
3220       * Fixed ordering problem with boundary-stamping
3221 
3222     V2.6.3 Sun May 19 08:17:58 1996  Doug Lea  (dl at gee)
3223       * Added pvalloc, as recommended by H.J. Liu
3224       * Added 64bit pointer support mainly from Wolfram Gloger
3225       * Added anonymously donated WIN32 sbrk emulation
3226       * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
3227       * malloc_extend_top: fix mask error that caused wastage after
3228 	foreign sbrks
3229       * Add linux mremap support code from HJ Liu
3230 
3231     V2.6.2 Tue Dec  5 06:52:55 1995  Doug Lea  (dl at gee)
3232       * Integrated most documentation with the code.
3233       * Add support for mmap, with help from
3234 	Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3235       * Use last_remainder in more cases.
3236       * Pack bins using idea from  colin@nyx10.cs.du.edu
3237       * Use ordered bins instead of best-fit threshhold
3238       * Eliminate block-local decls to simplify tracing and debugging.
3239       * Support another case of realloc via move into top
3240       * Fix error occuring when initial sbrk_base not word-aligned.
3241       * Rely on page size for units instead of SBRK_UNIT to
3242 	avoid surprises about sbrk alignment conventions.
3243       * Add mallinfo, mallopt. Thanks to Raymond Nijssen
3244 	(raymond@es.ele.tue.nl) for the suggestion.
3245       * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
3246       * More precautions for cases where other routines call sbrk,
3247 	courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
3248       * Added macros etc., allowing use in linux libc from
3249 	H.J. Lu (hjl@gnu.ai.mit.edu)
3250       * Inverted this history list
3251 
3252     V2.6.1 Sat Dec  2 14:10:57 1995  Doug Lea  (dl at gee)
3253       * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
3254       * Removed all preallocation code since under current scheme
3255 	the work required to undo bad preallocations exceeds
3256 	the work saved in good cases for most test programs.
3257       * No longer use return list or unconsolidated bins since
3258 	no scheme using them consistently outperforms those that don't
3259 	given above changes.
3260       * Use best fit for very large chunks to prevent some worst-cases.
3261       * Added some support for debugging
3262 
3263     V2.6.0 Sat Nov  4 07:05:23 1995  Doug Lea  (dl at gee)
3264       * Removed footers when chunks are in use. Thanks to
3265 	Paul Wilson (wilson@cs.texas.edu) for the suggestion.
3266 
3267     V2.5.4 Wed Nov  1 07:54:51 1995  Doug Lea  (dl at gee)
3268       * Added malloc_trim, with help from Wolfram Gloger
3269 	(wmglo@Dent.MED.Uni-Muenchen.DE).
3270 
3271     V2.5.3 Tue Apr 26 10:16:01 1994  Doug Lea  (dl at g)
3272 
3273     V2.5.2 Tue Apr  5 16:20:40 1994  Doug Lea  (dl at g)
3274       * realloc: try to expand in both directions
3275       * malloc: swap order of clean-bin strategy;
3276       * realloc: only conditionally expand backwards
3277       * Try not to scavenge used bins
3278       * Use bin counts as a guide to preallocation
3279       * Occasionally bin return list chunks in first scan
3280       * Add a few optimizations from colin@nyx10.cs.du.edu
3281 
3282     V2.5.1 Sat Aug 14 15:40:43 1993  Doug Lea  (dl at g)
3283       * faster bin computation & slightly different binning
3284       * merged all consolidations to one part of malloc proper
3285 	 (eliminating old malloc_find_space & malloc_clean_bin)
3286       * Scan 2 returns chunks (not just 1)
3287       * Propagate failure in realloc if malloc returns 0
3288       * Add stuff to allow compilation on non-ANSI compilers
3289 	  from kpv@research.att.com
3290 
3291     V2.5 Sat Aug  7 07:41:59 1993  Doug Lea  (dl at g.oswego.edu)
3292       * removed potential for odd address access in prev_chunk
3293       * removed dependency on getpagesize.h
3294       * misc cosmetics and a bit more internal documentation
3295       * anticosmetics: mangled names in macros to evade debugger strangeness
3296       * tested on sparc, hp-700, dec-mips, rs6000
3297 	  with gcc & native cc (hp, dec only) allowing
3298 	  Detlefs & Zorn comparison study (in SIGPLAN Notices.)
3299 
3300     Trial version Fri Aug 28 13:14:29 1992  Doug Lea  (dl at g.oswego.edu)
3301       * Based loosely on libg++-1.2X malloc. (It retains some of the overall
3302 	 structure of old version,  but most details differ.)
3303 
3304 */
3305