WebSVN – HelenOS-historic – Diff – /uspace/trunk/libc/malloc/malloc.c


/*
  This is a version (aka dlmalloc) of malloc/free/realloc written by
  Doug Lea and released to the public domain, as explained at
  http://creativecommons.org/licenses/publicdomain.  Send questions,
  comments, complaints, performance data, etc to dl@cs.oswego.edu
 
* Version 2.8.3 Thu Sep 22 11:16:15 2005  Doug Lea  (dl at gee)
 
   Note: There may be an updated version of this malloc obtainable at
           ftp://gee.cs.oswego.edu/pub/misc/malloc.c
         Check before installing!
 
* Quickstart
 
  This library is all in one file to simplify the most common usage:
  ftp it, compile it (-O3), and link it into another program. All of
  the compile-time options default to reasonable values for use on
  most platforms.  You might later want to step through various
  compile-time and dynamic tuning options.
 
  For convenience, an include file for code using this malloc is at:
     ftp://gee.cs.oswego.edu/pub/misc/malloc-2.8.3.h
  You don't really need this .h file unless you call functions not
  defined in your system include files.  The .h file contains only the
  excerpts from this file needed for using this malloc on ANSI C/C++
  systems, so long as you haven't changed compile-time options about
  naming and tuning parameters.  If you do, then you can create your
  own malloc.h that does include all settings by cutting at the point
  indicated below. Note that you may already by default be using a C
  library containing a malloc that is based on some version of this
  malloc (for example in linux). You might still want to use the one
  in this file to customize settings or to avoid overheads associated
  with library versions.
 
* Vital statistics:
 
  Supported pointer/size_t representation:       4 or 8 bytes
       size_t MUST be an unsigned type of the same width as
       pointers. (If you are using an ancient system that declares
       size_t as a signed type, or need it to be a different width
       than pointers, you can use a previous release of this malloc
       (e.g. 2.7.2) supporting these.)
 
  Alignment:                                     8 bytes (default)
       This suffices for nearly all current machines and C compilers.
       However, you can define MALLOC_ALIGNMENT to be wider than this
       if necessary (up to 128bytes), at the expense of using more space.
 
  Minimum overhead per allocated chunk:   4 or  8 bytes (if 4byte sizes)
                                          8 or 16 bytes (if 8byte sizes)
       Each malloced chunk has a hidden word of overhead holding size
       and status information, and additional cross-check word
       if FOOTERS is defined.
 
  Minimum allocated size: 4-byte ptrs:  16 bytes    (including overhead)
                          8-byte ptrs:  32 bytes    (including overhead)
 
       Even a request for zero bytes (i.e., malloc(0)) returns a
       pointer to something of the minimum allocatable size.
       The maximum overhead wastage (i.e., number of extra bytes
       allocated than were requested in malloc) is less than or equal
       to the minimum size, except for requests >= mmap_threshold that
       are serviced via mmap(), where the worst case wastage is about
       32 bytes plus the remainder from a system page (the minimal
       mmap unit); typically 4096 or 8192 bytes.
 
  Security: static-safe; optionally more or less
       The "security" of malloc refers to the ability of malicious
       code to accentuate the effects of errors (for example, freeing
       space that is not currently malloc'ed or overwriting past the
       ends of chunks) in code that calls malloc.  This malloc
       guarantees not to modify any memory locations below the base of
       heap, i.e., static variables, even in the presence of usage
       errors.  The routines additionally detect most improper frees
       and reallocs.  All this holds as long as the static bookkeeping
       for malloc itself is not corrupted by some other means.  This
       is only one aspect of security -- these checks do not, and
       cannot, detect all possible programming errors.
 
       If FOOTERS is defined nonzero, then each allocated chunk
       carries an additional check word to verify that it was malloced
       from its space.  These check words are the same within each
       execution of a program using malloc, but differ across
       executions, so externally crafted fake chunks cannot be
       freed. This improves security by rejecting frees/reallocs that
       could corrupt heap memory, in addition to the checks preventing
       writes to statics that are always on.  This may further improve
       security at the expense of time and space overhead.  (Note that
       FOOTERS may also be worth using with MSPACES.)
 
       By default detected errors cause the program to abort (calling
       "abort()"). You can override this to instead proceed past
       errors by defining PROCEED_ON_ERROR.  In this case, a bad free
       has no effect, and a malloc that encounters a bad address
       caused by user overwrites will ignore the bad address by
       dropping pointers and indices to all known memory. This may
       be appropriate for programs that should continue if at all
       possible in the face of programming errors, although they may
       run out of memory because dropped memory is never reclaimed.
 
       If you don't like either of these options, you can define
       CORRUPTION_ERROR_ACTION and USAGE_ERROR_ACTION to do anything
       else. And if if you are sure that your program using malloc has
       no errors or vulnerabilities, you can define INSECURE to 1,
       which might (or might not) provide a small performance improvement.
 
  Thread-safety: NOT thread-safe unless USE_LOCKS defined
       When USE_LOCKS is defined, each public call to malloc, free,
       etc is surrounded with either a pthread mutex or a win32
       spinlock (depending on WIN32). This is not especially fast, and
       can be a major bottleneck.  It is designed only to provide
       minimal protection in concurrent environments, and to provide a
       basis for extensions.  If you are using malloc in a concurrent
       program, consider instead using ptmalloc, which is derived from
       a version of this malloc. (See http://www.malloc.de).
 
  System requirements: Any combination of MORECORE and/or MMAP/MUNMAP
       This malloc can use unix sbrk or any emulation (invoked using
       the CALL_MORECORE macro) and/or mmap/munmap or any emulation
       (invoked using CALL_MMAP/CALL_MUNMAP) to get and release system
       memory.  On most unix systems, it tends to work best if both
       MORECORE and MMAP are enabled.  On Win32, it uses emulations
       based on VirtualAlloc. It also uses common C library functions
       like memset.
 
  Compliance: I believe it is compliant with the Single Unix Specification
       (See http://www.unix.org). Also SVID/XPG, ANSI C, and probably
       others as well.
 
* Overview of algorithms
 
  This is not the fastest, most space-conserving, most portable, or
  most tunable malloc ever written. However it is among the fastest
  while also being among the most space-conserving, portable and
  tunable.  Consistent balance across these factors results in a good
  general-purpose allocator for malloc-intensive programs.
 
  In most ways, this malloc is a best-fit allocator. Generally, it
  chooses the best-fitting existing chunk for a request, with ties
  broken in approximately least-recently-used order. (This strategy
  normally maintains low fragmentation.) However, for requests less
  than 256bytes, it deviates from best-fit when there is not an
  exactly fitting available chunk by preferring to use space adjacent
  to that used for the previous small request, as well as by breaking
  ties in approximately most-recently-used order. (These enhance
  locality of series of small allocations.)  And for very large requests
  (>= 256Kb by default), it relies on system memory mapping
  facilities, if supported.  (This helps avoid carrying around and
  possibly fragmenting memory used only for large chunks.)
 
  All operations (except malloc_stats and mallinfo) have execution
  times that are bounded by a constant factor of the number of bits in
  a size_t, not counting any clearing in calloc or copying in realloc,
  or actions surrounding MORECORE and MMAP that have times
  proportional to the number of non-contiguous regions returned by
  system allocation routines, which is often just 1.
 
  The implementation is not very modular and seriously overuses
  macros. Perhaps someday all C compilers will do as good a job
  inlining modular code as can now be done by brute-force expansion,
  but now, enough of them seem not to.
 
  Some compilers issue a lot of warnings about code that is
  dead/unreachable only on some platforms, and also about intentional
  uses of negation on unsigned types. All known cases of each can be
  ignored.
 
  For a longer but out of date high-level description, see
     http://gee.cs.oswego.edu/dl/html/malloc.html
 
* MSPACES
  If MSPACES is defined, then in addition to malloc, free, etc.,
  this file also defines mspace_malloc, mspace_free, etc. These
  are versions of malloc routines that take an "mspace" argument
  obtained using create_mspace, to control all internal bookkeeping.
  If ONLY_MSPACES is defined, only these versions are compiled.
  So if you would like to use this allocator for only some allocations,
  and your system malloc for others, you can compile with
  ONLY_MSPACES and then do something like...
    static mspace mymspace = create_mspace(0,0); // for example
    #define mymalloc(bytes)  mspace_malloc(mymspace, bytes)
 
  (Note: If you only need one instance of an mspace, you can instead
  use "USE_DL_PREFIX" to relabel the global malloc.)
 
  You can similarly create thread-local allocators by storing
  mspaces as thread-locals. For example:
    static __thread mspace tlms = 0;
    void*  tlmalloc(size_t bytes) {
      if (tlms == 0) tlms = create_mspace(0, 0);
      return mspace_malloc(tlms, bytes);
    }
    void  tlfree(void* mem) { mspace_free(tlms, mem); }
 
  Unless FOOTERS is defined, each mspace is completely independent.
  You cannot allocate from one and free to another (although
  conformance is only weakly checked, so usage errors are not always
  caught). If FOOTERS is defined, then each chunk carries around a tag
  indicating its originating mspace, and frees are directed to their
  originating spaces.
 
 -------------------------  Compile-time options ---------------------------
 
Be careful in setting #define values for numerical constants of type
size_t. On some systems, literal values are not automatically extended
to size_t precision unless they are explicitly casted.
 
WIN32                    default: defined if _WIN32 defined
  Defining WIN32 sets up defaults for MS environment and compilers.
  Otherwise defaults are for unix.
 
MALLOC_ALIGNMENT         default: (size_t)8
  Controls the minimum alignment for malloc'ed chunks.  It must be a
  power of two and at least 8, even on machines for which smaller
  alignments would suffice. It may be defined as larger than this
  though. Note however that code and data structures are optimized for
  the case of 8-byte alignment.
 
MSPACES                  default: 0 (false)
  If true, compile in support for independent allocation spaces.
  This is only supported if HAVE_MMAP is true.
 
ONLY_MSPACES             default: 0 (false)
  If true, only compile in mspace versions, not regular versions.
 
USE_LOCKS                default: 0 (false)
  Causes each call to each public routine to be surrounded with
  pthread or WIN32 mutex lock/unlock. (If set true, this can be
  overridden on a per-mspace basis for mspace versions.)
 
FOOTERS                  default: 0
  If true, provide extra checking and dispatching by placing
  information in the footers of allocated chunks. This adds
  space and time overhead.
 
INSECURE                 default: 0
  If true, omit checks for usage errors and heap space overwrites.
 
USE_DL_PREFIX            default: NOT defined
  Causes compiler to prefix all public routines with the string 'dl'.
  This can be useful when you only want to use this malloc in one part
  of a program, using your regular system malloc elsewhere.
 
ABORT                    default: defined as abort()
  Defines how to abort on failed checks.  On most systems, a failed
  check cannot die with an "assert" or even print an informative
  message, because the underlying print routines in turn call malloc,
  which will fail again.  Generally, the best policy is to simply call
  abort(). It's not very useful to do more than this because many
  errors due to overwriting will show up as address faults (null, odd
  addresses etc) rather than malloc-triggered checks, so will also
  abort.  Also, most compilers know that abort() does not return, so
  can better optimize code conditionally calling it.
 
PROCEED_ON_ERROR           default: defined as 0 (false)
  Controls whether detected bad addresses cause them to bypassed
  rather than aborting. If set, detected bad arguments to free and
  realloc are ignored. And all bookkeeping information is zeroed out
  upon a detected overwrite of freed heap space, thus losing the
  ability to ever return it from malloc again, but enabling the
  application to proceed. If PROCEED_ON_ERROR is defined, the
  static variable malloc_corruption_error_count is compiled in
  and can be examined to see if errors have occurred. This option
  generates slower code than the default abort policy.
 
DEBUG                    default: NOT defined
  The DEBUG setting is mainly intended for people trying to modify
  this code or diagnose problems when porting to new platforms.
  However, it may also be able to better isolate user errors than just
  using runtime checks.  The assertions in the check routines spell
  out in more detail the assumptions and invariants underlying the
  algorithms.  The checking is fairly extensive, and will slow down
  execution noticeably. Calling malloc_stats or mallinfo with DEBUG
  set will attempt to check every non-mmapped allocated and free chunk
  in the course of computing the summaries.
 
ABORT_ON_ASSERT_FAILURE   default: defined as 1 (true)
  Debugging assertion failures can be nearly impossible if your
  version of the assert macro causes malloc to be called, which will
  lead to a cascade of further failures, blowing the runtime stack.
  ABORT_ON_ASSERT_FAILURE cause assertions failures to call abort(),
  which will usually make debugging easier.
 
MALLOC_FAILURE_ACTION     default: sets errno to ENOMEM, or no-op on win32
  The action to take before "return 0" when malloc fails to be able to
  return memory because there is none available.
 
HAVE_MORECORE             default: 1 (true) unless win32 or ONLY_MSPACES
  True if this system supports sbrk or an emulation of it.
 
MORECORE                  default: sbrk
  The name of the sbrk-style system routine to call to obtain more
  memory.  See below for guidance on writing custom MORECORE
  functions. The type of the argument to sbrk/MORECORE varies across
  systems.  It cannot be size_t, because it supports negative
  arguments, so it is normally the signed type of the same width as
  size_t (sometimes declared as "intptr_t").  It doesn't much matter
  though. Internally, we only call it with arguments less than half
  the max value of a size_t, which should work across all reasonable
  possibilities, although sometimes generating compiler warnings.  See
  near the end of this file for guidelines for creating a custom
  version of MORECORE.
 
MORECORE_CONTIGUOUS       default: 1 (true)
  If true, take advantage of fact that consecutive calls to MORECORE
  with positive arguments always return contiguous increasing
  addresses.  This is true of unix sbrk. It does not hurt too much to
  set it true anyway, since malloc copes with non-contiguities.
  Setting it false when definitely non-contiguous saves time
  and possibly wasted space it would take to discover this though.
 
MORECORE_CANNOT_TRIM      default: NOT defined
  True if MORECORE cannot release space back to the system when given
  negative arguments. This is generally necessary only if you are
  using a hand-crafted MORECORE function that cannot handle negative
  arguments.
 
HAVE_MMAP                 default: 1 (true)
  True if this system supports mmap or an emulation of it.  If so, and
  HAVE_MORECORE is not true, MMAP is used for all system
  allocation. If set and HAVE_MORECORE is true as well, MMAP is
  primarily used to directly allocate very large blocks. It is also
  used as a backup strategy in cases where MORECORE fails to provide
  space from system. Note: A single call to MUNMAP is assumed to be
  able to unmap memory that may have be allocated using multiple calls
  to MMAP, so long as they are adjacent.
 
HAVE_MREMAP               default: 1 on linux, else 0
  If true realloc() uses mremap() to re-allocate large blocks and
  extend or shrink allocation spaces.
 
MMAP_CLEARS               default: 1 on unix
  True if mmap clears memory so calloc doesn't need to. This is true
  for standard unix mmap using /dev/zero.
 
USE_BUILTIN_FFS            default: 0 (i.e., not used)
  Causes malloc to use the builtin ffs() function to compute indices.
  Some compilers may recognize and intrinsify ffs to be faster than the
  supplied C version. Also, the case of x86 using gcc is special-cased
  to an asm instruction, so is already as fast as it can be, and so
  this setting has no effect. (On most x86s, the asm version is only
  slightly faster than the C version.)
 
malloc_getpagesize         default: derive from system includes, or 4096.
  The system page size. To the extent possible, this malloc manages
  memory from the system in page-size units.  This may be (and
  usually is) a function rather than a constant. This is ignored
  if WIN32, where page size is determined using getSystemInfo during
  initialization.
 
USE_DEV_RANDOM             default: 0 (i.e., not used)
  Causes malloc to use /dev/random to initialize secure magic seed for
  stamping footers. Otherwise, the current time is used.
 
NO_MALLINFO                default: 0
  If defined, don't compile "mallinfo". This can be a simple way
  of dealing with mismatches between system declarations and
  those in this file.
 
MALLINFO_FIELD_TYPE        default: size_t
  The type of the fields in the mallinfo struct. This was originally
  defined as "int" in SVID etc, but is more usefully defined as
  size_t. The value is used only if  HAVE_USR_INCLUDE_MALLOC_H is not set
 
REALLOC_ZERO_BYTES_FREES    default: not defined
  This should be set if a call to realloc with zero bytes should
  be the same as a call to free. Some people think it should. Otherwise,
  since this malloc returns a unique pointer for malloc(0), so does
  realloc(p, 0).
 
LACKS_UNISTD_H, LACKS_FCNTL_H, LACKS_SYS_PARAM_H, LACKS_SYS_MMAN_H
LACKS_STRINGS_H, LACKS_STRING_H, LACKS_SYS_TYPES_H,  LACKS_ERRNO_H
LACKS_STDLIB_H                default: NOT defined unless on WIN32
  Define these if your system does not have these header files.
  You might need to manually insert some of the declarations they provide.
 
DEFAULT_GRANULARITY        default: page size if MORECORE_CONTIGUOUS,
                                system_info.dwAllocationGranularity in WIN32,
                                otherwise 64K.
      Also settable using mallopt(M_GRANULARITY, x)
  The unit for allocating and deallocating memory from the system.  On
  most systems with contiguous MORECORE, there is no reason to
  make this more than a page. However, systems with MMAP tend to
  either require or encourage larger granularities.  You can increase
  this value to prevent system allocation functions to be called so
  often, especially if they are slow.  The value must be at least one
  page and must be a power of two.  Setting to 0 causes initialization
  to either page size or win32 region size.  (Note: In previous
  versions of malloc, the equivalent of this option was called
  "TOP_PAD")
 
DEFAULT_TRIM_THRESHOLD    default: 2MB
      Also settable using mallopt(M_TRIM_THRESHOLD, x)
  The maximum amount of unused top-most memory to keep before
  releasing via malloc_trim in free().  Automatic trimming is mainly
  useful in long-lived programs using contiguous MORECORE.  Because
  trimming via sbrk can be slow on some systems, and can sometimes be
  wasteful (in cases where programs immediately afterward allocate
  more large chunks) the value should be high enough so that your
  overall system performance would improve by releasing this much
  memory.  As a rough guide, you might set to a value close to the
  average size of a process (program) running on your system.
  Releasing this much memory would allow such a process to run in
  memory.  Generally, it is worth tuning trim thresholds when a
  program undergoes phases where several large chunks are allocated
  and released in ways that can reuse each other's storage, perhaps
  mixed with phases where there are no such chunks at all. The trim
  value must be greater than page size to have any useful effect.  To
  disable trimming completely, you can set to MAX_SIZE_T. Note that the trick
  some people use of mallocing a huge space and then freeing it at
  program startup, in an attempt to reserve system memory, doesn't
  have the intended effect under automatic trimming, since that memory
  will immediately be returned to the system.
 
DEFAULT_MMAP_THRESHOLD       default: 256K
      Also settable using mallopt(M_MMAP_THRESHOLD, x)
  The request size threshold for using MMAP to directly service a
  request. Requests of at least this size that cannot be allocated
  using already-existing space will be serviced via mmap.  (If enough
  normal freed space already exists it is used instead.)  Using mmap
  segregates relatively large chunks of memory so that they can be
  individually obtained and released from the host system. A request
  serviced through mmap is never reused by any other request (at least
  not directly; the system may just so happen to remap successive
  requests to the same locations).  Segregating space in this way has
  the benefits that: Mmapped space can always be individually released
  back to the system, which helps keep the system level memory demands
  of a long-lived program low.  Also, mapped memory doesn't become
  `locked' between other chunks, as can happen with normally allocated
  chunks, which means that even trimming via malloc_trim would not
  release them.  However, it has the disadvantage that the space
  cannot be reclaimed, consolidated, and then used to service later
  requests, as happens with normal chunks.  The advantages of mmap
  nearly always outweigh disadvantages for "large" chunks, but the
  value of "large" may vary across systems.  The default is an
  empirically derived value that works well in most systems. You can
  disable mmap by setting to MAX_SIZE_T.
 
*/
 
#include <sys/types.h>  /* For size_t */
 
/** Non-default helenos customizations */
#define LACKS_FCNTL_H
#define LACKS_SYS_MMAN_H
#define LACKS_SYS_PARAM_H
#undef HAVE_MMAP
#define HAVE_MMAP 0
#define LACKS_ERRNO_H
/* Set errno? */
#undef MALLOC_FAILURE_ACTION
#define MALLOC_FAILURE_ACTION
 
/* The maximum possible size_t value has all bits set */
#define MAX_SIZE_T           (~(size_t)0)
 
#define ONLY_MSPACES 0
#define MSPACES 0
#define MALLOC_ALIGNMENT ((size_t)8U)
#define FOOTERS 0
#define ABORT  abort()
#define ABORT_ON_ASSERT_FAILURE 1
#define PROCEED_ON_ERROR 0
#define USE_LOCKS 0
#define INSECURE 0
#define HAVE_MMAP 0
 
#define MMAP_CLEARS 1
 
#define HAVE_MORECORE 1
#define MORECORE_CONTIGUOUS 1
#define MORECORE sbrk
#define DEFAULT_GRANULARITY (0)  /* 0 means to compute in init_mparams */
 
#ifndef DEFAULT_TRIM_THRESHOLD
#ifndef MORECORE_CANNOT_TRIM
#define DEFAULT_TRIM_THRESHOLD ((size_t)2U * (size_t)1024U * (size_t)1024U)
#else   /* MORECORE_CANNOT_TRIM */
#define DEFAULT_TRIM_THRESHOLD MAX_SIZE_T
#endif  /* MORECORE_CANNOT_TRIM */
#endif  /* DEFAULT_TRIM_THRESHOLD */
#ifndef DEFAULT_MMAP_THRESHOLD
#if HAVE_MMAP
#define DEFAULT_MMAP_THRESHOLD ((size_t)256U * (size_t)1024U)
#else   /* HAVE_MMAP */
#define DEFAULT_MMAP_THRESHOLD MAX_SIZE_T
#endif  /* HAVE_MMAP */
#endif  /* DEFAULT_MMAP_THRESHOLD */
#ifndef USE_BUILTIN_FFS
#define USE_BUILTIN_FFS 0
#endif  /* USE_BUILTIN_FFS */
#ifndef USE_DEV_RANDOM
#define USE_DEV_RANDOM 0
#endif  /* USE_DEV_RANDOM */
#ifndef NO_MALLINFO
#define NO_MALLINFO 0
#endif  /* NO_MALLINFO */
#ifndef MALLINFO_FIELD_TYPE
#define MALLINFO_FIELD_TYPE size_t
#endif  /* MALLINFO_FIELD_TYPE */
 
/*
  mallopt tuning options.  SVID/XPG defines four standard parameter
  numbers for mallopt, normally defined in malloc.h.  None of these
  are used in this malloc, so setting them has no effect. But this
  malloc does support the following options.
*/
 
#define M_TRIM_THRESHOLD     (-1)
#define M_GRANULARITY        (-2)
#define M_MMAP_THRESHOLD     (-3)
 
/*
  ========================================================================
  To make a fully customizable malloc.h header file, cut everything
  above this line, put into file malloc.h, edit to suit, and #include it
  on the next line, as well as in programs that use this malloc.
  ========================================================================
*/
 
#include "malloc.h"
 
/*------------------------------ internal #includes ---------------------- */
 
#include <stdio.h>       /* for printing in malloc_stats */
#include <string.h>
 
#ifndef LACKS_ERRNO_H
#include <errno.h>       /* for MALLOC_FAILURE_ACTION */
#endif /* LACKS_ERRNO_H */
#if FOOTERS
#include <time.h>        /* for magic initialization */
#endif /* FOOTERS */
#ifndef LACKS_STDLIB_H
#include <stdlib.h>      /* for abort() */
#endif /* LACKS_STDLIB_H */
#ifdef DEBUG
#if ABORT_ON_ASSERT_FAILURE
#define assert(x) {if(!(x)) {printf(#x);ABORT;}}
#else /* ABORT_ON_ASSERT_FAILURE */
#include <assert.h>
#endif /* ABORT_ON_ASSERT_FAILURE */
#else  /* DEBUG */
#define assert(x)
#endif /* DEBUG */
#if USE_BUILTIN_FFS
#ifndef LACKS_STRINGS_H
#include <strings.h>     /* for ffs */
#endif /* LACKS_STRINGS_H */
#endif /* USE_BUILTIN_FFS */
#if HAVE_MMAP
#ifndef LACKS_SYS_MMAN_H
#include <sys/mman.h>    /* for mmap */
#endif /* LACKS_SYS_MMAN_H */
#ifndef LACKS_FCNTL_H
#include <fcntl.h>
#endif /* LACKS_FCNTL_H */
#endif /* HAVE_MMAP */
#if HAVE_MORECORE
#ifndef LACKS_UNISTD_H
#include <unistd.h>     /* for sbrk */
#else /* LACKS_UNISTD_H */
#if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
extern void*     sbrk(ptrdiff_t);
#endif /* FreeBSD etc */
#endif /* LACKS_UNISTD_H */
#endif /* HAVE_MMAP */
 
#ifndef WIN32
#ifndef malloc_getpagesize
#  ifdef _SC_PAGESIZE         /* some SVR4 systems omit an underscore */
#    ifndef _SC_PAGE_SIZE
#      define _SC_PAGE_SIZE _SC_PAGESIZE
#    endif
#  endif
#  ifdef _SC_PAGE_SIZE
#    define malloc_getpagesize sysconf(_SC_PAGE_SIZE)
#  else
#    if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE)
       extern size_t getpagesize();
#      define malloc_getpagesize getpagesize()
#    else
#      ifdef WIN32 /* use supplied emulation of getpagesize */
#        define malloc_getpagesize getpagesize()
#      else
#        ifndef LACKS_SYS_PARAM_H
#          include <sys/param.h>
#        endif
#        ifdef EXEC_PAGESIZE
#          define malloc_getpagesize EXEC_PAGESIZE
#        else
#          ifdef NBPG
#            ifndef CLSIZE
#              define malloc_getpagesize NBPG
#            else
#              define malloc_getpagesize (NBPG * CLSIZE)
#            endif
#          else
#            ifdef NBPC
#              define malloc_getpagesize NBPC
#            else
#              ifdef PAGESIZE
#                define malloc_getpagesize PAGESIZE
#              else /* just guess */
#                define malloc_getpagesize ((size_t)4096U)
#              endif
#            endif
#          endif
#        endif
#      endif
#    endif
#  endif
#endif
#endif
 
/* ------------------- size_t and alignment properties -------------------- */
 
/* The byte and bit size of a size_t */
#define SIZE_T_SIZE         (sizeof(size_t))
#define SIZE_T_BITSIZE      (sizeof(size_t) << 3)
 
/* Some constants coerced to size_t */
/* Annoying but necessary to avoid errors on some plaftorms */
#define SIZE_T_ZERO         ((size_t)0)
#define SIZE_T_ONE          ((size_t)1)
#define SIZE_T_TWO          ((size_t)2)
#define TWO_SIZE_T_SIZES    (SIZE_T_SIZE<<1)
#define FOUR_SIZE_T_SIZES   (SIZE_T_SIZE<<2)
#define SIX_SIZE_T_SIZES    (FOUR_SIZE_T_SIZES+TWO_SIZE_T_SIZES)
#define HALF_MAX_SIZE_T     (MAX_SIZE_T / 2U)
 
/* The bit mask value corresponding to MALLOC_ALIGNMENT */
#define CHUNK_ALIGN_MASK    (MALLOC_ALIGNMENT - SIZE_T_ONE)
 
/* True if address a has acceptable alignment */
#define is_aligned(A)       (((size_t)((A)) & (CHUNK_ALIGN_MASK)) == 0)
 
/* the number of bytes to offset an address to align it */
#define align_offset(A)\
 ((((size_t)(A) & CHUNK_ALIGN_MASK) == 0)? 0 :\
  ((MALLOC_ALIGNMENT - ((size_t)(A) & CHUNK_ALIGN_MASK)) & CHUNK_ALIGN_MASK))
 
/* -------------------------- MMAP preliminaries ------------------------- */
 
/*
   If HAVE_MORECORE or HAVE_MMAP are false, we just define calls and
   checks to fail so compiler optimizer can delete code rather than
   using so many "#if"s.
*/
 
 
/* MORECORE and MMAP must return MFAIL on failure */
#define MFAIL                ((void*)(MAX_SIZE_T))
#define CMFAIL               ((char*)(MFAIL)) /* defined for convenience */
 
#if !HAVE_MMAP
#define IS_MMAPPED_BIT       (SIZE_T_ZERO)
#define USE_MMAP_BIT         (SIZE_T_ZERO)
#define CALL_MMAP(s)         MFAIL
#define CALL_MUNMAP(a, s)    (-1)
#define DIRECT_MMAP(s)       MFAIL
 
#else /* HAVE_MMAP */
#define IS_MMAPPED_BIT       (SIZE_T_ONE)
#define USE_MMAP_BIT         (SIZE_T_ONE)
 
#ifndef WIN32
#define CALL_MUNMAP(a, s)    munmap((a), (s))
#define MMAP_PROT            (PROT_READ|PROT_WRITE)
#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
#define MAP_ANONYMOUS        MAP_ANON
#endif /* MAP_ANON */
#ifdef MAP_ANONYMOUS
#define MMAP_FLAGS           (MAP_PRIVATE|MAP_ANONYMOUS)
#define CALL_MMAP(s)         mmap(0, (s), MMAP_PROT, MMAP_FLAGS, -1, 0)
#else /* MAP_ANONYMOUS */
/*
   Nearly all versions of mmap support MAP_ANONYMOUS, so the following
   is unlikely to be needed, but is supplied just in case.
*/
#define MMAP_FLAGS           (MAP_PRIVATE)
static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */
#define CALL_MMAP(s) ((dev_zero_fd < 0) ? \
           (dev_zero_fd = open("/dev/zero", O_RDWR), \
            mmap(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0)) : \
            mmap(0, (s), MMAP_PROT, MMAP_FLAGS, dev_zero_fd, 0))
#endif /* MAP_ANONYMOUS */
 
#define DIRECT_MMAP(s)       CALL_MMAP(s)
#else /* WIN32 */
 
/* Win32 MMAP via VirtualAlloc */
static void* win32mmap(size_t size) {
  void* ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT, PAGE_READWRITE);
  return (ptr != 0)? ptr: MFAIL;
}
 
/* For direct MMAP, use MEM_TOP_DOWN to minimize interference */
static void* win32direct_mmap(size_t size) {
  void* ptr = VirtualAlloc(0, size, MEM_RESERVE|MEM_COMMIT|MEM_TOP_DOWN,
                           PAGE_READWRITE);
  return (ptr != 0)? ptr: MFAIL;
}
 
/* This function supports releasing coalesed segments */
static int win32munmap(void* ptr, size_t size) {
  MEMORY_BASIC_INFORMATION minfo;
  char* cptr = ptr;
  while (size) {
    if (VirtualQuery(cptr, &minfo, sizeof(minfo)) == 0)
      return -1;
    if (minfo.BaseAddress != cptr || minfo.AllocationBase != cptr ||
        minfo.State != MEM_COMMIT || minfo.RegionSize > size)
      return -1;
    if (VirtualFree(cptr, 0, MEM_RELEASE) == 0)
      return -1;
    cptr += minfo.RegionSize;
    size -= minfo.RegionSize;
  }
  return 0;
}
 
#define CALL_MMAP(s)         win32mmap(s)
#define CALL_MUNMAP(a, s)    win32munmap((a), (s))
#define DIRECT_MMAP(s)       win32direct_mmap(s)
#endif /* WIN32 */
#endif /* HAVE_MMAP */
 
#if HAVE_MMAP && HAVE_MREMAP
#define CALL_MREMAP(addr, osz, nsz, mv) mremap((addr), (osz), (nsz), (mv))
#else  /* HAVE_MMAP && HAVE_MREMAP */
#define CALL_MREMAP(addr, osz, nsz, mv) MFAIL
#endif /* HAVE_MMAP && HAVE_MREMAP */
 
#if HAVE_MORECORE
#define CALL_MORECORE(S)     MORECORE(S)
#else  /* HAVE_MORECORE */
#define CALL_MORECORE(S)     MFAIL
#endif /* HAVE_MORECORE */
 
/* mstate bit set if continguous morecore disabled or failed */
#define USE_NONCONTIGUOUS_BIT (4U)
 
/* segment bit set in create_mspace_with_base */
#define EXTERN_BIT            (8U)
 
 
/* --------------------------- Lock preliminaries ------------------------ */
 
#if USE_LOCKS
 
/*
  When locks are defined, there are up to two global locks:
 
  * If HAVE_MORECORE, morecore_mutex protects sequences of calls to
    MORECORE.  In many cases sys_alloc requires two calls, that should
    not be interleaved with calls by other threads.  This does not
    protect against direct calls to MORECORE by other threads not
    using this lock, so there is still code to cope the best we can on
    interference.
 
  * magic_init_mutex ensures that mparams.magic and other
    unique mparams values are initialized only once.
*/
 
#ifndef WIN32
/* By default use posix locks */
#include <pthread.h>
#define MLOCK_T pthread_mutex_t
#define INITIAL_LOCK(l)      pthread_mutex_init(l, NULL)
#define ACQUIRE_LOCK(l)      pthread_mutex_lock(l)
#define RELEASE_LOCK(l)      pthread_mutex_unlock(l)
 
#if HAVE_MORECORE
static MLOCK_T morecore_mutex = PTHREAD_MUTEX_INITIALIZER;
#endif /* HAVE_MORECORE */
 
static MLOCK_T magic_init_mutex = PTHREAD_MUTEX_INITIALIZER;
 
#else /* WIN32 */
/*
   Because lock-protected regions have bounded times, and there
   are no recursive lock calls, we can use simple spinlocks.
*/
 
#define MLOCK_T long
static int win32_acquire_lock (MLOCK_T *sl) {
  for (;;) {
#ifdef InterlockedCompareExchangePointer
    if (!InterlockedCompareExchange(sl, 1, 0))
      return 0;
#else  /* Use older void* version */
    if (!InterlockedCompareExchange((void**)sl, (void*)1, (void*)0))
      return 0;
#endif /* InterlockedCompareExchangePointer */
    Sleep (0);
  }
}
 
static void win32_release_lock (MLOCK_T *sl) {
  InterlockedExchange (sl, 0);
}
 
#define INITIAL_LOCK(l)      *(l)=0
#define ACQUIRE_LOCK(l)      win32_acquire_lock(l)
#define RELEASE_LOCK(l)      win32_release_lock(l)
#if HAVE_MORECORE
static MLOCK_T morecore_mutex;
#endif /* HAVE_MORECORE */
static MLOCK_T magic_init_mutex;
#endif /* WIN32 */
 
#define USE_LOCK_BIT               (2U)
#else  /* USE_LOCKS */
#define USE_LOCK_BIT               (0U)
#define INITIAL_LOCK(l)
#endif /* USE_LOCKS */
 
#if USE_LOCKS && HAVE_MORECORE
#define ACQUIRE_MORECORE_LOCK()    ACQUIRE_LOCK(&morecore_mutex);
#define RELEASE_MORECORE_LOCK()    RELEASE_LOCK(&morecore_mutex);
#else /* USE_LOCKS && HAVE_MORECORE */
#define ACQUIRE_MORECORE_LOCK()
#define RELEASE_MORECORE_LOCK()
#endif /* USE_LOCKS && HAVE_MORECORE */
 
#if USE_LOCKS
#define ACQUIRE_MAGIC_INIT_LOCK()  ACQUIRE_LOCK(&magic_init_mutex);
#define RELEASE_MAGIC_INIT_LOCK()  RELEASE_LOCK(&magic_init_mutex);
#else  /* USE_LOCKS */
#define ACQUIRE_MAGIC_INIT_LOCK()
#define RELEASE_MAGIC_INIT_LOCK()
#endif /* USE_LOCKS */
 
 
/* -----------------------  Chunk representations ------------------------ */
 
/*
  (The following includes lightly edited explanations by Colin Plumb.)
 
  The malloc_chunk declaration below is misleading (but accurate and
  necessary).  It declares a "view" into memory allowing access to
  necessary fields at known offsets from a given base.
 
  Chunks of memory are maintained using a `boundary tag' method as
  originally described by Knuth.  (See the paper by Paul Wilson
  ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a survey of such
  techniques.)  Sizes of free chunks are stored both in the front of
  each chunk and at the end.  This makes consolidating fragmented
  chunks into bigger chunks fast.  The head fields also hold bits
  representing whether chunks are free or in use.
 
  Here are some pictures to make it clearer.  They are "exploded" to
  show that the state of a chunk can be thought of as extending from
  the high 31 bits of the head field of its header through the
  prev_foot and PINUSE_BIT bit of the following chunk header.
 
  A chunk that's in use looks like:
 
   chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
           | Size of previous chunk (if P = 1)                             |
           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|
         | Size of this chunk                                         1| +-+
   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         |                                                               |
         +-                                                             -+
         |                                                               |
         +-                                                             -+
         |                                                               :
         +-      size - sizeof(size_t) available payload bytes          -+
         :                                                               |
 chunk-> +-                                                             -+
         |                                                               |
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1|
       | Size of next chunk (may or may not be in use)               | +-+
 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 
    And if it's free, it looks like this:
 
   chunk-> +-                                                             -+
           | User payload (must be in use, or we would have merged!)       |
           +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |P|
         | Size of this chunk                                         0| +-+
   mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         | Next pointer                                                  |
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         | Prev pointer                                                  |
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         |                                                               :
         +-      size - sizeof(struct chunk) unused bytes               -+
         :                                                               |
 chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
         | Size of this chunk                                            |
         +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |0|
       | Size of next chunk (must be in use, or we would have merged)| +-+
 mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
       |                                                               :
       +- User payload                                                -+
       :                                                               |
       +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
                                                                     |0|
                                                                     +-+
  Note that since we always merge adjacent free chunks, the chunks
  adjacent to a free chunk must be in use.
 
  Given a pointer to a chunk (which can be derived trivially from the
  payload pointer) we can, in O(1) time, find out whether the adjacent
  chunks are free, and if so, unlink them from the lists that they
  are on and merge them with the current chunk.
 
  Chunks always begin on even word boundaries, so the mem portion
  (which is returned to the user) is also on an even word boundary, and
  thus at least double-word aligned.
 
  The P (PINUSE_BIT) bit, stored in the unused low-order bit of the
  chunk size (which is always a multiple of two words), is an in-use
  bit for the *previous* chunk.  If that bit is *clear*, then the
  word before the current chunk size contains the previous chunk
  size, and can be used to find the front of the previous chunk.
  The very first chunk allocated always has this bit set, preventing
  access to non-existent (or non-owned) memory. If pinuse is set for
  any given chunk, then you CANNOT determine the size of the
  previous chunk, and might even get a memory addressing fault when
  trying to do so.
 
  The C (CINUSE_BIT) bit, stored in the unused second-lowest bit of
  the chunk size redundantly records whether the current chunk is
  inuse. This redundancy enables usage checks within free and realloc,
  and reduces indirection when freeing and consolidating chunks.
 
  Each freshly allocated chunk must have both cinuse and pinuse set.
  That is, each allocated chunk borders either a previously allocated
  and still in-use chunk, or the base of its memory arena. This is
  ensured by making all allocations from the the `lowest' part of any
  found chunk.  Further, no free chunk physically borders another one,
  so each free chunk is known to be preceded and followed by either
  inuse chunks or the ends of memory.
 
  Note that the `foot' of the current chunk is actually represented
  as the prev_foot of the NEXT chunk. This makes it easier to
  deal with alignments etc but can be very confusing when trying
  to extend or adapt this code.
 
  The exceptions to all this are
 
     1. The special chunk `top' is the top-most available chunk (i.e.,
        the one bordering the end of available memory). It is treated
        specially.  Top is never included in any bin, is used only if
        no other chunk is available, and is released back to the
        system if it is very large (see M_TRIM_THRESHOLD).  In effect,
        the top chunk is treated as larger (and thus less well
        fitting) than any other available chunk.  The top chunk
        doesn't update its trailing size field since there is no next
        contiguous chunk that would have to index off it. However,
        space is still allocated for it (TOP_FOOT_SIZE) to enable
        separation or merging when space is extended.
 
     3. Chunks allocated via mmap, which have the lowest-order bit
        (IS_MMAPPED_BIT) set in their prev_foot fields, and do not set
        PINUSE_BIT in their head fields.  Because they are allocated
        one-by-one, each must carry its own prev_foot field, which is
        also used to hold the offset this chunk has within its mmapped
        region, which is needed to preserve alignment. Each mmapped
        chunk is trailed by the first two fields of a fake next-chunk
        for sake of usage checks.
 
*/
 
struct malloc_chunk {
  size_t               prev_foot;  /* Size of previous chunk (if free).  */
  size_t               head;       /* Size and inuse bits. */
  struct malloc_chunk* fd;         /* double links -- used only if free. */
  struct malloc_chunk* bk;
};
 
typedef struct malloc_chunk  mchunk;
typedef struct malloc_chunk* mchunkptr;
typedef struct malloc_chunk* sbinptr;  /* The type of bins of chunks */
typedef unsigned int bindex_t;         /* Described below */
typedef unsigned int binmap_t;         /* Described below */
typedef unsigned int flag_t;           /* The type of various bit flag sets */
 
/* ------------------- Chunks sizes and alignments ----------------------- */
 
#define MCHUNK_SIZE         (sizeof(mchunk))
 
#if FOOTERS
#define CHUNK_OVERHEAD      (TWO_SIZE_T_SIZES)
#else /* FOOTERS */
#define CHUNK_OVERHEAD      (SIZE_T_SIZE)
#endif /* FOOTERS */
 
/* MMapped chunks need a second word of overhead ... */
#define MMAP_CHUNK_OVERHEAD (TWO_SIZE_T_SIZES)
/* ... and additional padding for fake next-chunk at foot */
#define MMAP_FOOT_PAD       (FOUR_SIZE_T_SIZES)
 
/* The smallest size we can malloc is an aligned minimal chunk */
#define MIN_CHUNK_SIZE\
  ((MCHUNK_SIZE + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)
 
/* conversion from malloc headers to user pointers, and back */
#define chunk2mem(p)        ((void*)((char*)(p)       + TWO_SIZE_T_SIZES))
#define mem2chunk(mem)      ((mchunkptr)((char*)(mem) - TWO_SIZE_T_SIZES))
/* chunk associated with aligned address A */
#define align_as_chunk(A)   (mchunkptr)((A) + align_offset(chunk2mem(A)))
 
/* Bounds on request (not chunk) sizes. */
#define MAX_REQUEST         ((-MIN_CHUNK_SIZE) << 2)
#define MIN_REQUEST         (MIN_CHUNK_SIZE - CHUNK_OVERHEAD - SIZE_T_ONE)
 
/* pad request bytes into a usable size */
#define pad_request(req) \
   (((req) + CHUNK_OVERHEAD + CHUNK_ALIGN_MASK) & ~CHUNK_ALIGN_MASK)
 
/* pad request, checking for minimum (but not maximum) */
#define request2size(req) \
  (((req) < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(req))
 
 
/* ------------------ Operations on head and foot fields ----------------- */
 
/*
  The head field of a chunk is or'ed with PINUSE_BIT when previous
  adjacent chunk in use, and or'ed with CINUSE_BIT if this chunk is in
  use. If the chunk was obtained with mmap, the prev_foot field has
  IS_MMAPPED_BIT set, otherwise holding the offset of the base of the
  mmapped region to the base of the chunk.
*/
 
#define PINUSE_BIT          (SIZE_T_ONE)
#define CINUSE_BIT          (SIZE_T_TWO)
#define INUSE_BITS          (PINUSE_BIT|CINUSE_BIT)
 
/* Head value for fenceposts */
#define FENCEPOST_HEAD      (INUSE_BITS|SIZE_T_SIZE)
 
/* extraction of fields from head words */
#define cinuse(p)           ((p)->head & CINUSE_BIT)
#define pinuse(p)           ((p)->head & PINUSE_BIT)
#define chunksize(p)        ((p)->head & ~(INUSE_BITS))
 
#define clear_pinuse(p)     ((p)->head &= ~PINUSE_BIT)
#define clear_cinuse(p)     ((p)->head &= ~CINUSE_BIT)
 
/* Treat space at ptr +/- offset as a chunk */
#define chunk_plus_offset(p, s)  ((mchunkptr)(((char*)(p)) + (s)))
#define chunk_minus_offset(p, s) ((mchunkptr)(((char*)(p)) - (s)))
 
/* Ptr to next or previous physical malloc_chunk. */
#define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->head & ~INUSE_BITS)))
#define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_foot) ))
 
/* extract next chunk's pinuse bit */
#define next_pinuse(p)  ((next_chunk(p)->head) & PINUSE_BIT)
 
/* Get/set size at footer */
#define get_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot)
#define set_foot(p, s)  (((mchunkptr)((char*)(p) + (s)))->prev_foot = (s))
 
/* Set size, pinuse bit, and foot */
#define set_size_and_pinuse_of_free_chunk(p, s)\
  ((p)->head = (s|PINUSE_BIT), set_foot(p, s))
 
/* Set size, pinuse bit, foot, and clear next pinuse */
#define set_free_with_pinuse(p, s, n)\
  (clear_pinuse(n), set_size_and_pinuse_of_free_chunk(p, s))
 
#define is_mmapped(p)\
  (!((p)->head & PINUSE_BIT) && ((p)->prev_foot & IS_MMAPPED_BIT))
 
/* Get the internal overhead associated with chunk p */
#define overhead_for(p)\
 (is_mmapped(p)? MMAP_CHUNK_OVERHEAD : CHUNK_OVERHEAD)
 
/* Return true if malloced space is not necessarily cleared */
#if MMAP_CLEARS
#define calloc_must_clear(p) (!is_mmapped(p))
#else /* MMAP_CLEARS */
#define calloc_must_clear(p) (1)
#endif /* MMAP_CLEARS */
 
/* ---------------------- Overlaid data structures ----------------------- */
 
/*
  When chunks are not in use, they are treated as nodes of either
  lists or trees.
 
  "Small"  chunks are stored in circular doubly-linked lists, and look
  like this:
 
    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Size of previous chunk                            |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    `head:' |             Size of chunk, in bytes                         |P|
      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Forward pointer to next chunk in list             |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Back pointer to previous chunk in list            |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Unused space (may be 0 bytes long)                .
            .                                                               .
            .                                                               |
nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    `foot:' |             Size of chunk, in bytes                           |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 
  Larger chunks are kept in a form of bitwise digital trees (aka
  tries) keyed on chunksizes.  Because malloc_tree_chunks are only for
  free chunks greater than 256 bytes, their size doesn't impose any
  constraints on user chunk sizes.  Each node looks like:
 
    chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Size of previous chunk                            |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    `head:' |             Size of chunk, in bytes                         |P|
      mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Forward pointer to next chunk of same size        |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Back pointer to previous chunk of same size       |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Pointer to left child (child[0])                  |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Pointer to right child (child[1])                 |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Pointer to parent                                 |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             bin index of this chunk                           |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
            |             Unused space                                      .
            .                                                               |
nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
    `foot:' |             Size of chunk, in bytes                           |
            +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
 
  Each tree holding treenodes is a tree of unique chunk sizes.  Chunks
  of the same size are arranged in a circularly-linked list, with only
  the oldest chunk (the next to be used, in our FIFO ordering)
  actually in the tree.  (Tree members are distinguished by a non-null
  parent pointer.)  If a chunk with the same size an an existing node
  is inserted, it is linked off the existing node using pointers that
  work in the same way as fd/bk pointers of small chunks.
 
  Each tree contains a power of 2 sized range of chunk sizes (the
  smallest is 0x100 <= x < 0x180), which is is divided in half at each
  tree level, with the chunks in the smaller half of the range (0x100
  <= x < 0x140 for the top nose) in the left subtree and the larger
  half (0x140 <= x < 0x180) in the right subtree.  This is, of course,
  done by inspecting individual bits.
 
  Using these rules, each node's left subtree contains all smaller
  sizes than its right subtree.  However, the node at the root of each
  subtree has no particular ordering relationship to either.  (The
  dividing line between the subtree sizes is based on trie relation.)
  If we remove the last chunk of a given size from the interior of the
  tree, we need to replace it with a leaf node.  The tree ordering
  rules permit a node to be replaced by any leaf below it.
 
  The smallest chunk in a tree (a common operation in a best-fit
  allocator) can be found by walking a path to the leftmost leaf in
  the tree.  Unlike a usual binary tree, where we follow left child
  pointers until we reach a null, here we follow the right child
  pointer any time the left one is null, until we reach a leaf with
  both child pointers null. The smallest chunk in the tree will be
  somewhere along that path.
 
  The worst case number of steps to add, find, or remove a node is
  bounded by the number of bits differentiating chunks within
  bins. Under current bin calculations, this ranges from 6 up to 21
  (for 32 bit sizes) or up to 53 (for 64 bit sizes). The typical case
  is of course much better.
*/
 
struct malloc_tree_chunk {
  /* The first four fields must be compatible with malloc_chunk */
  size_t                    prev_foot;
  size_t                    head;
  struct malloc_tree_chunk* fd;
  struct malloc_tree_chunk* bk;
 
  struct malloc_tree_chunk* child[2];
  struct malloc_tree_chunk* parent;
  bindex_t                  index;
};
 
typedef struct malloc_tree_chunk  tchunk;
typedef struct malloc_tree_chunk* tchunkptr;
typedef struct malloc_tree_chunk* tbinptr; /* The type of bins of trees */
 
/* A little helper macro for trees */
#define leftmost_child(t) ((t)->child[0] != 0? (t)->child[0] : (t)->child[1])
 
/* ----------------------------- Segments -------------------------------- */
 
/*
  Each malloc space may include non-contiguous segments, held in a
  list headed by an embedded malloc_segment record representing the
  top-most space. Segments also include flags holding properties of
  the space. Large chunks that are directly allocated by mmap are not
  included in this list. They are instead independently created and
  destroyed without otherwise keeping track of them.
 
  Segment management mainly comes into play for spaces allocated by
  MMAP.  Any call to MMAP might or might not return memory that is
  adjacent to an existing segment.  MORECORE normally contiguously
  extends the current space, so this space is almost always adjacent,
  which is simpler and faster to deal with. (This is why MORECORE is
  used preferentially to MMAP when both are available -- see
  sys_alloc.)  When allocating using MMAP, we don't use any of the
  hinting mechanisms (inconsistently) supported in various
  implementations of unix mmap, or distinguish reserving from
  committing memory. Instead, we just ask for space, and exploit
  contiguity when we get it.  It is probably possible to do
  better than this on some systems, but no general scheme seems
  to be significantly better.
 
  Management entails a simpler variant of the consolidation scheme
  used for chunks to reduce fragmentation -- new adjacent memory is
  normally prepended or appended to an existing segment. However,
  there are limitations compared to chunk consolidation that mostly
  reflect the fact that segment processing is relatively infrequent
  (occurring only when getting memory from system) and that we
  don't expect to have huge numbers of segments:
 
  * Segments are not indexed, so traversal requires linear scans.  (It
    would be possible to index these, but is not worth the extra
    overhead and complexity for most programs on most platforms.)
  * New segments are only appended to old ones when holding top-most
    memory; if they cannot be prepended to others, they are held in
    different segments.
 
  Except for the top-most segment of an mstate, each segment record
  is kept at the tail of its segment. Segments are added by pushing
  segment records onto the list headed by &mstate.seg for the
  containing mstate.
 
  Segment flags control allocation/merge/deallocation policies:
  * If EXTERN_BIT set, then we did not allocate this segment,
    and so should not try to deallocate or merge with others.
    (This currently holds only for the initial segment passed
    into create_mspace_with_base.)
  * If IS_MMAPPED_BIT set, the segment may be merged with
    other surrounding mmapped segments and trimmed/de-allocated
    using munmap.
  * If neither bit is set, then the segment was obtained using
    MORECORE so can be merged with surrounding MORECORE'd segments
    and deallocated/trimmed using MORECORE with negative arguments.
*/
 
struct malloc_segment {
  char*        base;             /* base address */
  size_t       size;             /* allocated size */
  struct malloc_segment* next;   /* ptr to next segment */
  flag_t       sflags;           /* mmap and extern flag */
};
 
#define is_mmapped_segment(S)  ((S)->sflags & IS_MMAPPED_BIT)
#define is_extern_segment(S)   ((S)->sflags & EXTERN_BIT)
 
typedef struct malloc_segment  msegment;
typedef struct malloc_segment* msegmentptr;
 
/* ---------------------------- malloc_state ----------------------------- */
 
/*
   A malloc_state holds all of the bookkeeping for a space.
   The main fields are:
 
  Top
    The topmost chunk of the currently active segment. Its size is
    cached in topsize.  The actual size of topmost space is
    topsize+TOP_FOOT_SIZE, which includes space reserved for adding
    fenceposts and segment records if necessary when getting more
    space from the system.  The size at which to autotrim top is
    cached from mparams in trim_check, except that it is disabled if
    an autotrim fails.
 
  Designated victim (dv)
    This is the preferred chunk for servicing small requests that
    don't have exact fits.  It is normally the chunk split off most
    recently to service another small request.  Its size is cached in
    dvsize. The link fields of this chunk are not maintained since it
    is not kept in a bin.
 
  SmallBins
    An array of bin headers for free chunks.  These bins hold chunks
    with sizes less than MIN_LARGE_SIZE bytes. Each bin contains
    chunks of all the same size, spaced 8 bytes apart.  To simplify
    use in double-linked lists, each bin header acts as a malloc_chunk
    pointing to the real first node, if it exists (else pointing to
    itself).  This avoids special-casing for headers.  But to avoid
    waste, we allocate only the fd/bk pointers of bins, and then use
    repositioning tricks to treat these as the fields of a chunk.
 
  TreeBins
    Treebins are pointers to the roots of trees holding a range of
    sizes. There are 2 equally spaced treebins for each power of two
    from TREE_SHIFT to TREE_SHIFT+16. The last bin holds anything
    larger.
 
  Bin maps
    There is one bit map for small bins ("smallmap") and one for
    treebins ("treemap).  Each bin sets its bit when non-empty, and
    clears the bit when empty.  Bit operations are then used to avoid
    bin-by-bin searching -- nearly all "search" is done without ever
    looking at bins that won't be selected.  The bit maps
    conservatively use 32 bits per map word, even if on 64bit system.
    For a good description of some of the bit-based techniques used
    here, see Henry S. Warren Jr's book "Hacker's Delight" (and
    supplement at http://hackersdelight.org/). Many of these are
    intended to reduce the branchiness of paths through malloc etc, as
    well as to reduce the number of memory locations read or written.
 
  Segments
    A list of segments headed by an embedded malloc_segment record
    representing the initial space.
 
  Address check support
    The least_addr field is the least address ever obtained from
    MORECORE or MMAP. Attempted frees and reallocs of any address less
    than this are trapped (unless INSECURE is defined).
 
  Magic tag
    A cross-check field that should always hold same value as mparams.magic.
 
  Flags
    Bits recording whether to use MMAP, locks, or contiguous MORECORE
 
  Statistics
    Each space keeps track of current and maximum system memory
    obtained via MORECORE or MMAP.
 
  Locking
    If USE_LOCKS is defined, the "mutex" lock is acquired and released
    around every public call using this mspace.
*/
 
/* Bin types, widths and sizes */
#define NSMALLBINS        (32U)
#define NTREEBINS         (32U)
#define SMALLBIN_SHIFT    (3U)
#define SMALLBIN_WIDTH    (SIZE_T_ONE << SMALLBIN_SHIFT)
#define TREEBIN_SHIFT     (8U)
#define MIN_LARGE_SIZE    (SIZE_T_ONE << TREEBIN_SHIFT)
#define MAX_SMALL_SIZE    (MIN_LARGE_SIZE - SIZE_T_ONE)
#define MAX_SMALL_REQUEST (MAX_SMALL_SIZE - CHUNK_ALIGN_MASK - CHUNK_OVERHEAD)
 
struct malloc_state {
  binmap_t   smallmap;
  binmap_t   treemap;
  size_t     dvsize;
  size_t     topsize;
  char*      least_addr;
  mchunkptr  dv;
  mchunkptr  top;
  size_t     trim_check;
  size_t     magic;
  mchunkptr  smallbins[(NSMALLBINS+1)*2];
  tbinptr    treebins[NTREEBINS];
  size_t     footprint;
  size_t     max_footprint;
  flag_t     mflags;
#if USE_LOCKS
  MLOCK_T    mutex;     /* locate lock among fields that rarely change */
#endif /* USE_LOCKS */
  msegment   seg;
};
 
typedef struct malloc_state*    mstate;
 
/* ------------- Global malloc_state and malloc_params ------------------- */
 
/*
  malloc_params holds global properties, including those that can be
  dynamically set using mallopt. There is a single instance, mparams,
  initialized in init_mparams.
*/
 
struct malloc_params {
  size_t magic;
  size_t page_size;
  size_t granularity;
  size_t mmap_threshold;
  size_t trim_threshold;
  flag_t default_mflags;
};
 
static struct malloc_params mparams;
 
/* The global malloc_state used for all non-"mspace" calls */
static struct malloc_state _gm_;
#define gm                 (&_gm_)
#define is_global(M)       ((M) == &_gm_)
#define is_initialized(M)  ((M)->top != 0)
 
/* -------------------------- system alloc setup ------------------------- */
 
/* Operations on mflags */
 
#define use_lock(M)           ((M)->mflags &   USE_LOCK_BIT)
#define enable_lock(M)        ((M)->mflags |=  USE_LOCK_BIT)
#define disable_lock(M)       ((M)->mflags &= ~USE_LOCK_BIT)
 
#define use_mmap(M)           ((M)->mflags &   USE_MMAP_BIT)
#define enable_mmap(M)        ((M)->mflags |=  USE_MMAP_BIT)
#define disable_mmap(M)       ((M)->mflags &= ~USE_MMAP_BIT)
 
#define use_noncontiguous(M)  ((M)->mflags &   USE_NONCONTIGUOUS_BIT)
#define disable_contiguous(M) ((M)->mflags |=  USE_NONCONTIGUOUS_BIT)
 
#define set_lock(M,L)\
 ((M)->mflags = (L)?\
  ((M)->mflags | USE_LOCK_BIT) :\
  ((M)->mflags & ~USE_LOCK_BIT))
 
/* page-align a size */
#define page_align(S)\
 (((S) + (mparams.page_size)) & ~(mparams.page_size - SIZE_T_ONE))
 
/* granularity-align a size */
#define granularity_align(S)\
  (((S) + (mparams.granularity)) & ~(mparams.granularity - SIZE_T_ONE))
 
#define is_page_aligned(S)\
   (((size_t)(S) & (mparams.page_size - SIZE_T_ONE)) == 0)
#define is_granularity_aligned(S)\
   (((size_t)(S) & (mparams.granularity - SIZE_T_ONE)) == 0)
 
/*  True if segment S holds address A */
#define segment_holds(S, A)\
  ((char*)(A) >= S->base && (char*)(A) < S->base + S->size)
 
/* Return segment holding given address */
static msegmentptr segment_holding(mstate m, char* addr) {
  msegmentptr sp = &m->seg;
  for (;;) {
    if (addr >= sp->base && addr < sp->base + sp->size)
      return sp;
    if ((sp = sp->next) == 0)
      return 0;
  }
}
 
/* Return true if segment contains a segment link */
static int has_segment_link(mstate m, msegmentptr ss) {
  msegmentptr sp = &m->seg;
  for (;;) {
    if ((char*)sp >= ss->base && (char*)sp < ss->base + ss->size)
      return 1;
    if ((sp = sp->next) == 0)
      return 0;
  }
}
 
#ifndef MORECORE_CANNOT_TRIM
#define should_trim(M,s)  ((s) > (M)->trim_check)
#else  /* MORECORE_CANNOT_TRIM */
#define should_trim(M,s)  (0)
#endif /* MORECORE_CANNOT_TRIM */
 
/*
  TOP_FOOT_SIZE is padding at the end of a segment, including space
  that may be needed to place segment records and fenceposts when new
  noncontiguous segments are added.
*/
#define TOP_FOOT_SIZE\
  (align_offset(chunk2mem(0))+pad_request(sizeof(struct malloc_segment))+MIN_CHUNK_SIZE)
 
 
/* -------------------------------  Hooks -------------------------------- */
 
/*
  PREACTION should be defined to return 0 on success, and nonzero on
  failure. If you are not using locking, you can redefine these to do
  anything you like.
*/
 
#if USE_LOCKS
 
/* Ensure locks are initialized */
#define GLOBALLY_INITIALIZE() (mparams.page_size == 0 && init_mparams())
 
#define PREACTION(M)  ((GLOBALLY_INITIALIZE() || use_lock(M))? ACQUIRE_LOCK(&(M)->mutex) : 0)
#define POSTACTION(M) { if (use_lock(M)) RELEASE_LOCK(&(M)->mutex); }
#else /* USE_LOCKS */
 
#ifndef PREACTION
#define PREACTION(M) (0)
#endif  /* PREACTION */
 
#ifndef POSTACTION
#define POSTACTION(M)
#endif  /* POSTACTION */
 
#endif /* USE_LOCKS */
 
/*
  CORRUPTION_ERROR_ACTION is triggered upon detected bad addresses.
  USAGE_ERROR_ACTION is triggered on detected bad frees and
  reallocs. The argument p is an address that might have triggered the
  fault. It is ignored by the two predefined actions, but might be
  useful in custom actions that try to help diagnose errors.
*/
 
#if PROCEED_ON_ERROR
 
/* A count of the number of corruption errors causing resets */
int malloc_corruption_error_count;
 
/* default corruption action */
static void reset_on_error(mstate m);
 
#define CORRUPTION_ERROR_ACTION(m)  reset_on_error(m)
#define USAGE_ERROR_ACTION(m, p)
 
#else /* PROCEED_ON_ERROR */
 
#ifndef CORRUPTION_ERROR_ACTION
#define CORRUPTION_ERROR_ACTION(m) ABORT
#endif /* CORRUPTION_ERROR_ACTION */
 
#ifndef USAGE_ERROR_ACTION
#define USAGE_ERROR_ACTION(m,p) ABORT
#endif /* USAGE_ERROR_ACTION */
 
#endif /* PROCEED_ON_ERROR */
 
/* -------------------------- Debugging setup ---------------------------- */
 
#if ! DEBUG
 
#define check_free_chunk(M,P)
#define check_inuse_chunk(M,P)
#define check_malloced_chunk(M,P,N)
#define check_mmapped_chunk(M,P)
#define check_malloc_state(M)
#define check_top_chunk(M,P)
 
#else /* DEBUG */
#define check_free_chunk(M,P)       do_check_free_chunk(M,P)
#define check_inuse_chunk(M,P)      do_check_inuse_chunk(M,P)
#define check_top_chunk(M,P)        do_check_top_chunk(M,P)
#define check_malloced_chunk(M,P,N) do_check_malloced_chunk(M,P,N)
#define check_mmapped_chunk(M,P)    do_check_mmapped_chunk(M,P)
#define check_malloc_state(M)       do_check_malloc_state(M)
 
static void   do_check_any_chunk(mstate m, mchunkptr p);
static void   do_check_top_chunk(mstate m, mchunkptr p);
static void   do_check_mmapped_chunk(mstate m, mchunkptr p);
static void   do_check_inuse_chunk(mstate m, mchunkptr p);
static void   do_check_free_chunk(mstate m, mchunkptr p);
static void   do_check_malloced_chunk(mstate m, void* mem, size_t s);
static void   do_check_tree(mstate m, tchunkptr t);
static void   do_check_treebin(mstate m, bindex_t i);
static void   do_check_smallbin(mstate m, bindex_t i);
static void   do_check_malloc_state(mstate m);
static int    bin_find(mstate m, mchunkptr x);
static size_t traverse_and_check(mstate m);
#endif /* DEBUG */
 
/* ---------------------------- Indexing Bins ---------------------------- */
 
#define is_small(s)         (((s) >> SMALLBIN_SHIFT) < NSMALLBINS)
#define small_index(s)      ((s)  >> SMALLBIN_SHIFT)
#define small_index2size(i) ((i)  << SMALLBIN_SHIFT)
#define MIN_SMALL_INDEX     (small_index(MIN_CHUNK_SIZE))
 
/* addressing by index. See above about smallbin repositioning */
#define smallbin_at(M, i)   ((sbinptr)((char*)&((M)->smallbins[(i)<<1])))
#define treebin_at(M,i)     (&((M)->treebins[i]))
 
/* assign tree index for size S to variable I */
#if defined(__GNUC__) && defined(i386)
#define compute_tree_index(S, I)\
{\
  size_t X = S >> TREEBIN_SHIFT;\
  if (X == 0)\
    I = 0;\
  else if (X > 0xFFFF)\
    I = NTREEBINS-1;\
  else {\
    unsigned int K;\
    __asm__("bsrl %1,%0\n\t" : "=r" (K) : "rm"  (X));\
    I =  (bindex_t)((K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1)));\
  }\
}
#else /* GNUC */
#define compute_tree_index(S, I)\
{\
  size_t X = S >> TREEBIN_SHIFT;\
  if (X == 0)\
    I = 0;\
  else if (X > 0xFFFF)\
    I = NTREEBINS-1;\
  else {\
    unsigned int Y = (unsigned int)X;\
    unsigned int N = ((Y - 0x100) >> 16) & 8;\
    unsigned int K = (((Y <<= N) - 0x1000) >> 16) & 4;\
    N += K;\
    N += K = (((Y <<= K) - 0x4000) >> 16) & 2;\
    K = 14 - N + ((Y <<= K) >> 15);\
    I = (K << 1) + ((S >> (K + (TREEBIN_SHIFT-1)) & 1));\
  }\
}
#endif /* GNUC */
 
/* Bit representing maximum resolved size in a treebin at i */
#define bit_for_tree_index(i) \
   (i == NTREEBINS-1)? (SIZE_T_BITSIZE-1) : (((i) >> 1) + TREEBIN_SHIFT - 2)
 
/* Shift placing maximum resolved bit in a treebin at i as sign bit */
#define leftshift_for_tree_index(i) \
   ((i == NTREEBINS-1)? 0 : \
    ((SIZE_T_BITSIZE-SIZE_T_ONE) - (((i) >> 1) + TREEBIN_SHIFT - 2)))
 
/* The size of the smallest chunk held in bin with index i */
#define minsize_for_tree_index(i) \
   ((SIZE_T_ONE << (((i) >> 1) + TREEBIN_SHIFT)) |  \
   (((size_t)((i) & SIZE_T_ONE)) << (((i) >> 1) + TREEBIN_SHIFT - 1)))
 
 
/* ------------------------ Operations on bin maps ----------------------- */
 
/* bit corresponding to given index */
#define idx2bit(i)              ((binmap_t)(1) << (i))
 
/* Mark/Clear bits with given index */
#define mark_smallmap(M,i)      ((M)->smallmap |=  idx2bit(i))
#define clear_smallmap(M,i)     ((M)->smallmap &= ~idx2bit(i))
#define smallmap_is_marked(M,i) ((M)->smallmap &   idx2bit(i))
 
#define mark_treemap(M,i)       ((M)->treemap  |=  idx2bit(i))
#define clear_treemap(M,i)      ((M)->treemap  &= ~idx2bit(i))
#define treemap_is_marked(M,i)  ((M)->treemap  &   idx2bit(i))
 
/* index corresponding to given bit */
 
#if defined(__GNUC__) && defined(i386)
#define compute_bit2idx(X, I)\
{\
  unsigned int J;\
  __asm__("bsfl %1,%0\n\t" : "=r" (J) : "rm" (X));\
  I = (bindex_t)J;\
}
 
#else /* GNUC */
#if  USE_BUILTIN_FFS
#define compute_bit2idx(X, I) I = ffs(X)-1
 
#else /* USE_BUILTIN_FFS */
#define compute_bit2idx(X, I)\
{\
  unsigned int Y = X - 1;\
  unsigned int K = Y >> (16-4) & 16;\
  unsigned int N = K;        Y >>= K;\
  N += K = Y >> (8-3) &  8;  Y >>= K;\
  N += K = Y >> (4-2) &  4;  Y >>= K;\
  N += K = Y >> (2-1) &  2;  Y >>= K;\
  N += K = Y >> (1-0) &  1;  Y >>= K;\
  I = (bindex_t)(N + Y);\
}
#endif /* USE_BUILTIN_FFS */
#endif /* GNUC */
 
/* isolate the least set bit of a bitmap */
#define least_bit(x)         ((x) & -(x))
 
/* mask with all bits to left of least bit of x on */
#define left_bits(x)         ((x<<1) | -(x<<1))
 
/* mask with all bits to left of or equal to least bit of x on */
#define same_or_left_bits(x) ((x) | -(x))
 
 
/* ----------------------- Runtime Check Support ------------------------- */
 
/*
  For security, the main invariant is that malloc/free/etc never
  writes to a static address other than malloc_state, unless static
  malloc_state itself has been corrupted, which cannot occur via
  malloc (because of these checks). In essence this means that we
  believe all pointers, sizes, maps etc held in malloc_state, but
  check all of those linked or offsetted from other embedded data
  structures.  These checks are interspersed with main code in a way
  that tends to minimize their run-time cost.
 
  When FOOTERS is defined, in addition to range checking, we also
  verify footer fields of inuse chunks, which can be used guarantee
  that the mstate controlling malloc/free is intact.  This is a
  streamlined version of the approach described by William Robertson
  et al in "Run-time Detection of Heap-based Overflows" LISA'03
  http://www.usenix.org/events/lisa03/tech/robertson.html The footer
  of an inuse chunk holds the xor of its mstate and a random seed,
  that is checked upon calls to free() and realloc().  This is
  (probablistically) unguessable from outside the program, but can be
  computed by any code successfully malloc'ing any chunk, so does not
  itself provide protection against code that has already broken
  security through some other means.  Unlike Robertson et al, we
  always dynamically check addresses of all offset chunks (previous,
  next, etc). This turns out to be cheaper than relying on hashes.
*/
 
#if !INSECURE
/* Check if address a is at least as high as any from MORECORE or MMAP */
#define ok_address(M, a) ((char*)(a) >= (M)->least_addr)
/* Check if address of next chunk n is higher than base chunk p */
#define ok_next(p, n)    ((char*)(p) < (char*)(n))
/* Check if p has its cinuse bit on */
#define ok_cinuse(p)     cinuse(p)
/* Check if p has its pinuse bit on */
#define ok_pinuse(p)     pinuse(p)
 
#else /* !INSECURE */
#define ok_address(M, a) (1)
#define ok_next(b, n)    (1)
#define ok_cinuse(p)     (1)
#define ok_pinuse(p)     (1)
#endif /* !INSECURE */
 
#if (FOOTERS && !INSECURE)
/* Check if (alleged) mstate m has expected magic field */
#define ok_magic(M)      ((M)->magic == mparams.magic)
#else  /* (FOOTERS && !INSECURE) */
#define ok_magic(M)      (1)
#endif /* (FOOTERS && !INSECURE) */
 
 
/* In gcc, use __builtin_expect to minimize impact of checks */
#if !INSECURE
#if defined(__GNUC__) && __GNUC__ >= 3
#define RTCHECK(e)  __builtin_expect(e, 1)
#else /* GNUC */
#define RTCHECK(e)  (e)
#endif /* GNUC */
#else /* !INSECURE */
#define RTCHECK(e)  (1)
#endif /* !INSECURE */
 
/* macros to set up inuse chunks with or without footers */
 
#if !FOOTERS
 
#define mark_inuse_foot(M,p,s)
 
/* Set cinuse bit and pinuse bit of next chunk */
#define set_inuse(M,p,s)\
  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\
  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)
 
/* Set cinuse and pinuse of this chunk and pinuse of next chunk */
#define set_inuse_and_pinuse(M,p,s)\
  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
  ((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT)
 
/* Set size, cinuse and pinuse bit of this chunk */
#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\
  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT))
 
#else /* FOOTERS */
 
/* Set foot of inuse chunk to be xor of mstate and seed */
#define mark_inuse_foot(M,p,s)\
  (((mchunkptr)((char*)(p) + (s)))->prev_foot = ((size_t)(M) ^ mparams.magic))
 
#define get_mstate_for(p)\
  ((mstate)(((mchunkptr)((char*)(p) +\
    (chunksize(p))))->prev_foot ^ mparams.magic))
 
#define set_inuse(M,p,s)\
  ((p)->head = (((p)->head & PINUSE_BIT)|s|CINUSE_BIT),\
  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT), \
  mark_inuse_foot(M,p,s))
 
#define set_inuse_and_pinuse(M,p,s)\
  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
  (((mchunkptr)(((char*)(p)) + (s)))->head |= PINUSE_BIT),\
 mark_inuse_foot(M,p,s))
 
#define set_size_and_pinuse_of_inuse_chunk(M, p, s)\
  ((p)->head = (s|PINUSE_BIT|CINUSE_BIT),\
  mark_inuse_foot(M, p, s))
 
#endif /* !FOOTERS */
 
/* ---------------------------- setting mparams -------------------------- */
 
/* Initialize mparams */
static int init_mparams(void) {
  if (mparams.page_size == 0) {
    size_t s;
 
    mparams.mmap_threshold = DEFAULT_MMAP_THRESHOLD;
    mparams.trim_threshold = DEFAULT_TRIM_THRESHOLD;
#if MORECORE_CONTIGUOUS
    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT;
#else  /* MORECORE_CONTIGUOUS */
    mparams.default_mflags = USE_LOCK_BIT|USE_MMAP_BIT|USE_NONCONTIGUOUS_BIT;
#endif /* MORECORE_CONTIGUOUS */
 
#if (FOOTERS && !INSECURE)
    {
#if USE_DEV_RANDOM
      int fd;
      unsigned char buf[sizeof(size_t)];
      /* Try to use /dev/urandom, else fall back on using time */
      if ((fd = open("/dev/urandom", O_RDONLY)) >= 0 &&
          read(fd, buf, sizeof(buf)) == sizeof(buf)) {
        s = *((size_t *) buf);
        close(fd);
      }
      else
#endif /* USE_DEV_RANDOM */
        s = (size_t)(time(0) ^ (size_t)0x55555555U);
 
      s |= (size_t)8U;    /* ensure nonzero */
      s &= ~(size_t)7U;   /* improve chances of fault for bad values */
 
    }
#else /* (FOOTERS && !INSECURE) */
    s = (size_t)0x58585858U;
#endif /* (FOOTERS && !INSECURE) */
    ACQUIRE_MAGIC_INIT_LOCK();
    if (mparams.magic == 0) {
      mparams.magic = s;
      /* Set up lock for main malloc area */
      INITIAL_LOCK(&gm->mutex);
      gm->mflags = mparams.default_mflags;
    }
    RELEASE_MAGIC_INIT_LOCK();
 
#ifndef WIN32
    mparams.page_size = malloc_getpagesize;
    mparams.granularity = ((DEFAULT_GRANULARITY != 0)?
                           DEFAULT_GRANULARITY : mparams.page_size);
#else /* WIN32 */
    {
      SYSTEM_INFO system_info;
      GetSystemInfo(&system_info);
      mparams.page_size = system_info.dwPageSize;
      mparams.granularity = system_info.dwAllocationGranularity;
    }
#endif /* WIN32 */
 
    /* Sanity-check configuration:
       size_t must be unsigned and as wide as pointer type.
       ints must be at least 4 bytes.
       alignment must be at least 8.
       Alignment, min chunk size, and page size must all be powers of 2.
    */
    if ((sizeof(size_t) != sizeof(char*)) ||
        (MAX_SIZE_T < MIN_CHUNK_SIZE)  ||
        (sizeof(int) < 4)  ||
        (MALLOC_ALIGNMENT < (size_t)8U) ||
        ((MALLOC_ALIGNMENT    & (MALLOC_ALIGNMENT-SIZE_T_ONE))    != 0) ||
        ((MCHUNK_SIZE         & (MCHUNK_SIZE-SIZE_T_ONE))         != 0) ||
        ((mparams.granularity & (mparams.granularity-SIZE_T_ONE)) != 0) ||
        ((mparams.page_size   & (mparams.page_size-SIZE_T_ONE))   != 0))
      ABORT;
  }
  return 0;
}
 
/* support for mallopt */
static int change_mparam(int param_number, int value) {
  size_t val = (size_t)value;
  init_mparams();
  switch(param_number) {
  case M_TRIM_THRESHOLD:
    mparams.trim_threshold = val;
    return 1;
  case M_GRANULARITY:
    if (val >= mparams.page_size && ((val & (val-1)) == 0)) {
      mparams.granularity = val;
      return 1;
    }
    else
      return 0;
  case M_MMAP_THRESHOLD:
    mparams.mmap_threshold = val;
    return 1;
  default:
    return 0;
  }
}
 
#if DEBUG
/* ------------------------- Debugging Support --------------------------- */
 
/* Check properties of any chunk, whether free, inuse, mmapped etc  */
static void do_check_any_chunk(mstate m, mchunkptr p) {
  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
  assert(ok_address(m, p));
}
 
/* Check properties of top chunk */
static void do_check_top_chunk(mstate m, mchunkptr p) {
  msegmentptr sp = segment_holding(m, (char*)p);
  size_t  sz = chunksize(p);
  assert(sp != 0);
  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
  assert(ok_address(m, p));
  assert(sz == m->topsize);
  assert(sz > 0);
  assert(sz == ((sp->base + sp->size) - (char*)p) - TOP_FOOT_SIZE);
  assert(pinuse(p));
  assert(!next_pinuse(p));
}
 
/* Check properties of (inuse) mmapped chunks */
static void do_check_mmapped_chunk(mstate m, mchunkptr p) {
  size_t  sz = chunksize(p);
  size_t len = (sz + (p->prev_foot & ~IS_MMAPPED_BIT) + MMAP_FOOT_PAD);
  assert(is_mmapped(p));
  assert(use_mmap(m));
  assert((is_aligned(chunk2mem(p))) || (p->head == FENCEPOST_HEAD));
  assert(ok_address(m, p));
  assert(!is_small(sz));
  assert((len & (mparams.page_size-SIZE_T_ONE)) == 0);
  assert(chunk_plus_offset(p, sz)->head == FENCEPOST_HEAD);
  assert(chunk_plus_offset(p, sz+SIZE_T_SIZE)->head == 0);
}
 
/* Check properties of inuse chunks */
static void do_check_inuse_chunk(mstate m, mchunkptr p) {
  do_check_any_chunk(m, p);
  assert(cinuse(p));
  assert(next_pinuse(p));
  /* If not pinuse and not mmapped, previous chunk has OK offset */
  assert(is_mmapped(p) || pinuse(p) || next_chunk(prev_chunk(p)) == p);
  if (is_mmapped(p))
    do_check_mmapped_chunk(m, p);
}
 
/* Check properties of free chunks */
static void do_check_free_chunk(mstate m, mchunkptr p) {
  size_t sz = p->head & ~(PINUSE_BIT|CINUSE_BIT);
  mchunkptr next = chunk_plus_offset(p, sz);
  do_check_any_chunk(m, p);
  assert(!cinuse(p));
  assert(!next_pinuse(p));
  assert (!is_mmapped(p));
  if (p != m->dv && p != m->top) {
    if (sz >= MIN_CHUNK_SIZE) {
      assert((sz & CHUNK_ALIGN_MASK) == 0);
      assert(is_aligned(chunk2mem(p)));
      assert(next->prev_foot == sz);
      assert(pinuse(p));
      assert (next == m->top || cinuse(next));
      assert(p->fd->bk == p);
      assert(p->bk->fd == p);
    }
    else  /* markers are always of size SIZE_T_SIZE */
      assert(sz == SIZE_T_SIZE);
  }
}
 
/* Check properties of malloced chunks at the point they are malloced */
static void do_check_malloced_chunk(mstate m, void* mem, size_t s) {
  if (mem != 0) {
    mchunkptr p = mem2chunk(mem);
    size_t sz = p->head & ~(PINUSE_BIT|CINUSE_BIT);
    do_check_inuse_chunk(m, p);
    assert((sz & CHUNK_ALIGN_MASK) == 0);
    assert(sz >= MIN_CHUNK_SIZE);
    assert(sz >= s);
    /* unless mmapped, size is less than MIN_CHUNK_SIZE more than request */
    assert(is_mmapped(p) || sz < (s + MIN_CHUNK_SIZE));
  }
}
 
/* Check a tree and its subtrees.  */
static void do_check_tree(mstate m, tchunkptr t) {
  tchunkptr head = 0;
  tchunkptr u = t;
  bindex_t tindex = t->index;
  size_t tsize = chunksize(t);
  bindex_t idx;
  compute_tree_index(tsize, idx);
  assert(tindex == idx);
  assert(tsize >= MIN_LARGE_SIZE);
  assert(tsize >= minsize_for_tree_index(idx));
  assert((idx == NTREEBINS-1) || (tsize < minsize_for_tree_index((idx+1))));
 
  do { /* traverse through chain of same-sized nodes */
    do_check_any_chunk(m, ((mchunkptr)u));
    assert(u->index == tindex);
    assert(chunksize(u) == tsize);
    assert(!cinuse(u));
    assert(!next_pinuse(u));
    assert(u->fd->bk == u);
    assert(u->bk->fd == u);
    if (u->parent == 0) {
      assert(u->child[0] == 0);
      assert(u->child[1] == 0);
    }
    else {
      assert(head == 0); /* only one node on chain has parent */
      head = u;
      assert(u->parent != u);
      assert (u->parent->child[0] == u ||
              u->parent->child[1] == u ||
              *((tbinptr*)(u->parent)) == u);
      if (u->child[0] != 0) {
        assert(u->child[0]->parent == u);
        assert(u->child[0] != u);
        do_check_tree(m, u->child[0]);
      }
      if (u->child[1] != 0) {
        assert(u->child[1]->parent == u);
        assert(u->child[1] != u);
        do_check_tree(m, u->child[1]);
      }
      if (u->child[0] != 0 && u->child[1] != 0) {
        assert(chunksize(u->child[0]) < chunksize(u->child[1]));
      }
    }
    u = u->fd;
  } while (u != t);
  assert(head != 0);
}
 
/*  Check all the chunks in a treebin.  */
static void do_check_treebin(mstate m, bindex_t i) {
  tbinptr* tb = treebin_at(m, i);
  tchunkptr t = *tb;
  int empty = (m->treemap & (1U << i)) == 0;
  if (t == 0)
    assert(empty);
  if (!empty)
    do_check_tree(m, t);
}
 
/*  Check all the chunks in a smallbin.  */
static void do_check_smallbin(mstate m, bindex_t i) {
  sbinptr b = smallbin_at(m, i);
  mchunkptr p = b->bk;
  unsigned int empty = (m->smallmap & (1U << i)) == 0;
  if (p == b)
    assert(empty);
  if (!empty) {
    for (; p != b; p = p->bk) {
      size_t size = chunksize(p);
      mchunkptr q;
      /* each chunk claims to be free */
      do_check_free_chunk(m, p);
      /* chunk belongs in bin */
      assert(small_index(size) == i);
      assert(p->bk == b || chunksize(p->bk) == chunksize(p));
      /* chunk is followed by an inuse chunk */
      q = next_chunk(p);
      if (q->head != FENCEPOST_HEAD)
        do_check_inuse_chunk(m, q);
    }
  }
}
 
/* Find x in a bin. Used in other check functions. */
static int bin_find(mstate m, mchunkptr x) {
  size_t size = chunksize(x);
  if (is_small(size)) {
    bindex_t sidx = small_index(size);
    sbinptr b = smallbin_at(m, sidx);
    if (smallmap_is_marked(m, sidx)) {
      mchunkptr p = b;
      do {
        if (p == x)
          return 1;
      } while ((p = p->fd) != b);
    }
  }
  else {
    bindex_t tidx;
    compute_tree_index(size, tidx);
    if (treemap_is_marked(m, tidx)) {
      tchunkptr t = *treebin_at(m, tidx);
      size_t sizebits = size << leftshift_for_tree_index(tidx);
      while (t != 0 && chunksize(t) != size) {
        t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];
        sizebits <<= 1;
      }
      if (t != 0) {
        tchunkptr u = t;
        do {
          if (u == (tchunkptr)x)
            return 1;
        } while ((u = u->fd) != t);
      }
    }
  }
  return 0;
}
 
/* Traverse each chunk and check it; return total */
static size_t traverse_and_check(mstate m) {
  size_t sum = 0;
  if (is_initialized(m)) {
    msegmentptr s = &m->seg;
    sum += m->topsize + TOP_FOOT_SIZE;
    while (s != 0) {
      mchunkptr q = align_as_chunk(s->base);
      mchunkptr lastq = 0;
      assert(pinuse(q));
      while (segment_holds(s, q) &&
             q != m->top && q->head != FENCEPOST_HEAD) {
        sum += chunksize(q);
        if (cinuse(q)) {
          assert(!bin_find(m, q));
          do_check_inuse_chunk(m, q);
        }
        else {
          assert(q == m->dv || bin_find(m, q));
          assert(lastq == 0 || cinuse(lastq)); /* Not 2 consecutive free */
          do_check_free_chunk(m, q);
        }
        lastq = q;
        q = next_chunk(q);
      }
      s = s->next;
    }
  }
  return sum;
}
 
/* Check all properties of malloc_state. */
static void do_check_malloc_state(mstate m) {
  bindex_t i;
  size_t total;
  /* check bins */
  for (i = 0; i < NSMALLBINS; ++i)
    do_check_smallbin(m, i);
  for (i = 0; i < NTREEBINS; ++i)
    do_check_treebin(m, i);
 
  if (m->dvsize != 0) { /* check dv chunk */
    do_check_any_chunk(m, m->dv);
    assert(m->dvsize == chunksize(m->dv));
    assert(m->dvsize >= MIN_CHUNK_SIZE);
    assert(bin_find(m, m->dv) == 0);
  }
 
  if (m->top != 0) {   /* check top chunk */
    do_check_top_chunk(m, m->top);
    assert(m->topsize == chunksize(m->top));
    assert(m->topsize > 0);
    assert(bin_find(m, m->top) == 0);
  }
 
  total = traverse_and_check(m);
  assert(total <= m->footprint);
  assert(m->footprint <= m->max_footprint);
}
#endif /* DEBUG */
 
/* ----------------------------- statistics ------------------------------ */
 
#if !NO_MALLINFO
static struct mallinfo internal_mallinfo(mstate m) {
  struct mallinfo nm = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
  if (!PREACTION(m)) {
    check_malloc_state(m);
    if (is_initialized(m)) {
      size_t nfree = SIZE_T_ONE; /* top always free */
      size_t mfree = m->topsize + TOP_FOOT_SIZE;
      size_t sum = mfree;
      msegmentptr s = &m->seg;
      while (s != 0) {
        mchunkptr q = align_as_chunk(s->base);
        while (segment_holds(s, q) &&
               q != m->top && q->head != FENCEPOST_HEAD) {
          size_t sz = chunksize(q);
          sum += sz;
          if (!cinuse(q)) {
            mfree += sz;
            ++nfree;
          }
          q = next_chunk(q);
        }
        s = s->next;
      }
 
      nm.arena    = sum;
      nm.ordblks  = nfree;
      nm.hblkhd   = m->footprint - sum;
      nm.usmblks  = m->max_footprint;
      nm.uordblks = m->footprint - mfree;
      nm.fordblks = mfree;
      nm.keepcost = m->topsize;
    }
 
    POSTACTION(m);
  }
  return nm;
}
#endif /* !NO_MALLINFO */
 
static void internal_malloc_stats(mstate m) {
  if (!PREACTION(m)) {
    size_t maxfp = 0;
    size_t fp = 0;
    size_t used = 0;
    check_malloc_state(m);
    if (is_initialized(m)) {
      msegmentptr s = &m->seg;
      maxfp = m->max_footprint;
      fp = m->footprint;
      used = fp - (m->topsize + TOP_FOOT_SIZE);
 
      while (s != 0) {
        mchunkptr q = align_as_chunk(s->base);
        while (segment_holds(s, q) &&
               q != m->top && q->head != FENCEPOST_HEAD) {
          if (!cinuse(q))
            used -= chunksize(q);
          q = next_chunk(q);
        }
        s = s->next;
      }
    }
 
    fprintf(stderr, "max system bytes = %10lu\n", (unsigned long)(maxfp));
    fprintf(stderr, "system bytes     = %10lu\n", (unsigned long)(fp));
    fprintf(stderr, "in use bytes     = %10lu\n", (unsigned long)(used));
 
    POSTACTION(m);
  }
}
 
/* ----------------------- Operations on smallbins ----------------------- */
 
/*
  Various forms of linking and unlinking are defined as macros.  Even
  the ones for trees, which are very long but have very short typical
  paths.  This is ugly but reduces reliance on inlining support of
  compilers.
*/
 
/* Link a free chunk into a smallbin  */
#define insert_small_chunk(M, P, S) {\
  bindex_t I  = small_index(S);\
  mchunkptr B = smallbin_at(M, I);\
  mchunkptr F = B;\
  assert(S >= MIN_CHUNK_SIZE);\
  if (!smallmap_is_marked(M, I))\
    mark_smallmap(M, I);\
  else if (RTCHECK(ok_address(M, B->fd)))\
    F = B->fd;\
  else {\
    CORRUPTION_ERROR_ACTION(M);\
  }\
  B->fd = P;\
  F->bk = P;\
  P->fd = F;\
  P->bk = B;\
}
 
/* Unlink a chunk from a smallbin  */
#define unlink_small_chunk(M, P, S) {\
  mchunkptr F = P->fd;\
  mchunkptr B = P->bk;\
  bindex_t I = small_index(S);\
  assert(P != B);\
  assert(P != F);\
  assert(chunksize(P) == small_index2size(I));\
  if (F == B)\
    clear_smallmap(M, I);\
  else if (RTCHECK((F == smallbin_at(M,I) || ok_address(M, F)) &&\
                   (B == smallbin_at(M,I) || ok_address(M, B)))) {\
    F->bk = B;\
    B->fd = F;\
  }\
  else {\
    CORRUPTION_ERROR_ACTION(M);\
  }\
}
 
/* Unlink the first chunk from a smallbin */
#define unlink_first_small_chunk(M, B, P, I) {\
  mchunkptr F = P->fd;\
  assert(P != B);\
  assert(P != F);\
  assert(chunksize(P) == small_index2size(I));\
  if (B == F)\
    clear_smallmap(M, I);\
  else if (RTCHECK(ok_address(M, F))) {\
    B->fd = F;\
    F->bk = B;\
  }\
  else {\
    CORRUPTION_ERROR_ACTION(M);\
  }\
}
 
/* Replace dv node, binning the old one */
/* Used only when dvsize known to be small */
#define replace_dv(M, P, S) {\
  size_t DVS = M->dvsize;\
  if (DVS != 0) {\
    mchunkptr DV = M->dv;\
    assert(is_small(DVS));\
    insert_small_chunk(M, DV, DVS);\
  }\
  M->dvsize = S;\
  M->dv = P;\
}
 
/* ------------------------- Operations on trees ------------------------- */
 
/* Insert chunk into tree */
#define insert_large_chunk(M, X, S) {\
  tbinptr* H;\
  bindex_t I;\
  compute_tree_index(S, I);\
  H = treebin_at(M, I);\
  X->index = I;\
  X->child[0] = X->child[1] = 0;\
  if (!treemap_is_marked(M, I)) {\
    mark_treemap(M, I);\
    *H = X;\
    X->parent = (tchunkptr)H;\
    X->fd = X->bk = X;\
  }\
  else {\
    tchunkptr T = *H;\
    size_t K = S << leftshift_for_tree_index(I);\
    for (;;) {\
      if (chunksize(T) != S) {\
        tchunkptr* C = &(T->child[(K >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1]);\
        K <<= 1;\
        if (*C != 0)\
          T = *C;\
        else if (RTCHECK(ok_address(M, C))) {\
          *C = X;\
          X->parent = T;\
          X->fd = X->bk = X;\
          break;\
        }\
        else {\
          CORRUPTION_ERROR_ACTION(M);\
          break;\
        }\
      }\
      else {\
        tchunkptr F = T->fd;\
        if (RTCHECK(ok_address(M, T) && ok_address(M, F))) {\
          T->fd = F->bk = X;\
          X->fd = F;\
          X->bk = T;\
          X->parent = 0;\
          break;\
        }\
        else {\
          CORRUPTION_ERROR_ACTION(M);\
          break;\
        }\
      }\
    }\
  }\
}
 
/*
  Unlink steps:
 
  1. If x is a chained node, unlink it from its same-sized fd/bk links
     and choose its bk node as its replacement.
  2. If x was the last node of its size, but not a leaf node, it must
     be replaced with a leaf node (not merely one with an open left or
     right), to make sure that lefts and rights of descendents
     correspond properly to bit masks.  We use the rightmost descendent
     of x.  We could use any other leaf, but this is easy to locate and
     tends to counteract removal of leftmosts elsewhere, and so keeps
     paths shorter than minimally guaranteed.  This doesn't loop much
     because on average a node in a tree is near the bottom.
  3. If x is the base of a chain (i.e., has parent links) relink
     x's parent and children to x's replacement (or null if none).
*/
 
#define unlink_large_chunk(M, X) {\
  tchunkptr XP = X->parent;\
  tchunkptr R;\
  if (X->bk != X) {\
    tchunkptr F = X->fd;\
    R = X->bk;\
    if (RTCHECK(ok_address(M, F))) {\
      F->bk = R;\
      R->fd = F;\
    }\
    else {\
      CORRUPTION_ERROR_ACTION(M);\
    }\
  }\
  else {\
    tchunkptr* RP;\
    if (((R = *(RP = &(X->child[1]))) != 0) ||\
        ((R = *(RP = &(X->child[0]))) != 0)) {\
      tchunkptr* CP;\
      while ((*(CP = &(R->child[1])) != 0) ||\
             (*(CP = &(R->child[0])) != 0)) {\
        R = *(RP = CP);\
      }\
      if (RTCHECK(ok_address(M, RP)))\
        *RP = 0;\
      else {\
        CORRUPTION_ERROR_ACTION(M);\
      }\
    }\
  }\
  if (XP != 0) {\
    tbinptr* H = treebin_at(M, X->index);\
    if (X == *H) {\
      if ((*H = R) == 0) \
        clear_treemap(M, X->index);\
    }\
    else if (RTCHECK(ok_address(M, XP))) {\
      if (XP->child[0] == X) \
        XP->child[0] = R;\
      else \
        XP->child[1] = R;\
    }\
    else\
      CORRUPTION_ERROR_ACTION(M);\
    if (R != 0) {\
      if (RTCHECK(ok_address(M, R))) {\
        tchunkptr C0, C1;\
        R->parent = XP;\
        if ((C0 = X->child[0]) != 0) {\
          if (RTCHECK(ok_address(M, C0))) {\
            R->child[0] = C0;\
            C0->parent = R;\
          }\
          else\
            CORRUPTION_ERROR_ACTION(M);\
        }\
        if ((C1 = X->child[1]) != 0) {\
          if (RTCHECK(ok_address(M, C1))) {\
            R->child[1] = C1;\
            C1->parent = R;\
          }\
          else\
            CORRUPTION_ERROR_ACTION(M);\
        }\
      }\
      else\
        CORRUPTION_ERROR_ACTION(M);\
    }\
  }\
}
 
/* Relays to large vs small bin operations */
 
#define insert_chunk(M, P, S)\
  if (is_small(S)) insert_small_chunk(M, P, S)\
  else { tchunkptr TP = (tchunkptr)(P); insert_large_chunk(M, TP, S); }
 
#define unlink_chunk(M, P, S)\
  if (is_small(S)) unlink_small_chunk(M, P, S)\
  else { tchunkptr TP = (tchunkptr)(P); unlink_large_chunk(M, TP); }
 
 
/* Relays to internal calls to malloc/free from realloc, memalign etc */
 
#if ONLY_MSPACES
#define internal_malloc(m, b) mspace_malloc(m, b)
#define internal_free(m, mem) mspace_free(m,mem);
#else /* ONLY_MSPACES */
#if MSPACES
#define internal_malloc(m, b)\
   (m == gm)? dlmalloc(b) : mspace_malloc(m, b)
#define internal_free(m, mem)\
   if (m == gm) dlfree(mem); else mspace_free(m,mem);
#else /* MSPACES */
#define internal_malloc(m, b) dlmalloc(b)
#define internal_free(m, mem) dlfree(mem)
#endif /* MSPACES */
#endif /* ONLY_MSPACES */
 
/* -----------------------  Direct-mmapping chunks ----------------------- */
 
/*
  Directly mmapped chunks are set up with an offset to the start of
  the mmapped region stored in the prev_foot field of the chunk. This
  allows reconstruction of the required argument to MUNMAP when freed,
  and also allows adjustment of the returned chunk to meet alignment
  requirements (especially in memalign).  There is also enough space
  allocated to hold a fake next chunk of size SIZE_T_SIZE to maintain
  the PINUSE bit so frees can be checked.
*/
 
/* Malloc using mmap */
static void* mmap_alloc(mstate m, size_t nb) {
  size_t mmsize = granularity_align(nb + SIX_SIZE_T_SIZES + CHUNK_ALIGN_MASK);
  if (mmsize > nb) {     /* Check for wrap around 0 */
    char* mm = (char*)(DIRECT_MMAP(mmsize));
    if (mm != CMFAIL) {
      size_t offset = align_offset(chunk2mem(mm));
      size_t psize = mmsize - offset - MMAP_FOOT_PAD;
      mchunkptr p = (mchunkptr)(mm + offset);
      p->prev_foot = offset | IS_MMAPPED_BIT;
      (p)->head = (psize|CINUSE_BIT);
      mark_inuse_foot(m, p, psize);
      chunk_plus_offset(p, psize)->head = FENCEPOST_HEAD;
      chunk_plus_offset(p, psize+SIZE_T_SIZE)->head = 0;
 
      if (mm < m->least_addr)
        m->least_addr = mm;
      if ((m->footprint += mmsize) > m->max_footprint)
        m->max_footprint = m->footprint;
      assert(is_aligned(chunk2mem(p)));
      check_mmapped_chunk(m, p);
      return chunk2mem(p);
    }
  }
  return 0;
}
 
/* Realloc using mmap */
static mchunkptr mmap_resize(mstate m, mchunkptr oldp, size_t nb) {
  size_t oldsize = chunksize(oldp);
  if (is_small(nb)) /* Can't shrink mmap regions below small size */
    return 0;
  /* Keep old chunk if big enough but not too big */
  if (oldsize >= nb + SIZE_T_SIZE &&
      (oldsize - nb) <= (mparams.granularity << 1))
    return oldp;
  else {
    size_t offset = oldp->prev_foot & ~IS_MMAPPED_BIT;
    size_t oldmmsize = oldsize + offset + MMAP_FOOT_PAD;
    size_t newmmsize = granularity_align(nb + SIX_SIZE_T_SIZES +
                                         CHUNK_ALIGN_MASK);
    char* cp = (char*)CALL_MREMAP((char*)oldp - offset,
                                  oldmmsize, newmmsize, 1);
    if (cp != CMFAIL) {
      mchunkptr newp = (mchunkptr)(cp + offset);
      size_t psize = newmmsize - offset - MMAP_FOOT_PAD;
      newp->head = (psize|CINUSE_BIT);
      mark_inuse_foot(m, newp, psize);
      chunk_plus_offset(newp, psize)->head = FENCEPOST_HEAD;
      chunk_plus_offset(newp, psize+SIZE_T_SIZE)->head = 0;
 
      if (cp < m->least_addr)
        m->least_addr = cp;
      if ((m->footprint += newmmsize - oldmmsize) > m->max_footprint)
        m->max_footprint = m->footprint;
      check_mmapped_chunk(m, newp);
      return newp;
    }
  }
  return 0;
}
 
/* -------------------------- mspace management -------------------------- */
 
/* Initialize top chunk and its size */
static void init_top(mstate m, mchunkptr p, size_t psize) {
  /* Ensure alignment */
  size_t offset = align_offset(chunk2mem(p));
  p = (mchunkptr)((char*)p + offset);
  psize -= offset;
 
  m->top = p;
  m->topsize = psize;
  p->head = psize | PINUSE_BIT;
  /* set size of fake trailing chunk holding overhead space only once */
  chunk_plus_offset(p, psize)->head = TOP_FOOT_SIZE;
  m->trim_check = mparams.trim_threshold; /* reset on each update */
}
 
/* Initialize bins for a new mstate that is otherwise zeroed out */
static void init_bins(mstate m) {
  /* Establish circular links for smallbins */
  bindex_t i;
  for (i = 0; i < NSMALLBINS; ++i) {
    sbinptr bin = smallbin_at(m,i);
    bin->fd = bin->bk = bin;
  }
}
 
#if PROCEED_ON_ERROR
 
/* default corruption action */
static void reset_on_error(mstate m) {
  int i;
  ++malloc_corruption_error_count;
  /* Reinitialize fields to forget about all memory */
  m->smallbins = m->treebins = 0;
  m->dvsize = m->topsize = 0;
  m->seg.base = 0;
  m->seg.size = 0;
  m->seg.next = 0;
  m->top = m->dv = 0;
  for (i = 0; i < NTREEBINS; ++i)
    *treebin_at(m, i) = 0;
  init_bins(m);
}
#endif /* PROCEED_ON_ERROR */
 
/* Allocate chunk and prepend remainder with chunk in successor base. */
static void* prepend_alloc(mstate m, char* newbase, char* oldbase,
                           size_t nb) {
  mchunkptr p = align_as_chunk(newbase);
  mchunkptr oldfirst = align_as_chunk(oldbase);
  size_t psize = (char*)oldfirst - (char*)p;
  mchunkptr q = chunk_plus_offset(p, nb);
  size_t qsize = psize - nb;
  set_size_and_pinuse_of_inuse_chunk(m, p, nb);
 
  assert((char*)oldfirst > (char*)q);
  assert(pinuse(oldfirst));
  assert(qsize >= MIN_CHUNK_SIZE);
 
  /* consolidate remainder with first chunk of old base */
  if (oldfirst == m->top) {
    size_t tsize = m->topsize += qsize;
    m->top = q;
    q->head = tsize | PINUSE_BIT;
    check_top_chunk(m, q);
  }
  else if (oldfirst == m->dv) {
    size_t dsize = m->dvsize += qsize;
    m->dv = q;
    set_size_and_pinuse_of_free_chunk(q, dsize);
  }
  else {
    if (!cinuse(oldfirst)) {
      size_t nsize = chunksize(oldfirst);
      unlink_chunk(m, oldfirst, nsize);
      oldfirst = chunk_plus_offset(oldfirst, nsize);
      qsize += nsize;
    }
    set_free_with_pinuse(q, qsize, oldfirst);
    insert_chunk(m, q, qsize);
    check_free_chunk(m, q);
  }
 
  check_malloced_chunk(m, chunk2mem(p), nb);
  return chunk2mem(p);
}
 
 
/* Add a segment to hold a new noncontiguous region */
static void add_segment(mstate m, char* tbase, size_t tsize, flag_t mmapped) {
  /* Determine locations and sizes of segment, fenceposts, old top */
  char* old_top = (char*)m->top;
  msegmentptr oldsp = segment_holding(m, old_top);
  char* old_end = oldsp->base + oldsp->size;
  size_t ssize = pad_request(sizeof(struct malloc_segment));
  char* rawsp = old_end - (ssize + FOUR_SIZE_T_SIZES + CHUNK_ALIGN_MASK);
  size_t offset = align_offset(chunk2mem(rawsp));
  char* asp = rawsp + offset;
  char* csp = (asp < (old_top + MIN_CHUNK_SIZE))? old_top : asp;
  mchunkptr sp = (mchunkptr)csp;
  msegmentptr ss = (msegmentptr)(chunk2mem(sp));
  mchunkptr tnext = chunk_plus_offset(sp, ssize);
  mchunkptr p = tnext;
  int nfences = 0;
 
  /* reset top to new space */
  init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);
 
  /* Set up segment record */
  assert(is_aligned(ss));
  set_size_and_pinuse_of_inuse_chunk(m, sp, ssize);
  *ss = m->seg; /* Push current record */
  m->seg.base = tbase;
  m->seg.size = tsize;
  m->seg.sflags = mmapped;
  m->seg.next = ss;
 
  /* Insert trailing fenceposts */
  for (;;) {
    mchunkptr nextp = chunk_plus_offset(p, SIZE_T_SIZE);
    p->head = FENCEPOST_HEAD;
    ++nfences;
    if ((char*)(&(nextp->head)) < old_end)
      p = nextp;
    else
      break;
  }
  assert(nfences >= 2);
 
  /* Insert the rest of old top into a bin as an ordinary free chunk */
  if (csp != old_top) {
    mchunkptr q = (mchunkptr)old_top;
    size_t psize = csp - old_top;
    mchunkptr tn = chunk_plus_offset(q, psize);
    set_free_with_pinuse(q, psize, tn);
    insert_chunk(m, q, psize);
  }
 
  check_top_chunk(m, m->top);
}
 
/* -------------------------- System allocation -------------------------- */
 
/* Get memory from system using MORECORE or MMAP */
static void* sys_alloc(mstate m, size_t nb) {
  char* tbase = CMFAIL;
  size_t tsize = 0;
  flag_t mmap_flag = 0;
 
  init_mparams();
 
  /* Directly map large chunks */
  if (use_mmap(m) && nb >= mparams.mmap_threshold) {
    void* mem = mmap_alloc(m, nb);
    if (mem != 0)
      return mem;
  }
 
  /*
    Try getting memory in any of three ways (in most-preferred to
    least-preferred order):
    1. A call to MORECORE that can normally contiguously extend memory.
       (disabled if not MORECORE_CONTIGUOUS or not HAVE_MORECORE or
       or main space is mmapped or a previous contiguous call failed)
    2. A call to MMAP new space (disabled if not HAVE_MMAP).
       Note that under the default settings, if MORECORE is unable to
       fulfill a request, and HAVE_MMAP is true, then mmap is
       used as a noncontiguous system allocator. This is a useful backup
       strategy for systems with holes in address spaces -- in this case
       sbrk cannot contiguously expand the heap, but mmap may be able to
       find space.
    3. A call to MORECORE that cannot usually contiguously extend memory.
       (disabled if not HAVE_MORECORE)
  */
 
  if (MORECORE_CONTIGUOUS && !use_noncontiguous(m)) {
    char* br = CMFAIL;
    msegmentptr ss = (m->top == 0)? 0 : segment_holding(m, (char*)m->top);
    size_t asize = 0;
    ACQUIRE_MORECORE_LOCK();
 
    if (ss == 0) {  /* First time through or recovery */
      char* base = (char*)CALL_MORECORE(0);
      if (base != CMFAIL) {
        asize = granularity_align(nb + TOP_FOOT_SIZE + SIZE_T_ONE);
        /* Adjust to end on a page boundary */
        if (!is_page_aligned(base))
          asize += (page_align((size_t)base) - (size_t)base);
        /* Can't call MORECORE if size is negative when treated as signed */
        if (asize < HALF_MAX_SIZE_T &&
            (br = (char*)(CALL_MORECORE(asize))) == base) {
          tbase = base;
          tsize = asize;
        }
      }
    }
    else {
      /* Subtract out existing available top space from MORECORE request. */
      asize = granularity_align(nb - m->topsize + TOP_FOOT_SIZE + SIZE_T_ONE);
      /* Use mem here only if it did continuously extend old space */
      if (asize < HALF_MAX_SIZE_T &&
          (br = (char*)(CALL_MORECORE(asize))) == ss->base+ss->size) {
        tbase = br;
        tsize = asize;
      }
    }
 
    if (tbase == CMFAIL) {    /* Cope with partial failure */
      if (br != CMFAIL) {    /* Try to use/extend the space we did get */
        if (asize < HALF_MAX_SIZE_T &&
            asize < nb + TOP_FOOT_SIZE + SIZE_T_ONE) {
          size_t esize = granularity_align(nb + TOP_FOOT_SIZE + SIZE_T_ONE - asize);
          if (esize < HALF_MAX_SIZE_T) {
            char* end = (char*)CALL_MORECORE(esize);
            if (end != CMFAIL)
              asize += esize;
            else {            /* Can't use; try to release */
              CALL_MORECORE(-asize);
              br = CMFAIL;
            }
          }
        }
      }
      if (br != CMFAIL) {    /* Use the space we did get */
        tbase = br;
        tsize = asize;
      }
      else
        disable_contiguous(m); /* Don't try contiguous path in the future */
    }
 
    RELEASE_MORECORE_LOCK();
  }
 
  if (HAVE_MMAP && tbase == CMFAIL) {  /* Try MMAP */
    size_t req = nb + TOP_FOOT_SIZE + SIZE_T_ONE;
    size_t rsize = granularity_align(req);
    if (rsize > nb) { /* Fail if wraps around zero */
      char* mp = (char*)(CALL_MMAP(rsize));
      if (mp != CMFAIL) {
        tbase = mp;
        tsize = rsize;
        mmap_flag = IS_MMAPPED_BIT;
      }
    }
  }
 
  if (HAVE_MORECORE && tbase == CMFAIL) { /* Try noncontiguous MORECORE */
    size_t asize = granularity_align(nb + TOP_FOOT_SIZE + SIZE_T_ONE);
    if (asize < HALF_MAX_SIZE_T) {
      char* br = CMFAIL;
      char* end = CMFAIL;
      ACQUIRE_MORECORE_LOCK();
      br = (char*)(CALL_MORECORE(asize));
      end = (char*)(CALL_MORECORE(0));
      RELEASE_MORECORE_LOCK();
      if (br != CMFAIL && end != CMFAIL && br < end) {
        size_t ssize = end - br;
        if (ssize > nb + TOP_FOOT_SIZE) {
          tbase = br;
          tsize = ssize;
        }
      }
    }
  }
 
  if (tbase != CMFAIL) {
 
    if ((m->footprint += tsize) > m->max_footprint)
      m->max_footprint = m->footprint;
 
    if (!is_initialized(m)) { /* first-time initialization */
      m->seg.base = m->least_addr = tbase;
      m->seg.size = tsize;
      m->seg.sflags = mmap_flag;
      m->magic = mparams.magic;
      init_bins(m);
      if (is_global(m))
        init_top(m, (mchunkptr)tbase, tsize - TOP_FOOT_SIZE);
      else {
        /* Offset top by embedded malloc_state */
        mchunkptr mn = next_chunk(mem2chunk(m));
        init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) -TOP_FOOT_SIZE);
      }
    }
 
    else {
      /* Try to merge with an existing segment */
      msegmentptr sp = &m->seg;
      while (sp != 0 && tbase != sp->base + sp->size)
        sp = sp->next;
      if (sp != 0 &&
          !is_extern_segment(sp) &&
          (sp->sflags & IS_MMAPPED_BIT) == mmap_flag &&
          segment_holds(sp, m->top)) { /* append */
        sp->size += tsize;
        init_top(m, m->top, m->topsize + tsize);
      }
      else {
        if (tbase < m->least_addr)
          m->least_addr = tbase;
        sp = &m->seg;
        while (sp != 0 && sp->base != tbase + tsize)
          sp = sp->next;
        if (sp != 0 &&
            !is_extern_segment(sp) &&
            (sp->sflags & IS_MMAPPED_BIT) == mmap_flag) {
          char* oldbase = sp->base;
          sp->base = tbase;
          sp->size += tsize;
          return prepend_alloc(m, tbase, oldbase, nb);
        }
        else
          add_segment(m, tbase, tsize, mmap_flag);
      }
    }
 
    if (nb < m->topsize) { /* Allocate from new or extended top space */
      size_t rsize = m->topsize -= nb;
      mchunkptr p = m->top;
      mchunkptr r = m->top = chunk_plus_offset(p, nb);
      r->head = rsize | PINUSE_BIT;
      set_size_and_pinuse_of_inuse_chunk(m, p, nb);
      check_top_chunk(m, m->top);
      check_malloced_chunk(m, chunk2mem(p), nb);
      return chunk2mem(p);
    }
  }
 
  MALLOC_FAILURE_ACTION;
  return 0;
}
 
/* -----------------------  system deallocation -------------------------- */
 
/* Unmap and unlink any mmapped segments that don't contain used chunks */
static size_t release_unused_segments(mstate m) {
  size_t released = 0;
  msegmentptr pred = &m->seg;
  msegmentptr sp = pred->next;
  while (sp != 0) {
    char* base = sp->base;
    size_t size = sp->size;
    msegmentptr next = sp->next;
    if (is_mmapped_segment(sp) && !is_extern_segment(sp)) {
      mchunkptr p = align_as_chunk(base);
      size_t psize = chunksize(p);
      /* Can unmap if first chunk holds entire segment and not pinned */
      if (!cinuse(p) && (char*)p + psize >= base + size - TOP_FOOT_SIZE) {
        tchunkptr tp = (tchunkptr)p;
        assert(segment_holds(sp, (char*)sp));
        if (p == m->dv) {
          m->dv = 0;
          m->dvsize = 0;
        }
        else {
          unlink_large_chunk(m, tp);
        }
        if (CALL_MUNMAP(base, size) == 0) {
          released += size;
          m->footprint -= size;
          /* unlink obsoleted record */
          sp = pred;
          sp->next = next;
        }
        else { /* back out if cannot unmap */
          insert_large_chunk(m, tp, psize);
        }
      }
    }
    pred = sp;
    sp = next;
  }
  return released;
}
 
static int sys_trim(mstate m, size_t pad) {
  size_t released = 0;
  if (pad < MAX_REQUEST && is_initialized(m)) {
    pad += TOP_FOOT_SIZE; /* ensure enough room for segment overhead */
 
    if (m->topsize > pad) {
      /* Shrink top space in granularity-size units, keeping at least one */
      size_t unit = mparams.granularity;
      size_t extra = ((m->topsize - pad + (unit - SIZE_T_ONE)) / unit -
                      SIZE_T_ONE) * unit;
      msegmentptr sp = segment_holding(m, (char*)m->top);
 
      if (!is_extern_segment(sp)) {
        if (is_mmapped_segment(sp)) {
          if (HAVE_MMAP &&
              sp->size >= extra &&
              !has_segment_link(m, sp)) { /* can't shrink if pinned */
            size_t newsize = sp->size - extra;
            /* Prefer mremap, fall back to munmap */
            if ((CALL_MREMAP(sp->base, sp->size, newsize, 0) != MFAIL) ||
                (CALL_MUNMAP(sp->base + newsize, extra) == 0)) {
              released = extra;
            }
          }
        }
        else if (HAVE_MORECORE) {
          if (extra >= HALF_MAX_SIZE_T) /* Avoid wrapping negative */
            extra = (HALF_MAX_SIZE_T) + SIZE_T_ONE - unit;
          ACQUIRE_MORECORE_LOCK();
          {
            /* Make sure end of memory is where we last set it. */
            char* old_br = (char*)(CALL_MORECORE(0));
            if (old_br == sp->base + sp->size) {
              char* rel_br = (char*)(CALL_MORECORE(-extra));
              char* new_br = (char*)(CALL_MORECORE(0));
              if (rel_br != CMFAIL && new_br < old_br)
                released = old_br - new_br;
            }
          }
          RELEASE_MORECORE_LOCK();
        }
      }
 
      if (released != 0) {
        sp->size -= released;
        m->footprint -= released;
        init_top(m, m->top, m->topsize - released);
        check_top_chunk(m, m->top);
      }
    }
 
    /* Unmap any unused mmapped segments */
    if (HAVE_MMAP)
      released += release_unused_segments(m);
 
    /* On failure, disable autotrim to avoid repeated failed future calls */
    if (released == 0)
      m->trim_check = MAX_SIZE_T;
  }
 
  return (released != 0)? 1 : 0;
}
 
/* ---------------------------- malloc support --------------------------- */
 
/* allocate a large request from the best fitting chunk in a treebin */
static void* tmalloc_large(mstate m, size_t nb) {
  tchunkptr v = 0;
  size_t rsize = -nb; /* Unsigned negation */
  tchunkptr t;
  bindex_t idx;
  compute_tree_index(nb, idx);
 
  if ((t = *treebin_at(m, idx)) != 0) {
    /* Traverse tree for this bin looking for node with size == nb */
    size_t sizebits = nb << leftshift_for_tree_index(idx);
    tchunkptr rst = 0;  /* The deepest untaken right subtree */
    for (;;) {
      tchunkptr rt;
      size_t trem = chunksize(t) - nb;
      if (trem < rsize) {
        v = t;
        if ((rsize = trem) == 0)
          break;
      }
      rt = t->child[1];
      t = t->child[(sizebits >> (SIZE_T_BITSIZE-SIZE_T_ONE)) & 1];
      if (rt != 0 && rt != t)
        rst = rt;
      if (t == 0) {
        t = rst; /* set t to least subtree holding sizes > nb */
        break;
      }
      sizebits <<= 1;
    }
  }
 
  if (t == 0 && v == 0) { /* set t to root of next non-empty treebin */
    binmap_t leftbits = left_bits(idx2bit(idx)) & m->treemap;
    if (leftbits != 0) {
      bindex_t i;
      binmap_t leastbit = least_bit(leftbits);
      compute_bit2idx(leastbit, i);
      t = *treebin_at(m, i);
    }
  }
 
  while (t != 0) { /* find smallest of tree or subtree */
    size_t trem = chunksize(t) - nb;
    if (trem < rsize) {
      rsize = trem;
      v = t;
    }
    t = leftmost_child(t);
  }
 
  /*  If dv is a better fit, return 0 so malloc will use it */
  if (v != 0 && rsize < (size_t)(m->dvsize - nb)) {
    if (RTCHECK(ok_address(m, v))) { /* split */
      mchunkptr r = chunk_plus_offset(v, nb);
      assert(chunksize(v) == rsize + nb);
      if (RTCHECK(ok_next(v, r))) {
        unlink_large_chunk(m, v);
        if (rsize < MIN_CHUNK_SIZE)
          set_inuse_and_pinuse(m, v, (rsize + nb));
        else {
          set_size_and_pinuse_of_inuse_chunk(m, v, nb);
          set_size_and_pinuse_of_free_chunk(r, rsize);
          insert_chunk(m, r, rsize);
        }
        return chunk2mem(v);
      }
    }
    CORRUPTION_ERROR_ACTION(m);
  }
  return 0;
}
 
/* allocate a small request from the best fitting chunk in a treebin */
static void* tmalloc_small(mstate m, size_t nb) {
  tchunkptr t, v;
  size_t rsize;
  bindex_t i;
  binmap_t leastbit = least_bit(m->treemap);
  compute_bit2idx(leastbit, i);
 
  v = t = *treebin_at(m, i);
  rsize = chunksize(t) - nb;
 
  while ((t = leftmost_child(t)) != 0) {
    size_t trem = chunksize(t) - nb;
    if (trem < rsize) {
      rsize = trem;
      v = t;
    }
  }
 
  if (RTCHECK(ok_address(m, v))) {
    mchunkptr r = chunk_plus_offset(v, nb);
    assert(chunksize(v) == rsize + nb);
    if (RTCHECK(ok_next(v, r))) {
      unlink_large_chunk(m, v);
      if (rsize < MIN_CHUNK_SIZE)
        set_inuse_and_pinuse(m, v, (rsize + nb));
      else {
        set_size_and_pinuse_of_inuse_chunk(m, v, nb);
        set_size_and_pinuse_of_free_chunk(r, rsize);
        replace_dv(m, r, rsize);
      }
      return chunk2mem(v);
    }
  }
 
  CORRUPTION_ERROR_ACTION(m);
  return 0;
}
 
/* --------------------------- realloc support --------------------------- */
 
static void* internal_realloc(mstate m, void* oldmem, size_t bytes) {
  if (bytes >= MAX_REQUEST) {
    MALLOC_FAILURE_ACTION;
    return 0;
  }
  if (!PREACTION(m)) {
    mchunkptr oldp = mem2chunk(oldmem);
    size_t oldsize = chunksize(oldp);
    mchunkptr next = chunk_plus_offset(oldp, oldsize);
    mchunkptr newp = 0;
    void* extra = 0;
 
    /* Try to either shrink or extend into top. Else malloc-copy-free */
 
    if (RTCHECK(ok_address(m, oldp) && ok_cinuse(oldp) &&
                ok_next(oldp, next) && ok_pinuse(next))) {
      size_t nb = request2size(bytes);
      if (is_mmapped(oldp))
        newp = mmap_resize(m, oldp, nb);
      else if (oldsize >= nb) { /* already big enough */
        size_t rsize = oldsize - nb;
        newp = oldp;
        if (rsize >= MIN_CHUNK_SIZE) {
          mchunkptr remainder = chunk_plus_offset(newp, nb);
          set_inuse(m, newp, nb);
          set_inuse(m, remainder, rsize);
          extra = chunk2mem(remainder);
        }
      }
      else if (next == m->top && oldsize + m->topsize > nb) {
        /* Expand into top */
        size_t newsize = oldsize + m->topsize;
        size_t newtopsize = newsize - nb;
        mchunkptr newtop = chunk_plus_offset(oldp, nb);
        set_inuse(m, oldp, nb);
        newtop->head = newtopsize |PINUSE_BIT;
        m->top = newtop;
        m->topsize = newtopsize;
        newp = oldp;
      }
    }
    else {
      USAGE_ERROR_ACTION(m, oldmem);
      POSTACTION(m);
      return 0;
    }
 
    POSTACTION(m);
 
    if (newp != 0) {
      if (extra != 0) {
        internal_free(m, extra);
      }
      check_inuse_chunk(m, newp);
      return chunk2mem(newp);
    }
    else {
      void* newmem = internal_malloc(m, bytes);
      if (newmem != 0) {
        size_t oc = oldsize - overhead_for(oldp);
        memcpy(newmem, oldmem, (oc < bytes)? oc : bytes);
        internal_free(m, oldmem);
      }
      return newmem;
    }
  }
  return 0;
}
 
/* --------------------------- memalign support -------------------------- */
 
static void* internal_memalign(mstate m, size_t alignment, size_t bytes) {
  if (alignment <= MALLOC_ALIGNMENT)    /* Can just use malloc */
    return internal_malloc(m, bytes);
  if (alignment <  MIN_CHUNK_SIZE) /* must be at least a minimum chunk size */
    alignment = MIN_CHUNK_SIZE;
  if ((alignment & (alignment-SIZE_T_ONE)) != 0) {/* Ensure a power of 2 */
    size_t a = MALLOC_ALIGNMENT << 1;
    while (a < alignment) a <<= 1;
    alignment = a;
  }
 
  if (bytes >= MAX_REQUEST - alignment) {
    if (m != 0)  { /* Test isn't needed but avoids compiler warning */
      MALLOC_FAILURE_ACTION;
    }
  }
  else {
    size_t nb = request2size(bytes);
    size_t req = nb + alignment + MIN_CHUNK_SIZE - CHUNK_OVERHEAD;
    char* mem = (char*)internal_malloc(m, req);
    if (mem != 0) {
      void* leader = 0;
      void* trailer = 0;
      mchunkptr p = mem2chunk(mem);
 
      if (PREACTION(m)) return 0;
      if ((((size_t)(mem)) % alignment) != 0) { /* misaligned */
        /*
          Find an aligned spot inside chunk.  Since we need to give
          back leading space in a chunk of at least MIN_CHUNK_SIZE, if
          the first calculation places us at a spot with less than
          MIN_CHUNK_SIZE leader, we can move to the next aligned spot.
          We've allocated enough total room so that this is always
          possible.
        */
        char* br = (char*)mem2chunk((size_t)(((size_t)(mem +
                                                       alignment -
                                                       SIZE_T_ONE)) &
                                             -alignment));
        char* pos = ((size_t)(br - (char*)(p)) >= MIN_CHUNK_SIZE)?
          br : br+alignment;
        mchunkptr newp = (mchunkptr)pos;
        size_t leadsize = pos - (char*)(p);
        size_t newsize = chunksize(p) - leadsize;
 
        if (is_mmapped(p)) { /* For mmapped chunks, just adjust offset */
          newp->prev_foot = p->prev_foot + leadsize;
          newp->head = (newsize|CINUSE_BIT);
        }
        else { /* Otherwise, give back leader, use the rest */
          set_inuse(m, newp, newsize);
          set_inuse(m, p, leadsize);
          leader = chunk2mem(p);
        }
        p = newp;
      }
 
      /* Give back spare room at the end */
      if (!is_mmapped(p)) {
        size_t size = chunksize(p);
        if (size > nb + MIN_CHUNK_SIZE) {
          size_t remainder_size = size - nb;
          mchunkptr remainder = chunk_plus_offset(p, nb);
          set_inuse(m, p, nb);
          set_inuse(m, remainder, remainder_size);
          trailer = chunk2mem(remainder);
        }
      }
 
      assert (chunksize(p) >= nb);
      assert((((size_t)(chunk2mem(p))) % alignment) == 0);
      check_inuse_chunk(m, p);
      POSTACTION(m);
      if (leader != 0) {
        internal_free(m, leader);
      }
      if (trailer != 0) {
        internal_free(m, trailer);
      }
      return chunk2mem(p);
    }
  }
  return 0;
}
 
/* ------------------------ comalloc/coalloc support --------------------- */
 
static void** ialloc(mstate m,
                     size_t n_elements,
                     size_t* sizes,
                     int opts,
                     void* chunks[]) {
  /*
    This provides common support for independent_X routines, handling
    all of the combinations that can result.
 
    The opts arg has:
    bit 0 set if all elements are same size (using sizes[0])
    bit 1 set if elements should be zeroed
  */
 
  size_t    element_size;   /* chunksize of each element, if all same */
  size_t    contents_size;  /* total size of elements */
  size_t    array_size;     /* request size of pointer array */
  void*     mem;            /* malloced aggregate space */
  mchunkptr p;              /* corresponding chunk */
  size_t    remainder_size; /* remaining bytes while splitting */
  void**    marray;         /* either "chunks" or malloced ptr array */
  mchunkptr array_chunk;    /* chunk for malloced ptr array */
  flag_t    was_enabled;    /* to disable mmap */
  size_t    size;
  size_t    i;
 
  /* compute array length, if needed */
  if (chunks != 0) {
    if (n_elements == 0)
      return chunks; /* nothing to do */
    marray = chunks;
    array_size = 0;
  }
  else {
    /* if empty req, must still return chunk representing empty array */
    if (n_elements == 0)
      return (void**)internal_malloc(m, 0);
    marray = 0;
    array_size = request2size(n_elements * (sizeof(void*)));
  }
 
  /* compute total element size */
  if (opts & 0x1) { /* all-same-size */
    element_size = request2size(*sizes);
    contents_size = n_elements * element_size;
  }
  else { /* add up all the sizes */
    element_size = 0;
    contents_size = 0;
    for (i = 0; i != n_elements; ++i)
      contents_size += request2size(sizes[i]);
  }
 
  size = contents_size + array_size;
 
  /*
     Allocate the aggregate chunk.  First disable direct-mmapping so
     malloc won't use it, since we would not be able to later
     free/realloc space internal to a segregated mmap region.
  */
  was_enabled = use_mmap(m);
  disable_mmap(m);
  mem = internal_malloc(m, size - CHUNK_OVERHEAD);
  if (was_enabled)
    enable_mmap(m);
  if (mem == 0)
    return 0;
 
  if (PREACTION(m)) return 0;
  p = mem2chunk(mem);
  remainder_size = chunksize(p);
 
  assert(!is_mmapped(p));
 
  if (opts & 0x2) {       /* optionally clear the elements */
    memset((size_t*)mem, 0, remainder_size - SIZE_T_SIZE - array_size);
  }
 
  /* If not provided, allocate the pointer array as final part of chunk */
  if (marray == 0) {
    size_t  array_chunk_size;
    array_chunk = chunk_plus_offset(p, contents_size);
    array_chunk_size = remainder_size - contents_size;
    marray = (void**) (chunk2mem(array_chunk));
    set_size_and_pinuse_of_inuse_chunk(m, array_chunk, array_chunk_size);
    remainder_size = contents_size;
  }
 
  /* split out elements */
  for (i = 0; ; ++i) {
    marray[i] = chunk2mem(p);
    if (i != n_elements-1) {
      if (element_size != 0)
        size = element_size;
      else
        size = request2size(sizes[i]);
      remainder_size -= size;
      set_size_and_pinuse_of_inuse_chunk(m, p, size);
      p = chunk_plus_offset(p, size);
    }
    else { /* the final element absorbs any overallocation slop */
      set_size_and_pinuse_of_inuse_chunk(m, p, remainder_size);
      break;
    }
  }
 
#if DEBUG
  if (marray != chunks) {
    /* final element must have exactly exhausted chunk */
    if (element_size != 0) {
      assert(remainder_size == element_size);
    }
    else {
      assert(remainder_size == request2size(sizes[i]));
    }
    check_inuse_chunk(m, mem2chunk(marray));
  }
  for (i = 0; i != n_elements; ++i)
    check_inuse_chunk(m, mem2chunk(marray[i]));
 
#endif /* DEBUG */
 
  POSTACTION(m);
  return marray;
}
 
 
/* -------------------------- public routines ---------------------------- */
 
#if !ONLY_MSPACES
 
void* dlmalloc(size_t bytes) {
  /*
     Basic algorithm:
     If a small request (< 256 bytes minus per-chunk overhead):
       1. If one exists, use a remainderless chunk in associated smallbin.
          (Remainderless means that there are too few excess bytes to
          represent as a chunk.)
       2. If it is big enough, use the dv chunk, which is normally the
          chunk adjacent to the one used for the most recent small request.
       3. If one exists, split the smallest available chunk in a bin,
          saving remainder in dv.
       4. If it is big enough, use the top chunk.
       5. If available, get memory from system and use it
     Otherwise, for a large request:
       1. Find the smallest available binned chunk that fits, and use it
          if it is better fitting than dv chunk, splitting if necessary.
       2. If better fitting than any binned chunk, use the dv chunk.
       3. If it is big enough, use the top chunk.
       4. If request size >= mmap threshold, try to directly mmap this chunk.
       5. If available, get memory from system and use it
 
     The ugly goto's here ensure that postaction occurs along all paths.
  */
 
  if (!PREACTION(gm)) {
    void* mem;
    size_t nb;
    if (bytes <= MAX_SMALL_REQUEST) {
      bindex_t idx;
      binmap_t smallbits;
      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);
      idx = small_index(nb);
      smallbits = gm->smallmap >> idx;
 
      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */
        mchunkptr b, p;
        idx += ~smallbits & 1;       /* Uses next bin if idx empty */
        b = smallbin_at(gm, idx);
        p = b->fd;
        assert(chunksize(p) == small_index2size(idx));
        unlink_first_small_chunk(gm, b, p, idx);
        set_inuse_and_pinuse(gm, p, small_index2size(idx));
        mem = chunk2mem(p);
        check_malloced_chunk(gm, mem, nb);
        goto postaction;
      }
 
      else if (nb > gm->dvsize) {
        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */
          mchunkptr b, p, r;
          size_t rsize;
          bindex_t i;
          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));
          binmap_t leastbit = least_bit(leftbits);
          compute_bit2idx(leastbit, i);
          b = smallbin_at(gm, i);
          p = b->fd;
          assert(chunksize(p) == small_index2size(i));
          unlink_first_small_chunk(gm, b, p, i);
          rsize = small_index2size(i) - nb;
          /* Fit here cannot be remainderless if 4byte sizes */
          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)
            set_inuse_and_pinuse(gm, p, small_index2size(i));
          else {
            set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
            r = chunk_plus_offset(p, nb);
            set_size_and_pinuse_of_free_chunk(r, rsize);
            replace_dv(gm, r, rsize);
          }
          mem = chunk2mem(p);
          check_malloced_chunk(gm, mem, nb);
          goto postaction;
        }
 
        else if (gm->treemap != 0 && (mem = tmalloc_small(gm, nb)) != 0) {
          check_malloced_chunk(gm, mem, nb);
          goto postaction;
        }
      }
    }
    else if (bytes >= MAX_REQUEST)
      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */
    else {
      nb = pad_request(bytes);
      if (gm->treemap != 0 && (mem = tmalloc_large(gm, nb)) != 0) {
        check_malloced_chunk(gm, mem, nb);
        goto postaction;
      }
    }
 
    if (nb <= gm->dvsize) {
      size_t rsize = gm->dvsize - nb;
      mchunkptr p = gm->dv;
      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */
        mchunkptr r = gm->dv = chunk_plus_offset(p, nb);
        gm->dvsize = rsize;
        set_size_and_pinuse_of_free_chunk(r, rsize);
        set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
      }
      else { /* exhaust dv */
        size_t dvs = gm->dvsize;
        gm->dvsize = 0;
        gm->dv = 0;
        set_inuse_and_pinuse(gm, p, dvs);
      }
      mem = chunk2mem(p);
      check_malloced_chunk(gm, mem, nb);
      goto postaction;
    }
 
    else if (nb < gm->topsize) { /* Split top */
      size_t rsize = gm->topsize -= nb;
      mchunkptr p = gm->top;
      mchunkptr r = gm->top = chunk_plus_offset(p, nb);
      r->head = rsize | PINUSE_BIT;
      set_size_and_pinuse_of_inuse_chunk(gm, p, nb);
      mem = chunk2mem(p);
      check_top_chunk(gm, gm->top);
      check_malloced_chunk(gm, mem, nb);
      goto postaction;
    }
 
    mem = sys_alloc(gm, nb);
 
  postaction:
    POSTACTION(gm);
    return mem;
  }
 
  return 0;
}
 
void dlfree(void* mem) {
  /*
     Consolidate freed chunks with preceeding or succeeding bordering
     free chunks, if they exist, and then place in a bin.  Intermixed
     with special cases for top, dv, mmapped chunks, and usage errors.
  */
 
  if (mem != 0) {
    mchunkptr p  = mem2chunk(mem);
#if FOOTERS
    mstate fm = get_mstate_for(p);
    if (!ok_magic(fm)) {
      USAGE_ERROR_ACTION(fm, p);
      return;
    }
#else /* FOOTERS */
#define fm gm
#endif /* FOOTERS */
    if (!PREACTION(fm)) {
      check_inuse_chunk(fm, p);
      if (RTCHECK(ok_address(fm, p) && ok_cinuse(p))) {
        size_t psize = chunksize(p);
        mchunkptr next = chunk_plus_offset(p, psize);
        if (!pinuse(p)) {
          size_t prevsize = p->prev_foot;
          if ((prevsize & IS_MMAPPED_BIT) != 0) {
            prevsize &= ~IS_MMAPPED_BIT;
            psize += prevsize + MMAP_FOOT_PAD;
            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)
              fm->footprint -= psize;
            goto postaction;
          }
          else {
            mchunkptr prev = chunk_minus_offset(p, prevsize);
            psize += prevsize;
            p = prev;
            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */
              if (p != fm->dv) {
                unlink_chunk(fm, p, prevsize);
              }
              else if ((next->head & INUSE_BITS) == INUSE_BITS) {
                fm->dvsize = psize;
                set_free_with_pinuse(p, psize, next);
                goto postaction;
              }
            }
            else
              goto erroraction;
          }
        }
 
        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {
          if (!cinuse(next)) {  /* consolidate forward */
            if (next == fm->top) {
              size_t tsize = fm->topsize += psize;
              fm->top = p;
              p->head = tsize | PINUSE_BIT;
              if (p == fm->dv) {
                fm->dv = 0;
                fm->dvsize = 0;
              }
              if (should_trim(fm, tsize))
                sys_trim(fm, 0);
              goto postaction;
            }
            else if (next == fm->dv) {
              size_t dsize = fm->dvsize += psize;
              fm->dv = p;
              set_size_and_pinuse_of_free_chunk(p, dsize);
              goto postaction;
            }
            else {
              size_t nsize = chunksize(next);
              psize += nsize;
              unlink_chunk(fm, next, nsize);
              set_size_and_pinuse_of_free_chunk(p, psize);
              if (p == fm->dv) {
                fm->dvsize = psize;
                goto postaction;
              }
            }
          }
          else
            set_free_with_pinuse(p, psize, next);
          insert_chunk(fm, p, psize);
          check_free_chunk(fm, p);
          goto postaction;
        }
      }
    erroraction:
      USAGE_ERROR_ACTION(fm, p);
    postaction:
      POSTACTION(fm);
    }
  }
#if !FOOTERS
#undef fm
#endif /* FOOTERS */
}
 
void* dlcalloc(size_t n_elements, size_t elem_size) {
  void* mem;
  size_t req = 0;
  if (n_elements != 0) {
    req = n_elements * elem_size;
    if (((n_elements | elem_size) & ~(size_t)0xffff) &&
        (req / n_elements != elem_size))
      req = MAX_SIZE_T; /* force downstream failure on overflow */
  }
  mem = dlmalloc(req);
  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))
    memset(mem, 0, req);
  return mem;
}
 
void* dlrealloc(void* oldmem, size_t bytes) {
  if (oldmem == 0)
    return dlmalloc(bytes);
#ifdef REALLOC_ZERO_BYTES_FREES
  if (bytes == 0) {
    dlfree(oldmem);
    return 0;
  }
#endif /* REALLOC_ZERO_BYTES_FREES */
  else {
#if ! FOOTERS
    mstate m = gm;
#else /* FOOTERS */
    mstate m = get_mstate_for(mem2chunk(oldmem));
    if (!ok_magic(m)) {
      USAGE_ERROR_ACTION(m, oldmem);
      return 0;
    }
#endif /* FOOTERS */
    return internal_realloc(m, oldmem, bytes);
  }
}
 
void* dlmemalign(size_t alignment, size_t bytes) {
  return internal_memalign(gm, alignment, bytes);
}
 
void** dlindependent_calloc(size_t n_elements, size_t elem_size,
                                 void* chunks[]) {
  size_t sz = elem_size; /* serves as 1-element array */
  return ialloc(gm, n_elements, &sz, 3, chunks);
}
 
void** dlindependent_comalloc(size_t n_elements, size_t sizes[],
                                   void* chunks[]) {
  return ialloc(gm, n_elements, sizes, 0, chunks);
}
 
void* dlvalloc(size_t bytes) {
  size_t pagesz;
  init_mparams();
  pagesz = mparams.page_size;
  return dlmemalign(pagesz, bytes);
}
 
void* dlpvalloc(size_t bytes) {
  size_t pagesz;
  init_mparams();
  pagesz = mparams.page_size;
  return dlmemalign(pagesz, (bytes + pagesz - SIZE_T_ONE) & ~(pagesz - SIZE_T_ONE));
}
 
int dlmalloc_trim(size_t pad) {
  int result = 0;
  if (!PREACTION(gm)) {
    result = sys_trim(gm, pad);
    POSTACTION(gm);
  }
  return result;
}
 
size_t dlmalloc_footprint(void) {
  return gm->footprint;
}
 
size_t dlmalloc_max_footprint(void) {
  return gm->max_footprint;
}
 
#if !NO_MALLINFO
struct mallinfo dlmallinfo(void) {
  return internal_mallinfo(gm);
}
#endif /* NO_MALLINFO */
 
void dlmalloc_stats() {
  internal_malloc_stats(gm);
}
 
size_t dlmalloc_usable_size(void* mem) {
  if (mem != 0) {
    mchunkptr p = mem2chunk(mem);
    if (cinuse(p))
      return chunksize(p) - overhead_for(p);
  }
  return 0;
}
 
int dlmallopt(int param_number, int value) {
  return change_mparam(param_number, value);
}
 
#endif /* !ONLY_MSPACES */
 
/* ----------------------------- user mspaces ---------------------------- */
 
#if MSPACES
 
static mstate init_user_mstate(char* tbase, size_t tsize) {
  size_t msize = pad_request(sizeof(struct malloc_state));
  mchunkptr mn;
  mchunkptr msp = align_as_chunk(tbase);
  mstate m = (mstate)(chunk2mem(msp));
  memset(m, 0, msize);
  INITIAL_LOCK(&m->mutex);
  msp->head = (msize|PINUSE_BIT|CINUSE_BIT);
  m->seg.base = m->least_addr = tbase;
  m->seg.size = m->footprint = m->max_footprint = tsize;
  m->magic = mparams.magic;
  m->mflags = mparams.default_mflags;
  disable_contiguous(m);
  init_bins(m);
  mn = next_chunk(mem2chunk(m));
  init_top(m, mn, (size_t)((tbase + tsize) - (char*)mn) - TOP_FOOT_SIZE);
  check_top_chunk(m, m->top);
  return m;
}
 
mspace create_mspace(size_t capacity, int locked) {
  mstate m = 0;
  size_t msize = pad_request(sizeof(struct malloc_state));
  init_mparams(); /* Ensure pagesize etc initialized */
 
  if (capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {
    size_t rs = ((capacity == 0)? mparams.granularity :
                 (capacity + TOP_FOOT_SIZE + msize));
    size_t tsize = granularity_align(rs);
    char* tbase = (char*)(CALL_MMAP(tsize));
    if (tbase != CMFAIL) {
      m = init_user_mstate(tbase, tsize);
      m->seg.sflags = IS_MMAPPED_BIT;
      set_lock(m, locked);
    }
  }
  return (mspace)m;
}
 
mspace create_mspace_with_base(void* base, size_t capacity, int locked) {
  mstate m = 0;
  size_t msize = pad_request(sizeof(struct malloc_state));
  init_mparams(); /* Ensure pagesize etc initialized */
 
  if (capacity > msize + TOP_FOOT_SIZE &&
      capacity < (size_t) -(msize + TOP_FOOT_SIZE + mparams.page_size)) {
    m = init_user_mstate((char*)base, capacity);
    m->seg.sflags = EXTERN_BIT;
    set_lock(m, locked);
  }
  return (mspace)m;
}
 
size_t destroy_mspace(mspace msp) {
  size_t freed = 0;
  mstate ms = (mstate)msp;
  if (ok_magic(ms)) {
    msegmentptr sp = &ms->seg;
    while (sp != 0) {
      char* base = sp->base;
      size_t size = sp->size;
      flag_t flag = sp->sflags;
      sp = sp->next;
      if ((flag & IS_MMAPPED_BIT) && !(flag & EXTERN_BIT) &&
          CALL_MUNMAP(base, size) == 0)
        freed += size;
    }
  }
  else {
    USAGE_ERROR_ACTION(ms,ms);
  }
  return freed;
}
 
/*
  mspace versions of routines are near-clones of the global
  versions. This is not so nice but better than the alternatives.
*/
 
 
void* mspace_malloc(mspace msp, size_t bytes) {
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
    return 0;
  }
  if (!PREACTION(ms)) {
    void* mem;
    size_t nb;
    if (bytes <= MAX_SMALL_REQUEST) {
      bindex_t idx;
      binmap_t smallbits;
      nb = (bytes < MIN_REQUEST)? MIN_CHUNK_SIZE : pad_request(bytes);
      idx = small_index(nb);
      smallbits = ms->smallmap >> idx;
 
      if ((smallbits & 0x3U) != 0) { /* Remainderless fit to a smallbin. */
        mchunkptr b, p;
        idx += ~smallbits & 1;       /* Uses next bin if idx empty */
        b = smallbin_at(ms, idx);
        p = b->fd;
        assert(chunksize(p) == small_index2size(idx));
        unlink_first_small_chunk(ms, b, p, idx);
        set_inuse_and_pinuse(ms, p, small_index2size(idx));
        mem = chunk2mem(p);
        check_malloced_chunk(ms, mem, nb);
        goto postaction;
      }
 
      else if (nb > ms->dvsize) {
        if (smallbits != 0) { /* Use chunk in next nonempty smallbin */
          mchunkptr b, p, r;
          size_t rsize;
          bindex_t i;
          binmap_t leftbits = (smallbits << idx) & left_bits(idx2bit(idx));
          binmap_t leastbit = least_bit(leftbits);
          compute_bit2idx(leastbit, i);
          b = smallbin_at(ms, i);
          p = b->fd;
          assert(chunksize(p) == small_index2size(i));
          unlink_first_small_chunk(ms, b, p, i);
          rsize = small_index2size(i) - nb;
          /* Fit here cannot be remainderless if 4byte sizes */
          if (SIZE_T_SIZE != 4 && rsize < MIN_CHUNK_SIZE)
            set_inuse_and_pinuse(ms, p, small_index2size(i));
          else {
            set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
            r = chunk_plus_offset(p, nb);
            set_size_and_pinuse_of_free_chunk(r, rsize);
            replace_dv(ms, r, rsize);
          }
          mem = chunk2mem(p);
          check_malloced_chunk(ms, mem, nb);
          goto postaction;
        }
 
        else if (ms->treemap != 0 && (mem = tmalloc_small(ms, nb)) != 0) {
          check_malloced_chunk(ms, mem, nb);
          goto postaction;
        }
      }
    }
    else if (bytes >= MAX_REQUEST)
      nb = MAX_SIZE_T; /* Too big to allocate. Force failure (in sys alloc) */
    else {
      nb = pad_request(bytes);
      if (ms->treemap != 0 && (mem = tmalloc_large(ms, nb)) != 0) {
        check_malloced_chunk(ms, mem, nb);
        goto postaction;
      }
    }
 
    if (nb <= ms->dvsize) {
      size_t rsize = ms->dvsize - nb;
      mchunkptr p = ms->dv;
      if (rsize >= MIN_CHUNK_SIZE) { /* split dv */
        mchunkptr r = ms->dv = chunk_plus_offset(p, nb);
        ms->dvsize = rsize;
        set_size_and_pinuse_of_free_chunk(r, rsize);
        set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
      }
      else { /* exhaust dv */
        size_t dvs = ms->dvsize;
        ms->dvsize = 0;
        ms->dv = 0;
        set_inuse_and_pinuse(ms, p, dvs);
      }
      mem = chunk2mem(p);
      check_malloced_chunk(ms, mem, nb);
      goto postaction;
    }
 
    else if (nb < ms->topsize) { /* Split top */
      size_t rsize = ms->topsize -= nb;
      mchunkptr p = ms->top;
      mchunkptr r = ms->top = chunk_plus_offset(p, nb);
      r->head = rsize | PINUSE_BIT;
      set_size_and_pinuse_of_inuse_chunk(ms, p, nb);
      mem = chunk2mem(p);
      check_top_chunk(ms, ms->top);
      check_malloced_chunk(ms, mem, nb);
      goto postaction;
    }
 
    mem = sys_alloc(ms, nb);
 
  postaction:
    POSTACTION(ms);
    return mem;
  }
 
  return 0;
}
 
void mspace_free(mspace msp, void* mem) {
  if (mem != 0) {
    mchunkptr p  = mem2chunk(mem);
#if FOOTERS
    mstate fm = get_mstate_for(p);
#else /* FOOTERS */
    mstate fm = (mstate)msp;
#endif /* FOOTERS */
    if (!ok_magic(fm)) {
      USAGE_ERROR_ACTION(fm, p);
      return;
    }
    if (!PREACTION(fm)) {
      check_inuse_chunk(fm, p);
      if (RTCHECK(ok_address(fm, p) && ok_cinuse(p))) {
        size_t psize = chunksize(p);
        mchunkptr next = chunk_plus_offset(p, psize);
        if (!pinuse(p)) {
          size_t prevsize = p->prev_foot;
          if ((prevsize & IS_MMAPPED_BIT) != 0) {
            prevsize &= ~IS_MMAPPED_BIT;
            psize += prevsize + MMAP_FOOT_PAD;
            if (CALL_MUNMAP((char*)p - prevsize, psize) == 0)
              fm->footprint -= psize;
            goto postaction;
          }
          else {
            mchunkptr prev = chunk_minus_offset(p, prevsize);
            psize += prevsize;
            p = prev;
            if (RTCHECK(ok_address(fm, prev))) { /* consolidate backward */
              if (p != fm->dv) {
                unlink_chunk(fm, p, prevsize);
              }
              else if ((next->head & INUSE_BITS) == INUSE_BITS) {
                fm->dvsize = psize;
                set_free_with_pinuse(p, psize, next);
                goto postaction;
              }
            }
            else
              goto erroraction;
          }
        }
 
        if (RTCHECK(ok_next(p, next) && ok_pinuse(next))) {
          if (!cinuse(next)) {  /* consolidate forward */
            if (next == fm->top) {
              size_t tsize = fm->topsize += psize;
              fm->top = p;
              p->head = tsize | PINUSE_BIT;
              if (p == fm->dv) {
                fm->dv = 0;
                fm->dvsize = 0;
              }
              if (should_trim(fm, tsize))
                sys_trim(fm, 0);
              goto postaction;
            }
            else if (next == fm->dv) {
              size_t dsize = fm->dvsize += psize;
              fm->dv = p;
              set_size_and_pinuse_of_free_chunk(p, dsize);
              goto postaction;
            }
            else {
              size_t nsize = chunksize(next);
              psize += nsize;
              unlink_chunk(fm, next, nsize);
              set_size_and_pinuse_of_free_chunk(p, psize);
              if (p == fm->dv) {
                fm->dvsize = psize;
                goto postaction;
              }
            }
          }
          else
            set_free_with_pinuse(p, psize, next);
          insert_chunk(fm, p, psize);
          check_free_chunk(fm, p);
          goto postaction;
        }
      }
    erroraction:
      USAGE_ERROR_ACTION(fm, p);
    postaction:
      POSTACTION(fm);
    }
  }
}
 
void* mspace_calloc(mspace msp, size_t n_elements, size_t elem_size) {
  void* mem;
  size_t req = 0;
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
    return 0;
  }
  if (n_elements != 0) {
    req = n_elements * elem_size;
    if (((n_elements | elem_size) & ~(size_t)0xffff) &&
        (req / n_elements != elem_size))
      req = MAX_SIZE_T; /* force downstream failure on overflow */
  }
  mem = internal_malloc(ms, req);
  if (mem != 0 && calloc_must_clear(mem2chunk(mem)))
    memset(mem, 0, req);
  return mem;
}
 
void* mspace_realloc(mspace msp, void* oldmem, size_t bytes) {
  if (oldmem == 0)
    return mspace_malloc(msp, bytes);
#ifdef REALLOC_ZERO_BYTES_FREES
  if (bytes == 0) {
    mspace_free(msp, oldmem);
    return 0;
  }
#endif /* REALLOC_ZERO_BYTES_FREES */
  else {
#if FOOTERS
    mchunkptr p  = mem2chunk(oldmem);
    mstate ms = get_mstate_for(p);
#else /* FOOTERS */
    mstate ms = (mstate)msp;
#endif /* FOOTERS */
    if (!ok_magic(ms)) {
      USAGE_ERROR_ACTION(ms,ms);
      return 0;
    }
    return internal_realloc(ms, oldmem, bytes);
  }
}
 
void* mspace_memalign(mspace msp, size_t alignment, size_t bytes) {
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
    return 0;
  }
  return internal_memalign(ms, alignment, bytes);
}
 
void** mspace_independent_calloc(mspace msp, size_t n_elements,
                                 size_t elem_size, void* chunks[]) {
  size_t sz = elem_size; /* serves as 1-element array */
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
    return 0;
  }
  return ialloc(ms, n_elements, &sz, 3, chunks);
}
 
void** mspace_independent_comalloc(mspace msp, size_t n_elements,
                                   size_t sizes[], void* chunks[]) {
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
    return 0;
  }
  return ialloc(ms, n_elements, sizes, 0, chunks);
}
 
int mspace_trim(mspace msp, size_t pad) {
  int result = 0;
  mstate ms = (mstate)msp;
  if (ok_magic(ms)) {
    if (!PREACTION(ms)) {
      result = sys_trim(ms, pad);
      POSTACTION(ms);
    }
  }
  else {
    USAGE_ERROR_ACTION(ms,ms);
  }
  return result;
}
 
void mspace_malloc_stats(mspace msp) {
  mstate ms = (mstate)msp;
  if (ok_magic(ms)) {
    internal_malloc_stats(ms);
  }
  else {
    USAGE_ERROR_ACTION(ms,ms);
  }
}
 
size_t mspace_footprint(mspace msp) {
  size_t result;
  mstate ms = (mstate)msp;
  if (ok_magic(ms)) {
    result = ms->footprint;
  }
  USAGE_ERROR_ACTION(ms,ms);
  return result;
}
 
 
size_t mspace_max_footprint(mspace msp) {
  size_t result;
  mstate ms = (mstate)msp;
  if (ok_magic(ms)) {
    result = ms->max_footprint;
  }
  USAGE_ERROR_ACTION(ms,ms);
  return result;
}
 
 
#if !NO_MALLINFO
struct mallinfo mspace_mallinfo(mspace msp) {
  mstate ms = (mstate)msp;
  if (!ok_magic(ms)) {
    USAGE_ERROR_ACTION(ms,ms);
  }
  return internal_mallinfo(ms);
}
#endif /* NO_MALLINFO */
 
int mspace_mallopt(int param_number, int value) {
  return change_mparam(param_number, value);
}
 
#endif /* MSPACES */
 
/* -------------------- Alternative MORECORE functions ------------------- */
 
/*
  Guidelines for creating a custom version of MORECORE:
 
  * For best performance, MORECORE should allocate in multiples of pagesize.
  * MORECORE may allocate more memory than requested. (Or even less,
      but this will usually result in a malloc failure.)
  * MORECORE must not allocate memory when given argument zero, but
      instead return one past the end address of memory from previous
      nonzero call.
  * For best performance, consecutive calls to MORECORE with positive
      arguments should return increasing addresses, indicating that
      space has been contiguously extended.
  * Even though consecutive calls to MORECORE need not return contiguous
      addresses, it must be OK for malloc'ed chunks to span multiple
      regions in those cases where they do happen to be contiguous.
  * MORECORE need not handle negative arguments -- it may instead
      just return MFAIL when given negative arguments.
      Negative arguments are always multiples of pagesize. MORECORE
      must not misinterpret negative args as large positive unsigned
      args. You can suppress all such calls from even occurring by defining
      MORECORE_CANNOT_TRIM,
 
  As an example alternative MORECORE, here is a custom allocator
  kindly contributed for pre-OSX macOS.  It uses virtually but not
  necessarily physically contiguous non-paged memory (locked in,
  present and won't get swapped out).  You can use it by uncommenting
  this section, adding some #includes, and setting up the appropriate
  defines above:
 
      #define MORECORE osMoreCore
 
  There is also a shutdown routine that should somehow be called for
  cleanup upon program exit.
 
  #define MAX_POOL_ENTRIES 100
  #define MINIMUM_MORECORE_SIZE  (64 * 1024U)
  static int next_os_pool;
  void *our_os_pools[MAX_POOL_ENTRIES];
 
  void *osMoreCore(int size)
  {
    void *ptr = 0;
    static void *sbrk_top = 0;
 
    if (size > 0)
    {
      if (size < MINIMUM_MORECORE_SIZE)
         size = MINIMUM_MORECORE_SIZE;
      if (CurrentExecutionLevel() == kTaskLevel)
         ptr = PoolAllocateResident(size + RM_PAGE_SIZE, 0);
      if (ptr == 0)
      {
        return (void *) MFAIL;
      }
      // save ptrs so they can be freed during cleanup
      our_os_pools[next_os_pool] = ptr;
      next_os_pool++;
      ptr = (void *) ((((size_t) ptr) + RM_PAGE_MASK) & ~RM_PAGE_MASK);
      sbrk_top = (char *) ptr + size;
      return ptr;
    }
    else if (size < 0)
    {
      // we don't currently support shrink behavior
      return (void *) MFAIL;
    }
    else
    {
      return sbrk_top;
    }
  }
 
  // cleanup any allocated memory pools
  // called as last thing before shutting down driver
 
  void osCleanupMem(void)
  {
    void **ptr;
 
    for (ptr = our_os_pools; ptr < &our_os_pools[MAX_POOL_ENTRIES]; ptr++)
      if (*ptr)
      {
         PoolDeallocate(*ptr);
         *ptr = 0;
      }
  }
 
*/
 
 
/* -----------------------------------------------------------------------
History:
    V2.8.3 Thu Sep 22 11:16:32 2005  Doug Lea  (dl at gee)
      * Add max_footprint functions
      * Ensure all appropriate literals are size_t
      * Fix conditional compilation problem for some #define settings
      * Avoid concatenating segments with the one provided
        in create_mspace_with_base
      * Rename some variables to avoid compiler shadowing warnings
      * Use explicit lock initialization.
      * Better handling of sbrk interference.
      * Simplify and fix segment insertion, trimming and mspace_destroy
      * Reinstate REALLOC_ZERO_BYTES_FREES option from 2.7.x
      * Thanks especially to Dennis Flanagan for help on these.
 
    V2.8.2 Sun Jun 12 16:01:10 2005  Doug Lea  (dl at gee)
      * Fix memalign brace error.
 
    V2.8.1 Wed Jun  8 16:11:46 2005  Doug Lea  (dl at gee)
      * Fix improper #endif nesting in C++
      * Add explicit casts needed for C++
 
    V2.8.0 Mon May 30 14:09:02 2005  Doug Lea  (dl at gee)
      * Use trees for large bins
      * Support mspaces
      * Use segments to unify sbrk-based and mmap-based system allocation,
        removing need for emulation on most platforms without sbrk.
      * Default safety checks
      * Optional footer checks. Thanks to William Robertson for the idea.
      * Internal code refactoring
      * Incorporate suggestions and platform-specific changes.
        Thanks to Dennis Flanagan, Colin Plumb, Niall Douglas,
        Aaron Bachmann,  Emery Berger, and others.
      * Speed up non-fastbin processing enough to remove fastbins.
      * Remove useless cfree() to avoid conflicts with other apps.
      * Remove internal memcpy, memset. Compilers handle builtins better.
      * Remove some options that no one ever used and rename others.
 
    V2.7.2 Sat Aug 17 09:07:30 2002  Doug Lea  (dl at gee)
      * Fix malloc_state bitmap array misdeclaration
 
    V2.7.1 Thu Jul 25 10:58:03 2002  Doug Lea  (dl at gee)
      * Allow tuning of FIRST_SORTED_BIN_SIZE
      * Use PTR_UINT as type for all ptr->int casts. Thanks to John Belmonte.
      * Better detection and support for non-contiguousness of MORECORE.
        Thanks to Andreas Mueller, Conal Walsh, and Wolfram Gloger
      * Bypass most of malloc if no frees. Thanks To Emery Berger.
      * Fix freeing of old top non-contiguous chunk im sysmalloc.
      * Raised default trim and map thresholds to 256K.
      * Fix mmap-related #defines. Thanks to Lubos Lunak.
      * Fix copy macros; added LACKS_FCNTL_H. Thanks to Neal Walfield.
      * Branch-free bin calculation
      * Default trim and mmap thresholds now 256K.
 
    V2.7.0 Sun Mar 11 14:14:06 2001  Doug Lea  (dl at gee)
      * Introduce independent_comalloc and independent_calloc.
        Thanks to Michael Pachos for motivation and help.
      * Make optional .h file available
      * Allow > 2GB requests on 32bit systems.
      * new WIN32 sbrk, mmap, munmap, lock code from <Walter@GeNeSys-e.de>.
        Thanks also to Andreas Mueller <a.mueller at paradatec.de>,
        and Anonymous.
      * Allow override of MALLOC_ALIGNMENT (Thanks to Ruud Waij for
        helping test this.)
      * memalign: check alignment arg
      * realloc: don't try to shift chunks backwards, since this
        leads to  more fragmentation in some programs and doesn't
        seem to help in any others.
      * Collect all cases in malloc requiring system memory into sysmalloc
      * Use mmap as backup to sbrk
      * Place all internal state in malloc_state
      * Introduce fastbins (although similar to 2.5.1)
      * Many minor tunings and cosmetic improvements
      * Introduce USE_PUBLIC_MALLOC_WRAPPERS, USE_MALLOC_LOCK
      * Introduce MALLOC_FAILURE_ACTION, MORECORE_CONTIGUOUS
        Thanks to Tony E. Bennett <tbennett@nvidia.com> and others.
      * Include errno.h to support default failure action.
 
    V2.6.6 Sun Dec  5 07:42:19 1999  Doug Lea  (dl at gee)
      * return null for negative arguments
      * Added Several WIN32 cleanups from Martin C. Fong <mcfong at yahoo.com>
         * Add 'LACKS_SYS_PARAM_H' for those systems without 'sys/param.h'
          (e.g. WIN32 platforms)
         * Cleanup header file inclusion for WIN32 platforms
         * Cleanup code to avoid Microsoft Visual C++ compiler complaints
         * Add 'USE_DL_PREFIX' to quickly allow co-existence with existing
           memory allocation routines
         * Set 'malloc_getpagesize' for WIN32 platforms (needs more work)
         * Use 'assert' rather than 'ASSERT' in WIN32 code to conform to
           usage of 'assert' in non-WIN32 code
         * Improve WIN32 'sbrk()' emulation's 'findRegion()' routine to
           avoid infinite loop
      * Always call 'fREe()' rather than 'free()'
 
    V2.6.5 Wed Jun 17 15:57:31 1998  Doug Lea  (dl at gee)
      * Fixed ordering problem with boundary-stamping
 
    V2.6.3 Sun May 19 08:17:58 1996  Doug Lea  (dl at gee)
      * Added pvalloc, as recommended by H.J. Liu
      * Added 64bit pointer support mainly from Wolfram Gloger
      * Added anonymously donated WIN32 sbrk emulation
      * Malloc, calloc, getpagesize: add optimizations from Raymond Nijssen
      * malloc_extend_top: fix mask error that caused wastage after
        foreign sbrks
      * Add linux mremap support code from HJ Liu
 
    V2.6.2 Tue Dec  5 06:52:55 1995  Doug Lea  (dl at gee)
      * Integrated most documentation with the code.
      * Add support for mmap, with help from
        Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
      * Use last_remainder in more cases.
      * Pack bins using idea from  colin@nyx10.cs.du.edu
      * Use ordered bins instead of best-fit threshhold
      * Eliminate block-local decls to simplify tracing and debugging.
      * Support another case of realloc via move into top
      * Fix error occuring when initial sbrk_base not word-aligned.
      * Rely on page size for units instead of SBRK_UNIT to
        avoid surprises about sbrk alignment conventions.
      * Add mallinfo, mallopt. Thanks to Raymond Nijssen
        (raymond@es.ele.tue.nl) for the suggestion.
      * Add `pad' argument to malloc_trim and top_pad mallopt parameter.
      * More precautions for cases where other routines call sbrk,
        courtesy of Wolfram Gloger (Gloger@lrz.uni-muenchen.de).
      * Added macros etc., allowing use in linux libc from
        H.J. Lu (hjl@gnu.ai.mit.edu)
      * Inverted this history list
 
    V2.6.1 Sat Dec  2 14:10:57 1995  Doug Lea  (dl at gee)
      * Re-tuned and fixed to behave more nicely with V2.6.0 changes.
      * Removed all preallocation code since under current scheme
        the work required to undo bad preallocations exceeds
        the work saved in good cases for most test programs.
      * No longer use return list or unconsolidated bins since
        no scheme using them consistently outperforms those that don't
        given above changes.
      * Use best fit for very large chunks to prevent some worst-cases.
      * Added some support for debugging
 
    V2.6.0 Sat Nov  4 07:05:23 1995  Doug Lea  (dl at gee)
      * Removed footers when chunks are in use. Thanks to
        Paul Wilson (wilson@cs.texas.edu) for the suggestion.
 
    V2.5.4 Wed Nov  1 07:54:51 1995  Doug Lea  (dl at gee)
      * Added malloc_trim, with help from Wolfram Gloger
        (wmglo@Dent.MED.Uni-Muenchen.DE).
 
    V2.5.3 Tue Apr 26 10:16:01 1994  Doug Lea  (dl at g)
 
    V2.5.2 Tue Apr  5 16:20:40 1994  Doug Lea  (dl at g)
      * realloc: try to expand in both directions
      * malloc: swap order of clean-bin strategy;
      * realloc: only conditionally expand backwards
      * Try not to scavenge used bins
      * Use bin counts as a guide to preallocation
      * Occasionally bin return list chunks in first scan
      * Add a few optimizations from colin@nyx10.cs.du.edu
 
    V2.5.1 Sat Aug 14 15:40:43 1993  Doug Lea  (dl at g)
      * faster bin computation & slightly different binning
      * merged all consolidations to one part of malloc proper
         (eliminating old malloc_find_space & malloc_clean_bin)
      * Scan 2 returns chunks (not just 1)
      * Propagate failure in realloc if malloc returns 0
      * Add stuff to allow compilation on non-ANSI compilers
          from kpv@research.att.com
 
    V2.5 Sat Aug  7 07:41:59 1993  Doug Lea  (dl at g.oswego.edu)
      * removed potential for odd address access in prev_chunk
      * removed dependency on getpagesize.h
      * misc cosmetics and a bit more internal documentation
      * anticosmetics: mangled names in macros to evade debugger strangeness
      * tested on sparc, hp-700, dec-mips, rs6000
          with gcc & native cc (hp, dec only) allowing
          Detlefs & Zorn comparison study (in SIGPLAN Notices.)
 
    Trial version Fri Aug 28 13:14:29 1992  Doug Lea  (dl at g.oswego.edu)
      * Based loosely on libg++-1.2X malloc. (It retains some of the overall
         structure of old version,  but most details differ.)
 
*/
 
Subversion Repositories HelenOS-historic

(root)/uspace/trunk/libc/malloc/malloc.c – Rev 999 → 1113
