Update.

2002-01-18  Wolfram Gloger  <wg@malloc.de>

	* malloc/malloc.c: Rewrite, adapted from Doug Lea's malloc-2.7.0.c.
	* malloc/malloc.h: Likewise.
	* malloc/arena.c: New file.
	* malloc/hooks.c: New file.
	* malloc/tst-mallocstate.c: New file.
	* malloc/Makefile: Add new testcase tst-mallocstate.
	Add arena.c and hooks.c to distribute.  Fix commented CPPFLAGS.

2002-01-28  Ulrich Drepper  <drepper@redhat.com>

	* stdlib/msort.c: Remove last patch.  The optimization violates the
	same rule which qsort.c had problems with.

2002-01-27  Paul Eggert  <eggert@twinsun.com>

	* stdlib/qsort.c (_quicksort): Do not apply the comparison function
	to a pivot element that lies outside the array to be sorted, as
	ISO C99 requires that the comparison function be called only with
	addresses of array elements [PR libc/2880].

9 files changed, 5705 insertions, 4268 deletions

diff --git a/ChangeLog b/ChangeLog
index 5577fb5ae8..8268adcbee 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,3 +1,25 @@
+2002-01-18  Wolfram Gloger  <wg@malloc.de>
+
+	* malloc/malloc.c: Rewrite, adapted from Doug Lea's malloc-2.7.0.c.
+	* malloc/malloc.h: Likewise.
+	* malloc/arena.c: New file.
+	* malloc/hooks.c: New file.
+	* malloc/tst-mallocstate.c: New file.
+	* malloc/Makefile: Add new testcase tst-mallocstate.
+	Add arena.c and hooks.c to distribute.  Fix commented CPPFLAGS.
+
+2002-01-28  Ulrich Drepper  <drepper@redhat.com>
+
+	* stdlib/msort.c: Remove last patch.  The optimization violates the
+	same rule which qsort.c had problems with.
+
+2002-01-27  Paul Eggert  <eggert@twinsun.com>
+
+	* stdlib/qsort.c (_quicksort): Do not apply the comparison function
+	to a pivot element that lies outside the array to be sorted, as
+	ISO C99 requires that the comparison function be called only with
+	addresses of array elements [PR libc/2880].
+
 2002-01-28  Ulrich Drepper  <drepper@redhat.com>
 
 	* elf/dl-load.c (_dl_map_object): Remove incorrect optimization
diff --git a/malloc/Makefile b/malloc/Makefile
index 3b5d991971..226ed99fc3 100644
--- a/malloc/Makefile
+++ b/malloc/Makefile
@@ -1,4 +1,4 @@
-# Copyright (C) 1991-1999, 2000, 2001 Free Software Foundation, Inc.
+# Copyright (C) 1991-1999, 2000, 2001, 2002 Free Software Foundation, Inc.
 # This file is part of the GNU C Library.
 
 # The GNU C Library is free software; you can redistribute it and/or
@@ -25,11 +25,12 @@ all:
 
 dist-headers := malloc.h
 headers := $(dist-headers) obstack.h mcheck.h
-tests := mallocbug tst-malloc tst-valloc tst-calloc tst-obstack
+tests := mallocbug tst-malloc tst-valloc tst-calloc tst-obstack \
+	 tst-mallocstate
 test-srcs = tst-mtrace
 
 distribute = thread-m.h mtrace.pl mcheck-init.c stackinfo.h memusage.h \
-	     memusage.sh memusagestat.c tst-mtrace.sh
+	     memusage.sh memusagestat.c tst-mtrace.sh arena.c hooks.c
 
 # Things which get pasted together into gmalloc.c.
 gmalloc-routines := malloc morecore
@@ -109,7 +110,7 @@ endif
 endif
 
 # Uncomment this for test releases.  For public releases it is too expensive.
-#CPPFLAGS-malloc.o += -DMALLOC_DEBUG
+#CPPFLAGS-malloc.o += -DMALLOC_DEBUG=1
 
 $(objpfx)mtrace: mtrace.pl
 	rm -f $@.new
@@ -126,3 +127,6 @@ $(objpfx)memusage: memusage.sh
 
 # The implementation uses `dlsym'
 $(objpfx)libmemusage.so: $(common-objpfx)dlfcn/libdl.so
+
+# Extra dependencies
+$(foreach o,$(all-object-suffixes),$(objpfx)malloc$(o)): arena.c hooks.c
diff --git a/malloc/arena.c b/malloc/arena.c
new file mode 100644
index 0000000000..8324b68689
--- /dev/null
+++ b/malloc/arena.c
@@ -0,0 +1,746 @@
+/* Malloc implementation for multiple threads without lock contention.
+   Copyright (C) 2001 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Wolfram Gloger <wg@malloc.de>, 2001.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+/* $Id$ */
+
+/* Compile-time constants.  */
+
+#define HEAP_MIN_SIZE (32*1024)
+#ifndef HEAP_MAX_SIZE
+#define HEAP_MAX_SIZE (1024*1024) /* must be a power of two */
+#endif
+
+/* HEAP_MIN_SIZE and HEAP_MAX_SIZE limit the size of mmap()ed heaps
+   that are dynamically created for multi-threaded programs.  The
+   maximum size must be a power of two, for fast determination of
+   which heap belongs to a chunk.  It should be much larger than the
+   mmap threshold, so that requests with a size just below that
+   threshold can be fulfilled without creating too many heaps.  */
+
+
+#ifndef THREAD_STATS
+#define THREAD_STATS 0
+#endif
+
+/* If THREAD_STATS is non-zero, some statistics on mutex locking are
+   computed.  */
+
+/***************************************************************************/
+
+#define top(ar_ptr) ((ar_ptr)->top)
+
+/* A heap is a single contiguous memory region holding (coalesceable)
+   malloc_chunks.  It is allocated with mmap() and always starts at an
+   address aligned to HEAP_MAX_SIZE.  Not used unless compiling with
+   USE_ARENAS. */
+
+typedef struct _heap_info {
+  mstate ar_ptr; /* Arena for this heap. */
+  struct _heap_info *prev; /* Previous heap. */
+  size_t size;   /* Current size in bytes. */
+  size_t pad;    /* Make sure the following data is properly aligned. */
+} heap_info;
+
+/* Thread specific data */
+
+static tsd_key_t arena_key;
+static mutex_t list_lock;
+
+#if THREAD_STATS
+static int stat_n_heaps;
+#define THREAD_STAT(x) x
+#else
+#define THREAD_STAT(x) do ; while(0)
+#endif
+
+/* Mapped memory in non-main arenas (reliable only for NO_THREADS). */
+static unsigned long arena_mem;
+
+/**************************************************************************/
+
+#if USE_ARENAS
+
+/* arena_get() acquires an arena and locks the corresponding mutex.
+   First, try the one last locked successfully by this thread.  (This
+   is the common case and handled with a macro for speed.)  Then, loop
+   once over the circularly linked list of arenas.  If no arena is
+   readily available, create a new one.  In this latter case, `size'
+   is just a hint as to how much memory will be required immediately
+   in the new arena. */
+
+#define arena_get(ptr, size) do { \
+  Void_t *vptr = NULL; \
+  ptr = (mstate)tsd_getspecific(arena_key, vptr); \
+  if(ptr && !mutex_trylock(&ptr->mutex)) { \
+    THREAD_STAT(++(ptr->stat_lock_direct)); \
+  } else \
+    ptr = arena_get2(ptr, (size)); \
+} while(0)
+
+/* find the heap and corresponding arena for a given ptr */
+
+#define heap_for_ptr(ptr) \
+ ((heap_info *)((unsigned long)(ptr) & ~(HEAP_MAX_SIZE-1)))
+#define arena_for_chunk(ptr) \
+ (chunk_non_main_arena(ptr) ? heap_for_ptr(ptr)->ar_ptr : &main_arena)
+
+#else /* !USE_ARENAS */
+
+/* There is only one arena, main_arena. */
+
+#if THREAD_STATS
+#define arena_get(ar_ptr, sz) do { \
+  ar_ptr = &main_arena; \
+  if(!mutex_trylock(&ar_ptr->mutex)) \
+    ++(ar_ptr->stat_lock_direct); \
+  else { \
+    (void)mutex_lock(&ar_ptr->mutex); \
+    ++(ar_ptr->stat_lock_wait); \
+  } \
+} while(0)
+#else
+#define arena_get(ar_ptr, sz) do { \
+  ar_ptr = &main_arena; \
+  (void)mutex_lock(&ar_ptr->mutex); \
+} while(0)
+#endif
+#define arena_for_chunk(ptr) (&main_arena)
+
+#endif /* USE_ARENAS */
+
+/**************************************************************************/
+
+#ifndef NO_THREADS
+
+/* atfork support.  */
+
+static __malloc_ptr_t (*save_malloc_hook) __MALLOC_P ((size_t __size,
+						       __const __malloc_ptr_t));
+static void           (*save_free_hook) __MALLOC_P ((__malloc_ptr_t __ptr,
+						     __const __malloc_ptr_t));
+static Void_t*        save_arena;
+
+/* Magic value for the thread-specific arena pointer when
+   malloc_atfork() is in use.  */
+
+#define ATFORK_ARENA_PTR ((Void_t*)-1)
+
+/* The following hooks are used while the `atfork' handling mechanism
+   is active. */
+
+static Void_t*
+malloc_atfork(size_t sz, const Void_t *caller)
+{
+  Void_t *vptr = NULL;
+  Void_t *victim;
+
+  tsd_getspecific(arena_key, vptr);
+  if(vptr == ATFORK_ARENA_PTR) {
+    /* We are the only thread that may allocate at all.  */
+    if(save_malloc_hook != malloc_check) {
+      return _int_malloc(&main_arena, sz);
+    } else {
+      if(top_check()<0)
+        return 0;
+      victim = _int_malloc(&main_arena, sz+1);
+      return mem2mem_check(victim, sz);
+    }
+  } else {
+    /* Suspend the thread until the `atfork' handlers have completed.
+       By that time, the hooks will have been reset as well, so that
+       mALLOc() can be used again. */
+    (void)mutex_lock(&list_lock);
+    (void)mutex_unlock(&list_lock);
+    return public_mALLOc(sz);
+  }
+}
+
+static void
+free_atfork(Void_t* mem, const Void_t *caller)
+{
+  Void_t *vptr = NULL;
+  mstate ar_ptr;
+  mchunkptr p;                          /* chunk corresponding to mem */
+
+  if (mem == 0)                              /* free(0) has no effect */
+    return;
+
+  p = mem2chunk(mem);         /* do not bother to replicate free_check here */
+
+#if HAVE_MMAP
+  if (chunk_is_mmapped(p))                       /* release mmapped memory. */
+  {
+    munmap_chunk(p);
+    return;
+  }
+#endif
+
+  ar_ptr = arena_for_chunk(p);
+  tsd_getspecific(arena_key, vptr);
+  if(vptr != ATFORK_ARENA_PTR)
+    (void)mutex_lock(&ar_ptr->mutex);
+  _int_free(ar_ptr, mem);
+  if(vptr != ATFORK_ARENA_PTR)
+    (void)mutex_unlock(&ar_ptr->mutex);
+}
+
+/* The following two functions are registered via thread_atfork() to
+   make sure that the mutexes remain in a consistent state in the
+   fork()ed version of a thread.  Also adapt the malloc and free hooks
+   temporarily, because the `atfork' handler mechanism may use
+   malloc/free internally (e.g. in LinuxThreads). */
+
+static void
+ptmalloc_lock_all __MALLOC_P((void))
+{
+  mstate ar_ptr;
+
+  (void)mutex_lock(&list_lock);
+  for(ar_ptr = &main_arena;;) {
+    (void)mutex_lock(&ar_ptr->mutex);
+    ar_ptr = ar_ptr->next;
+    if(ar_ptr == &main_arena) break;
+  }
+  save_malloc_hook = __malloc_hook;
+  save_free_hook = __free_hook;
+  __malloc_hook = malloc_atfork;
+  __free_hook = free_atfork;
+  /* Only the current thread may perform malloc/free calls now. */
+  tsd_getspecific(arena_key, save_arena);
+  tsd_setspecific(arena_key, ATFORK_ARENA_PTR);
+}
+
+static void
+ptmalloc_unlock_all __MALLOC_P((void))
+{
+  mstate ar_ptr;
+
+  tsd_setspecific(arena_key, save_arena);
+  __malloc_hook = save_malloc_hook;
+  __free_hook = save_free_hook;
+  for(ar_ptr = &main_arena;;) {
+    (void)mutex_unlock(&ar_ptr->mutex);
+    ar_ptr = ar_ptr->next;
+    if(ar_ptr == &main_arena) break;
+  }
+  (void)mutex_unlock(&list_lock);
+}
+
+#ifdef __linux__
+
+/* In LinuxThreads, unlocking a mutex in the child process after a
+   fork() is currently unsafe, whereas re-initializing it is safe and
+   does not leak resources.  Therefore, a special atfork handler is
+   installed for the child. */
+
+static void
+ptmalloc_unlock_all2 __MALLOC_P((void))
+{
+  mstate ar_ptr;
+
+#if defined _LIBC || defined MALLOC_HOOKS
+  tsd_setspecific(arena_key, save_arena);
+  __malloc_hook = save_malloc_hook;
+  __free_hook = save_free_hook;
+#endif
+  for(ar_ptr = &main_arena;;) {
+    (void)mutex_init(&ar_ptr->mutex);
+    ar_ptr = ar_ptr->next;
+    if(ar_ptr == &main_arena) break;
+  }
+  (void)mutex_init(&list_lock);
+}
+
+#else
+
+#define ptmalloc_unlock_all2 ptmalloc_unlock_all
+
+#endif
+
+#endif /* !defined NO_THREADS */
+
+/* Already initialized? */
+int __malloc_initialized = -1;
+
+/* Initialization routine. */
+#ifdef _LIBC
+#include <string.h>
+extern char **_environ;
+
+static char *
+internal_function
+next_env_entry (char ***position)
+{
+  char **current = *position;
+  char *result = NULL;
+
+  while (*current != NULL)
+    {
+      if (__builtin_expect ((*current)[0] == 'M', 0)
+	  && (*current)[1] == 'A'
+	  && (*current)[2] == 'L'
+	  && (*current)[3] == 'L'
+	  && (*current)[4] == 'O'
+	  && (*current)[5] == 'C'
+	  && (*current)[6] == '_')
+	{
+	  result = &(*current)[7];
+
+	  /* Save current position for next visit.  */
+	  *position = ++current;
+
+	  break;
+	}
+
+      ++current;
+    }
+
+  return result;
+}
+#endif /* _LIBC */
+
+static void
+ptmalloc_init __MALLOC_P((void))
+{
+#if __STD_C
+  const char* s;
+#else
+  char* s;
+#endif
+  int secure = 0;
+
+  if(__malloc_initialized >= 0) return;
+  __malloc_initialized = 0;
+
+  mp_.top_pad        = DEFAULT_TOP_PAD;
+  mp_.n_mmaps_max    = DEFAULT_MMAP_MAX;
+  mp_.mmap_threshold = DEFAULT_MMAP_THRESHOLD;
+  mp_.trim_threshold = DEFAULT_TRIM_THRESHOLD;
+  mp_.pagesize       = malloc_getpagesize;
+
+#ifndef NO_THREADS
+  /* With some threads implementations, creating thread-specific data
+     or initializing a mutex may call malloc() itself.  Provide a
+     simple starter version (realloc() won't work). */
+  save_malloc_hook = __malloc_hook;
+  save_free_hook = __free_hook;
+  __malloc_hook = malloc_starter;
+  __free_hook = free_starter;
+#ifdef _LIBC
+  /* Initialize the pthreads interface. */
+  if (__pthread_initialize != NULL)
+    __pthread_initialize();
+#endif
+#endif /* !defined NO_THREADS */
+  mutex_init(&main_arena.mutex);
+  main_arena.next = &main_arena;
+
+  mutex_init(&list_lock);
+  tsd_key_create(&arena_key, NULL);
+  tsd_setspecific(arena_key, (Void_t *)&main_arena);
+  thread_atfork(ptmalloc_lock_all, ptmalloc_unlock_all, ptmalloc_unlock_all2);
+#ifndef NO_THREADS
+  __malloc_hook = save_malloc_hook;
+  __free_hook = save_free_hook;
+#endif
+#ifdef _LIBC
+  secure = __libc_enable_secure;
+  s = NULL;
+  {
+    char **runp = _environ;
+    char *envline;
+
+    while (__builtin_expect ((envline = next_env_entry (&runp)) != NULL,
+			     0))
+      {
+	size_t len = strcspn (envline, "=");
+
+	if (envline[len] != '=')
+	  /* This is a "MALLOC_" variable at the end of the string
+	     without a '=' character.  Ignore it since otherwise we
+	     will access invalid memory below.  */
+	  continue;
+
+	switch (len)
+	  {
+	  case 6:
+	    if (memcmp (envline, "CHECK_", 6) == 0)
+	      s = &envline[7];
+	    break;
+	  case 8:
+	    if (! secure && memcmp (envline, "TOP_PAD_", 8) == 0)
+	      mALLOPt(M_TOP_PAD, atoi(&envline[9]));
+	    break;
+	  case 9:
+	    if (! secure && memcmp (envline, "MMAP_MAX_", 9) == 0)
+	      mALLOPt(M_MMAP_MAX, atoi(&envline[10]));
+	    break;
+	  case 15:
+	    if (! secure)
+	      {
+		if (memcmp (envline, "TRIM_THRESHOLD_", 15) == 0)
+		  mALLOPt(M_TRIM_THRESHOLD, atoi(&envline[16]));
+		else if (memcmp (envline, "MMAP_THRESHOLD_", 15) == 0)
+		  mALLOPt(M_MMAP_THRESHOLD, atoi(&envline[16]));
+	      }
+	    break;
+	  default:
+	    break;
+	  }
+      }
+  }
+#else
+  if (! secure)
+    {
+      if((s = getenv("MALLOC_TRIM_THRESHOLD_")))
+	mALLOPt(M_TRIM_THRESHOLD, atoi(s));
+      if((s = getenv("MALLOC_TOP_PAD_")))
+	mALLOPt(M_TOP_PAD, atoi(s));
+      if((s = getenv("MALLOC_MMAP_THRESHOLD_")))
+	mALLOPt(M_MMAP_THRESHOLD, atoi(s));
+      if((s = getenv("MALLOC_MMAP_MAX_")))
+	mALLOPt(M_MMAP_MAX, atoi(s));
+    }
+  s = getenv("MALLOC_CHECK_");
+#endif
+  if(s) {
+    if(s[0]) mALLOPt(M_CHECK_ACTION, (int)(s[0] - '0'));
+    __malloc_check_init();
+  }
+  if(__malloc_initialize_hook != NULL)
+    (*__malloc_initialize_hook)();
+  __malloc_initialized = 1;
+}
+
+/* There are platforms (e.g. Hurd) with a link-time hook mechanism. */
+#ifdef thread_atfork_static
+thread_atfork_static(ptmalloc_lock_all, ptmalloc_unlock_all, \
+                     ptmalloc_unlock_all2)
+#endif
+
+
+
+/* Managing heaps and arenas (for concurrent threads) */
+
+#if USE_ARENAS
+
+#if MALLOC_DEBUG > 1
+
+/* Print the complete contents of a single heap to stderr. */
+
+static void
+#if __STD_C
+dump_heap(heap_info *heap)
+#else
+dump_heap(heap) heap_info *heap;
+#endif
+{
+  char *ptr;
+  mchunkptr p;
+
+  fprintf(stderr, "Heap %p, size %10lx:\n", heap, (long)heap->size);
+  ptr = (heap->ar_ptr != (mstate)(heap+1)) ?
+    (char*)(heap + 1) : (char*)(heap + 1) + sizeof(struct malloc_state);
+  p = (mchunkptr)(((unsigned long)ptr + MALLOC_ALIGN_MASK) &
+                  ~MALLOC_ALIGN_MASK);
+  for(;;) {
+    fprintf(stderr, "chunk %p size %10lx", p, (long)p->size);
+    if(p == top(heap->ar_ptr)) {
+      fprintf(stderr, " (top)\n");
+      break;
+    } else if(p->size == (0|PREV_INUSE)) {
+      fprintf(stderr, " (fence)\n");
+      break;
+    }
+    fprintf(stderr, "\n");
+    p = next_chunk(p);
+  }
+}
+
+#endif /* MALLOC_DEBUG > 1 */
+
+/* Create a new heap.  size is automatically rounded up to a multiple
+   of the page size. */
+
+static heap_info *
+internal_function
+#if __STD_C
+new_heap(size_t size, size_t top_pad)
+#else
+new_heap(size, top_pad) size_t size, top_pad;
+#endif
+{
+  size_t page_mask = malloc_getpagesize - 1;
+  char *p1, *p2;
+  unsigned long ul;
+  heap_info *h;
+
+  if(size+top_pad < HEAP_MIN_SIZE)
+    size = HEAP_MIN_SIZE;
+  else if(size+top_pad <= HEAP_MAX_SIZE)
+    size += top_pad;
+  else if(size > HEAP_MAX_SIZE)
+    return 0;
+  else
+    size = HEAP_MAX_SIZE;
+  size = (size + page_mask) & ~page_mask;
+
+  /* A memory region aligned to a multiple of HEAP_MAX_SIZE is needed.
+     No swap space needs to be reserved for the following large
+     mapping (on Linux, this is the case for all non-writable mappings
+     anyway). */
+  p1 = (char *)MMAP(0, HEAP_MAX_SIZE<<1, PROT_NONE, MAP_PRIVATE|MAP_NORESERVE);
+  if(p1 != MAP_FAILED) {
+    p2 = (char *)(((unsigned long)p1 + (HEAP_MAX_SIZE-1)) & ~(HEAP_MAX_SIZE-1));
+    ul = p2 - p1;
+    munmap(p1, ul);
+    munmap(p2 + HEAP_MAX_SIZE, HEAP_MAX_SIZE - ul);
+  } else {
+    /* Try to take the chance that an allocation of only HEAP_MAX_SIZE
+       is already aligned. */
+    p2 = (char *)MMAP(0, HEAP_MAX_SIZE, PROT_NONE, MAP_PRIVATE|MAP_NORESERVE);
+    if(p2 == MAP_FAILED)
+      return 0;
+    if((unsigned long)p2 & (HEAP_MAX_SIZE-1)) {
+      munmap(p2, HEAP_MAX_SIZE);
+      return 0;
+    }
+  }
+  if(mprotect(p2, size, PROT_READ|PROT_WRITE) != 0) {
+    munmap(p2, HEAP_MAX_SIZE);
+    return 0;
+  }
+  h = (heap_info *)p2;
+  h->size = size;
+  THREAD_STAT(stat_n_heaps++);
+  return h;
+}
+
+/* Grow or shrink a heap.  size is automatically rounded up to a
+   multiple of the page size if it is positive. */
+
+static int
+#if __STD_C
+grow_heap(heap_info *h, long diff)
+#else
+grow_heap(h, diff) heap_info *h; long diff;
+#endif
+{
+  size_t page_mask = malloc_getpagesize - 1;
+  long new_size;
+
+  if(diff >= 0) {
+    diff = (diff + page_mask) & ~page_mask;
+    new_size = (long)h->size + diff;
+    if(new_size > HEAP_MAX_SIZE)
+      return -1;
+    if(mprotect((char *)h + h->size, diff, PROT_READ|PROT_WRITE) != 0)
+      return -2;
+  } else {
+    new_size = (long)h->size + diff;
+    if(new_size < (long)sizeof(*h))
+      return -1;
+    /* Try to re-map the extra heap space freshly to save memory, and
+       make it inaccessible. */
+    if((char *)MMAP((char *)h + new_size, -diff, PROT_NONE,
+                    MAP_PRIVATE|MAP_FIXED) == (char *) MAP_FAILED)
+      return -2;
+    /*fprintf(stderr, "shrink %p %08lx\n", h, new_size);*/
+  }
+  h->size = new_size;
+  return 0;
+}
+
+/* Delete a heap. */
+
+#define delete_heap(heap) munmap((char*)(heap), HEAP_MAX_SIZE)
+
+static int
+internal_function
+#if __STD_C
+heap_trim(heap_info *heap, size_t pad)
+#else
+heap_trim(heap, pad) heap_info *heap; size_t pad;
+#endif
+{
+  mstate ar_ptr = heap->ar_ptr;
+  unsigned long pagesz = mp_.pagesize;
+  mchunkptr top_chunk = top(ar_ptr), p, bck, fwd;
+  heap_info *prev_heap;
+  long new_size, top_size, extra;
+
+  /* Can this heap go away completely? */
+  while(top_chunk == chunk_at_offset(heap, sizeof(*heap))) {
+    prev_heap = heap->prev;
+    p = chunk_at_offset(prev_heap, prev_heap->size - (MINSIZE-2*SIZE_SZ));
+    assert(p->size == (0|PREV_INUSE)); /* must be fencepost */
+    p = prev_chunk(p);
+    new_size = chunksize(p) + (MINSIZE-2*SIZE_SZ);
+    assert(new_size>0 && new_size<(long)(2*MINSIZE));
+    if(!prev_inuse(p))
+      new_size += p->prev_size;
+    assert(new_size>0 && new_size<HEAP_MAX_SIZE);
+    if(new_size + (HEAP_MAX_SIZE - prev_heap->size) < pad + MINSIZE + pagesz)
+      break;
+    ar_ptr->system_mem -= heap->size;
+    arena_mem -= heap->size;
+    delete_heap(heap);
+    heap = prev_heap;
+    if(!prev_inuse(p)) { /* consolidate backward */
+      p = prev_chunk(p);
+      unlink(p, bck, fwd);
+    }
+    assert(((unsigned long)((char*)p + new_size) & (pagesz-1)) == 0);
+    assert( ((char*)p + new_size) == ((char*)heap + heap->size) );
+    top(ar_ptr) = top_chunk = p;
+    set_head(top_chunk, new_size | PREV_INUSE);
+    /*check_chunk(ar_ptr, top_chunk);*/
+  }
+  top_size = chunksize(top_chunk);
+  extra = ((top_size - pad - MINSIZE + (pagesz-1))/pagesz - 1) * pagesz;
+  if(extra < (long)pagesz)
+    return 0;
+  /* Try to shrink. */
+  if(grow_heap(heap, -extra) != 0)
+    return 0;
+  ar_ptr->system_mem -= extra;
+  arena_mem -= extra;
+
+  /* Success. Adjust top accordingly. */
+  set_head(top_chunk, (top_size - extra) | PREV_INUSE);
+  /*check_chunk(ar_ptr, top_chunk);*/
+  return 1;
+}
+
+static mstate
+internal_function
+#if __STD_C
+arena_get2(mstate a_tsd, size_t size)
+#else
+arena_get2(a_tsd, size) mstate a_tsd; size_t size;
+#endif
+{
+  mstate a;
+  int err;
+
+  if(!a_tsd)
+    a = a_tsd = &main_arena;
+  else {
+    a = a_tsd->next;
+    if(!a) {
+      /* This can only happen while initializing the new arena. */
+      (void)mutex_lock(&main_arena.mutex);
+      THREAD_STAT(++(main_arena.stat_lock_wait));
+      return &main_arena;
+    }
+  }
+
+  /* Check the global, circularly linked list for available arenas. */
+ repeat:
+  do {
+    if(!mutex_trylock(&a->mutex)) {
+      THREAD_STAT(++(a->stat_lock_loop));
+      tsd_setspecific(arena_key, (Void_t *)a);
+      return a;
+    }
+    a = a->next;
+  } while(a != a_tsd);
+
+  /* If not even the list_lock can be obtained, try again.  This can
+     happen during `atfork', or for example on systems where thread
+     creation makes it temporarily impossible to obtain _any_
+     locks. */
+  if(mutex_trylock(&list_lock)) {
+    a = a_tsd;
+    goto repeat;
+  }
+  (void)mutex_unlock(&list_lock);
+
+  /* Nothing immediately available, so generate a new arena.  */
+  a = _int_new_arena(size);
+  if(!a)
+    return 0;
+
+  tsd_setspecific(arena_key, (Void_t *)a);
+  mutex_init(&a->mutex);
+  err = mutex_lock(&a->mutex); /* remember result */
+
+  /* Add the new arena to the global list.  */
+  (void)mutex_lock(&list_lock);
+  a->next = main_arena.next;
+  main_arena.next = a;
+  (void)mutex_unlock(&list_lock);
+
+  if(err) /* locking failed; keep arena for further attempts later */
+    return 0;
+
+  THREAD_STAT(++(a->stat_lock_loop));
+  return a;
+}
+
+/* Create a new arena with initial size "size".  */
+
+mstate
+_int_new_arena(size_t size)
+{
+  mstate a;
+  heap_info *h;
+  char *ptr;
+  unsigned long misalign;
+
+  h = new_heap(size + (sizeof(*h) + sizeof(*a) + MALLOC_ALIGNMENT),
+	       mp_.top_pad);
+  if(!h) {
+    /* Maybe size is too large to fit in a single heap.  So, just try
+       to create a minimally-sized arena and let _int_malloc() attempt
+       to deal with the large request via mmap_chunk().  */
+    h = new_heap(sizeof(*h) + sizeof(*a) + MALLOC_ALIGNMENT, mp_.top_pad);
+    if(!h)
+      return 0;
+  }
+  a = h->ar_ptr = (mstate)(h+1);
+  malloc_init_state(a);
+  /*a->next = NULL;*/
+  a->system_mem = a->max_system_mem = h->size;
+  arena_mem += h->size;
+#ifdef NO_THREADS
+  if((unsigned long)(mp_.mmapped_mem + arena_mem + main_arena.system_mem) >
+     mp_.max_total_mem)
+    mp_.max_total_mem = mp_.mmapped_mem + arena_mem + main_arena.system_mem;
+#endif
+
+  /* Set up the top chunk, with proper alignment. */
+  ptr = (char *)(a + 1);
+  misalign = (unsigned long)chunk2mem(ptr) & MALLOC_ALIGN_MASK;
+  if (misalign > 0)
+    ptr += MALLOC_ALIGNMENT - misalign;
+  top(a) = (mchunkptr)ptr;
+  set_head(top(a), (((char*)h + h->size) - ptr) | PREV_INUSE);
+
+  return a;
+}
+
+#endif /* USE_ARENAS */
+
+/*
+ * Local variables:
+ * c-basic-offset: 2
+ * End:
+ */
diff --git a/malloc/hooks.c b/malloc/hooks.c
new file mode 100644
index 0000000000..28629eff80
--- /dev/null
+++ b/malloc/hooks.c
@@ -0,0 +1,631 @@
+/* Malloc implementation for multiple threads without lock contention.
+   Copyright (C) 2001 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Wolfram Gloger <wg@malloc.de>, 2001.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Library General Public License for more details.
+
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
+
+/* $Id$ */
+
+#ifndef weak_variable
+#define weak_variable /**/
+#endif
+
+#ifndef DEFAULT_CHECK_ACTION
+#define DEFAULT_CHECK_ACTION 1
+#endif
+
+/* What to do if the standard debugging hooks are in place and a
+   corrupt pointer is detected: do nothing (0), print an error message
+   (1), or call abort() (2). */
+
+/* Hooks for debugging versions.  The initial hooks just call the
+   initialization routine, then do the normal work. */
+
+static Void_t*
+#if __STD_C
+malloc_hook_ini(size_t sz, const __malloc_ptr_t caller)
+#else
+malloc_hook_ini(sz, caller)
+     size_t sz; const __malloc_ptr_t caller;
+#endif
+{
+  __malloc_hook = NULL;
+  ptmalloc_init();
+  return public_mALLOc(sz);
+}
+
+static Void_t*
+#if __STD_C
+realloc_hook_ini(Void_t* ptr, size_t sz, const __malloc_ptr_t caller)
+#else
+realloc_hook_ini(ptr, sz, caller)
+     Void_t* ptr; size_t sz; const __malloc_ptr_t caller;
+#endif
+{
+  __malloc_hook = NULL;
+  __realloc_hook = NULL;
+  ptmalloc_init();
+  return public_rEALLOc(ptr, sz);
+}
+
+static Void_t*
+#if __STD_C
+memalign_hook_ini(size_t alignment, size_t sz, const __malloc_ptr_t caller)
+#else
+memalign_hook_ini(alignment, sz, caller)
+     size_t alignment; size_t sz; const __malloc_ptr_t caller;
+#endif
+{
+  __memalign_hook = NULL;
+  ptmalloc_init();
+  return public_mEMALIGn(alignment, sz);
+}
+
+void weak_variable (*__malloc_initialize_hook) __MALLOC_P ((void)) = NULL;
+void weak_variable (*__free_hook) __MALLOC_P ((__malloc_ptr_t __ptr,
+					       const __malloc_ptr_t)) = NULL;
+__malloc_ptr_t weak_variable (*__malloc_hook)
+ __MALLOC_P ((size_t __size, const __malloc_ptr_t)) = malloc_hook_ini;
+__malloc_ptr_t weak_variable (*__realloc_hook)
+ __MALLOC_P ((__malloc_ptr_t __ptr, size_t __size, const __malloc_ptr_t))
+     = realloc_hook_ini;
+__malloc_ptr_t weak_variable (*__memalign_hook)
+ __MALLOC_P ((size_t __alignment, size_t __size, const __malloc_ptr_t))
+     = memalign_hook_ini;
+void weak_variable (*__after_morecore_hook) __MALLOC_P ((void)) = NULL;
+
+
+static int check_action = DEFAULT_CHECK_ACTION;
+
+/* Whether we are using malloc checking.  */
+static int using_malloc_checking;
+
+/* A flag that is set by malloc_set_state, to signal that malloc checking
+   must not be enabled on the request from the user (via the MALLOC_CHECK_
+   environment variable).  It is reset by __malloc_check_init to tell
+   malloc_set_state that the user has requested malloc checking.
+
+   The purpose of this flag is to make sure that malloc checking is not
+   enabled when the heap to be restored was constructed without malloc
+   checking, and thus does not contain the required magic bytes.
+   Otherwise the heap would be corrupted by calls to free and realloc.  If
+   it turns out that the heap was created with malloc checking and the
+   user has requested it malloc_set_state just calls __malloc_check_init
+   again to enable it.  On the other hand, reusing such a heap without
+   further malloc checking is safe.  */
+static int disallow_malloc_check;
+
+/* Activate a standard set of debugging hooks. */
+void
+__malloc_check_init()
+{
+  if (disallow_malloc_check) {
+    disallow_malloc_check = 0;
+    return;
+  }
+  using_malloc_checking = 1;
+  __malloc_hook = malloc_check;
+  __free_hook = free_check;
+  __realloc_hook = realloc_check;
+  __memalign_hook = memalign_check;
+  if(check_action & 1)
+    fprintf(stderr, "malloc: using debugging hooks\n");
+}
+
+/* A simple, standard set of debugging hooks.  Overhead is `only' one
+   byte per chunk; still this will catch most cases of double frees or
+   overruns.  The goal here is to avoid obscure crashes due to invalid
+   usage, unlike in the MALLOC_DEBUG code. */
+
+#define MAGICBYTE(p) ( ( ((size_t)p >> 3) ^ ((size_t)p >> 11)) & 0xFF )
+
+/* Instrument a chunk with overrun detector byte(s) and convert it
+   into a user pointer with requested size sz. */
+
+static Void_t*
+internal_function
+#if __STD_C
+mem2mem_check(Void_t *ptr, size_t sz)
+#else
+mem2mem_check(ptr, sz) Void_t *ptr; size_t sz;
+#endif
+{
+  mchunkptr p;
+  unsigned char* m_ptr = (unsigned char*)BOUNDED_N(ptr, sz);
+  size_t i;
+
+  if (!ptr)
+    return ptr;
+  p = mem2chunk(ptr);
+  for(i = chunksize(p) - (chunk_is_mmapped(p) ? 2*SIZE_SZ+1 : SIZE_SZ+1);
+      i > sz;
+      i -= 0xFF) {
+    if(i-sz < 0x100) {
+      m_ptr[i] = (unsigned char)(i-sz);
+      break;
+    }
+    m_ptr[i] = 0xFF;
+  }
+  m_ptr[sz] = MAGICBYTE(p);
+  return (Void_t*)m_ptr;
+}
+
+/* Convert a pointer to be free()d or realloc()ed to a valid chunk
+   pointer.  If the provided pointer is not valid, return NULL. */
+
+static mchunkptr
+internal_function
+#if __STD_C
+mem2chunk_check(Void_t* mem)
+#else
+mem2chunk_check(mem) Void_t* mem;
+#endif
+{
+  mchunkptr p;
+  INTERNAL_SIZE_T sz, c;
+  unsigned char magic;
+
+  p = mem2chunk(mem);
+  if(!aligned_OK(p)) return NULL;
+  if( (char*)p>=mp_.sbrk_base &&
+      (char*)p<(mp_.sbrk_base+main_arena.system_mem) ) {
+    /* Must be a chunk in conventional heap memory. */
+    if(chunk_is_mmapped(p) ||
+       ( (sz = chunksize(p)),
+	 ((char*)p + sz)>=(mp_.sbrk_base+main_arena.system_mem) ) ||
+       sz<MINSIZE || sz&MALLOC_ALIGN_MASK || !inuse(p) ||
+       ( !prev_inuse(p) && (p->prev_size&MALLOC_ALIGN_MASK ||
+                            (long)prev_chunk(p)<(long)mp_.sbrk_base ||
+                            next_chunk(prev_chunk(p))!=p) ))
+      return NULL;
+    magic = MAGICBYTE(p);
+    for(sz += SIZE_SZ-1; (c = ((unsigned char*)p)[sz]) != magic; sz -= c) {
+      if(c<=0 || sz<(c+2*SIZE_SZ)) return NULL;
+    }
+    ((unsigned char*)p)[sz] ^= 0xFF;
+  } else {
+    unsigned long offset, page_mask = malloc_getpagesize-1;
+
+    /* mmap()ed chunks have MALLOC_ALIGNMENT or higher power-of-two
+       alignment relative to the beginning of a page.  Check this
+       first. */
+    offset = (unsigned long)mem & page_mask;
+    if((offset!=MALLOC_ALIGNMENT && offset!=0 && offset!=0x10 &&
+        offset!=0x20 && offset!=0x40 && offset!=0x80 && offset!=0x100 &&
+        offset!=0x200 && offset!=0x400 && offset!=0x800 && offset!=0x1000 &&
+        offset<0x2000) ||
+       !chunk_is_mmapped(p) || (p->size & PREV_INUSE) ||
+       ( (((unsigned long)p - p->prev_size) & page_mask) != 0 ) ||
+       ( (sz = chunksize(p)), ((p->prev_size + sz) & page_mask) != 0 ) )
+      return NULL;
+    magic = MAGICBYTE(p);
+    for(sz -= 1; (c = ((unsigned char*)p)[sz]) != magic; sz -= c) {
+      if(c<=0 || sz<(c+2*SIZE_SZ)) return NULL;
+    }
+    ((unsigned char*)p)[sz] ^= 0xFF;
+  }
+  return p;
+}
+
+/* Check for corruption of the top chunk, and try to recover if
+   necessary. */
+
+static int
+internal_function
+#if __STD_C
+top_check(void)
+#else
+top_check()
+#endif
+{
+  mchunkptr t = top(&main_arena);
+  char* brk, * new_brk;
+  INTERNAL_SIZE_T front_misalign, sbrk_size;
+  unsigned long pagesz = malloc_getpagesize;
+
+  if((char*)t + chunksize(t) == mp_.sbrk_base + main_arena.system_mem ||
+     t == initial_top(&main_arena)) return 0;
+
+  if(check_action & 1)
+    fprintf(stderr, "malloc: top chunk is corrupt\n");
+  if(check_action & 2)
+    abort();
+
+  /* Try to set up a new top chunk. */
+  brk = MORECORE(0);
+  front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
+  if (front_misalign > 0)
+    front_misalign = MALLOC_ALIGNMENT - front_misalign;
+  sbrk_size = front_misalign + mp_.top_pad + MINSIZE;
+  sbrk_size += pagesz - ((unsigned long)(brk + sbrk_size) & (pagesz - 1));
+  new_brk = (char*)(MORECORE (sbrk_size));
+  if (new_brk == (char*)(MORECORE_FAILURE)) return -1;
+  /* Call the `morecore' hook if necessary.  */
+  if (__after_morecore_hook)
+    (*__after_morecore_hook) ();
+  main_arena.system_mem = (new_brk - mp_.sbrk_base) + sbrk_size;
+
+  top(&main_arena) = (mchunkptr)(brk + front_misalign);
+  set_head(top(&main_arena), (sbrk_size - front_misalign) | PREV_INUSE);
+
+  return 0;
+}
+
+static Void_t*
+#if __STD_C
+malloc_check(size_t sz, const Void_t *caller)
+#else
+malloc_check(sz, caller) size_t sz; const Void_t *caller;
+#endif
+{
+  Void_t *victim;
+
+  (void)mutex_lock(&main_arena.mutex);
+  victim = (top_check() >= 0) ? _int_malloc(&main_arena, sz+1) : NULL;
+  (void)mutex_unlock(&main_arena.mutex);
+  return mem2mem_check(victim, sz);
+}
+
+static void
+#if __STD_C
+free_check(Void_t* mem, const Void_t *caller)
+#else
+free_check(mem, caller) Void_t* mem; const Void_t *caller;
+#endif
+{
+  mchunkptr p;
+
+  if(!mem) return;
+  (void)mutex_lock(&main_arena.mutex);
+  p = mem2chunk_check(mem);
+  if(!p) {
+    (void)mutex_unlock(&main_arena.mutex);
+    if(check_action & 1)
+      fprintf(stderr, "free(): invalid pointer %p!\n", mem);
+    if(check_action & 2)
+      abort();
+    return;
+  }
+#if HAVE_MMAP
+  if (chunk_is_mmapped(p)) {
+    (void)mutex_unlock(&main_arena.mutex);
+    munmap_chunk(p);
+    return;
+  }
+#endif
+#if 0 /* Erase freed memory. */
+  memset(mem, 0, chunksize(p) - (SIZE_SZ+1));
+#endif
+  _int_free(&main_arena, mem);
+  (void)mutex_unlock(&main_arena.mutex);
+}
+
+static Void_t*
+#if __STD_C
+realloc_check(Void_t* oldmem, size_t bytes, const Void_t *caller)
+#else
+realloc_check(oldmem, bytes, caller)
+     Void_t* oldmem; size_t bytes; const Void_t *caller;
+#endif
+{
+  mchunkptr oldp, newp = 0;
+  INTERNAL_SIZE_T nb, oldsize;
+  Void_t* newmem = 0;
+
+  if (oldmem == 0) return malloc_check(bytes, NULL);
+  (void)mutex_lock(&main_arena.mutex);
+  oldp = mem2chunk_check(oldmem);
+  (void)mutex_unlock(&main_arena.mutex);
+  if(!oldp) {
+    if(check_action & 1)
+      fprintf(stderr, "realloc(): invalid pointer %p!\n", oldmem);
+    if(check_action & 2)
+      abort();
+    return malloc_check(bytes, NULL);
+  }
+  oldsize = chunksize(oldp);
+
+  checked_request2size(bytes+1, nb);
+  (void)mutex_lock(&main_arena.mutex);
+
+#if HAVE_MMAP
+  if (chunk_is_mmapped(oldp)) {
+#if HAVE_MREMAP
+    newp = mremap_chunk(oldp, nb);
+    if(!newp) {
+#endif
+      /* Note the extra SIZE_SZ overhead. */
+      if(oldsize - SIZE_SZ >= nb)
+	newmem = oldmem; /* do nothing */
+      else {
+        /* Must alloc, copy, free. */
+        if (top_check() >= 0)
+	  newmem = _int_malloc(&main_arena, bytes+1);
+        if (newmem) {
+          MALLOC_COPY(BOUNDED_N(newmem, bytes+1), oldmem, oldsize - 2*SIZE_SZ);
+          munmap_chunk(oldp);
+        }
+      }
+#if HAVE_MREMAP
+    }
+#endif
+  } else {
+#endif /* HAVE_MMAP */
+    if (top_check() >= 0)
+      newmem = _int_realloc(&main_arena, oldmem, bytes+1);
+#if 0 /* Erase freed memory. */
+    if(newmem)
+      newp = mem2chunk(newmem);
+    nb = chunksize(newp);
+    if(oldp<newp || oldp>=chunk_at_offset(newp, nb)) {
+      memset((char*)oldmem + 2*sizeof(mbinptr), 0,
+             oldsize - (2*sizeof(mbinptr)+2*SIZE_SZ+1));
+    } else if(nb > oldsize+SIZE_SZ) {
+      memset((char*)BOUNDED_N(chunk2mem(newp), bytes) + oldsize,
+	     0, nb - (oldsize+SIZE_SZ));
+    }
+#endif
+#if HAVE_MMAP
+  }
+#endif
+  (void)mutex_unlock(&main_arena.mutex);
+
+  return mem2mem_check(newmem, bytes);
+}
+
+static Void_t*
+#if __STD_C
+memalign_check(size_t alignment, size_t bytes, const Void_t *caller)
+#else
+memalign_check(alignment, bytes, caller)
+     size_t alignment; size_t bytes; const Void_t *caller;
+#endif
+{
+  INTERNAL_SIZE_T nb;
+  Void_t* mem;
+
+  if (alignment <= MALLOC_ALIGNMENT) return malloc_check(bytes, NULL);
+  if (alignment <  MINSIZE) alignment = MINSIZE;
+
+  checked_request2size(bytes+1, nb);
+  (void)mutex_lock(&main_arena.mutex);
+  mem = (top_check() >= 0) ? _int_memalign(&main_arena, alignment, bytes+1) :
+    NULL;
+  (void)mutex_unlock(&main_arena.mutex);
+  return mem2mem_check(mem, bytes);
+}
+
+#ifndef NO_THREADS
+
+/* The following hooks are used when the global initialization in
+   ptmalloc_init() hasn't completed yet. */
+
+static Void_t*
+#if __STD_C
+malloc_starter(size_t sz, const Void_t *caller)
+#else
+malloc_starter(sz, caller) size_t sz; const Void_t *caller;
+#endif
+{
+  Void_t* victim;
+
+  victim = _int_malloc(&main_arena, sz);
+
+  return victim ? BOUNDED_N(victim, sz) : 0;
+}
+
+static void
+#if __STD_C
+free_starter(Void_t* mem, const Void_t *caller)
+#else
+free_starter(mem, caller) Void_t* mem; const Void_t *caller;
+#endif
+{
+  mchunkptr p;
+
+  if(!mem) return;
+  p = mem2chunk(mem);
+#if HAVE_MMAP
+  if (chunk_is_mmapped(p)) {
+    munmap_chunk(p);
+    return;
+  }
+#endif
+  _int_free(&main_arena, mem);
+}
+
+#endif /* NO_THREADS */
+
+
+/* Get/set state: malloc_get_state() records the current state of all
+   malloc variables (_except_ for the actual heap contents and `hook'
+   function pointers) in a system dependent, opaque data structure.
+   This data structure is dynamically allocated and can be free()d
+   after use.  malloc_set_state() restores the state of all malloc
+   variables to the previously obtained state.  This is especially
+   useful when using this malloc as part of a shared library, and when
+   the heap contents are saved/restored via some other method.  The
+   primary example for this is GNU Emacs with its `dumping' procedure.
+   `Hook' function pointers are never saved or restored by these
+   functions, with two exceptions: If malloc checking was in use when
+   malloc_get_state() was called, then malloc_set_state() calls
+   __malloc_check_init() if possible; if malloc checking was not in
+   use in the recorded state but the user requested malloc checking,
+   then the hooks are reset to 0.  */
+
+#define MALLOC_STATE_MAGIC   0x444c4541l
+#define MALLOC_STATE_VERSION (0*0x100l + 2l) /* major*0x100 + minor */
+
+struct malloc_save_state {
+  long          magic;
+  long          version;
+  mbinptr       av[NBINS * 2 + 2];
+  char*         sbrk_base;
+  int           sbrked_mem_bytes;
+  unsigned long trim_threshold;
+  unsigned long top_pad;
+  unsigned int  n_mmaps_max;
+  unsigned long mmap_threshold;
+  int           check_action;
+  unsigned long max_sbrked_mem;
+  unsigned long max_total_mem;
+  unsigned int  n_mmaps;
+  unsigned int  max_n_mmaps;
+  unsigned long mmapped_mem;
+  unsigned long max_mmapped_mem;
+  int           using_malloc_checking;
+};
+
+Void_t*
+public_gET_STATe(void)
+{
+  struct malloc_save_state* ms;
+  int i;
+  mbinptr b;
+
+  ms = (struct malloc_save_state*)public_mALLOc(sizeof(*ms));
+  if (!ms)
+    return 0;
+  (void)mutex_lock(&main_arena.mutex);
+  malloc_consolidate(&main_arena);
+  ms->magic = MALLOC_STATE_MAGIC;
+  ms->version = MALLOC_STATE_VERSION;
+  ms->av[0] = 0;
+  ms->av[1] = 0; /* used to be binblocks, now no longer used */
+  ms->av[2] = top(&main_arena);
+  ms->av[3] = 0; /* used to be undefined */
+  for(i=1; i<NBINS; i++) {
+    b = bin_at(&main_arena, i);
+    if(first(b) == b)
+      ms->av[2*i+2] = ms->av[2*i+3] = 0; /* empty bin */
+    else {
+      ms->av[2*i+2] = first(b);
+      ms->av[2*i+3] = last(b);
+    }
+  }
+  ms->sbrk_base = mp_.sbrk_base;
+  ms->sbrked_mem_bytes = main_arena.system_mem;
+  ms->trim_threshold = mp_.trim_threshold;
+  ms->top_pad = mp_.top_pad;
+  ms->n_mmaps_max = mp_.n_mmaps_max;
+  ms->mmap_threshold = mp_.mmap_threshold;
+  ms->check_action = check_action;
+  ms->max_sbrked_mem = main_arena.max_system_mem;
+#ifdef NO_THREADS
+  ms->max_total_mem = max_total_mem;
+#else
+  ms->max_total_mem = 0;
+#endif
+  ms->n_mmaps = mp_.n_mmaps;
+  ms->max_n_mmaps = mp_.max_n_mmaps;
+  ms->mmapped_mem = mp_.mmapped_mem;
+  ms->max_mmapped_mem = mp_.max_mmapped_mem;
+  ms->using_malloc_checking = using_malloc_checking;
+  (void)mutex_unlock(&main_arena.mutex);
+  return (Void_t*)ms;
+}
+
+int
+public_sET_STATe(Void_t* msptr)
+{
+  struct malloc_save_state* ms = (struct malloc_save_state*)msptr;
+  int i;
+  mbinptr b;
+
+  disallow_malloc_check = 1;
+  ptmalloc_init();
+  if(ms->magic != MALLOC_STATE_MAGIC) return -1;
+  /* Must fail if the major version is too high. */
+  if((ms->version & ~0xffl) > (MALLOC_STATE_VERSION & ~0xffl)) return -2;
+  (void)mutex_lock(&main_arena.mutex);
+  /* There are no fastchunks.  */
+  clear_fastchunks(&main_arena);
+  set_max_fast(&main_arena, DEFAULT_MXFAST);
+  for (i=0; i<NFASTBINS; ++i)
+    main_arena.fastbins[i] = 0;
+  for (i=0; i<BINMAPSIZE; ++i)
+    main_arena.binmap[i] = 0;
+  top(&main_arena) = ms->av[2];
+  main_arena.last_remainder = 0;
+  for(i=1; i<NBINS; i++) {
+    b = bin_at(&main_arena, i);
+    if(ms->av[2*i+2] == 0) {
+      assert(ms->av[2*i+3] == 0);
+      first(b) = last(b) = b;
+    } else {
+      if(i<NSMALLBINS || (largebin_index(chunksize(ms->av[2*i+2]))==i &&
+			  largebin_index(chunksize(ms->av[2*i+3]))==i)) {
+	first(b) = ms->av[2*i+2];
+	last(b) = ms->av[2*i+3];
+	/* Make sure the links to the bins within the heap are correct.  */
+	first(b)->bk = b;
+	last(b)->fd = b;
+	/* Set bit in binblocks.  */
+	mark_bin(&main_arena, i);
+      } else {
+	/* Oops, index computation from chunksize must have changed.
+           Link the whole list into unsorted_chunks.  */
+	first(b) = last(b) = b;
+	b = unsorted_chunks(&main_arena);
+	ms->av[2*i+2]->bk = b;
+	ms->av[2*i+3]->fd = b->fd;
+	b->fd->bk = ms->av[2*i+3];
+	b->fd = ms->av[2*i+2];
+      }
+    }
+  }
+  mp_.sbrk_base = ms->sbrk_base;
+  main_arena.system_mem = ms->sbrked_mem_bytes;
+  mp_.trim_threshold = ms->trim_threshold;
+  mp_.top_pad = ms->top_pad;
+  mp_.n_mmaps_max = ms->n_mmaps_max;
+  mp_.mmap_threshold = ms->mmap_threshold;
+  check_action = ms->check_action;
+  main_arena.max_system_mem = ms->max_sbrked_mem;
+#ifdef NO_THREADS
+  mp_.max_total_mem = ms->max_total_mem;
+#endif
+  mp_.n_mmaps = ms->n_mmaps;
+  mp_.max_n_mmaps = ms->max_n_mmaps;
+  mp_.mmapped_mem = ms->mmapped_mem;
+  mp_.max_mmapped_mem = ms->max_mmapped_mem;
+  /* add version-dependent code here */
+  if (ms->version >= 1) {
+    /* Check whether it is safe to enable malloc checking, or whether
+       it is necessary to disable it.  */
+    if (ms->using_malloc_checking && !using_malloc_checking &&
+        !disallow_malloc_check)
+      __malloc_check_init ();
+    else if (!ms->using_malloc_checking && using_malloc_checking) {
+      __malloc_hook = 0;
+      __free_hook = 0;
+      __realloc_hook = 0;
+      __memalign_hook = 0;
+      using_malloc_checking = 0;
+    }
+  }
+  check_malloc_state(&main_arena);
+
+  (void)mutex_unlock(&main_arena.mutex);
+  return 0;
+}
+
+/*
+ * Local variables:
+ * c-basic-offset: 2
+ * End:
+ */
diff --git a/malloc/malloc.c b/malloc/malloc.c
index 8279ddaf22..e663f84707 100644
--- a/malloc/malloc.c
+++ b/malloc/malloc.c
@@ -1,32 +1,47 @@
 /* Malloc implementation for multiple threads without lock contention.
    Copyright (C) 1996,1997,1998,1999,2000,2001 Free Software Foundation, Inc.
    This file is part of the GNU C Library.
-   Contributed by Wolfram Gloger <wmglo@dent.med.uni-muenchen.de>
-   and Doug Lea <dl@cs.oswego.edu>, 1996.
+   Contributed by Wolfram Gloger <wg@malloc.de>
+   and Doug Lea <dl@cs.oswego.edu>, 2001.
 
    The GNU C Library is free software; you can redistribute it and/or
-   modify it under the terms of the GNU Lesser General Public
-   License as published by the Free Software Foundation; either
-   version 2.1 of the License, or (at your option) any later version.
+   modify it under the terms of the GNU Library General Public License as
+   published by the Free Software Foundation; either version 2 of the
+   License, or (at your option) any later version.
 
    The GNU C Library is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
-   Lesser General Public License for more details.
+   Library General Public License for more details.
 
-   You should have received a copy of the GNU Lesser General Public
-   License along with the GNU C Library; if not, write to the Free
-   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
-   02111-1307 USA.  */
+   You should have received a copy of the GNU Library General Public
+   License along with the GNU C Library; see the file COPYING.LIB.  If not,
+   write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
+   Boston, MA 02111-1307, USA.  */
 
-/* $Id$
+/*
+  This is a version (aka ptmalloc2) of malloc/free/realloc written by
+  Doug Lea and adapted to multiple threads/arenas by Wolfram Gloger.
+
+* Version ptmalloc2-20011215
+  $Id$
+  based on:
+  VERSION 2.7.0 Sun Mar 11 14:14:06 2001  Doug Lea  (dl at gee)
 
-  This work is mainly derived from malloc-2.6.4 by Doug Lea
-  <dl@cs.oswego.edu>, which is available from:
+   Note: There may be an updated version of this malloc obtainable at
+           http://www.malloc.de/malloc/ptmalloc2.tar.gz
+         Check before installing!
 
-                 ftp://g.oswego.edu/pub/misc/malloc.c
+* Quickstart
 
-  Most of the original comments are reproduced in the code below.
+  In order to compile this implementation, a Makefile is provided with
+  the ptmalloc2 distribution, which has pre-defined targets for some
+  popular systems (e.g. "make posix" for Posix threads).  All that is
+  typically required with regard to compiler flags is the selection of
+  the thread package via defining one out of USE_PTHREADS, USE_THR or
+  USE_SPROC.  Check the thread-m.h file for what effects this has.
+  Many/most systems will additionally require USE_TSD_DATA_HACK to be
+  defined, so this is the default for "make posix".
 
 * Why use this malloc?
 
@@ -34,85 +49,62 @@
   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 a high-level description, see
-     http://g.oswego.edu/dl/html/malloc.html
-
-  On many systems, the standard malloc implementation is by itself not
-  thread-safe, and therefore wrapped with a single global lock around
-  all malloc-related functions.  In some applications, especially with
-  multiple available processors, this can lead to contention problems
-  and bad performance.  This malloc version was designed with the goal
-  to avoid waiting for locks as much as possible.  Statistics indicate
-  that this goal is achieved in many cases.
-
-* Synopsis of public routines
-
-  (Much fuller descriptions are contained in the program documentation below.)
-
-  ptmalloc_init();
-     Initialize global configuration.  When compiled for multiple threads,
-     this function must be called once before any other function in the
-     package.  It is not required otherwise.  It is called automatically
-     in the Linux/GNU C libray or when compiling with MALLOC_HOOKS.
-  malloc(size_t n);
-     Return a pointer to a newly allocated chunk of at least n bytes, or null
-     if no space is available.
-  free(Void_t* p);
-     Release the chunk of memory pointed to by p, or no effect if p is null.
-  realloc(Void_t* p, size_t n);
-     Return a pointer to a chunk of size n that contains the same data
-     as does chunk p up to the minimum of (n, p's size) bytes, or null
-     if no space is available. The returned pointer may or may not be
-     the same as p. If p is null, equivalent to malloc.  Unless the
-     #define REALLOC_ZERO_BYTES_FREES below is set, realloc with a
-     size argument of zero (re)allocates a minimum-sized chunk.
-  memalign(size_t alignment, size_t n);
-     Return a pointer to a newly allocated chunk of n bytes, aligned
-     in accord with the alignment argument, which must be a power of
-     two.
-  valloc(size_t n);
-     Equivalent to memalign(pagesize, n), where pagesize is the page
-     size of the system (or as near to this as can be figured out from
-     all the includes/defines below.)
-  pvalloc(size_t n);
-     Equivalent to valloc(minimum-page-that-holds(n)), that is,
-     round up n to nearest pagesize.
-  calloc(size_t unit, size_t quantity);
-     Returns a pointer to quantity * unit bytes, with all locations
-     set to zero.
-  cfree(Void_t* p);
-     Equivalent to free(p).
-  malloc_trim(size_t pad);
-     Release all but pad bytes of freed top-most memory back
-     to the system. Return 1 if successful, else 0.
-  malloc_usable_size(Void_t* p);
-     Report the number usable allocated bytes associated with allocated
-     chunk p. This may or may not report more bytes than were requested,
-     due to alignment and minimum size constraints.
-  malloc_stats();
-     Prints brief summary statistics on stderr.
-  mallinfo()
-     Returns (by copy) a struct containing various summary statistics.
-  mallopt(int parameter_number, int parameter_value)
-     Changes one of the tunable parameters described below. Returns
-     1 if successful in changing the parameter, else 0.
+  allocator for malloc-intensive programs.
+
+  The main properties of the algorithms are:
+  * For large (>= 512 bytes) requests, it is a pure best-fit allocator,
+    with ties normally decided via FIFO (i.e. least recently used).
+  * For small (<= 64 bytes by default) requests, it is a caching
+    allocator, that maintains pools of quickly recycled chunks.
+  * In between, and for combinations of large and small requests, it does
+    the best it can trying to meet both goals at once.
+  * For very large requests (>= 128KB by default), it relies on system
+    memory mapping facilities, if supported.
+
+  For a longer but slightly out of date high-level description, see
+     http://gee.cs.oswego.edu/dl/html/malloc.html
+
+  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 in order to
+  customize settings or to avoid overheads associated with library
+  versions.
+
+* Contents, described in more detail in "description of public routines" below.
+
+  Standard (ANSI/SVID/...)  functions:
+    malloc(size_t n);
+    calloc(size_t n_elements, size_t element_size);
+    free(Void_t* p);
+    realloc(Void_t* p, size_t n);
+    memalign(size_t alignment, size_t n);
+    valloc(size_t n);
+    mallinfo()
+    mallopt(int parameter_number, int parameter_value)
+
+  Additional functions:
+    independent_calloc(size_t n_elements, size_t size, Void_t* chunks[]);
+    independent_comalloc(size_t n_elements, size_t sizes[], Void_t* chunks[]);
+    pvalloc(size_t n);
+    cfree(Void_t* p);
+    malloc_trim(size_t pad);
+    malloc_usable_size(Void_t* p);
+    malloc_stats();
 
 * Vital statistics:
 
-  Alignment:                            8-byte
-       8 byte alignment is currently hardwired into the design.  This
-       seems to suffice for all current machines and C compilers.
-
-  Assumed pointer representation:       4 or 8 bytes
-       Code for 8-byte pointers is untested by me but has worked
-       reliably by Wolfram Gloger, who contributed most of the
-       changes supporting this.
-
-  Assumed size_t  representation:       4 or 8 bytes
+  Supported pointer representation:       4 or 8 bytes
+  Supported size_t  representation:       4 or 8 bytes 
        Note that size_t is allowed to be 4 bytes even if pointers are 8.
+       You can adjust this by defining INTERNAL_SIZE_T
 
-  Minimum overhead per allocated chunk: 4 or 8 bytes
-       Each malloced chunk has a hidden overhead of 4 bytes holding size
+  Alignment:                              2 * sizeof(size_t) (default)
+       (i.e., 8 byte alignment with 4byte size_t). This suffices for
+       nearly all current machines and C compilers. However, you can
+       define MALLOC_ALIGNMENT to be wider than this if necessary.
+
+  Minimum overhead per allocated chunk:   4 or 8 bytes
+       Each malloced chunk has a hidden word of overhead holding size
        and status information.
 
   Minimum allocated size: 4-byte ptrs:  16 bytes    (including 4 overhead)
@@ -120,182 +112,136 @@
 
        When a chunk is freed, 12 (for 4byte ptrs) or 20 (for 8 byte
        ptrs but 4 byte size) or 24 (for 8/8) additional bytes are
-       needed; 4 (8) for a trailing size field
-       and 8 (16) bytes for free list pointers. Thus, the minimum
-       allocatable size is 16/24/32 bytes.
+       needed; 4 (8) for a trailing size field and 8 (16) bytes for
+       free list pointers. Thus, the minimum allocatable size is
+       16/24/32 bytes.
 
        Even a request for zero bytes (i.e., malloc(0)) returns a
        pointer to something of the minimum allocatable size.
 
-  Maximum allocated size: 4-byte size_t: 2^31 -  8 bytes
-                          8-byte size_t: 2^63 - 16 bytes
+       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 2 *
+       sizeof(size_t) bytes plus the remainder from a system page (the
+       minimal mmap unit); typically 4096 or 8192 bytes.
+
+  Maximum allocated size:  4-byte size_t: 2^32 minus about two pages 
+                           8-byte size_t: 2^64 minus about two pages
 
-       It is assumed that (possibly signed) size_t bit values suffice to
+       It is assumed that (possibly signed) size_t values suffice to
        represent chunk sizes. `Possibly signed' is due to the fact
        that `size_t' may be defined on a system as either a signed or
-       an unsigned type. To be conservative, values that would appear
-       as negative numbers are avoided.
-       Requests for sizes with a negative sign bit will return a
-       minimum-sized chunk.
-
-  Maximum overhead wastage per allocated chunk: normally 15 bytes
-
-       Alignment demands, plus the minimum allocatable size restriction
-       make the normal worst-case wastage 15 bytes (i.e., up to 15
-       more bytes will be allocated than were requested in malloc), with
-       two exceptions:
-         1. Because requests for zero bytes allocate non-zero space,
-            the worst case wastage for a request of zero bytes is 24 bytes.
-         2. For requests >= mmap_threshold that are serviced via
-            mmap(), the worst case wastage is 8 bytes plus the remainder
-            from a system page (the minimal mmap unit); typically 4096 bytes.
-
-* Limitations
-
-    Here are some features that are NOT currently supported
-
-    * No automated mechanism for fully checking that all accesses
-      to malloced memory stay within their bounds.
-    * No support for compaction.
+       an unsigned type. The ISO C standard says that it must be
+       unsigned, but a few systems are known not to adhere to this.
+       Additionally, even when size_t is unsigned, sbrk (which is by
+       default used to obtain memory from system) accepts signed
+       arguments, and may not be able to handle size_t-wide arguments
+       with negative sign bit.  Generally, values that would
+       appear as negative after accounting for overhead and alignment
+       are supported only via mmap(), which does not have this
+       limitation.
+
+       Requests for sizes outside the allowed range will perform an optional
+       failure action and then return null. (Requests may also
+       also fail because a system is out of memory.)
+
+  Thread-safety: thread-safe unless NO_THREADS is defined
+
+  Compliance: I believe it is compliant with the 1997 Single Unix Specification
+       (See http://www.opennc.org). Also SVID/XPG, ANSI C, and probably 
+       others as well.
 
 * Synopsis of compile-time options:
 
     People have reported using previous versions of this malloc on all
     versions of Unix, sometimes by tweaking some of the defines
     below. It has been tested most extensively on Solaris and
-    Linux. People have also reported adapting this malloc for use in
-    stand-alone embedded systems.
-
-    The implementation is in straight, hand-tuned ANSI C.  Among other
-    consequences, it uses a lot of macros.  Because of this, to be at
-    all usable, this code should be compiled using an optimizing compiler
-    (for example gcc -O2) that can simplify expressions and control
-    paths.
-
-  __STD_C                  (default: derived from C compiler defines)
-     Nonzero if using ANSI-standard C compiler, a C++ compiler, or
-     a C compiler sufficiently close to ANSI to get away with it.
-  MALLOC_DEBUG             (default: NOT defined)
-     Define to enable debugging. Adds fairly extensive assertion-based
-     checking to help track down memory errors, but noticeably slows down
-     execution.
-  MALLOC_HOOKS             (default: NOT defined)
-     Define to enable support run-time replacement of the allocation
-     functions through user-defined `hooks'.
-  REALLOC_ZERO_BYTES_FREES (default: defined)
-     Define this if you think that realloc(p, 0) should be equivalent
-     to free(p).  (The C standard requires this behaviour, therefore
-     it is the default.)  Otherwise, since malloc returns a unique
-     pointer for malloc(0), so does realloc(p, 0).
-  HAVE_MEMCPY               (default: defined)
-     Define if you are not otherwise using ANSI STD C, but still
-     have memcpy and memset in your C library and want to use them.
-     Otherwise, simple internal versions are supplied.
-  USE_MEMCPY               (default: 1 if HAVE_MEMCPY is defined, 0 otherwise)
-     Define as 1 if you want the C library versions of memset and
-     memcpy called in realloc and calloc (otherwise macro versions are used).
-     At least on some platforms, the simple macro versions usually
-     outperform libc versions.
-  HAVE_MMAP                 (default: defined as 1)
-     Define to non-zero to optionally make malloc() use mmap() to
-     allocate very large blocks.
-  HAVE_MREMAP                 (default: defined as 0 unless Linux libc set)
-     Define to non-zero to optionally make realloc() use mremap() to
-     reallocate very large blocks.
-  USE_ARENAS                (default: the same as HAVE_MMAP)
-     Enable support for multiple arenas, allocated using mmap().
-  malloc_getpagesize        (default: derived from system #includes)
-     Either a constant or routine call returning the system page size.
-  HAVE_USR_INCLUDE_MALLOC_H (default: NOT defined)
-     Optionally define if you are on a system with a /usr/include/malloc.h
-     that declares struct mallinfo. It is not at all necessary to
-     define this even if you do, but will ensure consistency.
-  INTERNAL_SIZE_T           (default: size_t)
-     Define to a 32-bit type (probably `unsigned int') if you are on a
-     64-bit machine, yet do not want or need to allow malloc requests of
-     greater than 2^31 to be handled. This saves space, especially for
-     very small chunks.
-  _LIBC                     (default: NOT defined)
-     Defined only when compiled as part of the Linux libc/glibc.
-     Also note that there is some odd internal name-mangling via defines
-     (for example, internally, `malloc' is named `mALLOc') needed
-     when compiling in this case. These look funny but don't otherwise
-     affect anything.
-  LACKS_UNISTD_H            (default: undefined)
-     Define this if your system does not have a <unistd.h>.
-  MORECORE                  (default: sbrk)
-     The name of the routine to call to obtain more memory from the system.
-  MORECORE_FAILURE          (default: -1)
-     The value returned upon failure of MORECORE.
-  MORECORE_CLEARS           (default 1)
-     The degree to which the routine mapped to MORECORE zeroes out
-     memory: never (0), only for newly allocated space (1) or always
-     (2).  The distinction between (1) and (2) is necessary because on
-     some systems, if the application first decrements and then
-     increments the break value, the contents of the reallocated space
-     are unspecified.
-  DEFAULT_TRIM_THRESHOLD
-  DEFAULT_TOP_PAD
-  DEFAULT_MMAP_THRESHOLD
-  DEFAULT_MMAP_MAX
-     Default values of tunable parameters (described in detail below)
-     controlling interaction with host system routines (sbrk, mmap, etc).
-     These values may also be changed dynamically via mallopt(). The
-     preset defaults are those that give best performance for typical
-     programs/systems.
-  DEFAULT_CHECK_ACTION
-     When the standard debugging hooks are in place, and a pointer is
-     detected as corrupt, do nothing (0), print an error message (1),
-     or call abort() (2).
-
-
-*/
+    Linux. It is also reported to work on WIN32 platforms.
+    People also report using it in stand-alone embedded systems.
+
+    The implementation is in straight, hand-tuned ANSI C.  It is not
+    at all modular. (Sorry!)  It uses a lot of macros.  To be at all
+    usable, this code should be compiled using an optimizing compiler
+    (for example gcc -O3) that can simplify expressions and control
+    paths. (FAQ: some macros import variables as arguments rather than
+    declare locals because people reported that some debuggers
+    otherwise get confused.)
+
+    OPTION                     DEFAULT VALUE
+
+    Compilation Environment options:
+
+    __STD_C                    derived from C compiler defines
+    WIN32                      NOT defined
+    HAVE_MEMCPY                defined
+    USE_MEMCPY                 1 if HAVE_MEMCPY is defined
+    HAVE_MMAP                  defined as 1 
+    MMAP_CLEARS                1
+    HAVE_MREMAP                0 unless linux defined
+    USE_ARENAS                 the same as HAVE_MMAP
+    malloc_getpagesize         derived from system #includes, or 4096 if not
+    HAVE_USR_INCLUDE_MALLOC_H  NOT defined
+    LACKS_UNISTD_H             NOT defined unless WIN32
+    LACKS_SYS_PARAM_H          NOT defined unless WIN32
+    LACKS_SYS_MMAN_H           NOT defined unless WIN32
+
+    Changing default word sizes:
+
+    INTERNAL_SIZE_T            size_t
+    MALLOC_ALIGNMENT           2 * sizeof(INTERNAL_SIZE_T)
+
+    Configuration and functionality options:
+
+    USE_DL_PREFIX              NOT defined
+    USE_PUBLIC_MALLOC_WRAPPERS NOT defined
+    USE_MALLOC_LOCK            NOT defined
+    MALLOC_DEBUG               NOT defined
+    REALLOC_ZERO_BYTES_FREES   1
+    MALLOC_FAILURE_ACTION      errno = ENOMEM, if __STD_C defined, else no-op
+    TRIM_FASTBINS              0
+
+    Options for customizing MORECORE:
+
+    MORECORE                   sbrk
+    MORECORE_FAILURE           -1
+    MORECORE_CONTIGUOUS        1 
+    MORECORE_CANNOT_TRIM       NOT defined
+    MORECORE_CLEARS            1
+    MMAP_AS_MORECORE_SIZE      (1024 * 1024) 
+
+    Tuning options that are also dynamically changeable via mallopt:
+
+    DEFAULT_MXFAST             64
+    DEFAULT_TRIM_THRESHOLD     128 * 1024
+    DEFAULT_TOP_PAD            0
+    DEFAULT_MMAP_THRESHOLD     128 * 1024
+    DEFAULT_MMAP_MAX           65536
+
+    There are several other #defined constants and macros that you
+    probably don't want to touch unless you are extending or adapting malloc.  */
 
 /*
-
-* Compile-time options for multiple threads:
-
-  USE_PTHREADS, USE_THR, USE_SPROC
-     Define one of these as 1 to select the thread interface:
-     POSIX threads, Solaris threads or SGI sproc's, respectively.
-     If none of these is defined as non-zero, you get a `normal'
-     malloc implementation which is not thread-safe.  Support for
-     multiple threads requires HAVE_MMAP=1.  As an exception, when
-     compiling for GNU libc, i.e. when _LIBC is defined, then none of
-     the USE_... symbols have to be defined.
-
-  HEAP_MIN_SIZE
-  HEAP_MAX_SIZE
-     When thread support is enabled, additional `heap's are created
-     with mmap calls.  These are limited in size; HEAP_MIN_SIZE should
-     be a multiple of the page size, while HEAP_MAX_SIZE must be a power
-     of two for alignment reasons.  HEAP_MAX_SIZE should be at least
-     twice as large as the mmap threshold.
-  THREAD_STATS
-     When this is defined as non-zero, some statistics on mutex locking
-     are computed.
-
+  __STD_C should be nonzero if using ANSI-standard C compiler, a C++
+  compiler, or a C compiler sufficiently close to ANSI to get away
+  with it.
 */
 
-
-
-
-/* Preliminaries */
-
 #ifndef __STD_C
-#if defined (__STDC__)
-#define __STD_C     1
-#else
-#if __cplusplus
+#if defined(__STDC__) || defined(__cplusplus)
 #define __STD_C     1
 #else
 #define __STD_C     0
-#endif /*__cplusplus*/
-#endif /*__STDC__*/
+#endif 
 #endif /*__STD_C*/
 
+
+/*
+  Void_t* is the pointer type that malloc should say it returns
+*/
+
 #ifndef Void_t
-#if __STD_C
+#if (__STD_C || defined(WIN32))
 #define Void_t      void
 #else
 #define Void_t      char
@@ -303,57 +249,59 @@
 #endif /*Void_t*/
 
 #if __STD_C
-# include <stddef.h>   /* for size_t */
-# if defined _LIBC || defined MALLOC_HOOKS
-#  include <stdlib.h>  /* for getenv(), abort() */
-# endif
+#include <stddef.h>   /* for size_t */
+#include <stdlib.h>   /* for getenv(), abort() */
 #else
-# include <sys/types.h>
-# if defined _LIBC || defined MALLOC_HOOKS
-extern char* getenv();
-# endif
+#include <sys/types.h>
 #endif
 
-/* Macros for handling mutexes and thread-specific data.  This is
-   included early, because some thread-related header files (such as
-   pthread.h) should be included before any others. */
-#include "thread-m.h"
-
 #ifdef __cplusplus
 extern "C" {
 #endif
 
-#include <errno.h>
-#include <stdio.h>    /* needed for malloc_stats */
+/* define LACKS_UNISTD_H if your system does not have a <unistd.h>. */
 
+/* #define  LACKS_UNISTD_H */
 
-/*
-  Compile-time options
-*/
+#ifndef LACKS_UNISTD_H
+#include <unistd.h>
+#endif
 
+/* define LACKS_SYS_PARAM_H if your system does not have a <sys/param.h>. */
+
+/* #define  LACKS_SYS_PARAM_H */
+
+
+#include <stdio.h>    /* needed for malloc_stats */
+#include <errno.h>    /* needed for optional MALLOC_FAILURE_ACTION */
 
-/*
-    Debugging:
-
-    Because freed chunks may be overwritten with link fields, this
-    malloc will often die when freed memory is overwritten by user
-    programs.  This can be very effective (albeit in an annoying way)
-    in helping track down dangling pointers.
-
-    If you compile with -DMALLOC_DEBUG, a number of assertion checks are
-    enabled that will catch more memory errors. You probably won't be
-    able to make much sense of the actual assertion errors, but they
-    should help you locate incorrectly overwritten memory.  The
-    checking is fairly extensive, and will slow down execution
-    noticeably. Calling malloc_stats or mallinfo with MALLOC_DEBUG set will
-    attempt to check every non-mmapped allocated and free chunk in the
-    course of computing the summaries. (By nature, mmapped regions
-    cannot be checked very much automatically.)
-
-    Setting MALLOC_DEBUG may also be helpful if you are trying to modify
-    this code. The assertions in the check routines spell out in more
-    detail the assumptions and invariants underlying the algorithms.
 
+/*
+  Debugging:
+
+  Because freed chunks may be overwritten with bookkeeping fields, this
+  malloc will often die when freed memory is overwritten by user
+  programs.  This can be very effective (albeit in an annoying way)
+  in helping track down dangling pointers.
+
+  If you compile with -DMALLOC_DEBUG, a number of assertion checks are
+  enabled that will catch more memory errors. You probably won't be
+  able to make much sense of the actual assertion errors, but they
+  should help you locate incorrectly overwritten memory.  The checking
+  is fairly extensive, and will slow down execution
+  noticeably. Calling malloc_stats or mallinfo with MALLOC_DEBUG set
+  will attempt to check every non-mmapped allocated and free chunk in
+  the course of computing the summmaries. (By nature, mmapped regions
+  cannot be checked very much automatically.)
+
+  Setting MALLOC_DEBUG may also be helpful if you are trying to modify
+  this code. The assertions in the check routines spell out in more
+  detail the assumptions and invariants underlying the algorithms.
+
+  Setting MALLOC_DEBUG does NOT provide an automated mechanism for
+  checking that all accesses to malloced memory stay within their
+  bounds. However, there are several add-ons and adaptations of this
+  or other mallocs available that do this.
 */
 
 #if MALLOC_DEBUG
@@ -365,42 +313,197 @@ extern "C" {
 
 /*
   INTERNAL_SIZE_T is the word-size used for internal bookkeeping
-  of chunk sizes. On a 64-bit machine, you can reduce malloc
-  overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int'
-  at the expense of not being able to handle requests greater than
-  2^31. This limitation is hardly ever a concern; you are encouraged
-  to set this. However, the default version is the same as size_t.
+  of chunk sizes.
+
+  The default version is the same as size_t.
+
+  While not strictly necessary, it is best to define this as an
+  unsigned type, even if size_t is a signed type. This may avoid some
+  artificial size limitations on some systems.
+
+  On a 64-bit machine, you may be able to reduce malloc overhead by
+  defining INTERNAL_SIZE_T to be a 32 bit `unsigned int' at the
+  expense of not being able to handle more than 2^32 of malloced
+  space. If this limitation is acceptable, you are encouraged to set
+  this unless you are on a platform requiring 16byte alignments. In
+  this case the alignment requirements turn out to negate any
+  potential advantages of decreasing size_t word size.
+
+  Implementors: Beware of the possible combinations of:
+     - INTERNAL_SIZE_T might be signed or unsigned, might be 32 or 64 bits,
+       and might be the same width as int or as long
+     - size_t might have different width and signedness as INTERNAL_SIZE_T
+     - int and long might be 32 or 64 bits, and might be the same width
+  To deal with this, most comparisons and difference computations
+  among INTERNAL_SIZE_Ts should cast them to unsigned long, being
+  aware of the fact that casting an unsigned int to a wider long does
+  not sign-extend. (This also makes checking for negative numbers
+  awkward.) Some of these casts result in harmless compiler warnings
+  on some systems.
 */
 
 #ifndef INTERNAL_SIZE_T
 #define INTERNAL_SIZE_T size_t
 #endif
 
+/* The corresponding word size */
+#define SIZE_SZ                (sizeof(INTERNAL_SIZE_T))
+
+
+/*
+  MALLOC_ALIGNMENT is the minimum alignment for malloc'ed chunks.
+  It must be a power of two at least 2 * SIZE_SZ, 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.
+*/
+
+
+#ifndef MALLOC_ALIGNMENT
+#define MALLOC_ALIGNMENT       (2 * SIZE_SZ)
+#endif
+
+/* The corresponding bit mask value */
+#define MALLOC_ALIGN_MASK      (MALLOC_ALIGNMENT - 1)
+
+
+
+/*
+  REALLOC_ZERO_BYTES_FREES should be set if a call to
+  realloc with zero bytes should be the same as a call to free.
+  This is required by the C standard. Otherwise, since this malloc
+  returns a unique pointer for malloc(0), so does realloc(p, 0).
+*/
+
+#ifndef REALLOC_ZERO_BYTES_FREES
+#define REALLOC_ZERO_BYTES_FREES 1
+#endif
+
 /*
-  REALLOC_ZERO_BYTES_FREES should be set if a call to realloc with
-  zero bytes should be the same as a call to free.  The C standard
-  requires this. Otherwise, since this malloc returns a unique pointer
-  for malloc(0), so does realloc(p, 0).
+  TRIM_FASTBINS controls whether free() of a very small chunk can
+  immediately lead to trimming. Setting to true (1) can reduce memory
+  footprint, but will almost always slow down programs that use a lot
+  of small chunks.
+
+  Define this only if you are willing to give up some speed to more
+  aggressively reduce system-level memory footprint when releasing
+  memory in programs that use many small chunks.  You can get
+  essentially the same effect by setting MXFAST to 0, but this can
+  lead to even greater slowdowns in programs using many small chunks.
+  TRIM_FASTBINS is an in-between compile-time option, that disables
+  only those chunks bordering topmost memory from being placed in
+  fastbins.
+*/
+
+#ifndef TRIM_FASTBINS
+#define TRIM_FASTBINS  0
+#endif
+
+
+/*
+  USE_DL_PREFIX will prefix all public routines with the string 'dl'.
+  This is necessary when you only want to use this malloc in one part 
+  of a program, using your regular system malloc elsewhere.
+*/
+
+/* #define USE_DL_PREFIX */
+
+
+/* 
+   Two-phase name translation.
+   All of the actual routines are given mangled names.
+   When wrappers are used, they become the public callable versions.
+   When DL_PREFIX is used, the callable names are prefixed.
 */
 
+#ifdef USE_DL_PREFIX
+#define public_cALLOc    dlcalloc
+#define public_fREe      dlfree
+#define public_cFREe     dlcfree
+#define public_mALLOc    dlmalloc
+#define public_mEMALIGn  dlmemalign
+#define public_rEALLOc   dlrealloc
+#define public_vALLOc    dlvalloc
+#define public_pVALLOc   dlpvalloc
+#define public_mALLINFo  dlmallinfo
+#define public_mALLOPt   dlmallopt
+#define public_mTRIm     dlmalloc_trim
+#define public_mSTATs    dlmalloc_stats
+#define public_mUSABLe   dlmalloc_usable_size
+#define public_iCALLOc   dlindependent_calloc
+#define public_iCOMALLOc dlindependent_comalloc
+#define public_gET_STATe dlget_state
+#define public_sET_STATe dlset_state
+#else /* USE_DL_PREFIX */
+#ifdef _LIBC
 
-#define REALLOC_ZERO_BYTES_FREES
+/* Special defines for the GNU C library.  */
+#define public_cALLOc    __libc_calloc
+#define public_fREe      __libc_free
+#define public_cFREe     __libc_cfree
+#define public_mALLOc    __libc_malloc
+#define public_mEMALIGn  __libc_memalign
+#define public_rEALLOc   __libc_realloc
+#define public_vALLOc    __libc_valloc
+#define public_pVALLOc   __libc_pvalloc
+#define public_mALLINFo  __libc_mallinfo
+#define public_mALLOPt   __libc_mallopt
+#define public_mTRIm     __malloc_trim
+#define public_mSTATs    __malloc_stats
+#define public_mUSABLe   __malloc_usable_size
+#define public_iCALLOc   __libc_independent_calloc
+#define public_iCOMALLOc __libc_independent_comalloc
+#define public_gET_STATe __malloc_get_state
+#define public_sET_STATe __malloc_set_state
+#define malloc_getpagesize __getpagesize()
+#define open             __open
+#define mmap             __mmap
+#define munmap           __munmap
+#define mremap           __mremap
+#define mprotect         __mprotect
+#define MORECORE         (*__morecore)
+#define MORECORE_FAILURE 0
+
+Void_t * __default_morecore (ptrdiff_t);
+Void_t *(*__morecore)(ptrdiff_t) = __default_morecore;
+
+#else /* !_LIBC */
+#define public_cALLOc    calloc
+#define public_fREe      free
+#define public_cFREe     cfree
+#define public_mALLOc    malloc
+#define public_mEMALIGn  memalign
+#define public_rEALLOc   realloc
+#define public_vALLOc    valloc
+#define public_pVALLOc   pvalloc
+#define public_mALLINFo  mallinfo
+#define public_mALLOPt   mallopt
+#define public_mTRIm     malloc_trim
+#define public_mSTATs    malloc_stats
+#define public_mUSABLe   malloc_usable_size
+#define public_iCALLOc   independent_calloc
+#define public_iCOMALLOc independent_comalloc
+#define public_gET_STATe malloc_get_state
+#define public_sET_STATe malloc_set_state
+#endif /* _LIBC */
+#endif /* USE_DL_PREFIX */
 
 
 /*
   HAVE_MEMCPY should be defined if you are not otherwise using
   ANSI STD C, but still have memcpy and memset in your C library
   and want to use them in calloc and realloc. Otherwise simple
-  macro versions are defined here.
+  macro versions are defined below.
 
   USE_MEMCPY should be defined as 1 if you actually want to
   have memset and memcpy called. People report that the macro
-  versions are often enough faster than libc versions on many
-  systems that it is better to use them.
-
+  versions are faster than libc versions on some systems.
+  
+  Even if USE_MEMCPY is set to 1, loops to copy/clear small chunks
+  (of <= 36 bytes) are manually unrolled in realloc and calloc.
 */
 
-#define HAVE_MEMCPY 1
+#define HAVE_MEMCPY
 
 #ifndef USE_MEMCPY
 #ifdef HAVE_MEMCPY
@@ -410,125 +513,161 @@ extern "C" {
 #endif
 #endif
 
+
 #if (__STD_C || defined(HAVE_MEMCPY))
 
+#ifdef WIN32
+/* On Win32 memset and memcpy are already declared in windows.h */
+#else
 #if __STD_C
 void* memset(void*, int, size_t);
 void* memcpy(void*, const void*, size_t);
-void* memmove(void*, const void*, size_t);
 #else
 Void_t* memset();
 Void_t* memcpy();
-Void_t* memmove();
+#endif
 #endif
 #endif
 
-/* The following macros are only invoked with (2n+1)-multiples of
-   INTERNAL_SIZE_T units, with a positive integer n. This is exploited
-   for fast inline execution when n is small.  If the regions to be
-   copied do overlap, the destination lies always _below_ the source.  */
+/*
+  MALLOC_FAILURE_ACTION is the action to take before "return 0" when
+  malloc fails to be able to return memory, either because memory is
+  exhausted or because of illegal arguments.
+  
+  By default, sets errno if running on STD_C platform, else does nothing.  
+*/
 
-#if USE_MEMCPY
+#ifndef MALLOC_FAILURE_ACTION
+#if __STD_C
+#define MALLOC_FAILURE_ACTION \
+   errno = ENOMEM;
 
-#define MALLOC_ZERO(charp, nbytes)                                            \
-do {                                                                          \
-  INTERNAL_SIZE_T mzsz = (nbytes);                                            \
-  if(mzsz <= 9*sizeof(mzsz)) {                                                \
-    INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp);                         \
-    if(mzsz >= 5*sizeof(mzsz)) {     *mz++ = 0;                               \
-                                     *mz++ = 0;                               \
-      if(mzsz >= 7*sizeof(mzsz)) {   *mz++ = 0;                               \
-                                     *mz++ = 0;                               \
-        if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0;                               \
-                                     *mz++ = 0; }}}                           \
-                                     *mz++ = 0;                               \
-                                     *mz++ = 0;                               \
-                                     *mz   = 0;                               \
-  } else memset((charp), 0, mzsz);                                            \
-} while(0)
+#else
+#define MALLOC_FAILURE_ACTION
+#endif
+#endif
 
-/* If the regions overlap, dest is always _below_ src.  */
+/*
+  MORECORE-related declarations. By default, rely on sbrk
+*/
 
-#define MALLOC_COPY(dest,src,nbytes,overlap)                                  \
-do {                                                                          \
-  INTERNAL_SIZE_T mcsz = (nbytes);                                            \
-  if(mcsz <= 9*sizeof(mcsz)) {                                                \
-    INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src);                        \
-    INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest);                       \
-    if(mcsz >= 5*sizeof(mcsz)) {     *mcdst++ = *mcsrc++;                     \
-                                     *mcdst++ = *mcsrc++;                     \
-      if(mcsz >= 7*sizeof(mcsz)) {   *mcdst++ = *mcsrc++;                     \
-                                     *mcdst++ = *mcsrc++;                     \
-        if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++;                     \
-                                     *mcdst++ = *mcsrc++; }}}                 \
-                                     *mcdst++ = *mcsrc++;                     \
-                                     *mcdst++ = *mcsrc++;                     \
-                                     *mcdst   = *mcsrc  ;                     \
-  } else if(overlap)                                                          \
-    memmove(dest, src, mcsz);                                                 \
-  else                                                                        \
-    memcpy(dest, src, mcsz);                                                  \
-} while(0)
 
-#else /* !USE_MEMCPY */
+#ifdef LACKS_UNISTD_H
+#if !defined(__FreeBSD__) && !defined(__OpenBSD__) && !defined(__NetBSD__)
+#if __STD_C
+extern Void_t*     sbrk(ptrdiff_t);
+#else
+extern Void_t*     sbrk();
+#endif
+#endif
+#endif
 
-/* Use Duff's device for good zeroing/copying performance. */
+/*
+  MORECORE is the name of the routine to call to obtain more memory
+  from the system.  See below for general guidance on writing
+  alternative MORECORE functions, as well as a version for WIN32 and a
+  sample version for pre-OSX macos.
+*/
 
-#define MALLOC_ZERO(charp, nbytes)                                            \
-do {                                                                          \
-  INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp);                           \
-  long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn;                         \
-  if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
-  switch (mctmp) {                                                            \
-    case 0: for(;;) { *mzp++ = 0;                                             \
-    case 7:           *mzp++ = 0;                                             \
-    case 6:           *mzp++ = 0;                                             \
-    case 5:           *mzp++ = 0;                                             \
-    case 4:           *mzp++ = 0;                                             \
-    case 3:           *mzp++ = 0;                                             \
-    case 2:           *mzp++ = 0;                                             \
-    case 1:           *mzp++ = 0; if(mcn <= 0) break; mcn--; }                \
-  }                                                                           \
-} while(0)
+#ifndef MORECORE
+#define MORECORE sbrk
+#endif
 
-/* If the regions overlap, dest is always _below_ src.  */
+/*
+  MORECORE_FAILURE is the value returned upon failure of MORECORE
+  as well as mmap. Since it cannot be an otherwise valid memory address,
+  and must reflect values of standard sys calls, you probably ought not
+  try to redefine it.
+*/
 
-#define MALLOC_COPY(dest,src,nbytes,overlap)                                  \
-do {                                                                          \
-  INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src;                            \
-  INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest;                           \
-  long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn;                         \
-  if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
-  switch (mctmp) {                                                            \
-    case 0: for(;;) { *mcdst++ = *mcsrc++;                                    \
-    case 7:           *mcdst++ = *mcsrc++;                                    \
-    case 6:           *mcdst++ = *mcsrc++;                                    \
-    case 5:           *mcdst++ = *mcsrc++;                                    \
-    case 4:           *mcdst++ = *mcsrc++;                                    \
-    case 3:           *mcdst++ = *mcsrc++;                                    \
-    case 2:           *mcdst++ = *mcsrc++;                                    \
-    case 1:           *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; }       \
-  }                                                                           \
-} while(0)
+#ifndef MORECORE_FAILURE
+#define MORECORE_FAILURE (-1)
+#endif
+
+/*
+  If MORECORE_CONTIGUOUS is true, take advantage of fact that
+  consecutive calls to MORECORE with positive arguments always return
+  contiguous increasing addresses.  This is true of unix sbrk.  Even
+  if not defined, when regions happen to be contiguous, malloc will
+  permit allocations spanning regions obtained from different
+  calls. But defining this when applicable enables some stronger
+  consistency checks and space efficiencies.
+*/
 
+#ifndef MORECORE_CONTIGUOUS
+#define MORECORE_CONTIGUOUS 1
 #endif
 
+/*
+  Define MORECORE_CANNOT_TRIM if your version of 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.
+*/
 
-#ifndef LACKS_UNISTD_H
-#  include <unistd.h>
+/* #define MORECORE_CANNOT_TRIM */
+
+/*  MORECORE_CLEARS           (default 1)
+     The degree to which the routine mapped to MORECORE zeroes out
+     memory: never (0), only for newly allocated space (1) or always
+     (2).  The distinction between (1) and (2) is necessary because on
+     some systems, if the application first decrements and then
+     increments the break value, the contents of the reallocated space
+     are unspecified.
+*/
+
+#ifndef MORECORE_CLEARS
+#define MORECORE_CLEARS 1
 #endif
 
+
 /*
-  Define HAVE_MMAP to optionally make malloc() use mmap() to allocate
-  very large blocks.  These will be returned to the operating system
-  immediately after a free().  HAVE_MMAP is also a prerequisite to
-  support multiple `arenas' (see USE_ARENAS below).
+  Define HAVE_MMAP as true to optionally make malloc() use mmap() to
+  allocate very large blocks.  These will be returned to the
+  operating system immediately after a free(). Also, if mmap
+  is available, it is used as a backup strategy in cases where
+  MORECORE fails to provide space from system.
+
+  This malloc is best tuned to work with mmap for large requests.
+  If you do not have mmap, operations involving very large chunks (1MB
+  or so) may be slower than you'd like.
 */
 
 #ifndef HAVE_MMAP
-# ifdef _POSIX_MAPPED_FILES
-#  define HAVE_MMAP 1
-# endif
+#define HAVE_MMAP 1
+
+/* 
+   Standard unix mmap using /dev/zero clears memory so calloc doesn't
+   need to.
+*/
+
+#ifndef MMAP_CLEARS
+#define MMAP_CLEARS 1
+#endif
+
+#else /* no mmap */
+#ifndef MMAP_CLEARS
+#define MMAP_CLEARS 0
+#endif
+#endif
+
+
+/* 
+   MMAP_AS_MORECORE_SIZE is the minimum mmap size argument to use if
+   sbrk fails, and mmap is used as a backup (which is done only if
+   HAVE_MMAP).  The value must be a multiple of page size.  This
+   backup strategy generally applies only when systems have "holes" in
+   address space, so sbrk cannot perform contiguous expansion, but
+   there is still space available on system.  On systems for which
+   this is known to be useful (i.e. most linux kernels), this occurs
+   only when programs allocate huge amounts of memory.  Between this,
+   and the fact that mmap regions tend to be limited, the size should
+   be large, to avoid too many mmap calls and thus avoid running out
+   of kernel resources.
+*/
+
+#ifndef MMAP_AS_MORECORE_SIZE
+#define MMAP_AS_MORECORE_SIZE (1024 * 1024)
 #endif
 
 /*
@@ -538,9 +677,14 @@ do {                                                                          \
 */
 
 #ifndef HAVE_MREMAP
-#define HAVE_MREMAP defined(__linux__)
+#ifdef linux
+#define HAVE_MREMAP 1
+#else
+#define HAVE_MREMAP 0
 #endif
 
+#endif /* HAVE_MMAP */
+
 /* Define USE_ARENAS to enable support for multiple `arenas'.  These
    are allocated using mmap(), are necessary for threads and
    occasionally useful to overcome address space limitations affecting
@@ -550,43 +694,32 @@ do {                                                                          \
 #define USE_ARENAS HAVE_MMAP
 #endif
 
-#if HAVE_MMAP
-
-#include <unistd.h>
-#include <fcntl.h>
-#include <sys/mman.h>
-
-#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
-#define MAP_ANONYMOUS MAP_ANON
-#endif
-#if !defined(MAP_FAILED)
-#define MAP_FAILED ((char*)-1)
-#endif
-
-#ifndef MAP_NORESERVE
-# ifdef MAP_AUTORESRV
-#  define MAP_NORESERVE MAP_AUTORESRV
-# else
-#  define MAP_NORESERVE 0
-# endif
-#endif
-
-#endif /* HAVE_MMAP */
 
 /*
-  Access to system page size. To the extent possible, this malloc
-  manages memory from the system in page-size units.
-
-  The following mechanics for getpagesize were adapted from
-  bsd/gnu getpagesize.h
+  The system page size. To the extent possible, this malloc manages
+  memory from the system in page-size units.  Note that this value is
+  cached during initialization into a field of malloc_state. So even
+  if malloc_getpagesize is a function, it is only called once.
+
+  The following mechanics for getpagesize were adapted from bsd/gnu
+  getpagesize.h. If none of the system-probes here apply, a value of
+  4096 is used, which should be OK: If they don't apply, then using
+  the actual value probably doesn't impact performance.
 */
 
+
 #ifndef malloc_getpagesize
+
+#ifndef LACKS_UNISTD_H
+#  include <unistd.h>
+#endif
+
 #  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
@@ -594,24 +727,30 @@ do {                                                                          \
        extern size_t getpagesize();
 #      define malloc_getpagesize getpagesize()
 #    else
-#      include <sys/param.h>
-#      ifdef EXEC_PAGESIZE
-#        define malloc_getpagesize EXEC_PAGESIZE
+#      ifdef WIN32 /* use supplied emulation of getpagesize */
+#        define malloc_getpagesize getpagesize() 
 #      else
-#        ifdef NBPG
-#          ifndef CLSIZE
-#            define malloc_getpagesize NBPG
-#          else
-#            define malloc_getpagesize (NBPG * CLSIZE)
-#          endif
+#        ifndef LACKS_SYS_PARAM_H
+#          include <sys/param.h>
+#        endif
+#        ifdef EXEC_PAGESIZE
+#          define malloc_getpagesize EXEC_PAGESIZE
 #        else
-#          ifdef NBPC
-#            define malloc_getpagesize NBPC
+#          ifdef NBPG
+#            ifndef CLSIZE
+#              define malloc_getpagesize NBPG
+#            else
+#              define malloc_getpagesize (NBPG * CLSIZE)
+#            endif
 #          else
-#            ifdef PAGESIZE
-#              define malloc_getpagesize PAGESIZE
+#            ifdef NBPC
+#              define malloc_getpagesize NBPC
 #            else
-#              define malloc_getpagesize (4096) /* just guess */
+#              ifdef PAGESIZE
+#                define malloc_getpagesize PAGESIZE
+#              else /* just guess */
+#                define malloc_getpagesize (4096) 
+#              endif
 #            endif
 #          endif
 #        endif
@@ -620,241 +759,656 @@ do {                                                                          \
 #  endif
 #endif
 
-
-
 /*
-
   This version of malloc supports the standard SVID/XPG mallinfo
-  routine that returns a struct containing the same kind of
-  information you can get from malloc_stats. It should work on
-  any SVID/XPG compliant system that has a /usr/include/malloc.h
-  defining struct mallinfo. (If you'd like to install such a thing
-  yourself, cut out the preliminary declarations as described above
-  and below and save them in a malloc.h file. But there's no
-  compelling reason to bother to do this.)
+  routine that returns a struct containing usage properties and
+  statistics. It should work on any SVID/XPG compliant system that has
+  a /usr/include/malloc.h defining struct mallinfo. (If you'd like to
+  install such a thing yourself, cut out the preliminary declarations
+  as described above and below and save them in a malloc.h file. But
+  there's no compelling reason to bother to do this.)
 
   The main declaration needed is the mallinfo struct that is returned
   (by-copy) by mallinfo().  The SVID/XPG malloinfo struct contains a
-  bunch of fields, most of which are not even meaningful in this
-  version of malloc. Some of these fields are are instead filled by
-  mallinfo() with other numbers that might possibly be of interest.
+  bunch of fields that are not even meaningful in this version of
+  malloc.  These fields are are instead filled by mallinfo() with
+  other numbers that might be of interest.
 
   HAVE_USR_INCLUDE_MALLOC_H should be set if you have a
   /usr/include/malloc.h file that includes a declaration of struct
   mallinfo.  If so, it is included; else an SVID2/XPG2 compliant
   version is declared below.  These must be precisely the same for
-  mallinfo() to work.
-
+  mallinfo() to work.  The original SVID version of this struct,
+  defined on most systems with mallinfo, declares all fields as
+  ints. But some others define as unsigned long. If your system
+  defines the fields using a type of different width than listed here,
+  you must #include your system version and #define
+  HAVE_USR_INCLUDE_MALLOC_H.
 */
 
 /* #define HAVE_USR_INCLUDE_MALLOC_H */
 
-#if HAVE_USR_INCLUDE_MALLOC_H
-# include "/usr/include/malloc.h"
+#ifdef HAVE_USR_INCLUDE_MALLOC_H
+#include "/usr/include/malloc.h"
+#endif
+
+
+/* ---------- description of public routines ------------ */
+
+/*
+  malloc(size_t n)
+  Returns a pointer to a newly allocated chunk of at least n bytes, or null
+  if no space is available. Additionally, on failure, errno is
+  set to ENOMEM on ANSI C systems.
+
+  If n is zero, malloc returns a minumum-sized chunk. (The minimum
+  size is 16 bytes on most 32bit systems, and 24 or 32 bytes on 64bit
+  systems.)  On most systems, size_t is an unsigned type, so calls
+  with negative arguments are interpreted as requests for huge amounts
+  of space, which will often fail. The maximum supported value of n
+  differs across systems, but is in all cases less than the maximum
+  representable value of a size_t.
+*/
+#if __STD_C
+Void_t*  public_mALLOc(size_t);
 #else
-# ifdef _LIBC
-#  include "malloc.h"
-# else
-#  include "ptmalloc.h"
-# endif
+Void_t*  public_mALLOc();
 #endif
 
-#include <bp-checks.h>
+/*
+  free(Void_t* p)
+  Releases the chunk of memory pointed to by p, that had been previously
+  allocated using malloc or a related routine such as realloc.
+  It has no effect if p is null. It can have arbitrary (i.e., bad!)
+  effects if p has already been freed.
+
+  Unless disabled (using mallopt), freeing very large spaces will
+  when possible, automatically trigger operations that give
+  back unused memory to the system, thus reducing program footprint.
+*/
+#if __STD_C
+void     public_fREe(Void_t*);
+#else
+void     public_fREe();
+#endif
 
-#ifndef DEFAULT_TRIM_THRESHOLD
-#define DEFAULT_TRIM_THRESHOLD (128 * 1024)
+/*
+  calloc(size_t n_elements, size_t element_size);
+  Returns a pointer to n_elements * element_size bytes, with all locations
+  set to zero.
+*/
+#if __STD_C
+Void_t*  public_cALLOc(size_t, size_t);
+#else
+Void_t*  public_cALLOc();
 #endif
 
 /*
-    M_TRIM_THRESHOLD is 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.
-      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.
-
-      The trim threshold and the mmap control parameters (see below)
-      can be traded off with one another. Trimming and mmapping are
-      two different ways of releasing unused memory back to the
-      system. Between these two, it is often possible to keep
-      system-level demands of a long-lived program down to a bare
-      minimum. For example, in one test suite of sessions measuring
-      the XF86 X server on Linux, using a trim threshold of 128K and a
-      mmap threshold of 192K led to near-minimal long term resource
-      consumption.
-
-      If you are using this malloc in a long-lived program, it should
-      pay to experiment with these values.  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's
-      worth it to tune for trimming rather than memory mapping 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.  And in well-behaved long-lived programs,
-      controlling release of large blocks via trimming versus mapping
-      is usually faster.
-
-      However, in most programs, these parameters serve mainly as
-      protection against the system-level effects of carrying around
-      massive amounts of unneeded memory. Since frequent calls to
-      sbrk, mmap, and munmap otherwise degrade performance, the default
-      parameters are set to relatively high values that serve only as
-      safeguards.
-
-      The default trim value is high enough to cause trimming only in
-      fairly extreme (by current memory consumption standards) cases.
-      It must be greater than page size to have any useful effect.  To
-      disable trimming completely, you can set to (unsigned long)(-1);
+  realloc(Void_t* p, size_t n)
+  Returns a pointer to a chunk of size n that contains the same data
+  as does chunk p up to the minimum of (n, p's size) bytes, or null
+  if no space is available. 
 
+  The returned pointer may or may not be the same as p. The algorithm
+  prefers extending p when possible, otherwise it employs the
+  equivalent of a malloc-copy-free sequence.
 
-*/
+  If p is null, realloc is equivalent to malloc.  
 
+  If space is not available, realloc returns null, errno is set (if on
+  ANSI) and p is NOT freed.
 
-#ifndef DEFAULT_TOP_PAD
-#define DEFAULT_TOP_PAD        (0)
+  if n is for fewer bytes than already held by p, the newly unused
+  space is lopped off and freed if possible.  Unless the #define
+  REALLOC_ZERO_BYTES_FREES is set, realloc with a size argument of
+  zero (re)allocates a minimum-sized chunk.
+
+  Large chunks that were internally obtained via mmap will always
+  be reallocated using malloc-copy-free sequences unless
+  the system supports MREMAP (currently only linux).
+
+  The old unix realloc convention of allowing the last-free'd chunk
+  to be used as an argument to realloc is not supported.
+*/
+#if __STD_C
+Void_t*  public_rEALLOc(Void_t*, size_t);
+#else
+Void_t*  public_rEALLOc();
 #endif
 
 /*
-    M_TOP_PAD is the amount of extra `padding' space to allocate or
-      retain whenever sbrk is called. It is used in two ways internally:
+  memalign(size_t alignment, size_t n);
+  Returns a pointer to a newly allocated chunk of n bytes, aligned
+  in accord with the alignment argument.
 
-      * When sbrk is called to extend the top of the arena to satisfy
-        a new malloc request, this much padding is added to the sbrk
-        request.
+  The alignment argument should be a power of two. If the argument is
+  not a power of two, the nearest greater power is used.
+  8-byte alignment is guaranteed by normal malloc calls, so don't
+  bother calling memalign with an argument of 8 or less.
 
-      * When malloc_trim is called automatically from free(),
-        it is used as the `pad' argument.
+  Overreliance on memalign is a sure way to fragment space.
+*/
+#if __STD_C
+Void_t*  public_mEMALIGn(size_t, size_t);
+#else
+Void_t*  public_mEMALIGn();
+#endif
 
-      In both cases, the actual amount of padding is rounded
-      so that the end of the arena is always a system page boundary.
+/*
+  valloc(size_t n);
+  Equivalent to memalign(pagesize, n), where pagesize is the page
+  size of the system. If the pagesize is unknown, 4096 is used.
+*/
+#if __STD_C
+Void_t*  public_vALLOc(size_t);
+#else
+Void_t*  public_vALLOc();
+#endif
 
-      The main reason for using padding is to avoid calling sbrk so
-      often. Having even a small pad greatly reduces the likelihood
-      that nearly every malloc request during program start-up (or
-      after trimming) will invoke sbrk, which needlessly wastes
-      time.
 
-      Automatic rounding-up to page-size units is normally sufficient
-      to avoid measurable overhead, so the default is 0.  However, in
-      systems where sbrk is relatively slow, it can pay to increase
-      this value, at the expense of carrying around more memory than
-      the program needs.
 
+/*
+  mallopt(int parameter_number, int parameter_value)
+  Sets tunable parameters The format is to provide a
+  (parameter-number, parameter-value) pair.  mallopt then sets the
+  corresponding parameter to the argument value if it can (i.e., so
+  long as the value is meaningful), and returns 1 if successful else
+  0.  SVID/XPG/ANSI defines four standard param numbers for mallopt,
+  normally defined in malloc.h.  Only one of these (M_MXFAST) is used
+  in this malloc. The others (M_NLBLKS, M_GRAIN, M_KEEP) don't apply,
+  so setting them has no effect. But this malloc also supports four
+  other options in mallopt. See below for details.  Briefly, supported
+  parameters are as follows (listed defaults are for "typical"
+  configurations).
+
+  Symbol            param #   default    allowed param values
+  M_MXFAST          1         64         0-80  (0 disables fastbins)
+  M_TRIM_THRESHOLD -1         128*1024   any   (-1U disables trimming)
+  M_TOP_PAD        -2         0          any  
+  M_MMAP_THRESHOLD -3         128*1024   any   (or 0 if no MMAP support)
+  M_MMAP_MAX       -4         65536      any   (0 disables use of mmap)
 */
+#if __STD_C
+int      public_mALLOPt(int, int);
+#else
+int      public_mALLOPt();
+#endif
 
 
-#ifndef DEFAULT_MMAP_THRESHOLD
-#define DEFAULT_MMAP_THRESHOLD (128 * 1024)
+/*
+  mallinfo()
+  Returns (by copy) a struct containing various summary statistics:
+
+  arena:     current total non-mmapped bytes allocated from system 
+  ordblks:   the number of free chunks 
+  smblks:    the number of fastbin blocks (i.e., small chunks that
+               have been freed but not use resused or consolidated)
+  hblks:     current number of mmapped regions 
+  hblkhd:    total bytes held in mmapped regions 
+  usmblks:   the maximum total allocated space. This will be greater
+                than current total if trimming has occurred.
+  fsmblks:   total bytes held in fastbin blocks 
+  uordblks:  current total allocated space (normal or mmapped)
+  fordblks:  total free space 
+  keepcost:  the maximum number of bytes that could ideally be released
+               back to system via malloc_trim. ("ideally" means that
+               it ignores page restrictions etc.)
+
+  Because these fields are ints, but internal bookkeeping may
+  be kept as longs, the reported values may wrap around zero and 
+  thus be inaccurate.
+*/
+#if __STD_C
+struct mallinfo public_mALLINFo(void);
+#else
+struct mallinfo public_mALLINFo();
 #endif
 
 /*
+  independent_calloc(size_t n_elements, size_t element_size, Void_t* chunks[]);
+
+  independent_calloc is similar to calloc, but instead of returning a
+  single cleared space, it returns an array of pointers to n_elements
+  independent elements that can hold contents of size elem_size, each
+  of which starts out cleared, and can be independently freed,
+  realloc'ed etc. The elements are guaranteed to be adjacently
+  allocated (this is not guaranteed to occur with multiple callocs or
+  mallocs), which may also improve cache locality in some
+  applications.
+
+  The "chunks" argument is optional (i.e., may be null, which is
+  probably the most typical usage). If it is null, the returned array
+  is itself dynamically allocated and should also be freed when it is
+  no longer needed. Otherwise, the chunks array must be of at least
+  n_elements in length. It is filled in with the pointers to the
+  chunks.
+
+  In either case, independent_calloc returns this pointer array, or
+  null if the allocation failed.  If n_elements is zero and "chunks"
+  is null, it returns a chunk representing an array with zero elements
+  (which should be freed if not wanted).
+
+  Each element must be individually freed when it is no longer
+  needed. If you'd like to instead be able to free all at once, you
+  should instead use regular calloc and assign pointers into this
+  space to represent elements.  (In this case though, you cannot
+  independently free elements.)
+  
+  independent_calloc simplifies and speeds up implementations of many
+  kinds of pools.  It may also be useful when constructing large data
+  structures that initially have a fixed number of fixed-sized nodes,
+  but the number is not known at compile time, and some of the nodes
+  may later need to be freed. For example:
+
+  struct Node { int item; struct Node* next; };
+  
+  struct Node* build_list() {
+    struct Node** pool;
+    int n = read_number_of_nodes_needed();
+    if (n <= 0) return 0;
+    pool = (struct Node**)(independent_calloc(n, sizeof(struct Node), 0);
+    if (pool == 0) die(); 
+    // organize into a linked list... 
+    struct Node* first = pool[0];
+    for (i = 0; i < n-1; ++i) 
+      pool[i]->next = pool[i+1];
+    free(pool);     // Can now free the array (or not, if it is needed later)
+    return first;
+  }
+*/
+#if __STD_C
+Void_t** public_iCALLOc(size_t, size_t, Void_t**);
+#else
+Void_t** public_iCALLOc();
+#endif
 
-    M_MMAP_THRESHOLD is the request size threshold for using mmap()
-      to 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 benefit 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. Mapped memory can never become `locked' between
-      other chunks, as can happen with normally allocated chunks, which
-      menas that even trimming via malloc_trim would not release them.
+/*
+  independent_comalloc(size_t n_elements, size_t sizes[], Void_t* chunks[]);
+
+  independent_comalloc allocates, all at once, a set of n_elements
+  chunks with sizes indicated in the "sizes" array.    It returns
+  an array of pointers to these elements, each of which can be
+  independently freed, realloc'ed etc. The elements are guaranteed to
+  be adjacently allocated (this is not guaranteed to occur with
+  multiple callocs or mallocs), which may also improve cache locality
+  in some applications.
+
+  The "chunks" argument is optional (i.e., may be null). If it is null
+  the returned array is itself dynamically allocated and should also
+  be freed when it is no longer needed. Otherwise, the chunks array
+  must be of at least n_elements in length. It is filled in with the
+  pointers to the chunks.
+
+  In either case, independent_comalloc returns this pointer array, or
+  null if the allocation failed.  If n_elements is zero and chunks is
+  null, it returns a chunk representing an array with zero elements
+  (which should be freed if not wanted).
+  
+  Each element must be individually freed when it is no longer
+  needed. If you'd like to instead be able to free all at once, you
+  should instead use a single regular malloc, and assign pointers at
+  particular offsets in the aggregate space. (In this case though, you 
+  cannot independently free elements.)
+
+  independent_comallac differs from independent_calloc in that each
+  element may have a different size, and also that it does not
+  automatically clear elements.
+
+  independent_comalloc can be used to speed up allocation in cases
+  where several structs or objects must always be allocated at the
+  same time.  For example:
+
+  struct Head { ... }
+  struct Foot { ... }
+
+  void send_message(char* msg) {
+    int msglen = strlen(msg);
+    size_t sizes[3] = { sizeof(struct Head), msglen, sizeof(struct Foot) };
+    void* chunks[3];
+    if (independent_comalloc(3, sizes, chunks) == 0)
+      die();
+    struct Head* head = (struct Head*)(chunks[0]);
+    char*        body = (char*)(chunks[1]);
+    struct Foot* foot = (struct Foot*)(chunks[2]);
+    // ...
+  }
 
-      However, it has the disadvantages that:
+  In general though, independent_comalloc is worth using only for
+  larger values of n_elements. For small values, you probably won't
+  detect enough difference from series of malloc calls to bother.
+
+  Overuse of independent_comalloc can increase overall memory usage,
+  since it cannot reuse existing noncontiguous small chunks that
+  might be available for some of the elements.
+*/
+#if __STD_C
+Void_t** public_iCOMALLOc(size_t, size_t*, Void_t**);
+#else
+Void_t** public_iCOMALLOc();
+#endif
 
-         1. The space cannot be reclaimed, consolidated, and then
-            used to service later requests, as happens with normal chunks.
-         2. It can lead to more wastage because of mmap page alignment
-            requirements
-         3. It causes malloc performance to be more dependent on host
-            system memory management support routines which may vary in
-            implementation quality and may impose arbitrary
-            limitations. Generally, servicing a request via normal
-            malloc steps is faster than going through a system's mmap.
 
-      All together, these considerations should lead you to use mmap
-      only for relatively large requests.
+/*
+  pvalloc(size_t n);
+  Equivalent to valloc(minimum-page-that-holds(n)), that is,
+  round up n to nearest pagesize.
+ */
+#if __STD_C
+Void_t*  public_pVALLOc(size_t);
+#else
+Void_t*  public_pVALLOc();
+#endif
+
+/*
+  cfree(Void_t* p);
+  Equivalent to free(p).
+
+  cfree is needed/defined on some systems that pair it with calloc,
+  for odd historical reasons (such as: cfree is used in example 
+  code in the first edition of K&R).
+*/
+#if __STD_C
+void     public_cFREe(Void_t*);
+#else
+void     public_cFREe();
+#endif
 
+/*
+  malloc_trim(size_t pad);
 
+  If possible, gives memory back to the system (via negative
+  arguments to sbrk) if there is unused memory at the `high' end of
+  the malloc pool. You can call this after freeing large blocks of
+  memory to potentially reduce the system-level memory requirements
+  of a program. However, it cannot guarantee to reduce memory. Under
+  some allocation patterns, some large free blocks of memory will be
+  locked between two used chunks, so they cannot be given back to
+  the system.
+  
+  The `pad' argument to malloc_trim represents the amount of free
+  trailing space to leave untrimmed. If this argument is zero,
+  only the minimum amount of memory to maintain internal data
+  structures will be left (one page or less). Non-zero arguments
+  can be supplied to maintain enough trailing space to service
+  future expected allocations without having to re-obtain memory
+  from the system.
+  
+  Malloc_trim returns 1 if it actually released any memory, else 0.
+  On systems that do not support "negative sbrks", it will always
+  rreturn 0.
 */
+#if __STD_C
+int      public_mTRIm(size_t);
+#else
+int      public_mTRIm();
+#endif
 
+/*
+  malloc_usable_size(Void_t* p);
 
+  Returns the number of bytes you can actually use in
+  an allocated chunk, which may be more than you requested (although
+  often not) due to alignment and minimum size constraints.
+  You can use this many bytes without worrying about
+  overwriting other allocated objects. This is not a particularly great
+  programming practice. malloc_usable_size can be more useful in
+  debugging and assertions, for example:
 
-#ifndef DEFAULT_MMAP_MAX
-#if HAVE_MMAP
-#define DEFAULT_MMAP_MAX       (1024)
+  p = malloc(n);
+  assert(malloc_usable_size(p) >= 256);
+
+*/
+#if __STD_C
+size_t   public_mUSABLe(Void_t*);
 #else
-#define DEFAULT_MMAP_MAX       (0)
+size_t   public_mUSABLe();
 #endif
+
+/*
+  malloc_stats();
+  Prints on stderr the amount of space obtained from the system (both
+  via sbrk and mmap), the maximum amount (which may be more than
+  current if malloc_trim and/or munmap got called), and the current
+  number of bytes allocated via malloc (or realloc, etc) but not yet
+  freed. Note that this is the number of bytes allocated, not the
+  number requested. It will be larger than the number requested
+  because of alignment and bookkeeping overhead. Because it includes
+  alignment wastage as being in use, this figure may be greater than
+  zero even when no user-level chunks are allocated.
+
+  The reported current and maximum system memory can be inaccurate if
+  a program makes other calls to system memory allocation functions
+  (normally sbrk) outside of malloc.
+
+  malloc_stats prints only the most commonly interesting statistics.
+  More information can be obtained by calling mallinfo.
+
+*/
+#if __STD_C
+void     public_mSTATs(void);
+#else
+void     public_mSTATs();
 #endif
 
+/* mallopt tuning options */
+
 /*
-    M_MMAP_MAX is the maximum number of requests to simultaneously
-      service using mmap. This parameter exists because:
-
-         1. Some systems have a limited number of internal tables for
-            use by mmap.
-         2. In most systems, overreliance on mmap can degrade overall
-            performance.
-         3. If a program allocates many large regions, it is probably
-            better off using normal sbrk-based allocation routines that
-            can reclaim and reallocate normal heap memory. Using a
-            small value allows transition into this mode after the
-            first few allocations.
-
-      Setting to 0 disables all use of mmap.  If HAVE_MMAP is not set,
-      the default value is 0, and attempts to set it to non-zero values
-      in mallopt will fail.
+  M_MXFAST is the maximum request size used for "fastbins", special bins
+  that hold returned chunks without consolidating their spaces. This
+  enables future requests for chunks of the same size to be handled
+  very quickly, but can increase fragmentation, and thus increase the
+  overall memory footprint of a program.
+
+  This malloc manages fastbins very conservatively yet still
+  efficiently, so fragmentation is rarely a problem for values less
+  than or equal to the default.  The maximum supported value of MXFAST
+  is 80. You wouldn't want it any higher than this anyway.  Fastbins
+  are designed especially for use with many small structs, objects or
+  strings -- the default handles structs/objects/arrays with sizes up
+  to 8 4byte fields, or small strings representing words, tokens,
+  etc. Using fastbins for larger objects normally worsens
+  fragmentation without improving speed.
+
+  M_MXFAST is set in REQUEST size units. It is internally used in
+  chunksize units, which adds padding and alignment.  You can reduce
+  M_MXFAST to 0 to disable all use of fastbins.  This causes the malloc
+  algorithm to be a closer approximation of fifo-best-fit in all cases,
+  not just for larger requests, but will generally cause it to be
+  slower.
 */
 
 
+/* M_MXFAST is a standard SVID/XPG tuning option, usually listed in malloc.h */
+#ifndef M_MXFAST
+#define M_MXFAST            1    
+#endif
 
-#ifndef DEFAULT_CHECK_ACTION
-#define DEFAULT_CHECK_ACTION 1
+#ifndef DEFAULT_MXFAST
+#define DEFAULT_MXFAST     64
 #endif
 
-/* What to do if the standard debugging hooks are in place and a
-   corrupt pointer is detected: do nothing (0), print an error message
-   (1), or call abort() (2). */
 
+/*
+  M_TRIM_THRESHOLD is 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.
+  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.
+
+  The trim threshold and the mmap control parameters (see below)
+  can be traded off with one another. Trimming and mmapping are
+  two different ways of releasing unused memory back to the
+  system. Between these two, it is often possible to keep
+  system-level demands of a long-lived program down to a bare
+  minimum. For example, in one test suite of sessions measuring
+  the XF86 X server on Linux, using a trim threshold of 128K and a
+  mmap threshold of 192K led to near-minimal long term resource
+  consumption.
+
+  If you are using this malloc in a long-lived program, it should
+  pay to experiment with these values.  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's
+  worth it to tune for trimming rather tham memory mapping 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.  And in well-behaved long-lived programs,
+  controlling release of large blocks via trimming versus mapping
+  is usually faster.
+
+  However, in most programs, these parameters serve mainly as
+  protection against the system-level effects of carrying around
+  massive amounts of unneeded memory. Since frequent calls to
+  sbrk, mmap, and munmap otherwise degrade performance, the default
+  parameters are set to relatively high values that serve only as
+  safeguards.
+
+  The trim value It must be greater than page size to have any useful
+  effect.  To disable trimming completely, you can set to 
+  (unsigned long)(-1)
+
+  Trim settings interact with fastbin (MXFAST) settings: Unless
+  TRIM_FASTBINS is defined, automatic trimming never takes place upon
+  freeing a chunk with size less than or equal to MXFAST. Trimming is
+  instead delayed until subsequent freeing of larger chunks. However,
+  you can still force an attempted trim by calling malloc_trim.
+
+  Also, trimming is not generally possible in cases where
+  the main arena is obtained via mmap.
+
+  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.
+*/
 
+#define M_TRIM_THRESHOLD       -1
 
-#define HEAP_MIN_SIZE (32*1024)
-#define HEAP_MAX_SIZE (1024*1024) /* must be a power of two */
+#ifndef DEFAULT_TRIM_THRESHOLD
+#define DEFAULT_TRIM_THRESHOLD (128 * 1024)
+#endif
+
+/*
+  M_TOP_PAD is the amount of extra `padding' space to allocate or
+  retain whenever sbrk is called. It is used in two ways internally:
+
+  * When sbrk is called to extend the top of the arena to satisfy
+  a new malloc request, this much padding is added to the sbrk
+  request.
+
+  * When malloc_trim is called automatically from free(),
+  it is used as the `pad' argument.
+
+  In both cases, the actual amount of padding is rounded
+  so that the end of the arena is always a system page boundary.
+
+  The main reason for using padding is to avoid calling sbrk so
+  often. Having even a small pad greatly reduces the likelihood
+  that nearly every malloc request during program start-up (or
+  after trimming) will invoke sbrk, which needlessly wastes
+  time.
+
+  Automatic rounding-up to page-size units is normally sufficient
+  to avoid measurable overhead, so the default is 0.  However, in
+  systems where sbrk is relatively slow, it can pay to increase
+  this value, at the expense of carrying around more memory than
+  the program needs.
+*/
+
+#define M_TOP_PAD              -2
+
+#ifndef DEFAULT_TOP_PAD
+#define DEFAULT_TOP_PAD        (0)
+#endif
 
-/* HEAP_MIN_SIZE and HEAP_MAX_SIZE limit the size of mmap()ed heaps
-      that are dynamically created for multi-threaded programs.  The
-      maximum size must be a power of two, for fast determination of
-      which heap belongs to a chunk.  It should be much larger than
-      the mmap threshold, so that requests with a size just below that
-      threshold can be fulfilled without creating too many heaps.
+/*
+  M_MMAP_THRESHOLD is the request size threshold for using mmap()
+  to 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:
+
+   1. 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. 
+   2. Mapped memory can never become `locked' between
+      other chunks, as can happen with normally allocated chunks, which
+      means that even trimming via malloc_trim would not release them.
+   3. On some systems with "holes" in address spaces, mmap can obtain
+      memory that sbrk cannot.
+
+  However, it has the disadvantages that:
+
+   1. The space cannot be reclaimed, consolidated, and then
+      used to service later requests, as happens with normal chunks.
+   2. It can lead to more wastage because of mmap page alignment
+      requirements
+   3. It causes malloc performance to be more dependent on host
+      system memory management support routines which may vary in
+      implementation quality and may impose arbitrary
+      limitations. Generally, servicing a request via normal
+      malloc steps is faster than going through a system's mmap.
+
+  The advantages of mmap nearly always outweigh disadvantages for
+  "large" chunks, but the value of "large" varies across systems.  The
+  default is an empirically derived value that works well in most
+  systems.
 */
 
+#define M_MMAP_THRESHOLD      -3
 
+#ifndef DEFAULT_MMAP_THRESHOLD
+#define DEFAULT_MMAP_THRESHOLD (128 * 1024)
+#endif
 
-#ifndef THREAD_STATS
-#define THREAD_STATS 0
+/*
+  M_MMAP_MAX is the maximum number of requests to simultaneously
+  service using mmap. This parameter exists because
+  some systems have a limited number of internal tables for
+  use by mmap, and using more than a few of them may degrade
+  performance.
+
+  The default is set to a value that serves only as a safeguard.
+  Setting to 0 disables use of mmap for servicing large requests.  If
+  HAVE_MMAP is not set, the default value is 0, and attempts to set it
+  to non-zero values in mallopt will fail.
+*/
+
+#define M_MMAP_MAX             -4
+
+#ifndef DEFAULT_MMAP_MAX
+#if HAVE_MMAP
+#define DEFAULT_MMAP_MAX       (65536)
+#else
+#define DEFAULT_MMAP_MAX       (0)
+#endif
 #endif
 
-/* If THREAD_STATS is non-zero, some statistics on mutex locking are
-   computed. */
+#ifdef __cplusplus
+};  /* end of extern "C" */
+#endif
 
+#include "malloc.h"
+#include "thread-m.h"
 
-/* Macro to set errno.  */
-#ifndef __set_errno
-# define __set_errno(val) errno = (val)
+#ifndef BOUNDED_N
+#define BOUNDED_N(ptr, sz) (ptr)
+#endif
+#ifndef RETURN_ADDRESS
+#define RETURN_ADDRESS(X_) (NULL)
 #endif
 
 /* On some platforms we can compile internal, not exported functions better.
@@ -864,179 +1418,198 @@ do {                                                                          \
 # define internal_function
 #endif
 
+/* Forward declarations.  */
+struct malloc_chunk;
+typedef struct malloc_chunk* mchunkptr;
 
-/*
+/* Internal routines.  */
 
-  Special defines for the Linux/GNU C library.
+#if __STD_C
 
-*/
+Void_t*         _int_malloc(mstate, size_t);
+void            _int_free(mstate, Void_t*);
+Void_t*         _int_realloc(mstate, Void_t*, size_t);
+Void_t*         _int_memalign(mstate, size_t, size_t);
+Void_t*         _int_valloc(mstate, size_t);
+static Void_t*  _int_pvalloc(mstate, size_t);
+/*static Void_t*  cALLOc(size_t, size_t);*/
+static Void_t** _int_icalloc(mstate, size_t, size_t, Void_t**);
+static Void_t** _int_icomalloc(mstate, size_t, size_t*, Void_t**);
+static int      mTRIm(size_t);
+static size_t   mUSABLe(Void_t*);
+static void     mSTATs(void);
+static int      mALLOPt(int, int);
+static struct mallinfo mALLINFo(mstate);
+
+static Void_t* internal_function mem2mem_check(Void_t *p, size_t sz);
+static int internal_function top_check(void);
+static void internal_function munmap_chunk(mchunkptr p);
+static mchunkptr internal_function mremap_chunk(mchunkptr p, size_t new_size);
 
+static Void_t*   malloc_check(size_t sz, const Void_t *caller);
+static void      free_check(Void_t* mem, const Void_t *caller);
+static Void_t*   realloc_check(Void_t* oldmem, size_t bytes,
+			       const Void_t *caller);
+static Void_t*   memalign_check(size_t alignment, size_t bytes,
+				const Void_t *caller);
+#ifndef NO_THREADS
+static Void_t*   malloc_starter(size_t sz, const Void_t *caller);
+static void      free_starter(Void_t* mem, const Void_t *caller);
+static Void_t*   malloc_atfork(size_t sz, const Void_t *caller);
+static void      free_atfork(Void_t* mem, const Void_t *caller);
+#endif
 
-#ifdef _LIBC
+#else
 
-#if __STD_C
+Void_t*         _int_malloc();
+void            _int_free();
+Void_t*         _int_realloc();
+Void_t*         _int_memalign();
+Void_t*         _int_valloc();
+Void_t*         _int_pvalloc();
+/*static Void_t*  cALLOc();*/
+static Void_t** _int_icalloc();
+static Void_t** _int_icomalloc();
+static int      mTRIm();
+static size_t   mUSABLe();
+static void     mSTATs();
+static int      mALLOPt();
+static struct mallinfo mALLINFo();
 
-Void_t * __default_morecore (ptrdiff_t);
-Void_t *(*__morecore)(ptrdiff_t) = __default_morecore;
+#endif
 
-#else
 
-Void_t * __default_morecore ();
-Void_t *(*__morecore)() = __default_morecore;
 
-#endif
 
-#define MORECORE (*__morecore)
-#define MORECORE_FAILURE 0
+/* ------------- Optional versions of memcopy ---------------- */
 
-#ifndef MORECORE_CLEARS
-#define MORECORE_CLEARS 1
-#endif
 
-static size_t __libc_pagesize;
+#if USE_MEMCPY
 
-#define access	__access
-#define mmap    __mmap
-#define munmap  __munmap
-#define mremap  __mremap
-#define mprotect __mprotect
-#undef malloc_getpagesize
-#define malloc_getpagesize __libc_pagesize
+/* 
+  Note: memcpy is ONLY invoked with non-overlapping regions,
+  so the (usually slower) memmove is not needed.
+*/
 
-#else /* _LIBC */
+#define MALLOC_COPY(dest, src, nbytes)  memcpy(dest, src, nbytes)
+#define MALLOC_ZERO(dest, nbytes)       memset(dest, 0,   nbytes)
 
-#if __STD_C
-extern Void_t*     sbrk(ptrdiff_t);
-#else
-extern Void_t*     sbrk();
-#endif
+#else /* !USE_MEMCPY */
 
-#ifndef MORECORE
-#define MORECORE sbrk
-#endif
+/* Use Duff's device for good zeroing/copying performance. */
 
-#ifndef MORECORE_FAILURE
-#define MORECORE_FAILURE -1
-#endif
+#define MALLOC_ZERO(charp, nbytes)                                            \
+do {                                                                          \
+  INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp);                           \
+  unsigned long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T);                     \
+  long mcn;                                                                   \
+  if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
+  switch (mctmp) {                                                            \
+    case 0: for(;;) { *mzp++ = 0;                                             \
+    case 7:           *mzp++ = 0;                                             \
+    case 6:           *mzp++ = 0;                                             \
+    case 5:           *mzp++ = 0;                                             \
+    case 4:           *mzp++ = 0;                                             \
+    case 3:           *mzp++ = 0;                                             \
+    case 2:           *mzp++ = 0;                                             \
+    case 1:           *mzp++ = 0; if(mcn <= 0) break; mcn--; }                \
+  }                                                                           \
+} while(0)
+
+#define MALLOC_COPY(dest,src,nbytes)                                          \
+do {                                                                          \
+  INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src;                            \
+  INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest;                           \
+  unsigned long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T);                     \
+  long mcn;                                                                   \
+  if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; }             \
+  switch (mctmp) {                                                            \
+    case 0: for(;;) { *mcdst++ = *mcsrc++;                                    \
+    case 7:           *mcdst++ = *mcsrc++;                                    \
+    case 6:           *mcdst++ = *mcsrc++;                                    \
+    case 5:           *mcdst++ = *mcsrc++;                                    \
+    case 4:           *mcdst++ = *mcsrc++;                                    \
+    case 3:           *mcdst++ = *mcsrc++;                                    \
+    case 2:           *mcdst++ = *mcsrc++;                                    \
+    case 1:           *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; }       \
+  }                                                                           \
+} while(0)
 
-#ifndef MORECORE_CLEARS
-#define MORECORE_CLEARS 1
 #endif
 
-#endif /* _LIBC */
+/* ------------------ MMAP support ------------------  */
 
-#ifdef _LIBC
 
-#define cALLOc          __libc_calloc
-#define fREe            __libc_free
-#define mALLOc          __libc_malloc
-#define mEMALIGn        __libc_memalign
-#define rEALLOc         __libc_realloc
-#define vALLOc          __libc_valloc
-#define pvALLOc         __libc_pvalloc
-#define mALLINFo        __libc_mallinfo
-#define mALLOPt         __libc_mallopt
-#define mALLOC_STATs    __malloc_stats
-#define mALLOC_USABLE_SIZe __malloc_usable_size
-#define mALLOC_TRIm     __malloc_trim
-#define mALLOC_GET_STATe __malloc_get_state
-#define mALLOC_SET_STATe __malloc_set_state
+#if HAVE_MMAP
 
-#else
+#include <fcntl.h>
+#ifndef LACKS_SYS_MMAN_H
+#include <sys/mman.h>
+#endif
 
-#define cALLOc          calloc
-#define fREe            free
-#define mALLOc          malloc
-#define mEMALIGn        memalign
-#define rEALLOc         realloc
-#define vALLOc          valloc
-#define pvALLOc         pvalloc
-#define mALLINFo        mallinfo
-#define mALLOPt         mallopt
-#define mALLOC_STATs    malloc_stats
-#define mALLOC_USABLE_SIZe malloc_usable_size
-#define mALLOC_TRIm     malloc_trim
-#define mALLOC_GET_STATe malloc_get_state
-#define mALLOC_SET_STATe malloc_set_state
+#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON)
+# define MAP_ANONYMOUS MAP_ANON
+#endif
+#if !defined(MAP_FAILED)
+# define MAP_FAILED ((char*)-1)
+#endif
 
+#ifndef MAP_NORESERVE
+# ifdef MAP_AUTORESRV
+#  define MAP_NORESERVE MAP_AUTORESRV
+# else
+#  define MAP_NORESERVE 0
+# endif
 #endif
 
-/* Public routines */
+/* 
+   Nearly all versions of mmap support MAP_ANONYMOUS, 
+   so the following is unlikely to be needed, but is
+   supplied just in case.
+*/
 
-#if __STD_C
+#ifndef MAP_ANONYMOUS
 
-#ifndef _LIBC
-void    ptmalloc_init(void);
-#endif
-Void_t* mALLOc(size_t);
-void    fREe(Void_t*);
-Void_t* rEALLOc(Void_t*, size_t);
-Void_t* mEMALIGn(size_t, size_t);
-Void_t* vALLOc(size_t);
-Void_t* pvALLOc(size_t);
-Void_t* cALLOc(size_t, size_t);
-void    cfree(Void_t*);
-int     mALLOC_TRIm(size_t);
-size_t  mALLOC_USABLE_SIZe(Void_t*);
-void    mALLOC_STATs(void);
-int     mALLOPt(int, int);
-struct mallinfo mALLINFo(void);
-Void_t* mALLOC_GET_STATe(void);
-int     mALLOC_SET_STATe(Void_t*);
-
-#else /* !__STD_C */
+static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */
 
-#ifndef _LIBC
-void    ptmalloc_init();
-#endif
-Void_t* mALLOc();
-void    fREe();
-Void_t* rEALLOc();
-Void_t* mEMALIGn();
-Void_t* vALLOc();
-Void_t* pvALLOc();
-Void_t* cALLOc();
-void    cfree();
-int     mALLOC_TRIm();
-size_t  mALLOC_USABLE_SIZe();
-void    mALLOC_STATs();
-int     mALLOPt();
-struct mallinfo mALLINFo();
-Void_t* mALLOC_GET_STATe();
-int     mALLOC_SET_STATe();
-
-#endif /* __STD_C */
+#define MMAP(addr, size, prot, flags) ((dev_zero_fd < 0) ? \
+ (dev_zero_fd = open("/dev/zero", O_RDWR), \
+  mmap((addr), (size), (prot), (flags), dev_zero_fd, 0)) : \
+   mmap((addr), (size), (prot), (flags), dev_zero_fd, 0))
 
+#else
 
-#ifdef __cplusplus
-} /* end of extern "C" */
-#endif
+#define MMAP(addr, size, prot, flags) \
+ (mmap((addr), (size), (prot), (flags)|MAP_ANONYMOUS, -1, 0))
 
-#if !defined(NO_THREADS) && !HAVE_MMAP
-"Can't have threads support without mmap"
-#endif
-#if USE_ARENAS && !HAVE_MMAP
-"Can't have multiple arenas without mmap"
 #endif
 
 
+#endif /* HAVE_MMAP */
+
+
 /*
-  Type declarations
+  -----------------------  Chunk representations -----------------------
 */
 
 
-struct malloc_chunk
-{
-  INTERNAL_SIZE_T prev_size; /* Size of previous chunk (if free). */
-  INTERNAL_SIZE_T size;      /* Size in bytes, including overhead. */
-  struct malloc_chunk* fd;   /* double links -- used only if free. */
+/*
+  This struct declaration is misleading (but accurate and necessary).
+  It declares a "view" into memory allowing access to necessary
+  fields at known offsets from a given base. See explanation below.
+*/
+
+struct malloc_chunk {
+
+  INTERNAL_SIZE_T      prev_size;  /* Size of previous chunk (if free).  */
+  INTERNAL_SIZE_T      size;       /* Size in bytes, including overhead. */
+
+  struct malloc_chunk* fd;         /* double links -- used only if free. */
   struct malloc_chunk* bk;
 };
 
-typedef struct malloc_chunk* mchunkptr;
 
 /*
-
    malloc_chunk details:
 
     (The following includes lightly edited explanations by Colin Plumb.)
@@ -1071,9 +1644,9 @@ nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     the malloc code, but "mem" is the pointer that is returned to the
     user.  "Nextchunk" is the beginning of the next contiguous chunk.
 
-    Chunks always begin on even word boundaries, so the mem portion
+    Chunks always begin on even word boundries, so the mem portion
     (which is returned to the user) is also on an even word boundary, and
-    thus double-word aligned.
+    thus at least double-word aligned.
 
     Free chunks are stored in circular doubly-linked lists, and look like this:
 
@@ -1098,3126 +1671,3333 @@ nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+
     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.)
+    The very first chunk allocated always has this bit set,
+    preventing access to non-existent (or non-owned) memory. If
+    prev_inuse 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.
 
     Note that the `foot' of the current chunk is actually represented
-    as the prev_size of the NEXT chunk. (This makes it easier to
-    deal with alignments etc).
+    as the prev_size 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 two exceptions to all this are
 
-     1. The special chunk `top', which doesn't bother using the
-        trailing size field since there is no
-        next contiguous chunk that would have to index off it. (After
-        initialization, `top' is forced to always exist.  If it would
-        become less than MINSIZE bytes long, it is replenished via
-        malloc_extend_top.)
+     1. The special chunk `top' doesn't bother using the
+        trailing size field since there is no next contiguous chunk
+        that would have to index off it. After initialization, `top'
+        is forced to always exist.  If it would become less than
+        MINSIZE bytes long, it is replenished.
 
      2. Chunks allocated via mmap, which have the second-lowest-order
         bit (IS_MMAPPED) set in their size fields.  Because they are
-        never merged or traversed from any other chunk, they have no
-        foot size or inuse information.
-
-    Available chunks are kept in any of several places (all declared below):
-
-    * `av': An array of chunks serving as bin headers for consolidated
-       chunks. Each bin is doubly linked.  The bins are approximately
-       proportionally (log) spaced.  There are a lot of these bins
-       (128). This may look excessive, but works very well in
-       practice.  All procedures maintain the invariant that no
-       consolidated chunk physically borders another one. Chunks in
-       bins are kept in size order, with ties going to the
-       approximately least recently used chunk.
-
-       The chunks in each bin are maintained in decreasing sorted order by
-       size.  This is irrelevant for the small bins, which all contain
-       the same-sized chunks, but facilitates best-fit allocation for
-       larger chunks. (These lists are just sequential. Keeping them in
-       order almost never requires enough traversal to warrant using
-       fancier ordered data structures.)  Chunks of the same size are
-       linked with the most recently freed at the front, and allocations
-       are taken from the back.  This results in LRU or FIFO allocation
-       order, which tends to give each chunk an equal opportunity to be
-       consolidated with adjacent freed chunks, resulting in larger free
-       chunks and less fragmentation.
-
-    * `top': The top-most available chunk (i.e., the one bordering the
-       end of available memory) is treated specially. It 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).
-
-    * `last_remainder': A bin holding only the remainder of the
-       most recently split (non-top) chunk. This bin is checked
-       before other non-fitting chunks, so as to provide better
-       locality for runs of sequentially allocated chunks.
-
-    *  Implicitly, through the host system's memory mapping tables.
-       If supported, requests greater than a threshold are usually
-       serviced via calls to mmap, and then later released via munmap.
+        allocated one-by-one, each must contain its own trailing size field.
 
 */
 
 /*
-   Bins
+  ---------- Size and alignment checks and conversions ----------
+*/
 
-    The bins are an array of pairs of pointers serving as the
-    heads of (initially empty) doubly-linked lists of chunks, laid out
-    in a way so that each pair can be treated as if it were in a
-    malloc_chunk. (This way, the fd/bk offsets for linking bin heads
-    and chunks are the same).
+/* conversion from malloc headers to user pointers, and back */
 
-    Bins for sizes < 512 bytes contain chunks of all the same size, spaced
-    8 bytes apart. Larger bins are approximately logarithmically
-    spaced. (See the table below.)
+#define chunk2mem(p)   ((Void_t*)((char*)(p) + 2*SIZE_SZ))
+#define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ))
 
-    Bin layout:
+/* The smallest possible chunk */
+#define MIN_CHUNK_SIZE        (sizeof(struct malloc_chunk))
 
-    64 bins of size       8
-    32 bins of size      64
-    16 bins of size     512
-     8 bins of size    4096
-     4 bins of size   32768
-     2 bins of size  262144
-     1 bin  of size what's left
+/* The smallest size we can malloc is an aligned minimal chunk */
 
-    There is actually a little bit of slop in the numbers in bin_index
-    for the sake of speed. This makes no difference elsewhere.
+#define MINSIZE  \
+  (unsigned long)(((MIN_CHUNK_SIZE+MALLOC_ALIGN_MASK) & ~MALLOC_ALIGN_MASK))
 
-    The special chunks `top' and `last_remainder' get their own bins,
-    (this is implemented via yet more trickery with the av array),
-    although `top' is never properly linked to its bin since it is
-    always handled specially.
+/* Check if m has acceptable alignment */
 
-*/
+#define aligned_OK(m)  (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
 
-#define NAV             128   /* number of bins */
 
-typedef struct malloc_chunk* mbinptr;
+/* 
+   Check if a request is so large that it would wrap around zero when
+   padded and aligned. To simplify some other code, the bound is made
+   low enough so that adding MINSIZE will also not wrap around zero.
+*/
 
-/* An arena is a configuration of malloc_chunks together with an array
-   of bins.  With multiple threads, it must be locked via a mutex
-   before changing its data structures.  One or more `heaps' are
-   associated with each arena, except for the main_arena, which is
-   associated only with the `main heap', i.e.  the conventional free
-   store obtained with calls to MORECORE() (usually sbrk).  The `av'
-   array is never mentioned directly in the code, but instead used via
-   bin access macros. */
-
-typedef struct _arena {
-  mbinptr av[2*NAV + 2];
-  struct _arena *next;
-  size_t size;
-#if THREAD_STATS
-  long stat_lock_direct, stat_lock_loop, stat_lock_wait;
-#endif
-  mutex_t mutex;
-} arena;
+#define REQUEST_OUT_OF_RANGE(req)                                 \
+  ((unsigned long)(req) >=                                        \
+   (unsigned long)(INTERNAL_SIZE_T)(-2 * MINSIZE))    
 
+/* pad request bytes into a usable size -- internal version */
 
-/* A heap is a single contiguous memory region holding (coalesceable)
-   malloc_chunks.  It is allocated with mmap() and always starts at an
-   address aligned to HEAP_MAX_SIZE.  Not used unless compiling with
-   USE_ARENAS. */
+#define request2size(req)                                         \
+  (((req) + SIZE_SZ + MALLOC_ALIGN_MASK < MINSIZE)  ?             \
+   MINSIZE :                                                      \
+   ((req) + SIZE_SZ + MALLOC_ALIGN_MASK) & ~MALLOC_ALIGN_MASK)
 
-typedef struct _heap_info {
-  arena *ar_ptr; /* Arena for this heap. */
-  struct _heap_info *prev; /* Previous heap. */
-  size_t size;   /* Current size in bytes. */
-  size_t pad;    /* Make sure the following data is properly aligned. */
-} heap_info;
+/*  Same, except also perform argument check */
 
+#define checked_request2size(req, sz)                             \
+  if (REQUEST_OUT_OF_RANGE(req)) {                                \
+    MALLOC_FAILURE_ACTION;                                        \
+    return 0;                                                     \
+  }                                                               \
+  (sz) = request2size(req);                                              
 
 /*
-  Static functions (forward declarations)
+  --------------- Physical chunk operations ---------------
 */
 
-#if __STD_C
-
-static void      chunk_free(arena *ar_ptr, mchunkptr p) internal_function;
-static mchunkptr chunk_alloc(arena *ar_ptr, INTERNAL_SIZE_T size)
-     internal_function;
-static mchunkptr chunk_realloc(arena *ar_ptr, mchunkptr oldp,
-                               INTERNAL_SIZE_T oldsize, INTERNAL_SIZE_T nb)
-     internal_function;
-static mchunkptr chunk_align(arena *ar_ptr, INTERNAL_SIZE_T nb,
-                             size_t alignment) internal_function;
-static int       main_trim(size_t pad) internal_function;
-#if USE_ARENAS
-static int       heap_trim(heap_info *heap, size_t pad) internal_function;
-#endif
-#if defined _LIBC || defined MALLOC_HOOKS
-static Void_t*   malloc_check(size_t sz, const Void_t *caller);
-static void      free_check(Void_t* mem, const Void_t *caller);
-static Void_t*   realloc_check(Void_t* oldmem, size_t bytes,
-			       const Void_t *caller);
-static Void_t*   memalign_check(size_t alignment, size_t bytes,
-				const Void_t *caller);
-#ifndef NO_THREADS
-static Void_t*   malloc_starter(size_t sz, const Void_t *caller);
-static void      free_starter(Void_t* mem, const Void_t *caller);
-static Void_t*   malloc_atfork(size_t sz, const Void_t *caller);
-static void      free_atfork(Void_t* mem, const Void_t *caller);
-#endif
-#endif
-
-#else
-
-static void      chunk_free();
-static mchunkptr chunk_alloc();
-static mchunkptr chunk_realloc();
-static mchunkptr chunk_align();
-static int       main_trim();
-#if USE_ARENAS
-static int       heap_trim();
-#endif
-#if defined _LIBC || defined MALLOC_HOOKS
-static Void_t*   malloc_check();
-static void      free_check();
-static Void_t*   realloc_check();
-static Void_t*   memalign_check();
-#ifndef NO_THREADS
-static Void_t*   malloc_starter();
-static void      free_starter();
-static Void_t*   malloc_atfork();
-static void      free_atfork();
-#endif
-#endif
-
-#endif
-
-
 
-/* sizes, alignments */
+/* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
+#define PREV_INUSE 0x1
 
-#define SIZE_SZ                (sizeof(INTERNAL_SIZE_T))
-/* Allow the default to be overwritten on the compiler command line.  */
-#ifndef MALLOC_ALIGNMENT
-# define MALLOC_ALIGNMENT      (SIZE_SZ + SIZE_SZ)
-#endif
-#define MALLOC_ALIGN_MASK      (MALLOC_ALIGNMENT - 1)
-#define MINSIZE                (sizeof(struct malloc_chunk))
+/* extract inuse bit of previous chunk */
+#define prev_inuse(p)       ((p)->size & PREV_INUSE)
 
-/* conversion from malloc headers to user pointers, and back */
 
-#define chunk2mem(p) ((Void_t*)((char*)(p) + 2*SIZE_SZ))
-#define mem2chunk(mem) chunk_at_offset((mem), -2*SIZE_SZ)
+/* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
+#define IS_MMAPPED 0x2
 
-/* pad request bytes into a usable size, return non-zero on overflow */
+/* check for mmap()'ed chunk */
+#define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
 
-#define request2size(req, nb) \
- ((nb = (req) + (SIZE_SZ + MALLOC_ALIGN_MASK)),\
-  ((long)nb <= 0 || nb < (INTERNAL_SIZE_T) (req) \
-   ? (__set_errno (ENOMEM), 1) \
-   : ((nb < (MINSIZE + MALLOC_ALIGN_MASK) \
-	   ? (nb = MINSIZE) : (nb &= ~MALLOC_ALIGN_MASK)), 0)))
 
-/* Check if m has acceptable alignment */
+/* size field is or'ed with NON_MAIN_ARENA if the chunk was obtained
+   from a non-main arena.  This is only set immediately before handing
+   the chunk to the user, if necessary.  */
+#define NON_MAIN_ARENA 0x4
 
-#define aligned_OK(m)    (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0)
+/* check for chunk from non-main arena */
+#define chunk_non_main_arena(p) ((p)->size & NON_MAIN_ARENA)
 
 
-
+/* 
+  Bits to mask off when extracting size 
 
-/*
-  Physical chunk operations
+  Note: IS_MMAPPED is intentionally not masked off from size field in
+  macros for which mmapped chunks should never be seen. This should
+  cause helpful core dumps to occur if it is tried by accident by
+  people extending or adapting this malloc.
 */
+#define SIZE_BITS (PREV_INUSE|IS_MMAPPED|NON_MAIN_ARENA)
 
-
-/* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */
-
-#define PREV_INUSE 0x1UL
-
-/* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */
-
-#define IS_MMAPPED 0x2UL
-
-/* Bits to mask off when extracting size */
-
-#define SIZE_BITS (PREV_INUSE|IS_MMAPPED)
+/* Get size, ignoring use bits */
+#define chunksize(p)         ((p)->size & ~(SIZE_BITS))
 
 
 /* Ptr to next physical malloc_chunk. */
-
-#define next_chunk(p) chunk_at_offset((p), (p)->size & ~PREV_INUSE)
+#define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~SIZE_BITS) ))
 
 /* Ptr to previous physical malloc_chunk */
-
-#define prev_chunk(p) chunk_at_offset((p), -(p)->prev_size)
-
+#define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) ))
 
 /* Treat space at ptr + offset as a chunk */
+#define chunk_at_offset(p, s)  ((mchunkptr)(((char*)(p)) + (s)))
 
-#define chunk_at_offset(p, s)  BOUNDED_1((mchunkptr)(((char*)(p)) + (s)))
+/* extract p's inuse bit */
+#define inuse(p)\
+((((mchunkptr)(((char*)(p))+((p)->size & ~SIZE_BITS)))->size) & PREV_INUSE)
 
+/* set/clear chunk as being inuse without otherwise disturbing */
+#define set_inuse(p)\
+((mchunkptr)(((char*)(p)) + ((p)->size & ~SIZE_BITS)))->size |= PREV_INUSE
 
-
+#define clear_inuse(p)\
+((mchunkptr)(((char*)(p)) + ((p)->size & ~SIZE_BITS)))->size &= ~(PREV_INUSE)
 
-/*
-  Dealing with use bits
-*/
 
-/* extract p's inuse bit */
+/* check/set/clear inuse bits in known places */
+#define inuse_bit_at_offset(p, s)\
+ (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE)
 
-#define inuse(p) (next_chunk(p)->size & PREV_INUSE)
+#define set_inuse_bit_at_offset(p, s)\
+ (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE)
 
-/* extract inuse bit of previous chunk */
+#define clear_inuse_bit_at_offset(p, s)\
+ (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE))
 
-#define prev_inuse(p)  ((p)->size & PREV_INUSE)
 
-/* check for mmap()'ed chunk */
+/* Set size at head, without disturbing its use bit */
+#define set_head_size(p, s)  ((p)->size = (((p)->size & SIZE_BITS) | (s)))
 
-#define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED)
+/* Set size/use field */
+#define set_head(p, s)       ((p)->size = (s))
 
-/* set/clear chunk as in use without otherwise disturbing */
+/* Set size at footer (only when chunk is not in use) */
+#define set_foot(p, s)       (((mchunkptr)((char*)(p) + (s)))->prev_size = (s))
 
-#define set_inuse(p) (next_chunk(p)->size |= PREV_INUSE)
 
-#define clear_inuse(p) (next_chunk(p)->size &= ~PREV_INUSE)
+/*
+  -------------------- Internal data structures --------------------
+
+   All internal state is held in an instance of malloc_state defined
+   below. There are no other static variables, except in two optional
+   cases: 
+   * If USE_MALLOC_LOCK is defined, the mALLOC_MUTEx declared above. 
+   * If HAVE_MMAP is true, but mmap doesn't support
+     MAP_ANONYMOUS, a dummy file descriptor for mmap.
+
+   Beware of lots of tricks that minimize the total bookkeeping space
+   requirements. The result is a little over 1K bytes (for 4byte
+   pointers and size_t.)
+*/
 
-/* check/set/clear inuse bits in known places */
+/*
+  Bins
+
+    An array of bin headers for free chunks. Each bin is doubly
+    linked.  The bins are approximately proportionally (log) spaced.
+    There are a lot of these bins (128). This may look excessive, but
+    works very well in practice.  Most bins hold sizes that are
+    unusual as malloc request sizes, but are more usual for fragments
+    and consolidated sets of chunks, which is what these bins hold, so
+    they can be found quickly.  All procedures maintain the invariant
+    that no consolidated chunk physically borders another one, so each
+    chunk in a list is known to be preceeded and followed by either
+    inuse chunks or the ends of memory.
+
+    Chunks in bins are kept in size order, with ties going to the
+    approximately least recently used chunk. Ordering isn't needed
+    for the small bins, which all contain the same-sized chunks, but
+    facilitates best-fit allocation for larger chunks. These lists
+    are just sequential. Keeping them in order almost never requires
+    enough traversal to warrant using fancier ordered data
+    structures.  
+
+    Chunks of the same size are linked with the most
+    recently freed at the front, and allocations are taken from the
+    back.  This results in LRU (FIFO) allocation order, which tends
+    to give each chunk an equal opportunity to be consolidated with
+    adjacent freed chunks, resulting in larger free chunks and less
+    fragmentation.
+
+    To simplify use in double-linked lists, each bin header acts
+    as a malloc_chunk. This avoids special-casing for headers.
+    But to conserve space and improve locality, we allocate
+    only the fd/bk pointers of bins, and then use repositioning tricks
+    to treat these as the fields of a malloc_chunk*.  
+*/
 
-#define inuse_bit_at_offset(p, s) \
-  (chunk_at_offset((p), (s))->size & PREV_INUSE)
+typedef struct malloc_chunk* mbinptr;
 
-#define set_inuse_bit_at_offset(p, s) \
-  (chunk_at_offset((p), (s))->size |= PREV_INUSE)
+/* addressing -- note that bin_at(0) does not exist */
+#define bin_at(m, i) ((mbinptr)((char*)&((m)->bins[(i)<<1]) - (SIZE_SZ<<1)))
 
-#define clear_inuse_bit_at_offset(p, s) \
-  (chunk_at_offset((p), (s))->size &= ~(PREV_INUSE))
+/* analog of ++bin */
+#define next_bin(b)  ((mbinptr)((char*)(b) + (sizeof(mchunkptr)<<1)))
 
+/* Reminders about list directionality within bins */
+#define first(b)     ((b)->fd)
+#define last(b)      ((b)->bk)
 
-
+/* Take a chunk off a bin list */
+#define unlink(P, BK, FD) {                                            \
+  FD = P->fd;                                                          \
+  BK = P->bk;                                                          \
+  FD->bk = BK;                                                         \
+  BK->fd = FD;                                                         \
+}
 
 /*
-  Dealing with size fields
-*/
+  Indexing
 
-/* Get size, ignoring use bits */
+    Bins for sizes < 512 bytes contain chunks of all the same size, spaced
+    8 bytes apart. Larger bins are approximately logarithmically spaced:
+
+    64 bins of size       8
+    32 bins of size      64
+    16 bins of size     512
+     8 bins of size    4096
+     4 bins of size   32768
+     2 bins of size  262144
+     1 bin  of size what's left
 
-#define chunksize(p)          ((p)->size & ~(SIZE_BITS))
+    There is actually a little bit of slop in the numbers in bin_index
+    for the sake of speed. This makes no difference elsewhere.
 
-/* Set size at head, without disturbing its use bit */
+    The bins top out around 1MB because we expect to service large
+    requests via mmap.
+*/
 
-#define set_head_size(p, s)   ((p)->size = (((p)->size & PREV_INUSE) | (s)))
+#define NBINS             128
+#define NSMALLBINS         64
+#define SMALLBIN_WIDTH      8
+#define MIN_LARGE_SIZE    512
 
-/* Set size/use ignoring previous bits in header */
+#define in_smallbin_range(sz)  \
+  ((unsigned long)(sz) < (unsigned long)MIN_LARGE_SIZE)
 
-#define set_head(p, s)        ((p)->size = (s))
+#define smallbin_index(sz)     (((unsigned)(sz)) >> 3)
 
-/* Set size at footer (only when chunk is not in use) */
+#define largebin_index(sz)                                                   \
+(((((unsigned long)(sz)) >>  6) <= 32)?  56 + (((unsigned long)(sz)) >>  6): \
+ ((((unsigned long)(sz)) >>  9) <= 20)?  91 + (((unsigned long)(sz)) >>  9): \
+ ((((unsigned long)(sz)) >> 12) <= 10)? 110 + (((unsigned long)(sz)) >> 12): \
+ ((((unsigned long)(sz)) >> 15) <=  4)? 119 + (((unsigned long)(sz)) >> 15): \
+ ((((unsigned long)(sz)) >> 18) <=  2)? 124 + (((unsigned long)(sz)) >> 18): \
+                                        126)
 
-#define set_foot(p, s)   (chunk_at_offset(p, s)->prev_size = (s))
+#define bin_index(sz) \
+ ((in_smallbin_range(sz)) ? smallbin_index(sz) : largebin_index(sz))
 
 
-
+/*
+  Unsorted chunks
 
+    All remainders from chunk splits, as well as all returned chunks,
+    are first placed in the "unsorted" bin. They are then placed
+    in regular bins after malloc gives them ONE chance to be used before
+    binning. So, basically, the unsorted_chunks list acts as a queue,
+    with chunks being placed on it in free (and malloc_consolidate),
+    and taken off (to be either used or placed in bins) in malloc.
 
-/* access macros */
+    The NON_MAIN_ARENA flag is never set for unsorted chunks, so it
+    does not have to be taken into account in size comparisons.
+*/
 
-#define bin_at(a, i)   BOUNDED_1(_bin_at(a, i))
-#define _bin_at(a, i)  ((mbinptr)((char*)&(((a)->av)[2*(i)+2]) - 2*SIZE_SZ))
-#define init_bin(a, i) ((a)->av[2*(i)+2] = (a)->av[2*(i)+3] = bin_at((a), (i)))
-#define next_bin(b)    ((mbinptr)((char*)(b) + 2 * sizeof(((arena*)0)->av[0])))
-#define prev_bin(b)    ((mbinptr)((char*)(b) - 2 * sizeof(((arena*)0)->av[0])))
+/* The otherwise unindexable 1-bin is used to hold unsorted chunks. */
+#define unsorted_chunks(M)          (bin_at(M, 1))
 
 /*
-   The first 2 bins are never indexed. The corresponding av cells are instead
-   used for bookkeeping. This is not to save space, but to simplify
-   indexing, maintain locality, and avoid some initialization tests.
+  Top
+
+    The top-most available chunk (i.e., the one bordering the end of
+    available memory) is treated specially. It 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).  Because top initially
+    points to its own bin with initial zero size, thus forcing
+    extension on the first malloc request, we avoid having any special
+    code in malloc to check whether it even exists yet. But we still
+    need to do so when getting memory from system, so we make
+    initial_top treat the bin as a legal but unusable chunk during the
+    interval between initialization and the first call to
+    sYSMALLOc. (This is somewhat delicate, since it relies on
+    the 2 preceding words to be zero during this interval as well.)
 */
 
-#define binblocks(a)      (bin_at(a,0)->size)/* bitvector of nonempty blocks */
-#define top(a)            (bin_at(a,0)->fd)  /* The topmost chunk */
-#define last_remainder(a) (bin_at(a,1))      /* remainder from last split */
+/* Conveniently, the unsorted bin can be used as dummy top on first call */
+#define initial_top(M)              (unsorted_chunks(M))
 
 /*
-   Because top initially points to its own bin with initial
-   zero size, thus forcing extension on the first malloc request,
-   we avoid having any special code in malloc to check whether
-   it even exists yet. But we still need to in malloc_extend_top.
-*/
+  Binmap
 
-#define initial_top(a)    ((mchunkptr)bin_at(a, 0))
+    To help compensate for the large number of bins, a one-level index
+    structure is used for bin-by-bin searching.  `binmap' is a
+    bitvector recording whether bins are definitely empty so they can
+    be skipped over during during traversals.  The bits are NOT always
+    cleared as soon as bins are empty, but instead only
+    when they are noticed to be empty during traversal in malloc.
+*/
 
-
+/* Conservatively use 32 bits per map word, even if on 64bit system */
+#define BINMAPSHIFT      5
+#define BITSPERMAP       (1U << BINMAPSHIFT)
+#define BINMAPSIZE       (NBINS / BITSPERMAP)
 
-/* field-extraction macros */
+#define idx2block(i)     ((i) >> BINMAPSHIFT)
+#define idx2bit(i)       ((1U << ((i) & ((1U << BINMAPSHIFT)-1))))
 
-#define first(b) ((b)->fd)
-#define last(b)  ((b)->bk)
+#define mark_bin(m,i)    ((m)->binmap[idx2block(i)] |=  idx2bit(i))
+#define unmark_bin(m,i)  ((m)->binmap[idx2block(i)] &= ~(idx2bit(i)))
+#define get_binmap(m,i)  ((m)->binmap[idx2block(i)] &   idx2bit(i))
 
 /*
-  Indexing into bins
+  Fastbins
+
+    An array of lists holding recently freed small chunks.  Fastbins
+    are not doubly linked.  It is faster to single-link them, and
+    since chunks are never removed from the middles of these lists,
+    double linking is not necessary. Also, unlike regular bins, they
+    are not even processed in FIFO order (they use faster LIFO) since
+    ordering doesn't much matter in the transient contexts in which
+    fastbins are normally used.
+
+    Chunks in fastbins keep their inuse bit set, so they cannot
+    be consolidated with other free chunks. malloc_consolidate
+    releases all chunks in fastbins and consolidates them with
+    other free chunks. 
 */
 
-#define bin_index(sz)                                                         \
-(((((unsigned long)(sz)) >> 9) ==    0) ?       (((unsigned long)(sz)) >>  3):\
- ((((unsigned long)(sz)) >> 9) <=    4) ?  56 + (((unsigned long)(sz)) >>  6):\
- ((((unsigned long)(sz)) >> 9) <=   20) ?  91 + (((unsigned long)(sz)) >>  9):\
- ((((unsigned long)(sz)) >> 9) <=   84) ? 110 + (((unsigned long)(sz)) >> 12):\
- ((((unsigned long)(sz)) >> 9) <=  340) ? 119 + (((unsigned long)(sz)) >> 15):\
- ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18):\
-                                          126)
-/*
-  bins for chunks < 512 are all spaced 8 bytes apart, and hold
-  identically sized chunks. This is exploited in malloc.
-*/
+typedef struct malloc_chunk* mfastbinptr;
 
-#define MAX_SMALLBIN         63
-#define MAX_SMALLBIN_SIZE   512
-#define SMALLBIN_WIDTH        8
+/* offset 2 to use otherwise unindexable first 2 bins */
+#define fastbin_index(sz)        ((((unsigned int)(sz)) >> 3) - 2)
 
-#define smallbin_index(sz)  (((unsigned long)(sz)) >> 3)
+/* The maximum fastbin request size we support */
+#define MAX_FAST_SIZE     80
+
+#define NFASTBINS  (fastbin_index(request2size(MAX_FAST_SIZE))+1)
 
 /*
-   Requests are `small' if both the corresponding and the next bin are small
+  FASTBIN_CONSOLIDATION_THRESHOLD is the size of a chunk in free()
+  that triggers automatic consolidation of possibly-surrounding
+  fastbin chunks. This is a heuristic, so the exact value should not
+  matter too much. It is defined at half the default trim threshold as a
+  compromise heuristic to only attempt consolidation if it is likely
+  to lead to trimming. However, it is not dynamically tunable, since
+  consolidation reduces fragmentation surrounding large chunks even 
+  if trimming is not used.
 */
 
-#define is_small_request(nb) ((nb) < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH)
-
-
+#define FASTBIN_CONSOLIDATION_THRESHOLD  (65536UL)
 
 /*
-    To help compensate for the large number of bins, a one-level index
-    structure is used for bin-by-bin searching.  `binblocks' is a
-    one-word bitvector recording whether groups of BINBLOCKWIDTH bins
-    have any (possibly) non-empty bins, so they can be skipped over
-    all at once during during traversals. The bits are NOT always
-    cleared as soon as all bins in a block are empty, but instead only
-    when all are noticed to be empty during traversal in malloc.
+  Since the lowest 2 bits in max_fast don't matter in size comparisons, 
+  they are used as flags.
 */
 
-#define BINBLOCKWIDTH     4   /* bins per block */
+/*
+  FASTCHUNKS_BIT held in max_fast indicates that there are probably
+  some fastbin chunks. It is set true on entering a chunk into any
+  fastbin, and cleared only in malloc_consolidate.
 
-/* bin<->block macros */
+  The truth value is inverted so that have_fastchunks will be true
+  upon startup (since statics are zero-filled), simplifying
+  initialization checks.
+*/
 
-#define idx2binblock(ix)      ((unsigned)1 << ((ix) / BINBLOCKWIDTH))
-#define mark_binblock(a, ii)  (binblocks(a) |= idx2binblock(ii))
-#define clear_binblock(a, ii) (binblocks(a) &= ~(idx2binblock(ii)))
+#define FASTCHUNKS_BIT        (1U)
 
+#define have_fastchunks(M)     (((M)->max_fast &  FASTCHUNKS_BIT) == 0)
+#define clear_fastchunks(M)    ((M)->max_fast |=  FASTCHUNKS_BIT)
+#define set_fastchunks(M)      ((M)->max_fast &= ~FASTCHUNKS_BIT)
 
-
+/*
+  NONCONTIGUOUS_BIT indicates that MORECORE does not return contiguous
+  regions.  Otherwise, contiguity is exploited in merging together,
+  when possible, results from consecutive MORECORE calls.
 
-/* Static bookkeeping data */
+  The initial value comes from MORECORE_CONTIGUOUS, but is
+  changed dynamically if mmap is ever used as an sbrk substitute.
+*/
 
-/* Helper macro to initialize bins */
-#define IAV(i) _bin_at(&main_arena, i), _bin_at(&main_arena, i)
+#define NONCONTIGUOUS_BIT     (2U)
 
-static arena main_arena = {
-    {
- 0, 0,
- IAV(0),   IAV(1),   IAV(2),   IAV(3),   IAV(4),   IAV(5),   IAV(6),   IAV(7),
- IAV(8),   IAV(9),   IAV(10),  IAV(11),  IAV(12),  IAV(13),  IAV(14),  IAV(15),
- IAV(16),  IAV(17),  IAV(18),  IAV(19),  IAV(20),  IAV(21),  IAV(22),  IAV(23),
- IAV(24),  IAV(25),  IAV(26),  IAV(27),  IAV(28),  IAV(29),  IAV(30),  IAV(31),
- IAV(32),  IAV(33),  IAV(34),  IAV(35),  IAV(36),  IAV(37),  IAV(38),  IAV(39),
- IAV(40),  IAV(41),  IAV(42),  IAV(43),  IAV(44),  IAV(45),  IAV(46),  IAV(47),
- IAV(48),  IAV(49),  IAV(50),  IAV(51),  IAV(52),  IAV(53),  IAV(54),  IAV(55),
- IAV(56),  IAV(57),  IAV(58),  IAV(59),  IAV(60),  IAV(61),  IAV(62),  IAV(63),
- IAV(64),  IAV(65),  IAV(66),  IAV(67),  IAV(68),  IAV(69),  IAV(70),  IAV(71),
- IAV(72),  IAV(73),  IAV(74),  IAV(75),  IAV(76),  IAV(77),  IAV(78),  IAV(79),
- IAV(80),  IAV(81),  IAV(82),  IAV(83),  IAV(84),  IAV(85),  IAV(86),  IAV(87),
- IAV(88),  IAV(89),  IAV(90),  IAV(91),  IAV(92),  IAV(93),  IAV(94),  IAV(95),
- IAV(96),  IAV(97),  IAV(98),  IAV(99),  IAV(100), IAV(101), IAV(102), IAV(103),
- IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111),
- IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119),
- IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127)
-    },
-    &main_arena, /* next */
-    0, /* size */
-#if THREAD_STATS
-    0, 0, 0, /* stat_lock_direct, stat_lock_loop, stat_lock_wait */
-#endif
-    MUTEX_INITIALIZER /* mutex */
-};
+#define contiguous(M)          (((M)->max_fast &  NONCONTIGUOUS_BIT) == 0)
+#define noncontiguous(M)       (((M)->max_fast &  NONCONTIGUOUS_BIT) != 0)
+#define set_noncontiguous(M)   ((M)->max_fast |=  NONCONTIGUOUS_BIT)
+#define set_contiguous(M)      ((M)->max_fast &= ~NONCONTIGUOUS_BIT)
 
-#undef IAV
+/* 
+   Set value of max_fast. 
+   Use impossibly small value if 0.
+   Precondition: there are no existing fastbin chunks.
+   Setting the value clears fastchunk bit but preserves noncontiguous bit.
+*/
 
-/* Thread specific data */
+#define set_max_fast(M, s) \
+  (M)->max_fast = (((s) == 0)? SMALLBIN_WIDTH: request2size(s)) | \
+  FASTCHUNKS_BIT | \
+  ((M)->max_fast &  NONCONTIGUOUS_BIT)
 
-static tsd_key_t arena_key;
-static mutex_t list_lock = MUTEX_INITIALIZER;
 
-#if THREAD_STATS
-static int stat_n_heaps;
-#define THREAD_STAT(x) x
-#else
-#define THREAD_STAT(x) do ; while(0)
-#endif
+/*
+   ----------- Internal state representation and initialization -----------
+*/
 
-/* variables holding tunable values */
+struct malloc_state {
+  /* Serialize access.  */
+  mutex_t mutex;
 
-static unsigned long trim_threshold   = DEFAULT_TRIM_THRESHOLD;
-static unsigned long top_pad          = DEFAULT_TOP_PAD;
-static unsigned int  n_mmaps_max      = DEFAULT_MMAP_MAX;
-static unsigned long mmap_threshold   = DEFAULT_MMAP_THRESHOLD;
-static int           check_action     = DEFAULT_CHECK_ACTION;
+  /* Statistics for locking.  Only used if THREAD_STATS is defined.  */
+  long stat_lock_direct, stat_lock_loop, stat_lock_wait;
+  long pad0_[1]; /* try to give the mutex its own cacheline */
 
-/* The first value returned from sbrk */
-static char* sbrk_base = (char*)(-1);
+  /* The maximum chunk size to be eligible for fastbin */
+  INTERNAL_SIZE_T  max_fast;   /* low 2 bits used as flags */
 
-/* The maximum memory obtained from system via sbrk */
-static unsigned long max_sbrked_mem;
+  /* Fastbins */
+  mfastbinptr      fastbins[NFASTBINS];
 
-/* The maximum via either sbrk or mmap (too difficult to track with threads) */
-#ifdef NO_THREADS
-static unsigned long max_total_mem;
-#endif
+  /* Base of the topmost chunk -- not otherwise kept in a bin */
+  mchunkptr        top;
 
-/* The total memory obtained from system via sbrk */
-#define sbrked_mem (main_arena.size)
+  /* The remainder from the most recent split of a small request */
+  mchunkptr        last_remainder;
 
-/* Tracking mmaps */
+  /* Normal bins packed as described above */
+  mchunkptr        bins[NBINS * 2];
 
-static unsigned int n_mmaps;
-static unsigned int max_n_mmaps;
-static unsigned long mmapped_mem;
-static unsigned long max_mmapped_mem;
+  /* Bitmap of bins */
+  unsigned int     binmap[BINMAPSIZE];
 
-/* Mapped memory in non-main arenas (reliable only for NO_THREADS). */
-static unsigned long arena_mem;
+  /* Linked list */
+  struct malloc_state *next;
 
+  /* Memory allocated from the system in this arena.  */
+  INTERNAL_SIZE_T system_mem;
+  INTERNAL_SIZE_T max_system_mem;
+};
 
-
-#ifndef _LIBC
-#define weak_variable
-#else
-/* In GNU libc we want the hook variables to be weak definitions to
-   avoid a problem with Emacs.  */
-#define weak_variable weak_function
-#endif
+struct malloc_par {
+  /* Tunable parameters */
+  unsigned long    trim_threshold;
+  INTERNAL_SIZE_T  top_pad;
+  INTERNAL_SIZE_T  mmap_threshold;
+
+  /* Memory map support */
+  int              n_mmaps;
+  int              n_mmaps_max;
+  int              max_n_mmaps;
+
+  /* Cache malloc_getpagesize */
+  unsigned int     pagesize;    
+
+  /* Statistics */
+  INTERNAL_SIZE_T  mmapped_mem;
+  /*INTERNAL_SIZE_T  sbrked_mem;*/
+  /*INTERNAL_SIZE_T  max_sbrked_mem;*/
+  INTERNAL_SIZE_T  max_mmapped_mem;
+  INTERNAL_SIZE_T  max_total_mem; /* only kept for NO_THREADS */
+
+  /* First address handed out by MORECORE/sbrk.  */
+  char*            sbrk_base;
+};
 
-/* Already initialized? */
-int __malloc_initialized = -1;
+/* There are several instances of this struct ("arenas") in this
+   malloc.  If you are adapting this malloc in a way that does NOT use
+   a static or mmapped malloc_state, you MUST explicitly zero-fill it
+   before using. This malloc relies on the property that malloc_state
+   is initialized to all zeroes (as is true of C statics).  */
 
+static struct malloc_state main_arena;
 
-#ifndef NO_THREADS
+/* There is only one instance of the malloc parameters.  */
 
-/* Magic value for the thread-specific arena pointer when
-   malloc_atfork() is in use.  */
+static struct malloc_par mp_;
 
-#define ATFORK_ARENA_PTR ((Void_t*)-1)
+/*
+  Initialize a malloc_state struct.
 
-/* The following two functions are registered via thread_atfork() to
-   make sure that the mutexes remain in a consistent state in the
-   fork()ed version of a thread.  Also adapt the malloc and free hooks
-   temporarily, because the `atfork' handler mechanism may use
-   malloc/free internally (e.g. in LinuxThreads). */
+  This is called only from within malloc_consolidate, which needs
+  be called in the same contexts anyway.  It is never called directly
+  outside of malloc_consolidate because some optimizing compilers try
+  to inline it at all call points, which turns out not to be an
+  optimization at all. (Inlining it in malloc_consolidate is fine though.)
+*/
 
-#if defined _LIBC || defined MALLOC_HOOKS
-static __malloc_ptr_t (*save_malloc_hook) __MALLOC_P ((size_t __size,
-						       const __malloc_ptr_t));
-static void           (*save_free_hook) __MALLOC_P ((__malloc_ptr_t __ptr,
-						     const __malloc_ptr_t));
-static Void_t*        save_arena;
+#if __STD_C
+static void malloc_init_state(mstate av)
+#else
+static void malloc_init_state(av) mstate av;
 #endif
-
-static void
-ptmalloc_lock_all __MALLOC_P((void))
 {
-  arena *ar_ptr;
-
-  (void)mutex_lock(&list_lock);
-  for(ar_ptr = &main_arena;;) {
-    (void)mutex_lock(&ar_ptr->mutex);
-    ar_ptr = ar_ptr->next;
-    if(ar_ptr == &main_arena) break;
+  int     i;
+  mbinptr bin;
+  
+  /* Establish circular links for normal bins */
+  for (i = 1; i < NBINS; ++i) { 
+    bin = bin_at(av,i);
+    bin->fd = bin->bk = bin;
   }
-#if defined _LIBC || defined MALLOC_HOOKS
-  save_malloc_hook = __malloc_hook;
-  save_free_hook = __free_hook;
-  __malloc_hook = malloc_atfork;
-  __free_hook = free_atfork;
-  /* Only the current thread may perform malloc/free calls now. */
-  tsd_getspecific(arena_key, save_arena);
-  tsd_setspecific(arena_key, ATFORK_ARENA_PTR);
+
+#if MORECORE_CONTIGUOUS
+  if (av != &main_arena)
 #endif
-}
+    set_noncontiguous(av);
 
-static void
-ptmalloc_unlock_all __MALLOC_P((void))
-{
-  arena *ar_ptr;
+  set_max_fast(av, DEFAULT_MXFAST);
 
-#if defined _LIBC || defined MALLOC_HOOKS
-  tsd_setspecific(arena_key, save_arena);
-  __malloc_hook = save_malloc_hook;
-  __free_hook = save_free_hook;
-#endif
-  for(ar_ptr = &main_arena;;) {
-    (void)mutex_unlock(&ar_ptr->mutex);
-    ar_ptr = ar_ptr->next;
-    if(ar_ptr == &main_arena) break;
-  }
-  (void)mutex_unlock(&list_lock);
+  av->top            = initial_top(av);
 }
 
-static void
-ptmalloc_init_all __MALLOC_P((void))
-{
-  arena *ar_ptr;
+/* 
+   Other internal utilities operating on mstates
+*/
 
-#if defined _LIBC || defined MALLOC_HOOKS
-  tsd_setspecific(arena_key, save_arena);
-  __malloc_hook = save_malloc_hook;
-  __free_hook = save_free_hook;
+#if __STD_C
+static Void_t*  sYSMALLOc(INTERNAL_SIZE_T, mstate);
+static int      sYSTRIm(size_t, mstate);
+static void     malloc_consolidate(mstate);
+static Void_t** iALLOc(mstate, size_t, size_t*, int, Void_t**);
+#else
+static Void_t*  sYSMALLOc();
+static int      sYSTRIm();
+static void     malloc_consolidate();
+static Void_t** iALLOc();
 #endif
-  for(ar_ptr = &main_arena;;) {
-    (void)mutex_init(&ar_ptr->mutex);
-    ar_ptr = ar_ptr->next;
-    if(ar_ptr == &main_arena) break;
-  }
-  (void)mutex_init(&list_lock);
-}
 
-#endif /* !defined NO_THREADS */
+/* ------------------- Support for multiple arenas -------------------- */
+#include "arena.c"
 
-/* Initialization routine. */
-#if defined(_LIBC)
-#if 0
-static void ptmalloc_init __MALLOC_P ((void)) __attribute__ ((constructor));
-#endif
+/*
+  Debugging support
 
-#ifdef _LIBC
-#include <string.h>
-extern char **_environ;
+  These routines make a number of assertions about the states
+  of data structures that should be true at all times. If any
+  are not true, it's very likely that a user program has somehow
+  trashed memory. (It's also possible that there is a coding error
+  in malloc. In which case, please report it!)
+*/
 
-static char *
-internal_function
-next_env_entry (char ***position)
-{
-  char **current = *position;
-  char *result = NULL;
+#if ! MALLOC_DEBUG
 
-  while (*current != NULL)
-    {
-      if (__builtin_expect ((*current)[0] == 'M', 0)
-	  && (*current)[1] == 'A'
-	  && (*current)[2] == 'L'
-	  && (*current)[3] == 'L'
-	  && (*current)[4] == 'O'
-	  && (*current)[5] == 'C'
-	  && (*current)[6] == '_')
-	{
-	  result = &(*current)[7];
-
-	  /* Save current position for next visit.  */
-	  *position = ++current;
-
-	  break;
-	}
+#define check_chunk(A,P)
+#define check_free_chunk(A,P)
+#define check_inuse_chunk(A,P)
+#define check_remalloced_chunk(A,P,N)
+#define check_malloced_chunk(A,P,N)
+#define check_malloc_state(A)
 
-      ++current;
-    }
+#else
 
-  return result;
-}
-#endif
+#define check_chunk(A,P)              do_check_chunk(A,P)
+#define check_free_chunk(A,P)         do_check_free_chunk(A,P)
+#define check_inuse_chunk(A,P)        do_check_inuse_chunk(A,P)
+#define check_remalloced_chunk(A,P,N) do_check_remalloced_chunk(A,P,N)
+#define check_malloced_chunk(A,P,N)   do_check_malloced_chunk(A,P,N)
+#define check_malloc_state(A)         do_check_malloc_state(A)
 
-static void
-ptmalloc_init __MALLOC_P((void))
+/*
+  Properties of all chunks
+*/
+
+#if __STD_C
+static void do_check_chunk(mstate av, mchunkptr p)
 #else
-void
-ptmalloc_init __MALLOC_P((void))
+static void do_check_chunk(av, p) mstate av; mchunkptr p;
 #endif
 {
-#if defined _LIBC || defined MALLOC_HOOKS
-# if __STD_C
-  const char* s;
-# else
-  char* s;
-# endif
-#endif
-  int secure;
-
-  if(__malloc_initialized >= 0) return;
-  __malloc_initialized = 0;
-#ifdef _LIBC
-  __libc_pagesize = __getpagesize();
-#endif
-#ifndef NO_THREADS
-#if defined _LIBC || defined MALLOC_HOOKS
-  /* With some threads implementations, creating thread-specific data
-     or initializing a mutex may call malloc() itself.  Provide a
-     simple starter version (realloc() won't work). */
-  save_malloc_hook = __malloc_hook;
-  save_free_hook = __free_hook;
-  __malloc_hook = malloc_starter;
-  __free_hook = free_starter;
-#endif
-#ifdef _LIBC
-  /* Initialize the pthreads interface. */
-  if (__pthread_initialize != NULL)
-    __pthread_initialize();
-#endif
-#endif /* !defined NO_THREADS */
-  mutex_init(&main_arena.mutex);
-  mutex_init(&list_lock);
-  tsd_key_create(&arena_key, NULL);
-  tsd_setspecific(arena_key, (Void_t *)&main_arena);
-  thread_atfork(ptmalloc_lock_all, ptmalloc_unlock_all, ptmalloc_init_all);
-#if defined _LIBC || defined MALLOC_HOOKS
-#ifndef NO_THREADS
-  __malloc_hook = save_malloc_hook;
-  __free_hook = save_free_hook;
-#endif
-  secure = __libc_enable_secure;
-#ifdef _LIBC
-  s = NULL;
-  if (_environ != NULL)
-    {
-      char **runp = _environ;
-      char *envline;
-
-      while (__builtin_expect ((envline = next_env_entry (&runp)) != NULL, 0))
-	{
-	  size_t len = strcspn (envline, "=");
-
-	  if (envline[len] != '=')
-	    /* This is a "MALLOC_" variable at the end of the string
-	       without a '=' character.  Ignore it since otherwise we
-	       will access invalid memory below.  */
-	    continue;
-
-	  switch (len)
-	    {
-	    case 6:
-	      if (memcmp (envline, "CHECK_", 6) == 0)
-		s = &envline[7];
-	      break;
-	    case 8:
-	      if (! secure && memcmp (envline, "TOP_PAD_", 8) == 0)
-		mALLOPt(M_TOP_PAD, atoi(&envline[9]));
-	      break;
-	    case 9:
-	      if (! secure && memcmp (envline, "MMAP_MAX_", 9) == 0)
-		mALLOPt(M_MMAP_MAX, atoi(&envline[10]));
-	      break;
-	    case 15:
-	      if (! secure)
-		{
-		  if (memcmp (envline, "TRIM_THRESHOLD_", 15) == 0)
-		    mALLOPt(M_TRIM_THRESHOLD, atoi(&envline[16]));
-		  else if (memcmp (envline, "MMAP_THRESHOLD_", 15) == 0)
-		    mALLOPt(M_MMAP_THRESHOLD, atoi(&envline[16]));
-		}
-	      break;
-	    default:
-	      break;
-	    }
-	}
+  unsigned long sz = chunksize(p);
+  /* min and max possible addresses assuming contiguous allocation */
+  char* max_address = (char*)(av->top) + chunksize(av->top);
+  char* min_address = max_address - av->system_mem;
+
+  if (!chunk_is_mmapped(p)) {
+    
+    /* Has legal address ... */
+    if (p != av->top) {
+      if (contiguous(av)) {
+        assert(((char*)p) >= min_address);
+        assert(((char*)p + sz) <= ((char*)(av->top)));
+      }
     }
-#else
-  if (! secure)
-    {
-      if((s = getenv("MALLOC_TRIM_THRESHOLD_")))
-	mALLOPt(M_TRIM_THRESHOLD, atoi(s));
-      if((s = getenv("MALLOC_TOP_PAD_")))
-	mALLOPt(M_TOP_PAD, atoi(s));
-      if((s = getenv("MALLOC_MMAP_THRESHOLD_")))
-	mALLOPt(M_MMAP_THRESHOLD, atoi(s));
-      if((s = getenv("MALLOC_MMAP_MAX_")))
-	mALLOPt(M_MMAP_MAX, atoi(s));
+    else {
+      /* top size is always at least MINSIZE */
+      assert((unsigned long)(sz) >= MINSIZE);
+      /* top predecessor always marked inuse */
+      assert(prev_inuse(p));
     }
-  s = getenv("MALLOC_CHECK_");
-#endif
-  if(s) {
-    if(s[0]) mALLOPt(M_CHECK_ACTION, (int)(s[0] - '0'));
-    __malloc_check_init();
+      
   }
-  if(__malloc_initialize_hook != NULL)
-    (*__malloc_initialize_hook)();
+  else {
+#if HAVE_MMAP
+    /* address is outside main heap  */
+    if (contiguous(av) && av->top != initial_top(av)) {
+      assert(((char*)p) < min_address || ((char*)p) > max_address);
+    }
+    /* chunk is page-aligned */
+    assert(((p->prev_size + sz) & (mp_.pagesize-1)) == 0);
+    /* mem is aligned */
+    assert(aligned_OK(chunk2mem(p)));
+#else
+    /* force an appropriate assert violation if debug set */
+    assert(!chunk_is_mmapped(p));
 #endif
-  __malloc_initialized = 1;
+  }
 }
 
-/* There are platforms (e.g. Hurd) with a link-time hook mechanism. */
-#ifdef thread_atfork_static
-thread_atfork_static(ptmalloc_lock_all, ptmalloc_unlock_all, \
-                     ptmalloc_init_all)
-#endif
-
-#if defined _LIBC || defined MALLOC_HOOKS
-
-/* Hooks for debugging versions.  The initial hooks just call the
-   initialization routine, then do the normal work. */
+/*
+  Properties of free chunks
+*/
 
-static Void_t*
 #if __STD_C
-malloc_hook_ini(size_t sz, const __malloc_ptr_t caller)
+static void do_check_free_chunk(mstate av, mchunkptr p)
 #else
-malloc_hook_ini(sz, caller)
-     size_t sz; const __malloc_ptr_t caller;
+static void do_check_free_chunk(av, p) mstate av; mchunkptr p;
 #endif
 {
-  __malloc_hook = NULL;
-  ptmalloc_init();
-  return mALLOc(sz);
-}
+  INTERNAL_SIZE_T sz = p->size & ~(PREV_INUSE|NON_MAIN_ARENA);
+  mchunkptr next = chunk_at_offset(p, sz);
 
-static Void_t*
-#if __STD_C
-realloc_hook_ini(Void_t* ptr, size_t sz, const __malloc_ptr_t caller)
-#else
-realloc_hook_ini(ptr, sz, caller)
-     Void_t* ptr; size_t sz; const __malloc_ptr_t caller;
-#endif
-{
-  __malloc_hook = NULL;
-  __realloc_hook = NULL;
-  ptmalloc_init();
-  return rEALLOc(ptr, sz);
-}
+  do_check_chunk(av, p);
 
-static Void_t*
-#if __STD_C
-memalign_hook_ini(size_t alignment, size_t sz, const __malloc_ptr_t caller)
-#else
-memalign_hook_ini(alignment, sz, caller)
-     size_t alignment; size_t sz; const __malloc_ptr_t caller;
-#endif
-{
-  __memalign_hook = NULL;
-  ptmalloc_init();
-  return mEMALIGn(alignment, sz);
-}
+  /* Chunk must claim to be free ... */
+  assert(!inuse(p));
+  assert (!chunk_is_mmapped(p));
 
-void weak_variable (*__malloc_initialize_hook) __MALLOC_P ((void)) = NULL;
-void weak_variable (*__free_hook) __MALLOC_P ((__malloc_ptr_t __ptr,
-					       const __malloc_ptr_t)) = NULL;
-__malloc_ptr_t weak_variable (*__malloc_hook)
- __MALLOC_P ((size_t __size, const __malloc_ptr_t)) = malloc_hook_ini;
-__malloc_ptr_t weak_variable (*__realloc_hook)
- __MALLOC_P ((__malloc_ptr_t __ptr, size_t __size, const __malloc_ptr_t))
-     = realloc_hook_ini;
-__malloc_ptr_t weak_variable (*__memalign_hook)
- __MALLOC_P ((size_t __alignment, size_t __size, const __malloc_ptr_t))
-     = memalign_hook_ini;
-void weak_variable (*__after_morecore_hook) __MALLOC_P ((void)) = NULL;
-
-/* Whether we are using malloc checking.  */
-static int using_malloc_checking;
-
-/* A flag that is set by malloc_set_state, to signal that malloc checking
-   must not be enabled on the request from the user (via the MALLOC_CHECK_
-   environment variable).  It is reset by __malloc_check_init to tell
-   malloc_set_state that the user has requested malloc checking.
-
-   The purpose of this flag is to make sure that malloc checking is not
-   enabled when the heap to be restored was constructed without malloc
-   checking, and thus does not contain the required magic bytes.
-   Otherwise the heap would be corrupted by calls to free and realloc.  If
-   it turns out that the heap was created with malloc checking and the
-   user has requested it malloc_set_state just calls __malloc_check_init
-   again to enable it.  On the other hand, reusing such a heap without
-   further malloc checking is safe.  */
-static int disallow_malloc_check;
-
-/* Activate a standard set of debugging hooks. */
-void
-__malloc_check_init()
-{
-  if (disallow_malloc_check) {
-    disallow_malloc_check = 0;
-    return;
+  /* Unless a special marker, must have OK fields */
+  if ((unsigned long)(sz) >= MINSIZE)
+  {
+    assert((sz & MALLOC_ALIGN_MASK) == 0);
+    assert(aligned_OK(chunk2mem(p)));
+    /* ... matching footer field */
+    assert(next->prev_size == sz);
+    /* ... and is fully consolidated */
+    assert(prev_inuse(p));
+    assert (next == av->top || inuse(next));
+
+    /* ... and has minimally sane links */
+    assert(p->fd->bk == p);
+    assert(p->bk->fd == p);
   }
-  using_malloc_checking = 1;
-  __malloc_hook = malloc_check;
-  __free_hook = free_check;
-  __realloc_hook = realloc_check;
-  __memalign_hook = memalign_check;
-  if(check_action & 1)
-    fprintf(stderr, "malloc: using debugging hooks\n");
+  else /* markers are always of size SIZE_SZ */
+    assert(sz == SIZE_SZ);
 }
 
-#endif
-
-
-
-
-
-/* Routines dealing with mmap(). */
-
-#if HAVE_MMAP
-
-#ifndef MAP_ANONYMOUS
-
-static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */
-
-#define MMAP(addr, size, prot, flags) ((dev_zero_fd < 0) ? \
- (dev_zero_fd = open("/dev/zero", O_RDWR), \
-  mmap((addr), (size), (prot), (flags), dev_zero_fd, 0)) : \
-   mmap((addr), (size), (prot), (flags), dev_zero_fd, 0))
-
-#else
-
-#define MMAP(addr, size, prot, flags) \
- (mmap((addr), (size), (prot), (flags)|MAP_ANONYMOUS, -1, 0))
-
-#endif
+/*
+  Properties of inuse chunks
+*/
 
-#if defined __GNUC__ && __GNUC__ >= 2
-/* This function is only called from one place, inline it.  */
-__inline__
-#endif
-static mchunkptr
-internal_function
 #if __STD_C
-mmap_chunk(size_t size)
+static void do_check_inuse_chunk(mstate av, mchunkptr p)
 #else
-mmap_chunk(size) size_t size;
+static void do_check_inuse_chunk(av, p) mstate av; mchunkptr p;
 #endif
 {
-  size_t page_mask = malloc_getpagesize - 1;
-  mchunkptr p;
+  mchunkptr next;
 
-  /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because
-   * there is no following chunk whose prev_size field could be used.
-   */
-  size = (size + SIZE_SZ + page_mask) & ~page_mask;
+  do_check_chunk(av, p);
 
-  p = (mchunkptr)MMAP(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE);
-  if(p == (mchunkptr) MAP_FAILED) return 0;
+  if (chunk_is_mmapped(p))
+    return; /* mmapped chunks have no next/prev */
 
-  n_mmaps++;
-  if (n_mmaps > max_n_mmaps) max_n_mmaps = n_mmaps;
+  /* Check whether it claims to be in use ... */
+  assert(inuse(p));
 
-  /* We demand that eight bytes into a page must be 8-byte aligned. */
-  assert(aligned_OK(chunk2mem(p)));
+  next = next_chunk(p);
 
-  /* The offset to the start of the mmapped region is stored
-   * in the prev_size field of the chunk; normally it is zero,
-   * but that can be changed in memalign().
-   */
-  p->prev_size = 0;
-  set_head(p, size|IS_MMAPPED);
+  /* ... and is surrounded by OK chunks.
+    Since more things can be checked with free chunks than inuse ones,
+    if an inuse chunk borders them and debug is on, it's worth doing them.
+  */
+  if (!prev_inuse(p))  {
+    /* Note that we cannot even look at prev unless it is not inuse */
+    mchunkptr prv = prev_chunk(p);
+    assert(next_chunk(prv) == p);
+    do_check_free_chunk(av, prv);
+  }
 
-  mmapped_mem += size;
-  if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
-    max_mmapped_mem = mmapped_mem;
-#ifdef NO_THREADS
-  if ((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
-    max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
-#endif
-  return p;
+  if (next == av->top) {
+    assert(prev_inuse(next));
+    assert(chunksize(next) >= MINSIZE);
+  }
+  else if (!inuse(next))
+    do_check_free_chunk(av, next);
 }
 
-static void
-internal_function
+/*
+  Properties of chunks recycled from fastbins
+*/
+
 #if __STD_C
-munmap_chunk(mchunkptr p)
+static void do_check_remalloced_chunk(mstate av, mchunkptr p, INTERNAL_SIZE_T s)
 #else
-munmap_chunk(p) mchunkptr p;
+static void do_check_remalloced_chunk(av, p, s)
+mstate av; mchunkptr p; INTERNAL_SIZE_T s;
 #endif
 {
-  INTERNAL_SIZE_T size = chunksize(p);
-  int ret;
-
-  assert (chunk_is_mmapped(p));
-  assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
-  assert((n_mmaps > 0));
-  assert(((p->prev_size + size) & (malloc_getpagesize-1)) == 0);
+  INTERNAL_SIZE_T sz = p->size & ~(PREV_INUSE|NON_MAIN_ARENA);
 
-  n_mmaps--;
-  mmapped_mem -= (size + p->prev_size);
+  if (!chunk_is_mmapped(p)) {
+    assert(av == arena_for_chunk(p));
+    if (chunk_non_main_arena(p))
+      assert(av != &main_arena);
+    else
+      assert(av == &main_arena);
+  }
 
-  ret = munmap((char *)p - p->prev_size, size + p->prev_size);
+  do_check_inuse_chunk(av, p);
 
-  /* munmap returns non-zero on failure */
-  assert(ret == 0);
+  /* Legal size ... */
+  assert((sz & MALLOC_ALIGN_MASK) == 0);
+  assert((unsigned long)(sz) >= MINSIZE);
+  /* ... and alignment */
+  assert(aligned_OK(chunk2mem(p)));
+  /* chunk is less than MINSIZE more than request */
+  assert((long)(sz) - (long)(s) >= 0);
+  assert((long)(sz) - (long)(s + MINSIZE) < 0);
 }
 
-#if HAVE_MREMAP
+/*
+  Properties of nonrecycled chunks at the point they are malloced
+*/
 
-static mchunkptr
-internal_function
 #if __STD_C
-mremap_chunk(mchunkptr p, size_t new_size)
+static void do_check_malloced_chunk(mstate av, mchunkptr p, INTERNAL_SIZE_T s)
 #else
-mremap_chunk(p, new_size) mchunkptr p; size_t new_size;
+static void do_check_malloced_chunk(av, p, s)
+mstate av; mchunkptr p; INTERNAL_SIZE_T s;
 #endif
 {
-  size_t page_mask = malloc_getpagesize - 1;
-  INTERNAL_SIZE_T offset = p->prev_size;
-  INTERNAL_SIZE_T size = chunksize(p);
-  char *cp;
-
-  assert (chunk_is_mmapped(p));
-  assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem));
-  assert((n_mmaps > 0));
-  assert(((size + offset) & (malloc_getpagesize-1)) == 0);
+  /* same as recycled case ... */
+  do_check_remalloced_chunk(av, p, s);
 
-  /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
-  new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
+  /*
+    ... plus,  must obey implementation invariant that prev_inuse is
+    always true of any allocated chunk; i.e., that 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.  This does not necessarily hold however for chunks
+    recycled via fastbins.
+  */
 
-  cp = (char *)mremap((char *)p - offset, size + offset, new_size,
-                      MREMAP_MAYMOVE);
+  assert(prev_inuse(p));
+}
 
-  if (cp == MAP_FAILED) return 0;
 
-  p = (mchunkptr)(cp + offset);
+/*
+  Properties of malloc_state.
 
-  assert(aligned_OK(chunk2mem(p)));
+  This may be useful for debugging malloc, as well as detecting user
+  programmer errors that somehow write into malloc_state.
 
-  assert((p->prev_size == offset));
-  set_head(p, (new_size - offset)|IS_MMAPPED);
-
-  mmapped_mem -= size + offset;
-  mmapped_mem += new_size;
-  if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem)
-    max_mmapped_mem = mmapped_mem;
-#ifdef NO_THREADS
-  if ((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
-    max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
-#endif
-  return p;
-}
+  If you are extending or experimenting with this malloc, you can
+  probably figure out how to hack this routine to print out or
+  display chunk addresses, sizes, bins, and other instrumentation.
+*/
 
-#endif /* HAVE_MREMAP */
+static void do_check_malloc_state(mstate av)
+{
+  int i;
+  mchunkptr p;
+  mchunkptr q;
+  mbinptr b;
+  unsigned int binbit;
+  int empty;
+  unsigned int idx;
+  INTERNAL_SIZE_T size;
+  unsigned long total = 0;
+  int max_fast_bin;
 
-#endif /* HAVE_MMAP */
+  /* internal size_t must be no wider than pointer type */
+  assert(sizeof(INTERNAL_SIZE_T) <= sizeof(char*));
 
-
+  /* alignment is a power of 2 */
+  assert((MALLOC_ALIGNMENT & (MALLOC_ALIGNMENT-1)) == 0);
 
-/* Managing heaps and arenas (for concurrent threads) */
+  /* cannot run remaining checks until fully initialized */
+  if (av->top == 0 || av->top == initial_top(av))
+    return;
 
-#if USE_ARENAS
+  /* pagesize is a power of 2 */
+  assert((mp_.pagesize & (mp_.pagesize-1)) == 0);
 
-/* Create a new heap.  size is automatically rounded up to a multiple
-   of the page size. */
+  /* A contiguous main_arena is consistent with sbrk_base.  */
+  if (av == &main_arena && contiguous(av))
+    assert((char*)mp_.sbrk_base + av->system_mem ==
+	   (char*)av->top + chunksize(av->top));
 
-static heap_info *
-internal_function
-#if __STD_C
-new_heap(size_t size)
-#else
-new_heap(size) size_t size;
-#endif
-{
-  size_t page_mask = malloc_getpagesize - 1;
-  char *p1, *p2;
-  unsigned long ul;
-  heap_info *h;
-
-  if(size+top_pad < HEAP_MIN_SIZE)
-    size = HEAP_MIN_SIZE;
-  else if(size+top_pad <= HEAP_MAX_SIZE)
-    size += top_pad;
-  else if(size > HEAP_MAX_SIZE)
-    return 0;
-  else
-    size = HEAP_MAX_SIZE;
-  size = (size + page_mask) & ~page_mask;
-
-  /* A memory region aligned to a multiple of HEAP_MAX_SIZE is needed.
-     No swap space needs to be reserved for the following large
-     mapping (on Linux, this is the case for all non-writable mappings
-     anyway). */
-  p1 = (char *)MMAP(0, HEAP_MAX_SIZE<<1, PROT_NONE, MAP_PRIVATE|MAP_NORESERVE);
-  if(p1 != MAP_FAILED) {
-    p2 = (char *)(((unsigned long)p1 + (HEAP_MAX_SIZE-1)) & ~(HEAP_MAX_SIZE-1));
-    ul = p2 - p1;
-    if (ul)
-      munmap(p1, ul);
-    munmap(p2 + HEAP_MAX_SIZE, HEAP_MAX_SIZE - ul);
-  } else {
-    /* Try to take the chance that an allocation of only HEAP_MAX_SIZE
-       is already aligned. */
-    p2 = (char *)MMAP(0, HEAP_MAX_SIZE, PROT_NONE, MAP_PRIVATE|MAP_NORESERVE);
-    if(p2 == MAP_FAILED)
-      return 0;
-    if((unsigned long)p2 & (HEAP_MAX_SIZE-1)) {
-      munmap(p2, HEAP_MAX_SIZE);
-      return 0;
-    }
-  }
-  if(MMAP(p2, size, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FIXED)
-     == (char *) MAP_FAILED) {
-    munmap(p2, HEAP_MAX_SIZE);
-    return 0;
-  }
-  h = (heap_info *)p2;
-  h->size = size;
-  THREAD_STAT(stat_n_heaps++);
-  return h;
-}
+  /* properties of fastbins */
 
-/* Grow or shrink a heap.  size is automatically rounded up to a
-   multiple of the page size if it is positive. */
+  /* max_fast is in allowed range */
+  assert((av->max_fast & ~1) <= request2size(MAX_FAST_SIZE));
 
-static int
-#if __STD_C
-grow_heap(heap_info *h, long diff)
-#else
-grow_heap(h, diff) heap_info *h; long diff;
-#endif
-{
-  size_t page_mask = malloc_getpagesize - 1;
-  long new_size;
-
-  if(diff >= 0) {
-    diff = (diff + page_mask) & ~page_mask;
-    new_size = (long)h->size + diff;
-    if(new_size > HEAP_MAX_SIZE)
-      return -1;
-    if(MMAP((char *)h + h->size, diff, PROT_READ|PROT_WRITE,
-	    MAP_PRIVATE|MAP_FIXED) == (char *) MAP_FAILED)
-      return -2;
-  } else {
-    new_size = (long)h->size + diff;
-    if(new_size < (long)sizeof(*h))
-      return -1;
-    /* Try to re-map the extra heap space freshly to save memory, and
-       make it inaccessible. */
-    if((char *)MMAP((char *)h + new_size, -diff, PROT_NONE,
-                    MAP_PRIVATE|MAP_FIXED) == (char *) MAP_FAILED)
-      return -2;
-  }
-  h->size = new_size;
-  return 0;
-}
+  max_fast_bin = fastbin_index(av->max_fast);
 
-/* Delete a heap. */
-
-#define delete_heap(heap) munmap((char*)(heap), HEAP_MAX_SIZE)
-
-/* arena_get() acquires an arena and locks the corresponding mutex.
-   First, try the one last locked successfully by this thread.  (This
-   is the common case and handled with a macro for speed.)  Then, loop
-   once over the circularly linked list of arenas.  If no arena is
-   readily available, create a new one.  In this latter case, `size'
-   is just a hint as to how much memory will be required immediately
-   in the new arena. */
-
-#define arena_get(ptr, size) do { \
-  Void_t *vptr = NULL; \
-  ptr = (arena *)tsd_getspecific(arena_key, vptr); \
-  if(ptr && !mutex_trylock(&ptr->mutex)) { \
-    THREAD_STAT(++(ptr->stat_lock_direct)); \
-  } else \
-    ptr = arena_get2(ptr, (size)); \
-} while(0)
+  for (i = 0; i < NFASTBINS; ++i) {
+    p = av->fastbins[i];
 
-static arena *
-internal_function
-#if __STD_C
-arena_get2(arena *a_tsd, size_t size)
-#else
-arena_get2(a_tsd, size) arena *a_tsd; size_t size;
-#endif
-{
-  arena *a;
-  heap_info *h;
-  char *ptr;
-  int i;
-  unsigned long misalign;
+    /* all bins past max_fast are empty */
+    if (i > max_fast_bin)
+      assert(p == 0);
 
-  if(!a_tsd)
-    a = a_tsd = &main_arena;
-  else {
-    a = a_tsd->next;
-    if(!a) {
-      /* This can only happen while initializing the new arena. */
-      (void)mutex_lock(&main_arena.mutex);
-      THREAD_STAT(++(main_arena.stat_lock_wait));
-      return &main_arena;
+    while (p != 0) {
+      /* each chunk claims to be inuse */
+      do_check_inuse_chunk(av, p);
+      total += chunksize(p);
+      /* chunk belongs in this bin */
+      assert(fastbin_index(chunksize(p)) == i);
+      p = p->fd;
     }
   }
 
-  /* Check the global, circularly linked list for available arenas. */
- repeat:
-  do {
-    if(!mutex_trylock(&a->mutex)) {
-      THREAD_STAT(++(a->stat_lock_loop));
-      tsd_setspecific(arena_key, (Void_t *)a);
-      return a;
+  if (total != 0)
+    assert(have_fastchunks(av));
+  else if (!have_fastchunks(av))
+    assert(total == 0);
+
+  /* check normal bins */
+  for (i = 1; i < NBINS; ++i) {
+    b = bin_at(av,i);
+
+    /* binmap is accurate (except for bin 1 == unsorted_chunks) */
+    if (i >= 2) {
+      binbit = get_binmap(av,i);
+      empty = last(b) == b;
+      if (!binbit)
+        assert(empty);
+      else if (!empty)
+        assert(binbit);
     }
-    a = a->next;
-  } while(a != a_tsd);
-
-  /* If not even the list_lock can be obtained, try again.  This can
-     happen during `atfork', or for example on systems where thread
-     creation makes it temporarily impossible to obtain _any_
-     locks. */
-  if(mutex_trylock(&list_lock)) {
-    a = a_tsd;
-    goto repeat;
-  }
-  (void)mutex_unlock(&list_lock);
-
-  /* Nothing immediately available, so generate a new arena. */
-  h = new_heap(size + (sizeof(*h) + sizeof(*a) + MALLOC_ALIGNMENT));
-  if(!h) {
-    /* Maybe size is too large to fit in a single heap.  So, just try
-       to create a minimally-sized arena and let chunk_alloc() attempt
-       to deal with the large request via mmap_chunk(). */
-    h = new_heap(sizeof(*h) + sizeof(*a) + MALLOC_ALIGNMENT);
-    if(!h)
-      return 0;
-  }
-  a = h->ar_ptr = (arena *)(h+1);
-  for(i=0; i<NAV; i++)
-    init_bin(a, i);
-  a->next = NULL;
-  a->size = h->size;
-  arena_mem += h->size;
-#ifdef NO_THREADS
-  if((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
-    max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
-#endif
-  tsd_setspecific(arena_key, (Void_t *)a);
-  mutex_init(&a->mutex);
-  i = mutex_lock(&a->mutex); /* remember result */
-
-  /* Set up the top chunk, with proper alignment. */
-  ptr = (char *)(a + 1);
-  misalign = (unsigned long)chunk2mem(ptr) & MALLOC_ALIGN_MASK;
-  if (misalign > 0)
-    ptr += MALLOC_ALIGNMENT - misalign;
-  top(a) = (mchunkptr)ptr;
-  set_head(top(a), (((char*)h + h->size) - ptr) | PREV_INUSE);
-
-  /* Add the new arena to the list. */
-  (void)mutex_lock(&list_lock);
-  a->next = main_arena.next;
-  main_arena.next = a;
-  (void)mutex_unlock(&list_lock);
-
-  if(i) /* locking failed; keep arena for further attempts later */
-    return 0;
 
-  THREAD_STAT(++(a->stat_lock_loop));
-  return a;
-}
+    for (p = last(b); p != b; p = p->bk) {
+      /* each chunk claims to be free */
+      do_check_free_chunk(av, p);
+      size = chunksize(p);
+      total += size;
+      if (i >= 2) {
+        /* chunk belongs in bin */
+        idx = bin_index(size);
+        assert(idx == i);
+        /* lists are sorted */
+        assert(p->bk == b || 
+               (unsigned long)chunksize(p->bk) >= (unsigned long)chunksize(p));
+      }
+      /* chunk is followed by a legal chain of inuse chunks */
+      for (q = next_chunk(p);
+           (q != av->top && inuse(q) && 
+             (unsigned long)(chunksize(q)) >= MINSIZE);
+           q = next_chunk(q))
+        do_check_inuse_chunk(av, q);
+    }
+  }
 
-/* find the heap and corresponding arena for a given ptr */
+  /* top chunk is OK */
+  check_chunk(av, av->top);
 
-#define heap_for_ptr(ptr) \
- ((heap_info *)((unsigned long)(ptr) & ~(HEAP_MAX_SIZE-1)))
-#define arena_for_ptr(ptr) \
- (((mchunkptr)(ptr) < top(&main_arena) && (char *)(ptr) >= sbrk_base) ? \
-  &main_arena : heap_for_ptr(ptr)->ar_ptr)
+  /* sanity checks for statistics */
 
-#else /* !USE_ARENAS */
+#ifdef NO_THREADS
+  assert(total <= (unsigned long)(mp_.max_total_mem));
+  assert(mp_.n_mmaps >= 0);
+#endif
+  assert(mp_.n_mmaps <= mp_.n_mmaps_max);
+  assert(mp_.n_mmaps <= mp_.max_n_mmaps);
 
-/* There is only one arena, main_arena. */
+  assert((unsigned long)(av->system_mem) <=
+         (unsigned long)(av->max_system_mem));
 
-#define arena_get(ptr, sz) (ptr = &main_arena)
-#define arena_for_ptr(ptr) (&main_arena)
+  assert((unsigned long)(mp_.mmapped_mem) <=
+         (unsigned long)(mp_.max_mmapped_mem));
 
-#endif /* USE_ARENAS */
+#ifdef NO_THREADS
+  assert((unsigned long)(mp_.max_total_mem) >=
+         (unsigned long)(mp_.mmapped_mem) + (unsigned long)(av->system_mem));
+#endif
+}
+#endif
 
-
 
-/*
-  Debugging support
-*/
+/* ----------------- Support for debugging hooks -------------------- */
+#include "hooks.c"
 
-#if MALLOC_DEBUG
 
+/* ----------- Routines dealing with system allocation -------------- */
 
 /*
-  These routines make a number of assertions about the states
-  of data structures that should be true at all times. If any
-  are not true, it's very likely that a user program has somehow
-  trashed memory. (It's also possible that there is a coding error
-  in malloc. In which case, please report it!)
+  sysmalloc handles malloc cases requiring more memory from the system.
+  On entry, it is assumed that av->top does not have enough
+  space to service request for nb bytes, thus requiring that av->top
+  be extended or replaced.
 */
 
 #if __STD_C
-static void do_check_chunk(arena *ar_ptr, mchunkptr p)
+static Void_t* sYSMALLOc(INTERNAL_SIZE_T nb, mstate av)
 #else
-static void do_check_chunk(ar_ptr, p) arena *ar_ptr; mchunkptr p;
+static Void_t* sYSMALLOc(nb, av) INTERNAL_SIZE_T nb; mstate av;
 #endif
 {
-  INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
+  mchunkptr       old_top;        /* incoming value of av->top */
+  INTERNAL_SIZE_T old_size;       /* its size */
+  char*           old_end;        /* its end address */
 
-  /* No checkable chunk is mmapped */
-  assert(!chunk_is_mmapped(p));
+  long            size;           /* arg to first MORECORE or mmap call */
+  char*           brk;            /* return value from MORECORE */
 
-#if USE_ARENAS
-  if(ar_ptr != &main_arena) {
-    heap_info *heap = heap_for_ptr(p);
-    assert(heap->ar_ptr == ar_ptr);
-    if(p != top(ar_ptr))
-      assert((char *)p + sz <= (char *)heap + heap->size);
-    else
-      assert((char *)p + sz == (char *)heap + heap->size);
-    return;
-  }
-#endif
+  long            correction;     /* arg to 2nd MORECORE call */
+  char*           snd_brk;        /* 2nd return val */
 
-  /* Check for legal address ... */
-  assert((char*)p >= sbrk_base);
-  if (p != top(ar_ptr))
-    assert((char*)p + sz <= (char*)top(ar_ptr));
-  else
-    assert((char*)p + sz <= sbrk_base + sbrked_mem);
+  INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of new space */
+  INTERNAL_SIZE_T end_misalign;   /* partial page left at end of new space */
+  char*           aligned_brk;    /* aligned offset into brk */
 
-}
+  mchunkptr       p;              /* the allocated/returned chunk */
+  mchunkptr       remainder;      /* remainder from allocation */
+  unsigned long   remainder_size; /* its size */
 
+  unsigned long   sum;            /* for updating stats */
 
-#if __STD_C
-static void do_check_free_chunk(arena *ar_ptr, mchunkptr p)
-#else
-static void do_check_free_chunk(ar_ptr, p) arena *ar_ptr; mchunkptr p;
-#endif
-{
-  INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
-  mchunkptr next = chunk_at_offset(p, sz);
+  size_t          pagemask  = mp_.pagesize - 1;
 
-  do_check_chunk(ar_ptr, p);
 
-  /* Check whether it claims to be free ... */
-  assert(!inuse(p));
+#if HAVE_MMAP
 
-  /* Must have OK size and fields */
-  assert((long)sz >= (long)MINSIZE);
-  assert((sz & MALLOC_ALIGN_MASK) == 0);
-  assert(aligned_OK(chunk2mem(p)));
-  /* ... matching footer field */
-  assert(next->prev_size == sz);
-  /* ... and is fully consolidated */
-  assert(prev_inuse(p));
-  assert (next == top(ar_ptr) || inuse(next));
+  /*
+    If have mmap, and the request size meets the mmap threshold, and
+    the system supports mmap, and there are few enough currently
+    allocated mmapped regions, try to directly map this request
+    rather than expanding top.
+  */
 
-  /* ... and has minimally sane links */
-  assert(p->fd->bk == p);
-  assert(p->bk->fd == p);
-}
+  if ((unsigned long)(nb) >= (unsigned long)(mp_.mmap_threshold) &&
+      (mp_.n_mmaps < mp_.n_mmaps_max)) {
 
-#if __STD_C
-static void do_check_inuse_chunk(arena *ar_ptr, mchunkptr p)
-#else
-static void do_check_inuse_chunk(ar_ptr, p) arena *ar_ptr; mchunkptr p;
+    char* mm;             /* return value from mmap call*/
+
+    /*
+      Round up size to nearest page.  For mmapped chunks, the overhead
+      is one SIZE_SZ unit larger than for normal chunks, because there
+      is no following chunk whose prev_size field could be used.
+    */
+    size = (nb + SIZE_SZ + MALLOC_ALIGN_MASK + pagemask) & ~pagemask;
+
+    /* Don't try if size wraps around 0 */
+    if ((unsigned long)(size) > (unsigned long)(nb)) {
+
+      mm = (char*)(MMAP(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE));
+      
+      if (mm != MAP_FAILED) {
+        
+        /*
+          The offset to the start of the mmapped region is stored
+          in the prev_size field of the chunk. This allows us to adjust
+          returned start address to meet alignment requirements here 
+          and in memalign(), and still be able to compute proper
+          address argument for later munmap in free() and realloc().
+        */
+        
+        front_misalign = (INTERNAL_SIZE_T)chunk2mem(mm) & MALLOC_ALIGN_MASK;
+        if (front_misalign > 0) {
+          correction = MALLOC_ALIGNMENT - front_misalign;
+          p = (mchunkptr)(mm + correction);
+          p->prev_size = correction;
+          set_head(p, (size - correction) |IS_MMAPPED);
+        }
+        else {
+          p = (mchunkptr)mm;
+          set_head(p, size|IS_MMAPPED);
+        }
+        
+        /* update statistics */
+        
+        if (++mp_.n_mmaps > mp_.max_n_mmaps) 
+          mp_.max_n_mmaps = mp_.n_mmaps;
+        
+        sum = mp_.mmapped_mem += size;
+        if (sum > (unsigned long)(mp_.max_mmapped_mem)) 
+          mp_.max_mmapped_mem = sum;
+#ifdef NO_THREADS
+        sum += av->system_mem;
+        if (sum > (unsigned long)(mp_.max_total_mem)) 
+          mp_.max_total_mem = sum;
 #endif
-{
-  mchunkptr next = next_chunk(p);
-  do_check_chunk(ar_ptr, p);
 
-  /* Check whether it claims to be in use ... */
-  assert(inuse(p));
+        check_chunk(av, p);
+        
+        return chunk2mem(p);
+      }
+    }
+  }
+#endif
 
-  /* ... whether its size is OK (it might be a fencepost) ... */
-  assert(chunksize(p) >= MINSIZE || next->size == (0|PREV_INUSE));
+  /* Record incoming configuration of top */
 
-  /* ... and is surrounded by OK chunks.
-    Since more things can be checked with free chunks than inuse ones,
-    if an inuse chunk borders them and debug is on, it's worth doing them.
+  old_top  = av->top;
+  old_size = chunksize(old_top);
+  old_end  = (char*)(chunk_at_offset(old_top, old_size));
+
+  brk = snd_brk = (char*)(MORECORE_FAILURE); 
+
+  /* 
+     If not the first time through, we require old_size to be
+     at least MINSIZE and to have prev_inuse set.
   */
-  if (!prev_inuse(p))
-  {
-    mchunkptr prv = prev_chunk(p);
-    assert(next_chunk(prv) == p);
-    do_check_free_chunk(ar_ptr, prv);
-  }
-  if (next == top(ar_ptr))
-  {
-    assert(prev_inuse(next));
-    assert(chunksize(next) >= MINSIZE);
-  }
-  else if (!inuse(next))
-    do_check_free_chunk(ar_ptr, next);
 
-}
+  assert((old_top == initial_top(av) && old_size == 0) || 
+         ((unsigned long) (old_size) >= MINSIZE &&
+          prev_inuse(old_top) &&
+	  ((unsigned long)old_end & pagemask) == 0));
 
-#if __STD_C
-static void do_check_malloced_chunk(arena *ar_ptr,
-                                    mchunkptr p, INTERNAL_SIZE_T s)
-#else
-static void do_check_malloced_chunk(ar_ptr, p, s)
-arena *ar_ptr; mchunkptr p; INTERNAL_SIZE_T s;
-#endif
-{
-  INTERNAL_SIZE_T sz = p->size & ~PREV_INUSE;
-  long room = sz - s;
+  /* Precondition: not enough current space to satisfy nb request */
+  assert((unsigned long)(old_size) < (unsigned long)(nb + MINSIZE));
 
-  do_check_inuse_chunk(ar_ptr, p);
+  /* Precondition: all fastbins are consolidated */
+  assert(!have_fastchunks(av));
 
-  /* Legal size ... */
-  assert((long)sz >= (long)MINSIZE);
-  assert((sz & MALLOC_ALIGN_MASK) == 0);
-  assert(room >= 0);
-  assert(room < (long)MINSIZE);
 
-  /* ... and alignment */
-  assert(aligned_OK(chunk2mem(p)));
+  if (av != &main_arena) {
 
+    heap_info *old_heap, *heap;
+    size_t old_heap_size;
 
-  /* ... and was allocated at front of an available chunk */
-  assert(prev_inuse(p));
+    /* First try to extend the current heap. */
+    old_heap = heap_for_ptr(old_top);
+    old_heap_size = old_heap->size;
+    if (grow_heap(old_heap, MINSIZE + nb - old_size) == 0) {
+      av->system_mem += old_heap->size - old_heap_size;
+      arena_mem += old_heap->size - old_heap_size;
+#if 0
+      if(mmapped_mem + arena_mem + sbrked_mem > max_total_mem)
+        max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
+#endif
+      set_head(old_top, (((char *)old_heap + old_heap->size) - (char *)old_top)
+	       | PREV_INUSE);
+    } else {
+      /* A new heap must be created. */
+      heap = new_heap(nb + (MINSIZE + sizeof(*heap)), mp_.top_pad);
+      if(heap) {
+	heap->ar_ptr = av;
+	heap->prev = old_heap;
+	av->system_mem += heap->size;
+	arena_mem += heap->size;
+#if 0
+	if((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
+	  max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
+#endif
+      }
 
-}
+      /* Set up the new top.  */
+      top(av) = chunk_at_offset(heap, sizeof(*heap));
+      set_head(top(av), (heap->size - sizeof(*heap)) | PREV_INUSE);
+
+      /* Setup fencepost and free the old top chunk. */
+      /* The fencepost takes at least MINSIZE bytes, because it might
+	 become the top chunk again later.  Note that a footer is set
+	 up, too, although the chunk is marked in use. */
+      old_size -= MINSIZE;
+      set_head(chunk_at_offset(old_top, old_size + 2*SIZE_SZ), 0|PREV_INUSE);
+      if (old_size >= MINSIZE) {
+	set_head(chunk_at_offset(old_top, old_size), (2*SIZE_SZ)|PREV_INUSE);
+	set_foot(chunk_at_offset(old_top, old_size), (2*SIZE_SZ));
+	set_head(old_top, old_size|PREV_INUSE|NON_MAIN_ARENA);
+	_int_free(av, chunk2mem(old_top));
+      } else {
+	set_head(old_top, (old_size + 2*SIZE_SZ)|PREV_INUSE);
+	set_foot(old_top, (old_size + 2*SIZE_SZ));
+      }
+    }
 
+  } else { /* av == main_arena */
 
-#define check_free_chunk(A,P) do_check_free_chunk(A,P)
-#define check_inuse_chunk(A,P) do_check_inuse_chunk(A,P)
-#define check_chunk(A,P) do_check_chunk(A,P)
-#define check_malloced_chunk(A,P,N) do_check_malloced_chunk(A,P,N)
-#else
-#define check_free_chunk(A,P)
-#define check_inuse_chunk(A,P)
-#define check_chunk(A,P)
-#define check_malloced_chunk(A,P,N)
-#endif
 
-
+  /* Request enough space for nb + pad + overhead */
 
-/*
-  Macro-based internal utilities
-*/
+  size = nb + mp_.top_pad + MINSIZE;
 
+  /*
+    If contiguous, we can subtract out existing space that we hope to
+    combine with new space. We add it back later only if
+    we don't actually get contiguous space.
+  */
 
-/*
-  Linking chunks in bin lists.
-  Call these only with variables, not arbitrary expressions, as arguments.
-*/
+  if (contiguous(av))
+    size -= old_size;
 
-/*
-  Place chunk p of size s in its bin, in size order,
-  putting it ahead of others of same size.
-*/
+  /*
+    Round to a multiple of page size.
+    If MORECORE is not contiguous, this ensures that we only call it
+    with whole-page arguments.  And if MORECORE is contiguous and
+    this is not first time through, this preserves page-alignment of
+    previous calls. Otherwise, we correct to page-align below.
+  */
 
+  size = (size + pagemask) & ~pagemask;
 
-#define frontlink(A, P, S, IDX, BK, FD)                                       \
-{                                                                             \
-  if (S < MAX_SMALLBIN_SIZE)                                                  \
-  {                                                                           \
-    IDX = smallbin_index(S);                                                  \
-    mark_binblock(A, IDX);                                                    \
-    BK = bin_at(A, IDX);                                                      \
-    FD = BK->fd;                                                              \
-    P->bk = BK;                                                               \
-    P->fd = FD;                                                               \
-    FD->bk = BK->fd = P;                                                      \
-  }                                                                           \
-  else                                                                        \
-  {                                                                           \
-    IDX = bin_index(S);                                                       \
-    BK = bin_at(A, IDX);                                                      \
-    FD = BK->fd;                                                              \
-    if (FD == BK) mark_binblock(A, IDX);                                      \
-    else                                                                      \
-    {                                                                         \
-      while (FD != BK && S < chunksize(FD)) FD = FD->fd;                      \
-      BK = FD->bk;                                                            \
-    }                                                                         \
-    P->bk = BK;                                                               \
-    P->fd = FD;                                                               \
-    FD->bk = BK->fd = P;                                                      \
-  }                                                                           \
-}
+  /*
+    Don't try to call MORECORE if argument is so big as to appear
+    negative. Note that since mmap takes size_t arg, it may succeed
+    below even if we cannot call MORECORE.
+  */
+
+  if (size > 0) 
+    brk = (char*)(MORECORE(size));
 
+  if (brk != (char*)(MORECORE_FAILURE)) {
+    /* Call the `morecore' hook if necessary.  */
+    if (__after_morecore_hook)
+      (*__after_morecore_hook) ();
+  } else {
+  /*
+    If have mmap, try using it as a backup when MORECORE fails or
+    cannot be used. This is worth doing on systems that have "holes" in
+    address space, so sbrk cannot extend to give contiguous space, but
+    space is available elsewhere.  Note that we ignore mmap max count
+    and threshold limits, since the space will not be used as a
+    segregated mmap region.
+  */
 
-/* take a chunk off a list */
+#if HAVE_MMAP
+    /* Cannot merge with old top, so add its size back in */
+    if (contiguous(av))
+      size = (size + old_size + pagemask) & ~pagemask;
+
+    /* If we are relying on mmap as backup, then use larger units */
+    if ((unsigned long)(size) < (unsigned long)(MMAP_AS_MORECORE_SIZE))
+      size = MMAP_AS_MORECORE_SIZE;
+
+    /* Don't try if size wraps around 0 */
+    if ((unsigned long)(size) > (unsigned long)(nb)) {
+
+      brk = (char*)(MMAP(0, size, PROT_READ|PROT_WRITE, MAP_PRIVATE));
+      
+      if (brk != MAP_FAILED) {
+        
+        /* We do not need, and cannot use, another sbrk call to find end */
+        snd_brk = brk + size;
+        
+        /* 
+           Record that we no longer have a contiguous sbrk region. 
+           After the first time mmap is used as backup, we do not
+           ever rely on contiguous space since this could incorrectly
+           bridge regions.
+        */
+        set_noncontiguous(av);
+      }
+    }
+#endif
+  }
 
-#define unlink(P, BK, FD)                                                     \
-{                                                                             \
-  BK = P->bk;                                                                 \
-  FD = P->fd;                                                                 \
-  FD->bk = BK;                                                                \
-  BK->fd = FD;                                                                \
-}                                                                             \
+  if (brk != (char*)(MORECORE_FAILURE)) {
+    if (mp_.sbrk_base == 0)
+      mp_.sbrk_base = brk;
+    av->system_mem += size;
+
+    /*
+      If MORECORE extends previous space, we can likewise extend top size.
+    */
+    
+    if (brk == old_end && snd_brk == (char*)(MORECORE_FAILURE))
+      set_head(old_top, (size + old_size) | PREV_INUSE);
 
-/* Place p as the last remainder */
+    else if (old_size && brk < old_end) {
+      /* Oops!  Someone else killed our space..  Can't touch anything.  */
+      assert(0);
+    }
 
-#define link_last_remainder(A, P)                                             \
-{                                                                             \
-  last_remainder(A)->fd = last_remainder(A)->bk = P;                          \
-  P->fd = P->bk = last_remainder(A);                                          \
-}
+    /*
+      Otherwise, make adjustments:
+      
+      * If the first time through or noncontiguous, we need to call sbrk
+        just to find out where the end of memory lies.
+
+      * We need to ensure that all returned chunks from malloc will meet
+        MALLOC_ALIGNMENT
+
+      * If there was an intervening foreign sbrk, we need to adjust sbrk
+        request size to account for fact that we will not be able to
+        combine new space with existing space in old_top.
+
+      * Almost all systems internally allocate whole pages at a time, in
+        which case we might as well use the whole last page of request.
+        So we allocate enough more memory to hit a page boundary now,
+        which in turn causes future contiguous calls to page-align.
+    */
+    
+    else {
+      /* Count foreign sbrk as system_mem.  */
+      if (old_size)
+	av->system_mem += brk - old_end;
+      front_misalign = 0;
+      end_misalign = 0;
+      correction = 0;
+      aligned_brk = brk;
+      
+      /* handle contiguous cases */
+      if (contiguous(av)) { 
+        
+        /* Guarantee alignment of first new chunk made from this space */
+
+        front_misalign = (INTERNAL_SIZE_T)chunk2mem(brk) & MALLOC_ALIGN_MASK;
+        if (front_misalign > 0) {
+
+          /*
+            Skip over some bytes to arrive at an aligned position.
+            We don't need to specially mark these wasted front bytes.
+            They will never be accessed anyway because
+            prev_inuse of av->top (and any chunk created from its start)
+            is always true after initialization.
+          */
+
+          correction = MALLOC_ALIGNMENT - front_misalign;
+          aligned_brk += correction;
+        }
+        
+        /*
+          If this isn't adjacent to existing space, then we will not
+          be able to merge with old_top space, so must add to 2nd request.
+        */
+        
+        correction += old_size;
+        
+        /* Extend the end address to hit a page boundary */
+        end_misalign = (INTERNAL_SIZE_T)(brk + size + correction);
+        correction += ((end_misalign + pagemask) & ~pagemask) - end_misalign;
+        
+        assert(correction >= 0);
+        snd_brk = (char*)(MORECORE(correction));
+        
+        /*
+          If can't allocate correction, try to at least find out current
+          brk.  It might be enough to proceed without failing.
+          
+          Note that if second sbrk did NOT fail, we assume that space
+          is contiguous with first sbrk. This is a safe assumption unless
+          program is multithreaded but doesn't use locks and a foreign sbrk
+          occurred between our first and second calls.
+        */
+        
+        if (snd_brk == (char*)(MORECORE_FAILURE)) {
+          correction = 0;
+          snd_brk = (char*)(MORECORE(0));
+        } else
+	  /* Call the `morecore' hook if necessary.  */
+	  if (__after_morecore_hook)
+	    (*__after_morecore_hook) ();
+      }
+      
+      /* handle non-contiguous cases */
+      else { 
+        /* MORECORE/mmap must correctly align */
+        assert(((unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK) == 0);
+        
+        /* Find out current end of memory */
+        if (snd_brk == (char*)(MORECORE_FAILURE)) {
+          snd_brk = (char*)(MORECORE(0));
+        }
+      }
+      
+      /* Adjust top based on results of second sbrk */
+      if (snd_brk != (char*)(MORECORE_FAILURE)) {
+        av->top = (mchunkptr)aligned_brk;
+        set_head(av->top, (snd_brk - aligned_brk + correction) | PREV_INUSE);
+        av->system_mem += correction;
+     
+        /*
+          If not the first time through, we either have a
+          gap due to foreign sbrk or a non-contiguous region.  Insert a
+          double fencepost at old_top to prevent consolidation with space
+          we don't own. These fenceposts are artificial chunks that are
+          marked as inuse and are in any case too small to use.  We need
+          two to make sizes and alignments work out.
+        */
+   
+        if (old_size != 0) {
+          /* 
+             Shrink old_top to insert fenceposts, keeping size a
+             multiple of MALLOC_ALIGNMENT. We know there is at least
+             enough space in old_top to do this.
+          */
+          old_size = (old_size - 4*SIZE_SZ) & ~MALLOC_ALIGN_MASK;
+          set_head(old_top, old_size | PREV_INUSE);
+          
+          /*
+            Note that the following assignments completely overwrite
+            old_top when old_size was previously MINSIZE.  This is
+            intentional. We need the fencepost, even if old_top otherwise gets
+            lost.
+          */
+          chunk_at_offset(old_top, old_size            )->size =
+            (2*SIZE_SZ)|PREV_INUSE;
+
+          chunk_at_offset(old_top, old_size + 2*SIZE_SZ)->size =
+            (2*SIZE_SZ)|PREV_INUSE;
+
+          /* If possible, release the rest. */
+          if (old_size >= MINSIZE) {
+            _int_free(av, chunk2mem(old_top));
+          }
 
-/* Clear the last_remainder bin */
+        }
+      }
+    }
+    
+    /* Update statistics */
+#ifdef NO_THREADS
+    sum = av->system_mem + mp_.mmapped_mem;
+    if (sum > (unsigned long)(mp_.max_total_mem))
+      mp_.max_total_mem = sum;
+#endif
 
-#define clear_last_remainder(A) \
-  (last_remainder(A)->fd = last_remainder(A)->bk = last_remainder(A))
+  }
 
+  } /* if (av !=  &main_arena) */
 
+  if ((unsigned long)av->system_mem > (unsigned long)(av->max_system_mem))
+    av->max_system_mem = av->system_mem;
+  check_malloc_state(av);
+    
+  /* finally, do the allocation */
+  p = av->top;
+  size = chunksize(p);
+
+  /* check that one of the above allocation paths succeeded */
+  if ((unsigned long)(size) >= (unsigned long)(nb + MINSIZE)) {
+    remainder_size = size - nb;
+    remainder = chunk_at_offset(p, nb);
+    av->top = remainder;
+    set_head(p, nb | PREV_INUSE | (av != &main_arena ? NON_MAIN_ARENA : 0));
+    set_head(remainder, remainder_size | PREV_INUSE);
+    check_malloced_chunk(av, p, nb);
+    return chunk2mem(p);
+  }
+
+  /* catch all failure paths */
+  MALLOC_FAILURE_ACTION;
+  return 0;
+}
 
-
 
 /*
-  Extend the top-most chunk by obtaining memory from system.
-  Main interface to sbrk (but see also malloc_trim).
+  sYSTRIm is an inverse of sorts to sYSMALLOc.  It gives memory back
+  to the system (via negative arguments to sbrk) if there is unused
+  memory at the `high' end of the malloc pool. It is called
+  automatically by free() when top space exceeds the trim
+  threshold. It is also called by the public malloc_trim routine.  It
+  returns 1 if it actually released any memory, else 0.
 */
 
-#if defined __GNUC__ && __GNUC__ >= 2
-/* This function is called only from one place, inline it.  */
-__inline__
+#if __STD_C
+static int sYSTRIm(size_t pad, mstate av)
+#else
+static int sYSTRIm(pad, av) size_t pad; mstate av;
 #endif
+{
+  long  top_size;        /* Amount of top-most memory */
+  long  extra;           /* Amount to release */
+  long  released;        /* Amount actually released */
+  char* current_brk;     /* address returned by pre-check sbrk call */
+  char* new_brk;         /* address returned by post-check sbrk call */
+  size_t pagesz;
+
+  pagesz = mp_.pagesize;
+  top_size = chunksize(av->top);
+  
+  /* Release in pagesize units, keeping at least one page */
+  extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
+  
+  if (extra > 0) {
+    
+    /*
+      Only proceed if end of memory is where we last set it.
+      This avoids problems if there were foreign sbrk calls.
+    */
+    current_brk = (char*)(MORECORE(0));
+    if (current_brk == (char*)(av->top) + top_size) {
+      
+      /*
+        Attempt to release memory. We ignore MORECORE return value,
+        and instead call again to find out where new end of memory is.
+        This avoids problems if first call releases less than we asked,
+        of if failure somehow altered brk value. (We could still
+        encounter problems if it altered brk in some very bad way,
+        but the only thing we can do is adjust anyway, which will cause
+        some downstream failure.)
+      */
+      
+      MORECORE(-extra);
+      /* Call the `morecore' hook if necessary.  */
+      if (__after_morecore_hook)
+	(*__after_morecore_hook) ();
+      new_brk = (char*)(MORECORE(0));
+      
+      if (new_brk != (char*)MORECORE_FAILURE) {
+        released = (long)(current_brk - new_brk);
+        
+        if (released != 0) {
+          /* Success. Adjust top. */
+          av->system_mem -= released;
+          set_head(av->top, (top_size - released) | PREV_INUSE);
+          check_malloc_state(av);
+          return 1;
+        }
+      }
+    }
+  }
+  return 0;
+}
+
+#ifdef HAVE_MMAP
+
 static void
 internal_function
 #if __STD_C
-malloc_extend_top(arena *ar_ptr, INTERNAL_SIZE_T nb)
+munmap_chunk(mchunkptr p)
 #else
-malloc_extend_top(ar_ptr, nb) arena *ar_ptr; INTERNAL_SIZE_T nb;
+munmap_chunk(p) mchunkptr p;
 #endif
 {
-  unsigned long pagesz   = malloc_getpagesize;
-  mchunkptr old_top      = top(ar_ptr);        /* Record state of old top */
-  INTERNAL_SIZE_T old_top_size = chunksize(old_top);
-  INTERNAL_SIZE_T top_size;                    /* new size of top chunk */
+  INTERNAL_SIZE_T size = chunksize(p);
+  int ret;
 
-#if USE_ARENAS
-  if(ar_ptr == &main_arena) {
+  assert (chunk_is_mmapped(p));
+#if 0
+  assert(! ((char*)p >= mp_.sbrk_base && (char*)p < mp_.sbrk_base + mp_.sbrked_mem));
+  assert((mp_.n_mmaps > 0));
 #endif
+  assert(((p->prev_size + size) & (mp_.pagesize-1)) == 0);
 
-    char*     brk;                  /* return value from sbrk */
-    INTERNAL_SIZE_T front_misalign; /* unusable bytes at front of sbrked space */
-    INTERNAL_SIZE_T correction;     /* bytes for 2nd sbrk call */
-    char*     new_brk;              /* return of 2nd sbrk call */
-    char*     old_end = (char*)(chunk_at_offset(old_top, old_top_size));
+  mp_.n_mmaps--;
+  mp_.mmapped_mem -= (size + p->prev_size);
 
-    /* Pad request with top_pad plus minimal overhead */
-    INTERNAL_SIZE_T sbrk_size = nb + top_pad + MINSIZE;
-
-    /* If not the first time through, round to preserve page boundary */
-    /* Otherwise, we need to correct to a page size below anyway. */
-    /* (We also correct below if an intervening foreign sbrk call.) */
-
-    if (sbrk_base != (char*)(-1))
-      sbrk_size = (sbrk_size + (pagesz - 1)) & ~(pagesz - 1);
+  ret = munmap((char *)p - p->prev_size, size + p->prev_size);
 
-    brk = (char*)(MORECORE (sbrk_size));
+  /* munmap returns non-zero on failure */
+  assert(ret == 0);
+}
 
-    /* Fail if sbrk failed or if a foreign sbrk call killed our space */
-    if (brk == (char*)(MORECORE_FAILURE) ||
-        (brk < old_end && old_top != initial_top(&main_arena)))
-      return;
+#if HAVE_MREMAP
 
-#if defined _LIBC || defined MALLOC_HOOKS
-    /* Call the `morecore' hook if necessary.  */
-    if (__after_morecore_hook)
-      (*__after_morecore_hook) ();
+static mchunkptr
+internal_function
+#if __STD_C
+mremap_chunk(mchunkptr p, size_t new_size)
+#else
+mremap_chunk(p, new_size) mchunkptr p; size_t new_size;
 #endif
+{
+  size_t page_mask = mp_.pagesize - 1;
+  INTERNAL_SIZE_T offset = p->prev_size;
+  INTERNAL_SIZE_T size = chunksize(p);
+  char *cp;
 
-    sbrked_mem += sbrk_size;
-
-    if (brk == old_end) { /* can just add bytes to current top */
-      top_size = sbrk_size + old_top_size;
-      set_head(old_top, top_size | PREV_INUSE);
-      old_top = 0; /* don't free below */
-    } else {
-      if (sbrk_base == (char*)(-1)) /* First time through. Record base */
-        sbrk_base = brk;
-      else
-        /* Someone else called sbrk().  Count those bytes as sbrked_mem. */
-        sbrked_mem += brk - (char*)old_end;
-
-      /* Guarantee alignment of first new chunk made from this space */
-      front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
-      if (front_misalign > 0) {
-        correction = (MALLOC_ALIGNMENT) - front_misalign;
-        brk += correction;
-      } else
-        correction = 0;
-
-      /* Guarantee the next brk will be at a page boundary */
-      correction += pagesz - ((unsigned long)(brk + sbrk_size) & (pagesz - 1));
-
-      /* Allocate correction */
-      new_brk = (char*)(MORECORE (correction));
-      if (new_brk == (char*)(MORECORE_FAILURE)) return;
-
-#if defined _LIBC || defined MALLOC_HOOKS
-      /* Call the `morecore' hook if necessary.  */
-      if (__after_morecore_hook)
-        (*__after_morecore_hook) ();
+  assert (chunk_is_mmapped(p));
+#if 0
+  assert(! ((char*)p >= mp_.sbrk_base && (char*)p < mp_.sbrk_base + mp_.sbrked_mem));
+  assert((mp_.n_mmaps > 0));
 #endif
+  assert(((size + offset) & (mp_.pagesize-1)) == 0);
 
-      sbrked_mem += correction;
-
-      top(&main_arena) = chunk_at_offset(brk, 0);
-      top_size = new_brk - brk + correction;
-      set_head(top(&main_arena), top_size | PREV_INUSE);
+  /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */
+  new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask;
 
-      if (old_top == initial_top(&main_arena))
-        old_top = 0; /* don't free below */
-    }
+  cp = (char *)mremap((char *)p - offset, size + offset, new_size,
+                      MREMAP_MAYMOVE);
 
-    if ((unsigned long)sbrked_mem > (unsigned long)max_sbrked_mem)
-      max_sbrked_mem = sbrked_mem;
-#ifdef NO_THREADS
-    if ((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
-      max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
-#endif
+  if (cp == MAP_FAILED) return 0;
 
-#if USE_ARENAS
-  } else { /* ar_ptr != &main_arena */
-    heap_info *old_heap, *heap;
-    size_t old_heap_size;
+  p = (mchunkptr)(cp + offset);
 
-    if(old_top_size < MINSIZE) /* this should never happen */
-      return;
+  assert(aligned_OK(chunk2mem(p)));
 
-    /* First try to extend the current heap. */
-    if(MINSIZE + nb <= old_top_size)
-      return;
-    old_heap = heap_for_ptr(old_top);
-    old_heap_size = old_heap->size;
-    if(grow_heap(old_heap, MINSIZE + nb - old_top_size) == 0) {
-      ar_ptr->size += old_heap->size - old_heap_size;
-      arena_mem += old_heap->size - old_heap_size;
-#ifdef NO_THREADS
-      if(mmapped_mem + arena_mem + sbrked_mem > max_total_mem)
-        max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
-#endif
-      top_size = ((char *)old_heap + old_heap->size) - (char *)old_top;
-      set_head(old_top, top_size | PREV_INUSE);
-      return;
-    }
+  assert((p->prev_size == offset));
+  set_head(p, (new_size - offset)|IS_MMAPPED);
 
-    /* A new heap must be created. */
-    heap = new_heap(nb + (MINSIZE + sizeof(*heap)));
-    if(!heap)
-      return;
-    heap->ar_ptr = ar_ptr;
-    heap->prev = old_heap;
-    ar_ptr->size += heap->size;
-    arena_mem += heap->size;
+  mp_.mmapped_mem -= size + offset;
+  mp_.mmapped_mem += new_size;
+  if ((unsigned long)mp_.mmapped_mem > (unsigned long)mp_.max_mmapped_mem)
+    mp_.max_mmapped_mem = mp_.mmapped_mem;
 #ifdef NO_THREADS
-    if((unsigned long)(mmapped_mem + arena_mem + sbrked_mem) > max_total_mem)
-      max_total_mem = mmapped_mem + arena_mem + sbrked_mem;
+  if ((unsigned long)(mp_.mmapped_mem + arena_mem + main_arena.system_mem) >
+      mp_.max_total_mem)
+    mp_.max_total_mem = mp_.mmapped_mem + arena_mem + main_arena.system_mem;
 #endif
-
-    /* Set up the new top, so we can safely use chunk_free() below. */
-    top(ar_ptr) = chunk_at_offset(heap, sizeof(*heap));
-    top_size = heap->size - sizeof(*heap);
-    set_head(top(ar_ptr), top_size | PREV_INUSE);
-  }
-#endif /* USE_ARENAS */
-
-  /* We always land on a page boundary */
-  assert(((unsigned long)((char*)top(ar_ptr) + top_size) & (pagesz-1)) == 0);
-
-  /* Setup fencepost and free the old top chunk. */
-  if(old_top) {
-    /* The fencepost takes at least MINSIZE bytes, because it might
-       become the top chunk again later.  Note that a footer is set
-       up, too, although the chunk is marked in use. */
-    old_top_size -= MINSIZE;
-    set_head(chunk_at_offset(old_top, old_top_size + 2*SIZE_SZ), 0|PREV_INUSE);
-    if(old_top_size >= MINSIZE) {
-      set_head(chunk_at_offset(old_top, old_top_size), (2*SIZE_SZ)|PREV_INUSE);
-      set_foot(chunk_at_offset(old_top, old_top_size), (2*SIZE_SZ));
-      set_head_size(old_top, old_top_size);
-      chunk_free(ar_ptr, old_top);
-    } else {
-      set_head(old_top, (old_top_size + 2*SIZE_SZ)|PREV_INUSE);
-      set_foot(old_top, (old_top_size + 2*SIZE_SZ));
-    }
-  }
+  return p;
 }
 
+#endif /* HAVE_MREMAP */
 
-
-
-/* Main public routines */
-
-
-/*
-  Malloc Algorithm:
-
-    The requested size is first converted into a usable form, `nb'.
-    This currently means to add 4 bytes overhead plus possibly more to
-    obtain 8-byte alignment and/or to obtain a size of at least
-    MINSIZE (currently 16, 24, or 32 bytes), the smallest allocatable
-    size.  (All fits are considered `exact' if they are within MINSIZE
-    bytes.)
-
-    From there, the first successful of the following steps is taken:
-
-      1. The bin corresponding to the request size is scanned, and if
-         a chunk of exactly the right size is found, it is taken.
-
-      2. The most recently remaindered chunk is used if it is big
-         enough.  This is a form of (roving) first fit, used only in
-         the absence of exact fits. Runs of consecutive requests use
-         the remainder of the chunk used for the previous such request
-         whenever possible. This limited use of a first-fit style
-         allocation strategy tends to give contiguous chunks
-         coextensive lifetimes, which improves locality and can reduce
-         fragmentation in the long run.
-
-      3. Other bins are scanned in increasing size order, using a
-         chunk big enough to fulfill the request, and splitting off
-         any remainder.  This search is strictly by best-fit; i.e.,
-         the smallest (with ties going to approximately the least
-         recently used) chunk that fits is selected.
-
-      4. If large enough, the chunk bordering the end of memory
-         (`top') is split off. (This use of `top' is in accord with
-         the best-fit search rule.  In effect, `top' is treated as
-         larger (and thus less well fitting) than any other available
-         chunk since it can be extended to be as large as necessary
-         (up to system limitations).
-
-      5. If the request size meets the mmap threshold and the
-         system supports mmap, and there are few enough currently
-         allocated mmapped regions, and a call to mmap succeeds,
-         the request is allocated via direct memory mapping.
-
-      6. Otherwise, the top of memory is extended by
-         obtaining more space from the system (normally using sbrk,
-         but definable to anything else via the MORECORE macro).
-         Memory is gathered from the system (in system page-sized
-         units) in a way that allows chunks obtained across different
-         sbrk calls to be consolidated, but does not require
-         contiguous memory. Thus, it should be safe to intersperse
-         mallocs with other sbrk calls.
-
-
-      All allocations are made from the `lowest' part of any found
-      chunk. (The implementation invariant is that prev_inuse is
-      always true of any allocated chunk; i.e., that each allocated
-      chunk borders either a previously allocated and still in-use chunk,
-      or the base of its memory arena.)
+#endif /* HAVE_MMAP */
 
-*/
+/*------------------------ Public wrappers. --------------------------------*/
 
-#if __STD_C
-Void_t* mALLOc(size_t bytes)
-#else
-Void_t* mALLOc(bytes) size_t bytes;
-#endif
+Void_t*
+public_mALLOc(size_t bytes)
 {
-  arena *ar_ptr;
-  INTERNAL_SIZE_T nb; /* padded request size */
-  mchunkptr victim;
-
-#if defined _LIBC || defined MALLOC_HOOKS
-  __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, __const __malloc_ptr_t)) =
-      __malloc_hook;
-  if (hook != NULL) {
-    Void_t* result;
+  mstate ar_ptr;
+  Void_t *victim;
 
-#if defined __GNUC__ && __GNUC__ >= 2
-    result = (*hook)(bytes, RETURN_ADDRESS (0));
-#else
-    result = (*hook)(bytes, NULL);
-#endif
-    return result;
-  }
-#endif
+  __malloc_ptr_t (*hook) __MALLOC_P ((size_t, __const __malloc_ptr_t)) =
+    __malloc_hook;
+  if (hook != NULL)
+    return (*hook)(bytes, RETURN_ADDRESS (0));
 
-  if(request2size(bytes, nb))
-    return 0;
-  arena_get(ar_ptr, nb);
+  arena_get(ar_ptr, bytes);
   if(!ar_ptr)
     return 0;
-  victim = chunk_alloc(ar_ptr, nb);
+  victim = _int_malloc(ar_ptr, bytes);
   if(!victim) {
     /* Maybe the failure is due to running out of mmapped areas. */
     if(ar_ptr != &main_arena) {
       (void)mutex_unlock(&ar_ptr->mutex);
       (void)mutex_lock(&main_arena.mutex);
-      victim = chunk_alloc(&main_arena, nb);
+      victim = _int_malloc(&main_arena, bytes);
       (void)mutex_unlock(&main_arena.mutex);
     } else {
 #if USE_ARENAS
       /* ... or sbrk() has failed and there is still a chance to mmap() */
-      ar_ptr = arena_get2(ar_ptr->next ? ar_ptr : 0, nb);
+      ar_ptr = arena_get2(ar_ptr->next ? ar_ptr : 0, bytes);
       (void)mutex_unlock(&main_arena.mutex);
       if(ar_ptr) {
-        victim = chunk_alloc(ar_ptr, nb);
+        victim = _int_malloc(ar_ptr, bytes);
         (void)mutex_unlock(&ar_ptr->mutex);
       }
 #endif
     }
-    if(!victim) return 0;
   } else
     (void)mutex_unlock(&ar_ptr->mutex);
-  return BOUNDED_N(chunk2mem(victim), bytes);
+  assert(!victim || chunk_is_mmapped(mem2chunk(victim)) ||
+	 ar_ptr == arena_for_chunk(mem2chunk(victim)));
+  return victim;
 }
 
-static mchunkptr
-internal_function
-#if __STD_C
-chunk_alloc(arena *ar_ptr, INTERNAL_SIZE_T nb)
-#else
-chunk_alloc(ar_ptr, nb) arena *ar_ptr; INTERNAL_SIZE_T nb;
-#endif
+void
+public_fREe(Void_t* mem)
 {
-  mchunkptr victim;                  /* inspected/selected chunk */
-  INTERNAL_SIZE_T victim_size;       /* its size */
-  int       idx;                     /* index for bin traversal */
-  mbinptr   bin;                     /* associated bin */
-  mchunkptr remainder;               /* remainder from a split */
-  long      remainder_size;          /* its size */
-  int       remainder_index;         /* its bin index */
-  unsigned long block;               /* block traverser bit */
-  int       startidx;                /* first bin of a traversed block */
-  mchunkptr fwd;                     /* misc temp for linking */
-  mchunkptr bck;                     /* misc temp for linking */
-  mbinptr q;                         /* misc temp */
-
-
-  /* Check for exact match in a bin */
-
-  if (is_small_request(nb))  /* Faster version for small requests */
-  {
-    idx = smallbin_index(nb);
+  mstate ar_ptr;
+  mchunkptr p;                          /* chunk corresponding to mem */
 
-    /* No traversal or size check necessary for small bins.  */
+  void (*hook) __MALLOC_P ((__malloc_ptr_t, __const __malloc_ptr_t)) =
+    __free_hook;
+  if (hook != NULL) {
+    (*hook)(mem, RETURN_ADDRESS (0));
+    return;
+  }
 
-    q = _bin_at(ar_ptr, idx);
-    victim = last(q);
+  if (mem == 0)                              /* free(0) has no effect */
+    return;
 
-    /* Also scan the next one, since it would have a remainder < MINSIZE */
-    if (victim == q)
-    {
-      q = next_bin(q);
-      victim = last(q);
-    }
-    if (victim != q)
-    {
-      victim_size = chunksize(victim);
-      unlink(victim, bck, fwd);
-      set_inuse_bit_at_offset(victim, victim_size);
-      check_malloced_chunk(ar_ptr, victim, nb);
-      return victim;
-    }
+  p = mem2chunk(mem);
 
-    idx += 2; /* Set for bin scan below. We've already scanned 2 bins. */
+#if HAVE_MMAP
+  if (chunk_is_mmapped(p))                       /* release mmapped memory. */
+  {
+    munmap_chunk(p);
+    return;
+  }
+#endif
 
+  ar_ptr = arena_for_chunk(p);
+#if THREAD_STATS
+  if(!mutex_trylock(&ar_ptr->mutex))
+    ++(ar_ptr->stat_lock_direct);
+  else {
+    (void)mutex_lock(&ar_ptr->mutex);
+    ++(ar_ptr->stat_lock_wait);
   }
-  else
-  {
-    idx = bin_index(nb);
-    bin = bin_at(ar_ptr, idx);
+#else
+  (void)mutex_lock(&ar_ptr->mutex);
+#endif
+  _int_free(ar_ptr, mem);
+  (void)mutex_unlock(&ar_ptr->mutex);
+}
 
-    for (victim = last(bin); victim != bin; victim = victim->bk)
-    {
-      victim_size = chunksize(victim);
-      remainder_size = victim_size - nb;
+Void_t*
+public_rEALLOc(Void_t* oldmem, size_t bytes)
+{
+  mstate ar_ptr;
+  INTERNAL_SIZE_T    nb;      /* padded request size */
 
-      if (remainder_size >= (long)MINSIZE) /* too big */
-      {
-        --idx; /* adjust to rescan below after checking last remainder */
-        break;
-      }
+  mchunkptr oldp;             /* chunk corresponding to oldmem */
+  INTERNAL_SIZE_T    oldsize; /* its size */
 
-      else if (remainder_size >= 0) /* exact fit */
-      {
-        unlink(victim, bck, fwd);
-        set_inuse_bit_at_offset(victim, victim_size);
-        check_malloced_chunk(ar_ptr, victim, nb);
-        return victim;
-      }
-    }
+  Void_t* newp;             /* chunk to return */
 
-    ++idx;
+  __malloc_ptr_t (*hook) __MALLOC_P ((__malloc_ptr_t, size_t,
+				      __const __malloc_ptr_t)) =
+    __realloc_hook;
+  if (hook != NULL)
+    return (*hook)(oldmem, bytes, RETURN_ADDRESS (0));
 
-  }
+#if REALLOC_ZERO_BYTES_FREES
+  if (bytes == 0 && oldmem != NULL) { public_fREe(oldmem); return 0; }
+#endif
 
-  /* Try to use the last split-off remainder */
+  /* realloc of null is supposed to be same as malloc */
+  if (oldmem == 0) return public_mALLOc(bytes);
+
+  oldp    = mem2chunk(oldmem);
+  oldsize = chunksize(oldp);
+
+  checked_request2size(bytes, nb);
 
-  if ( (victim = last_remainder(ar_ptr)->fd) != last_remainder(ar_ptr))
+#if HAVE_MMAP
+  if (chunk_is_mmapped(oldp))
   {
-    victim_size = chunksize(victim);
-    remainder_size = victim_size - nb;
+    Void_t* newmem;
 
-    if (remainder_size >= (long)MINSIZE) /* re-split */
-    {
-      remainder = chunk_at_offset(victim, nb);
-      set_head(victim, nb | PREV_INUSE);
-      link_last_remainder(ar_ptr, remainder);
-      set_head(remainder, remainder_size | PREV_INUSE);
-      set_foot(remainder, remainder_size);
-      check_malloced_chunk(ar_ptr, victim, nb);
-      return victim;
-    }
+#if HAVE_MREMAP
+    newp = mremap_chunk(oldp, nb);
+    if(newp) return chunk2mem(newp);
+#endif
+    /* Note the extra SIZE_SZ overhead. */
+    if(oldsize - SIZE_SZ >= nb) return oldmem; /* do nothing */
+    /* Must alloc, copy, free. */
+    newmem = public_mALLOc(bytes);
+    if (newmem == 0) return 0; /* propagate failure */
+    MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ);
+    munmap_chunk(oldp);
+    return newmem;
+  }
+#endif
 
-    clear_last_remainder(ar_ptr);
+  ar_ptr = arena_for_chunk(oldp);
+#if THREAD_STATS
+  if(!mutex_trylock(&ar_ptr->mutex))
+    ++(ar_ptr->stat_lock_direct);
+  else {
+    (void)mutex_lock(&ar_ptr->mutex);
+    ++(ar_ptr->stat_lock_wait);
+  }
+#else
+  (void)mutex_lock(&ar_ptr->mutex);
+#endif
 
-    if (remainder_size >= 0)  /* exhaust */
-    {
-      set_inuse_bit_at_offset(victim, victim_size);
-      check_malloced_chunk(ar_ptr, victim, nb);
-      return victim;
-    }
+#ifndef NO_THREADS
+  /* As in malloc(), remember this arena for the next allocation. */
+  tsd_setspecific(arena_key, (Void_t *)ar_ptr);
+#endif
 
-    /* Else place in bin */
+  newp = _int_realloc(ar_ptr, oldmem, bytes);
 
-    frontlink(ar_ptr, victim, victim_size, remainder_index, bck, fwd);
-  }
+  (void)mutex_unlock(&ar_ptr->mutex);
+  assert(!newp || chunk_is_mmapped(mem2chunk(newp)) ||
+	 ar_ptr == arena_for_chunk(mem2chunk(newp)));
+  return newp;
+}
 
-  /*
-     If there are any possibly nonempty big-enough blocks,
-     search for best fitting chunk by scanning bins in blockwidth units.
-  */
+Void_t*
+public_mEMALIGn(size_t alignment, size_t bytes)
+{
+  mstate ar_ptr;
+  Void_t *p;
 
-  if ( (block = idx2binblock(idx)) <= binblocks(ar_ptr))
-  {
+  __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, size_t,
+					__const __malloc_ptr_t)) =
+    __memalign_hook;
+  if (hook != NULL)
+    return (*hook)(alignment, bytes, RETURN_ADDRESS (0));
+
+  /* If need less alignment than we give anyway, just relay to malloc */
+  if (alignment <= MALLOC_ALIGNMENT) return public_mALLOc(bytes);
 
-    /* Get to the first marked block */
+  /* Otherwise, ensure that it is at least a minimum chunk size */
+  if (alignment <  MINSIZE) alignment = MINSIZE;
 
-    if ( (block & binblocks(ar_ptr)) == 0)
-    {
-      /* force to an even block boundary */
-      idx = (idx & ~(BINBLOCKWIDTH - 1)) + BINBLOCKWIDTH;
-      block <<= 1;
-      while ((block & binblocks(ar_ptr)) == 0)
-      {
-        idx += BINBLOCKWIDTH;
-        block <<= 1;
+  arena_get(ar_ptr, bytes + alignment + MINSIZE);
+  if(!ar_ptr)
+    return 0;
+  p = _int_memalign(ar_ptr, alignment, bytes);
+  (void)mutex_unlock(&ar_ptr->mutex);
+  if(!p) {
+    /* Maybe the failure is due to running out of mmapped areas. */
+    if(ar_ptr != &main_arena) {
+      (void)mutex_lock(&main_arena.mutex);
+      p = _int_memalign(&main_arena, alignment, bytes);
+      (void)mutex_unlock(&main_arena.mutex);
+    } else {
+#if USE_ARENAS
+      /* ... or sbrk() has failed and there is still a chance to mmap() */
+      ar_ptr = arena_get2(ar_ptr->next ? ar_ptr : 0, bytes);
+      if(ar_ptr) {
+        p = _int_memalign(ar_ptr, alignment, bytes);
+        (void)mutex_unlock(&ar_ptr->mutex);
       }
+#endif
     }
+  }
+  assert(!p || chunk_is_mmapped(mem2chunk(p)) ||
+	 ar_ptr == arena_for_chunk(mem2chunk(p)));
+  return p;
+}
 
-    /* For each possibly nonempty block ... */
-    for (;;)
-    {
-      startidx = idx;          /* (track incomplete blocks) */
-      q = bin = _bin_at(ar_ptr, idx);
+Void_t*
+public_vALLOc(size_t bytes)
+{
+  mstate ar_ptr;
+  Void_t *p;
 
-      /* For each bin in this block ... */
-      do
-      {
-        /* Find and use first big enough chunk ... */
-
-        for (victim = last(bin); victim != bin; victim = victim->bk)
-        {
-          victim_size = chunksize(victim);
-          remainder_size = victim_size - nb;
-
-          if (remainder_size >= (long)MINSIZE) /* split */
-          {
-            remainder = chunk_at_offset(victim, nb);
-            set_head(victim, nb | PREV_INUSE);
-            unlink(victim, bck, fwd);
-            link_last_remainder(ar_ptr, remainder);
-            set_head(remainder, remainder_size | PREV_INUSE);
-            set_foot(remainder, remainder_size);
-            check_malloced_chunk(ar_ptr, victim, nb);
-            return victim;
-          }
+  if(__malloc_initialized < 0)
+    ptmalloc_init ();
+  arena_get(ar_ptr, bytes + mp_.pagesize + MINSIZE);
+  if(!ar_ptr)
+    return 0;
+  p = _int_valloc(ar_ptr, bytes);
+  (void)mutex_unlock(&ar_ptr->mutex);
+  return p;
+}
 
-          else if (remainder_size >= 0)  /* take */
-          {
-            set_inuse_bit_at_offset(victim, victim_size);
-            unlink(victim, bck, fwd);
-            check_malloced_chunk(ar_ptr, victim, nb);
-            return victim;
-          }
+Void_t*
+public_pVALLOc(size_t bytes)
+{
+  mstate ar_ptr;
+  Void_t *p;
 
-        }
+  if(__malloc_initialized < 0)
+    ptmalloc_init ();
+  arena_get(ar_ptr, bytes + 2*mp_.pagesize + MINSIZE);
+  p = _int_pvalloc(ar_ptr, bytes);
+  (void)mutex_unlock(&ar_ptr->mutex);
+  return p;
+}
+
+Void_t*
+public_cALLOc(size_t n, size_t elem_size)
+{
+  mstate av;
+  mchunkptr oldtop, p;
+  INTERNAL_SIZE_T sz, csz, oldtopsize;
+  Void_t* mem;
+  unsigned long clearsize;
+  unsigned long nclears;
+  INTERNAL_SIZE_T* d;
+
+  __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, __const __malloc_ptr_t)) =
+    __malloc_hook;
+  if (hook != NULL) {
+    sz = n * elem_size;
+    mem = (*hook)(sz, RETURN_ADDRESS (0));
+    if(mem == 0)
+      return 0;
+#ifdef HAVE_MEMCPY
+    return memset(mem, 0, sz);
+#else
+    while(sz > 0) ((char*)mem)[--sz] = 0; /* rather inefficient */
+    return mem;
+#endif
+  }
 
-       bin = next_bin(bin);
+  /* FIXME: check for overflow on multiplication.  */
+  sz = n * elem_size;
 
-      } while ((++idx & (BINBLOCKWIDTH - 1)) != 0);
+  arena_get(av, sz);
+  if(!av)
+    return 0;
 
-      /* Clear out the block bit. */
+  /* Check if we hand out the top chunk, in which case there may be no
+     need to clear. */
+#if MORECORE_CLEARS
+  oldtop = top(av);
+  oldtopsize = chunksize(top(av));
+#if MORECORE_CLEARS < 2
+  /* Only newly allocated memory is guaranteed to be cleared.  */
+  if (av == &main_arena &&
+      oldtopsize < mp_.sbrk_base + av->max_system_mem - (char *)oldtop)
+    oldtopsize = (mp_.sbrk_base + av->max_system_mem - (char *)oldtop);
+#endif
+#endif
+  mem = _int_malloc(av, sz);
 
-      do   /* Possibly backtrack to try to clear a partial block */
-      {
-        if ((startidx & (BINBLOCKWIDTH - 1)) == 0)
-        {
-          binblocks(ar_ptr) &= ~block;
-          break;
-        }
-        --startidx;
-        q = prev_bin(q);
-      } while (first(q) == q);
+  /* Only clearing follows, so we can unlock early. */
+  (void)mutex_unlock(&av->mutex);
 
-      /* Get to the next possibly nonempty block */
+  assert(!mem || chunk_is_mmapped(mem2chunk(mem)) ||
+	 av == arena_for_chunk(mem2chunk(mem)));
 
-      if ( (block <<= 1) <= binblocks(ar_ptr) && (block != 0) )
-      {
-        while ((block & binblocks(ar_ptr)) == 0)
-        {
-          idx += BINBLOCKWIDTH;
-          block <<= 1;
-        }
+  if (mem == 0) {
+    /* Maybe the failure is due to running out of mmapped areas. */
+    if(av != &main_arena) {
+      (void)mutex_lock(&main_arena.mutex);
+      mem = _int_malloc(&main_arena, sz);
+      (void)mutex_unlock(&main_arena.mutex);
+    } else {
+#if USE_ARENAS
+      /* ... or sbrk() has failed and there is still a chance to mmap() */
+      (void)mutex_lock(&main_arena.mutex);
+      av = arena_get2(av->next ? av : 0, sz);
+      (void)mutex_unlock(&main_arena.mutex);
+      if(av) {
+        mem = _int_malloc(av, sz);
+        (void)mutex_unlock(&av->mutex);
       }
-      else
-        break;
+#endif
     }
+    if (mem == 0) return 0;
   }
+  p = mem2chunk(mem);
 
+  /* Two optional cases in which clearing not necessary */
+#if HAVE_MMAP
+  if (chunk_is_mmapped(p))
+    return mem;
+#endif
 
-  /* Try to use top chunk */
-
-  /* Require that there be a remainder, ensuring top always exists  */
-  if ( (remainder_size = chunksize(top(ar_ptr)) - nb) < (long)MINSIZE)
-  {
+  csz = chunksize(p);
 
-#if HAVE_MMAP
-    /* If the request is big and there are not yet too many regions,
-       and we would otherwise need to extend, try to use mmap instead.  */
-    if ((unsigned long)nb >= (unsigned long)mmap_threshold &&
-        n_mmaps < n_mmaps_max &&
-        (victim = mmap_chunk(nb)) != 0)
-      return victim;
+#if MORECORE_CLEARS
+  if (p == oldtop && csz > oldtopsize) {
+    /* clear only the bytes from non-freshly-sbrked memory */
+    csz = oldtopsize;
+  }
 #endif
 
-    /* Try to extend */
-    malloc_extend_top(ar_ptr, nb);
-    if ((remainder_size = chunksize(top(ar_ptr)) - nb) < (long)MINSIZE)
-    {
-#if HAVE_MMAP
-      /* A last attempt: when we are out of address space in a
-         non-main arena, try mmap anyway, as long as it is allowed at
-         all.  */
-      if (ar_ptr != &main_arena &&
-          n_mmaps_max > 0 &&
-          (victim = mmap_chunk(nb)) != 0)
-        return victim;
-#endif
-      return 0; /* propagate failure */
+  /* Unroll clear of <= 36 bytes (72 if 8byte sizes).  We know that
+     contents have an odd number of INTERNAL_SIZE_T-sized words;
+     minimally 3.  */
+  d = (INTERNAL_SIZE_T*)mem;
+  clearsize = csz - SIZE_SZ;
+  nclears = clearsize / sizeof(INTERNAL_SIZE_T);
+  assert(nclears >= 3);
+
+  if (nclears > 9)
+    MALLOC_ZERO(d, clearsize);
+
+  else {
+    *(d+0) = 0;
+    *(d+1) = 0;
+    *(d+2) = 0;
+    if (nclears > 4) {
+      *(d+3) = 0;
+      *(d+4) = 0;
+      if (nclears > 6) {
+	*(d+5) = 0;
+	*(d+6) = 0;
+	if (nclears > 8) {
+	  *(d+7) = 0;
+	  *(d+8) = 0;
+	}
+      }
     }
   }
 
-  victim = top(ar_ptr);
-  set_head(victim, nb | PREV_INUSE);
-  top(ar_ptr) = chunk_at_offset(victim, nb);
-  set_head(top(ar_ptr), remainder_size | PREV_INUSE);
-  check_malloced_chunk(ar_ptr, victim, nb);
-  return victim;
-
+  return mem;
 }
 
+Void_t**
+public_iCALLOc(size_t n, size_t elem_size, Void_t** chunks)
+{
+  mstate ar_ptr;
+  Void_t** m;
 
-
+  arena_get(ar_ptr, n*elem_size);
+  if(!ar_ptr)
+    return 0;
 
-/*
+  m = _int_icalloc(ar_ptr, n, elem_size, chunks);
+  (void)mutex_unlock(&ar_ptr->mutex);
+  return m;
+}
 
-  free() algorithm :
+Void_t**
+public_iCOMALLOc(size_t n, size_t sizes[], Void_t** chunks)
+{
+  mstate ar_ptr;
+  Void_t** m;
 
-    cases:
+  arena_get(ar_ptr, 0);
+  if(!ar_ptr)
+    return 0;
 
-       1. free(0) has no effect.
+  m = _int_icomalloc(ar_ptr, n, sizes, chunks);
+  (void)mutex_unlock(&ar_ptr->mutex);
+  return m;
+}
 
-       2. If the chunk was allocated via mmap, it is released via munmap().
+#ifndef _LIBC
 
-       3. If a returned chunk borders the current high end of memory,
-          it is consolidated into the top, and if the total unused
-          topmost memory exceeds the trim threshold, malloc_trim is
-          called.
+void
+public_cFREe(Void_t* m)
+{
+  public_fREe(m);
+}
 
-       4. Other chunks are consolidated as they arrive, and
-          placed in corresponding bins. (This includes the case of
-          consolidating with the current `last_remainder').
+#endif /* _LIBC */
 
-*/
+int
+public_mTRIm(size_t s)
+{
+  int result;
 
+  (void)mutex_lock(&main_arena.mutex);
+  result = mTRIm(s);
+  (void)mutex_unlock(&main_arena.mutex);
+  return result;
+}
 
-#if __STD_C
-void fREe(Void_t* mem)
-#else
-void fREe(mem) Void_t* mem;
-#endif
+size_t
+public_mUSABLe(Void_t* m)
 {
-  arena *ar_ptr;
-  mchunkptr p;                          /* chunk corresponding to mem */
-
-#if defined _LIBC || defined MALLOC_HOOKS
-  void (*hook) __MALLOC_PMT ((__malloc_ptr_t, __const __malloc_ptr_t)) =
-    __free_hook;
+  size_t result;
 
-  if (hook != NULL) {
-#if defined __GNUC__ && __GNUC__ >= 2
-    (*hook)(mem, RETURN_ADDRESS (0));
-#else
-    (*hook)(mem, NULL);
-#endif
-    return;
-  }
-#endif
+  result = mUSABLe(m);
+  return result;
+}
 
-  if (mem == 0)                              /* free(0) has no effect */
-    return;
+void
+public_mSTATs()
+{
+  mSTATs();
+}
 
-  p = mem2chunk(mem);
+struct mallinfo public_mALLINFo()
+{
+  struct mallinfo m;
 
-#if HAVE_MMAP
-  if (chunk_is_mmapped(p))                       /* release mmapped memory. */
-  {
-    munmap_chunk(p);
-    return;
-  }
-#endif
+  (void)mutex_lock(&main_arena.mutex);
+  m = mALLINFo(&main_arena);
+  (void)mutex_unlock(&main_arena.mutex);
+  return m;
+}
 
-  ar_ptr = arena_for_ptr(p);
-#if THREAD_STATS
-  if(!mutex_trylock(&ar_ptr->mutex))
-    ++(ar_ptr->stat_lock_direct);
-  else {
-    (void)mutex_lock(&ar_ptr->mutex);
-    ++(ar_ptr->stat_lock_wait);
-  }
-#else
-  (void)mutex_lock(&ar_ptr->mutex);
-#endif
-  chunk_free(ar_ptr, p);
-  (void)mutex_unlock(&ar_ptr->mutex);
+int
+public_mALLOPt(int p, int v)
+{
+  int result;
+  result = mALLOPt(p, v);
+  return result;
 }
 
-static void
-internal_function
-#if __STD_C
-chunk_free(arena *ar_ptr, mchunkptr p)
-#else
-chunk_free(ar_ptr, p) arena *ar_ptr; mchunkptr p;
-#endif
+/*
+  ------------------------------ malloc ------------------------------
+*/
+
+Void_t*
+_int_malloc(mstate av, size_t bytes)
 {
-  INTERNAL_SIZE_T hd = p->size; /* its head field */
-  INTERNAL_SIZE_T sz;  /* its size */
-  int       idx;       /* its bin index */
-  mchunkptr next;      /* next contiguous chunk */
-  INTERNAL_SIZE_T nextsz; /* its size */
-  INTERNAL_SIZE_T prevsz; /* size of previous contiguous chunk */
-  mchunkptr bck;       /* misc temp for linking */
-  mchunkptr fwd;       /* misc temp for linking */
-  int       islr;      /* track whether merging with last_remainder */
-
-  check_inuse_chunk(ar_ptr, p);
-
-  sz = hd & ~PREV_INUSE;
-  next = chunk_at_offset(p, sz);
-  nextsz = chunksize(next);
-
-  if (next == top(ar_ptr))                         /* merge with top */
-  {
-    sz += nextsz;
+  INTERNAL_SIZE_T nb;               /* normalized request size */
+  unsigned int    idx;              /* associated bin index */
+  mbinptr         bin;              /* associated bin */
+  mfastbinptr*    fb;               /* associated fastbin */
 
-    if (!(hd & PREV_INUSE))                    /* consolidate backward */
-    {
-      prevsz = p->prev_size;
-      p = chunk_at_offset(p, -(long)prevsz);
-      sz += prevsz;
-      unlink(p, bck, fwd);
-    }
+  mchunkptr       victim;           /* inspected/selected chunk */
+  INTERNAL_SIZE_T size;             /* its size */
+  int             victim_index;     /* its bin index */
 
-    set_head(p, sz | PREV_INUSE);
-    top(ar_ptr) = p;
+  mchunkptr       remainder;        /* remainder from a split */
+  unsigned long   remainder_size;   /* its size */
 
-#if USE_ARENAS
-    if(ar_ptr == &main_arena) {
-#endif
-      if ((unsigned long)(sz) >= (unsigned long)trim_threshold)
-        main_trim(top_pad);
-#if USE_ARENAS
-    } else {
-      heap_info *heap = heap_for_ptr(p);
+  unsigned int    block;            /* bit map traverser */
+  unsigned int    bit;              /* bit map traverser */
+  unsigned int    map;              /* current word of binmap */
 
-      assert(heap->ar_ptr == ar_ptr);
+  mchunkptr       fwd;              /* misc temp for linking */
+  mchunkptr       bck;              /* misc temp for linking */
 
-      /* Try to get rid of completely empty heaps, if possible. */
-      if((unsigned long)(sz) >= (unsigned long)trim_threshold ||
-         p == chunk_at_offset(heap, sizeof(*heap)))
-        heap_trim(heap, top_pad);
-    }
-#endif
-    return;
-  }
+  /*
+    Convert request size to internal form by adding SIZE_SZ bytes
+    overhead plus possibly more to obtain necessary alignment and/or
+    to obtain a size of at least MINSIZE, the smallest allocatable
+    size. Also, checked_request2size traps (returning 0) request sizes
+    that are so large that they wrap around zero when padded and
+    aligned.
+  */
 
-  islr = 0;
+  checked_request2size(bytes, nb);
 
-  if (!(hd & PREV_INUSE))                    /* consolidate backward */
-  {
-    prevsz = p->prev_size;
-    p = chunk_at_offset(p, -(long)prevsz);
-    sz += prevsz;
+  /*
+    If the size qualifies as a fastbin, first check corresponding bin.
+    This code is safe to execute even if av is not yet initialized, so we
+    can try it without checking, which saves some time on this fast path.
+  */
 
-    if (p->fd == last_remainder(ar_ptr))     /* keep as last_remainder */
-      islr = 1;
-    else
-      unlink(p, bck, fwd);
+  if ((unsigned long)(nb) <= (unsigned long)(av->max_fast)) { 
+    fb = &(av->fastbins[(fastbin_index(nb))]);
+    if ( (victim = *fb) != 0) {
+      *fb = victim->fd;
+      check_remalloced_chunk(av, victim, nb);
+      return chunk2mem(victim);
+    }
   }
 
-  if (!(inuse_bit_at_offset(next, nextsz)))   /* consolidate forward */
-  {
-    sz += nextsz;
+  /*
+    If a small request, check regular bin.  Since these "smallbins"
+    hold one size each, no searching within bins is necessary.
+    (For a large request, we need to wait until unsorted chunks are
+    processed to find best fit. But for small ones, fits are exact
+    anyway, so we can check now, which is faster.)
+  */
 
-    if (!islr && next->fd == last_remainder(ar_ptr))
-                                              /* re-insert last_remainder */
-    {
-      islr = 1;
-      link_last_remainder(ar_ptr, p);
-    }
-    else
-      unlink(next, bck, fwd);
+  if (in_smallbin_range(nb)) {
+    idx = smallbin_index(nb);
+    bin = bin_at(av,idx);
 
-    next = chunk_at_offset(p, sz);
+    if ( (victim = last(bin)) != bin) {
+      if (victim == 0) /* initialization check */
+        malloc_consolidate(av);
+      else {
+        bck = victim->bk;
+        set_inuse_bit_at_offset(victim, nb);
+        bin->bk = bck;
+        bck->fd = bin;
+
+        if (av != &main_arena)
+	  victim->size |= NON_MAIN_ARENA;
+        check_malloced_chunk(av, victim, nb);
+        return chunk2mem(victim);
+      }
+    }
   }
-  else
-    set_head(next, nextsz);                  /* clear inuse bit */
 
-  set_head(p, sz | PREV_INUSE);
-  next->prev_size = sz;
-  if (!islr)
-    frontlink(ar_ptr, p, sz, idx, bck, fwd);
+  /* 
+     If this is a large request, consolidate fastbins before continuing.
+     While it might look excessive to kill all fastbins before
+     even seeing if there is space available, this avoids
+     fragmentation problems normally associated with fastbins.
+     Also, in practice, programs tend to have runs of either small or
+     large requests, but less often mixtures, so consolidation is not 
+     invoked all that often in most programs. And the programs that
+     it is called frequently in otherwise tend to fragment.
+  */
 
-#if USE_ARENAS
-  /* Check whether the heap containing top can go away now. */
-  if(next->size < MINSIZE &&
-     (unsigned long)sz > trim_threshold &&
-     ar_ptr != &main_arena) {                /* fencepost */
-    heap_info *heap = heap_for_ptr(top(ar_ptr));
-
-    if(top(ar_ptr) == chunk_at_offset(heap, sizeof(*heap)) &&
-       heap->prev == heap_for_ptr(p))
-      heap_trim(heap, top_pad);
+  else {
+    idx = largebin_index(nb);
+    if (have_fastchunks(av)) 
+      malloc_consolidate(av);
   }
-#endif
-}
 
+  /*
+    Process recently freed or remaindered chunks, taking one only if
+    it is exact fit, or, if this a small request, the chunk is remainder from
+    the most recent non-exact fit.  Place other traversed chunks in
+    bins.  Note that this step is the only place in any routine where
+    chunks are placed in bins.
+
+    The outer loop here is needed because we might not realize until
+    near the end of malloc that we should have consolidated, so must
+    do so and retry. This happens at most once, and only when we would
+    otherwise need to expand memory to service a "small" request.
+  */
+    
+  for(;;) {    
+    
+    while ( (victim = unsorted_chunks(av)->bk) != unsorted_chunks(av)) {
+      bck = victim->bk;
+      size = chunksize(victim);
+
+      /* 
+         If a small request, try to use last remainder if it is the
+         only chunk in unsorted bin.  This helps promote locality for
+         runs of consecutive small requests. This is the only
+         exception to best-fit, and applies only when there is
+         no exact fit for a small chunk.
+      */
+
+      if (in_smallbin_range(nb) && 
+          bck == unsorted_chunks(av) &&
+          victim == av->last_remainder &&
+          (unsigned long)(size) > (unsigned long)(nb + MINSIZE)) {
+
+        /* split and reattach remainder */
+        remainder_size = size - nb;
+        remainder = chunk_at_offset(victim, nb);
+        unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
+        av->last_remainder = remainder; 
+        remainder->bk = remainder->fd = unsorted_chunks(av);
+        
+        set_head(victim, nb | PREV_INUSE |
+		 (av != &main_arena ? NON_MAIN_ARENA : 0));
+        set_head(remainder, remainder_size | PREV_INUSE);
+        set_foot(remainder, remainder_size);
+        
+        check_malloced_chunk(av, victim, nb);
+        return chunk2mem(victim);
+      }
 
-
+      /* remove from unsorted list */
+      unsorted_chunks(av)->bk = bck;
+      bck->fd = unsorted_chunks(av);
+      
+      /* Take now instead of binning if exact fit */
+      
+      if (size == nb) {
+        set_inuse_bit_at_offset(victim, size);
+	if (av != &main_arena)
+	  victim->size |= NON_MAIN_ARENA;
+        check_malloced_chunk(av, victim, nb);
+        return chunk2mem(victim);
+      }
+      
+      /* place chunk in bin */
+      
+      if (in_smallbin_range(size)) {
+        victim_index = smallbin_index(size);
+        bck = bin_at(av, victim_index);
+        fwd = bck->fd;
+      }
+      else {
+        victim_index = largebin_index(size);
+        bck = bin_at(av, victim_index);
+        fwd = bck->fd;
+
+        /* maintain large bins in sorted order */
+        if (fwd != bck) {
+	  /* Or with inuse bit to speed comparisons */
+          size |= PREV_INUSE;
+          /* if smaller than smallest, bypass loop below */
+	  assert((bck->bk->size & NON_MAIN_ARENA) == 0);
+          if ((unsigned long)(size) <= (unsigned long)(bck->bk->size)) {
+            fwd = bck;
+            bck = bck->bk;
+          }
+          else {
+	    assert((fwd->size & NON_MAIN_ARENA) == 0);
+            while ((unsigned long)(size) < (unsigned long)(fwd->size)) {
+              fwd = fwd->fd;
+	      assert((fwd->size & NON_MAIN_ARENA) == 0);
+	    }
+            bck = fwd->bk;
+          }
+        }
+      }
+      
+      mark_bin(av, victim_index);
+      victim->bk = bck;
+      victim->fd = fwd;
+      fwd->bk = victim;
+      bck->fd = victim;
+    }
+   
+    /*
+      If a large request, scan through the chunks of current bin in
+      sorted order to find smallest that fits.  This is the only step
+      where an unbounded number of chunks might be scanned without doing
+      anything useful with them. However the lists tend to be short.
+    */
+      
+    if (!in_smallbin_range(nb)) {
+      bin = bin_at(av, idx);
 
+      /* skip scan if empty or largest chunk is too small */
+      if ((victim = last(bin)) != bin &&
+          (unsigned long)(first(bin)->size) >= (unsigned long)(nb)) {
 
-/*
+        while (((unsigned long)(size = chunksize(victim)) < 
+                (unsigned long)(nb)))
+          victim = victim->bk;
 
-  Realloc algorithm:
+        remainder_size = size - nb;
+        unlink(victim, bck, fwd);
+        
+        /* Exhaust */
+        if (remainder_size < MINSIZE)  {
+          set_inuse_bit_at_offset(victim, size);
+	  if (av != &main_arena)
+	    victim->size |= NON_MAIN_ARENA;
+          check_malloced_chunk(av, victim, nb);
+          return chunk2mem(victim);
+        }
+        /* Split */
+        else {
+          remainder = chunk_at_offset(victim, nb);
+          unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
+          remainder->bk = remainder->fd = unsorted_chunks(av);
+          set_head(victim, nb | PREV_INUSE |
+		   (av != &main_arena ? NON_MAIN_ARENA : 0));
+          set_head(remainder, remainder_size | PREV_INUSE);
+          set_foot(remainder, remainder_size);
+          check_malloced_chunk(av, victim, nb);
+          return chunk2mem(victim);
+        } 
+      }
+    }    
 
-    Chunks that were obtained via mmap cannot be extended or shrunk
-    unless HAVE_MREMAP is defined, in which case mremap is used.
-    Otherwise, if their reallocation is for additional space, they are
-    copied.  If for less, they are just left alone.
+    /*
+      Search for a chunk by scanning bins, starting with next largest
+      bin. This search is strictly by best-fit; i.e., the smallest
+      (with ties going to approximately the least recently used) chunk
+      that fits is selected.
+      
+      The bitmap avoids needing to check that most blocks are nonempty.
+      The particular case of skipping all bins during warm-up phases
+      when no chunks have been returned yet is faster than it might look.
+    */
+    
+    ++idx;
+    bin = bin_at(av,idx);
+    block = idx2block(idx);
+    map = av->binmap[block];
+    bit = idx2bit(idx);
+    
+    for (;;) {
+
+      /* Skip rest of block if there are no more set bits in this block.  */
+      if (bit > map || bit == 0) {
+        do {
+          if (++block >= BINMAPSIZE)  /* out of bins */
+            goto use_top;
+        } while ( (map = av->binmap[block]) == 0);
+
+        bin = bin_at(av, (block << BINMAPSHIFT));
+        bit = 1;
+      }
+      
+      /* Advance to bin with set bit. There must be one. */
+      while ((bit & map) == 0) {
+        bin = next_bin(bin);
+        bit <<= 1;
+        assert(bit != 0);
+      }
+      
+      /* Inspect the bin. It is likely to be non-empty */
+      victim = last(bin);
+      
+      /*  If a false alarm (empty bin), clear the bit. */
+      if (victim == bin) {
+        av->binmap[block] = map &= ~bit; /* Write through */
+        bin = next_bin(bin);
+        bit <<= 1;
+      }
+      
+      else {
+        size = chunksize(victim);
+
+        /*  We know the first chunk in this bin is big enough to use. */
+        assert((unsigned long)(size) >= (unsigned long)(nb));
+
+        remainder_size = size - nb;
+        
+        /* unlink */
+        bck = victim->bk;
+        bin->bk = bck;
+        bck->fd = bin;
+        
+        /* Exhaust */
+        if (remainder_size < MINSIZE) {
+          set_inuse_bit_at_offset(victim, size);
+	  if (av != &main_arena)
+	    victim->size |= NON_MAIN_ARENA;
+          check_malloced_chunk(av, victim, nb);
+          return chunk2mem(victim);
+        }
+        
+        /* Split */
+        else {
+          remainder = chunk_at_offset(victim, nb);
+          
+          unsorted_chunks(av)->bk = unsorted_chunks(av)->fd = remainder;
+          remainder->bk = remainder->fd = unsorted_chunks(av);
+          /* advertise as last remainder */
+          if (in_smallbin_range(nb)) 
+            av->last_remainder = remainder; 
+          
+          set_head(victim, nb | PREV_INUSE |
+		   (av != &main_arena ? NON_MAIN_ARENA : 0));
+          set_head(remainder, remainder_size | PREV_INUSE);
+          set_foot(remainder, remainder_size);
+          check_malloced_chunk(av, victim, nb);
+          return chunk2mem(victim);
+        }
+      }
+    }
 
-    Otherwise, if the reallocation is for additional space, and the
-    chunk can be extended, it is, else a malloc-copy-free sequence is
-    taken.  There are several different ways that a chunk could be
-    extended. All are tried:
+  use_top:    
+    /*
+      If large enough, split off the chunk bordering the end of memory
+      (held in av->top). Note that this is in accord with the best-fit
+      search rule.  In effect, av->top is treated as larger (and thus
+      less well fitting) than any other available chunk since it can
+      be extended to be as large as necessary (up to system
+      limitations).
+
+      We require that av->top always exists (i.e., has size >=
+      MINSIZE) after initialization, so if it would otherwise be
+      exhuasted by current request, it is replenished. (The main
+      reason for ensuring it exists is that we may need MINSIZE space
+      to put in fenceposts in sysmalloc.)
+    */
 
-       * Extending forward into following adjacent free chunk.
-       * Shifting backwards, joining preceding adjacent space
-       * Both shifting backwards and extending forward.
-       * Extending into newly sbrked space
+    victim = av->top;
+    size = chunksize(victim);
+   
+    if ((unsigned long)(size) >= (unsigned long)(nb + MINSIZE)) {
+      remainder_size = size - nb;
+      remainder = chunk_at_offset(victim, nb);
+      av->top = remainder;
+      set_head(victim, nb | PREV_INUSE |
+	       (av != &main_arena ? NON_MAIN_ARENA : 0));
+      set_head(remainder, remainder_size | PREV_INUSE);
 
-    Unless the #define REALLOC_ZERO_BYTES_FREES is set, realloc with a
-    size argument of zero (re)allocates a minimum-sized chunk.
+      check_malloced_chunk(av, victim, nb);
+      return chunk2mem(victim);
+    }
 
-    If the reallocation is for less space, and the new request is for
-    a `small' (<512 bytes) size, then the newly unused space is lopped
-    off and freed.
+    /*
+      If there is space available in fastbins, consolidate and retry,
+      to possibly avoid expanding memory. This can occur only if nb is
+      in smallbin range so we didn't consolidate upon entry.
+    */
 
-    The old unix realloc convention of allowing the last-free'd chunk
-    to be used as an argument to realloc is no longer supported.
-    I don't know of any programs still relying on this feature,
-    and allowing it would also allow too many other incorrect
-    usages of realloc to be sensible.
+    else if (have_fastchunks(av)) {
+      assert(in_smallbin_range(nb));
+      malloc_consolidate(av);
+      idx = smallbin_index(nb); /* restore original bin index */
+    }
 
+    /* 
+       Otherwise, relay to handle system-dependent cases 
+    */
+    else 
+      return sYSMALLOc(nb, av);    
+  }
+}
 
+/*
+  ------------------------------ free ------------------------------
 */
 
-
-#if __STD_C
-Void_t* rEALLOc(Void_t* oldmem, size_t bytes)
-#else
-Void_t* rEALLOc(oldmem, bytes) Void_t* oldmem; size_t bytes;
-#endif
+void
+_int_free(mstate av, Void_t* mem)
 {
-  arena *ar_ptr;
-  INTERNAL_SIZE_T    nb;      /* padded request size */
-
-  mchunkptr oldp;             /* chunk corresponding to oldmem */
-  INTERNAL_SIZE_T    oldsize; /* its size */
+  mchunkptr       p;           /* chunk corresponding to mem */
+  INTERNAL_SIZE_T size;        /* its size */
+  mfastbinptr*    fb;          /* associated fastbin */
+  mchunkptr       nextchunk;   /* next contiguous chunk */
+  INTERNAL_SIZE_T nextsize;    /* its size */
+  int             nextinuse;   /* true if nextchunk is used */
+  INTERNAL_SIZE_T prevsize;    /* size of previous contiguous chunk */
+  mchunkptr       bck;         /* misc temp for linking */
+  mchunkptr       fwd;         /* misc temp for linking */
+
+
+  /* free(0) has no effect */
+  if (mem != 0) {
+    p = mem2chunk(mem);
+    size = chunksize(p);
 
-  mchunkptr newp;             /* chunk to return */
+    check_inuse_chunk(av, p);
 
-#if defined _LIBC || defined MALLOC_HOOKS
-  __malloc_ptr_t (*hook) __MALLOC_PMT ((__malloc_ptr_t, size_t,
-                                        __const __malloc_ptr_t)) =
-    __realloc_hook;
-  if (hook != NULL) {
-    Void_t* result;
+    /*
+      If eligible, place chunk on a fastbin so it can be found
+      and used quickly in malloc.
+    */
 
-#if defined __GNUC__ && __GNUC__ >= 2
-    result = (*hook)(oldmem, bytes, RETURN_ADDRESS (0));
-#else
-    result = (*hook)(oldmem, bytes, NULL);
-#endif
-    return result;
-  }
-#endif
+    if ((unsigned long)(size) <= (unsigned long)(av->max_fast)
 
-#ifdef REALLOC_ZERO_BYTES_FREES
-  if (bytes == 0 && oldmem != NULL) { fREe(oldmem); return 0; }
+#if TRIM_FASTBINS
+        /* 
+           If TRIM_FASTBINS set, don't place chunks
+           bordering top into fastbins
+        */
+        && (chunk_at_offset(p, size) != av->top)
 #endif
+        ) {
 
-  /* realloc of null is supposed to be same as malloc */
-  if (oldmem == 0) return mALLOc(bytes);
+      set_fastchunks(av);
+      fb = &(av->fastbins[fastbin_index(size)]);
+      p->fd = *fb;
+      *fb = p;
+    }
 
-  oldp    = mem2chunk(oldmem);
-  oldsize = chunksize(oldp);
+    /*
+       Consolidate other non-mmapped chunks as they arrive.
+    */
 
-  if(request2size(bytes, nb))
-    return 0;
+    else if (!chunk_is_mmapped(p)) {
+      nextchunk = chunk_at_offset(p, size);
+      nextsize = chunksize(nextchunk);
+      assert(nextsize > 0);
+
+      /* consolidate backward */
+      if (!prev_inuse(p)) {
+        prevsize = p->prev_size;
+        size += prevsize;
+        p = chunk_at_offset(p, -((long) prevsize));
+        unlink(p, bck, fwd);
+      }
 
-#if HAVE_MMAP
-  if (chunk_is_mmapped(oldp))
-  {
-    Void_t* newmem;
+      if (nextchunk != av->top) {
+        /* get and clear inuse bit */
+        nextinuse = inuse_bit_at_offset(nextchunk, nextsize);
+
+        /* consolidate forward */
+        if (!nextinuse) {
+          unlink(nextchunk, bck, fwd);
+          size += nextsize;
+        } else
+	  clear_inuse_bit_at_offset(nextchunk, 0);
+
+        /*
+          Place the chunk in unsorted chunk list. Chunks are
+          not placed into regular bins until after they have
+          been given one chance to be used in malloc.
+        */
+
+        bck = unsorted_chunks(av);
+        fwd = bck->fd;
+        p->bk = bck;
+        p->fd = fwd;
+        bck->fd = p;
+        fwd->bk = p;
+
+        set_head(p, size | PREV_INUSE);
+        set_foot(p, size);
+        
+        check_free_chunk(av, p);
+      }
 
-#if HAVE_MREMAP
-    newp = mremap_chunk(oldp, nb);
-    if(newp)
-      return BOUNDED_N(chunk2mem(newp), bytes);
-#endif
-    /* Note the extra SIZE_SZ overhead. */
-    if(oldsize - SIZE_SZ >= nb) return oldmem; /* do nothing */
-    /* Must alloc, copy, free. */
-    newmem = mALLOc(bytes);
-    if (newmem == 0) return 0; /* propagate failure */
-    MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ, 0);
-    munmap_chunk(oldp);
-    return newmem;
-  }
-#endif
+      /*
+         If the chunk borders the current high end of memory,
+         consolidate into top
+      */
 
-  ar_ptr = arena_for_ptr(oldp);
-#if THREAD_STATS
-  if(!mutex_trylock(&ar_ptr->mutex))
-    ++(ar_ptr->stat_lock_direct);
-  else {
-    (void)mutex_lock(&ar_ptr->mutex);
-    ++(ar_ptr->stat_lock_wait);
-  }
-#else
-  (void)mutex_lock(&ar_ptr->mutex);
-#endif
+      else {
+        size += nextsize;
+        set_head(p, size | PREV_INUSE);
+        av->top = p;
+        check_chunk(av, p);
+      }
 
-#ifndef NO_THREADS
-  /* As in malloc(), remember this arena for the next allocation. */
-  tsd_setspecific(arena_key, (Void_t *)ar_ptr);
-#endif
+      /*
+        If freeing a large space, consolidate possibly-surrounding
+        chunks. Then, if the total unused topmost memory exceeds trim
+        threshold, ask malloc_trim to reduce top.
+
+        Unless max_fast is 0, we don't know if there are fastbins
+        bordering top, so we cannot tell for sure whether threshold
+        has been reached unless fastbins are consolidated.  But we
+        don't want to consolidate on each free.  As a compromise,
+        consolidation is performed if FASTBIN_CONSOLIDATION_THRESHOLD
+        is reached.
+      */
+
+      if ((unsigned long)(size) >= FASTBIN_CONSOLIDATION_THRESHOLD) { 
+        if (have_fastchunks(av)) 
+          malloc_consolidate(av);
+
+	if (av == &main_arena) {
+#ifndef MORECORE_CANNOT_TRIM        
+	  if ((unsigned long)(chunksize(av->top)) >= 
+	      (unsigned long)(mp_.trim_threshold))
+	    sYSTRIm(mp_.top_pad, av);
+#endif
+	} else {
+	  /* Always try heap_trim(), even if the top chunk is not
+             large, because the corresponding heap might go away.  */
+	  heap_info *heap = heap_for_ptr(top(av));
+
+	  assert(heap->ar_ptr == av);
+	  heap_trim(heap, mp_.top_pad);
+	}
+      }
 
-  newp = chunk_realloc(ar_ptr, oldp, oldsize, nb);
+    }
+    /*
+      If the chunk was allocated via mmap, release via munmap(). Note
+      that if HAVE_MMAP is false but chunk_is_mmapped is true, then
+      user must have overwritten memory. There's nothing we can do to
+      catch this error unless MALLOC_DEBUG is set, in which case
+      check_inuse_chunk (above) will have triggered error.
+    */
 
-  (void)mutex_unlock(&ar_ptr->mutex);
-  return newp ? BOUNDED_N(chunk2mem(newp), bytes) : NULL;
+    else {
+#if HAVE_MMAP
+      int ret;
+      INTERNAL_SIZE_T offset = p->prev_size;
+      mp_.n_mmaps--;
+      mp_.mmapped_mem -= (size + offset);
+      ret = munmap((char*)p - offset, size + offset);
+      /* munmap returns non-zero on failure */
+      assert(ret == 0);
+#endif
+    }
+  }
 }
 
-static mchunkptr
-internal_function
+/*
+  ------------------------- malloc_consolidate -------------------------
+
+  malloc_consolidate is a specialized version of free() that tears
+  down chunks held in fastbins.  Free itself cannot be used for this
+  purpose since, among other things, it might place chunks back onto
+  fastbins.  So, instead, we need to use a minor variant of the same
+  code.
+  
+  Also, because this routine needs to be called the first time through
+  malloc anyway, it turns out to be the perfect place to trigger
+  initialization code.
+*/
+
 #if __STD_C
-chunk_realloc(arena* ar_ptr, mchunkptr oldp, INTERNAL_SIZE_T oldsize,
-              INTERNAL_SIZE_T nb)
+static void malloc_consolidate(mstate av)
 #else
-chunk_realloc(ar_ptr, oldp, oldsize, nb)
-arena* ar_ptr; mchunkptr oldp; INTERNAL_SIZE_T oldsize, nb;
+static void malloc_consolidate(av) mstate av;
 #endif
 {
-  mchunkptr newp = oldp;      /* chunk to return */
-  INTERNAL_SIZE_T newsize = oldsize; /* its size */
+  mfastbinptr*    fb;                 /* current fastbin being consolidated */
+  mfastbinptr*    maxfb;              /* last fastbin (for loop control) */
+  mchunkptr       p;                  /* current chunk being consolidated */
+  mchunkptr       nextp;              /* next chunk to consolidate */
+  mchunkptr       unsorted_bin;       /* bin header */
+  mchunkptr       first_unsorted;     /* chunk to link to */
+
+  /* These have same use as in free() */
+  mchunkptr       nextchunk;
+  INTERNAL_SIZE_T size;
+  INTERNAL_SIZE_T nextsize;
+  INTERNAL_SIZE_T prevsize;
+  int             nextinuse;
+  mchunkptr       bck;
+  mchunkptr       fwd;
 
-  mchunkptr next;             /* next contiguous chunk after oldp */
-  INTERNAL_SIZE_T  nextsize;  /* its size */
+  /*
+    If max_fast is 0, we know that av hasn't
+    yet been initialized, in which case do so below
+  */
 
-  mchunkptr prev;             /* previous contiguous chunk before oldp */
-  INTERNAL_SIZE_T  prevsize;  /* its size */
+  if (av->max_fast != 0) {
+    clear_fastchunks(av);
 
-  mchunkptr remainder;        /* holds split off extra space from newp */
-  INTERNAL_SIZE_T  remainder_size;   /* its size */
+    unsorted_bin = unsorted_chunks(av);
 
-  mchunkptr bck;              /* misc temp for linking */
-  mchunkptr fwd;              /* misc temp for linking */
+    /*
+      Remove each chunk from fast bin and consolidate it, placing it
+      then in unsorted bin. Among other reasons for doing this,
+      placing in unsorted bin avoids needing to calculate actual bins
+      until malloc is sure that chunks aren't immediately going to be
+      reused anyway.
+    */
+    
+    maxfb = &(av->fastbins[fastbin_index(av->max_fast)]);
+    fb = &(av->fastbins[0]);
+    do {
+      if ( (p = *fb) != 0) {
+        *fb = 0;
+        
+        do {
+          check_inuse_chunk(av, p);
+          nextp = p->fd;
+          
+          /* Slightly streamlined version of consolidation code in free() */
+          size = p->size & ~(PREV_INUSE|NON_MAIN_ARENA);
+          nextchunk = chunk_at_offset(p, size);
+          nextsize = chunksize(nextchunk);
+          
+          if (!prev_inuse(p)) {
+            prevsize = p->prev_size;
+            size += prevsize;
+            p = chunk_at_offset(p, -((long) prevsize));
+            unlink(p, bck, fwd);
+          }
+          
+          if (nextchunk != av->top) {
+            nextinuse = inuse_bit_at_offset(nextchunk, nextsize);
+            
+            if (!nextinuse) {
+              size += nextsize;
+              unlink(nextchunk, bck, fwd);
+            } else
+	      clear_inuse_bit_at_offset(nextchunk, 0);
+            
+            first_unsorted = unsorted_bin->fd;
+            unsorted_bin->fd = p;
+            first_unsorted->bk = p;
+            
+            set_head(p, size | PREV_INUSE);
+            p->bk = unsorted_bin;
+            p->fd = first_unsorted;
+            set_foot(p, size);
+          }
+          
+          else {
+            size += nextsize;
+            set_head(p, size | PREV_INUSE);
+            av->top = p;
+          }
+          
+        } while ( (p = nextp) != 0);
+        
+      }
+    } while (fb++ != maxfb);
+  }
+  else {
+    malloc_init_state(av);
+    check_malloc_state(av);
+  }
+}
 
-  check_inuse_chunk(ar_ptr, oldp);
+/*
+  ------------------------------ realloc ------------------------------
+*/
 
-  if ((long)(oldsize) < (long)(nb))
-  {
-    Void_t* oldmem = BOUNDED_N(chunk2mem(oldp), oldsize);
+Void_t*
+_int_realloc(mstate av, Void_t* oldmem, size_t bytes)
+{
+  INTERNAL_SIZE_T  nb;              /* padded request size */
 
-    /* Try expanding forward */
+  mchunkptr        oldp;            /* chunk corresponding to oldmem */
+  INTERNAL_SIZE_T  oldsize;         /* its size */
 
-    next = chunk_at_offset(oldp, oldsize);
-    if (next == top(ar_ptr) || !inuse(next))
-    {
-      nextsize = chunksize(next);
+  mchunkptr        newp;            /* chunk to return */
+  INTERNAL_SIZE_T  newsize;         /* its size */
+  Void_t*          newmem;          /* corresponding user mem */
 
-      /* Forward into top only if a remainder */
-      if (next == top(ar_ptr))
-      {
-        if ((long)(nextsize + newsize) >= (long)(nb + MINSIZE))
-        {
-          newsize += nextsize;
-          top(ar_ptr) = chunk_at_offset(oldp, nb);
-          set_head(top(ar_ptr), (newsize - nb) | PREV_INUSE);
-          set_head_size(oldp, nb);
-          return oldp;
-        }
-      }
+  mchunkptr        next;            /* next contiguous chunk after oldp */
 
-      /* Forward into next chunk */
-      else if (((long)(nextsize + newsize) >= (long)(nb)))
-      {
-        unlink(next, bck, fwd);
-        newsize  += nextsize;
-        goto split;
-      }
-    }
-    else
-    {
-      next = 0;
-      nextsize = 0;
-    }
+  mchunkptr        remainder;       /* extra space at end of newp */
+  unsigned long    remainder_size;  /* its size */
 
-    oldsize -= SIZE_SZ;
+  mchunkptr        bck;             /* misc temp for linking */
+  mchunkptr        fwd;             /* misc temp for linking */
 
-    /* Try shifting backwards. */
+  unsigned long    copysize;        /* bytes to copy */
+  unsigned int     ncopies;         /* INTERNAL_SIZE_T words to copy */
+  INTERNAL_SIZE_T* s;               /* copy source */ 
+  INTERNAL_SIZE_T* d;               /* copy destination */
 
-    if (!prev_inuse(oldp))
-    {
-      prev = prev_chunk(oldp);
-      prevsize = chunksize(prev);
 
-      /* try forward + backward first to save a later consolidation */
+#if REALLOC_ZERO_BYTES_FREES
+  if (bytes == 0) {
+    _int_free(av, oldmem);
+    return 0;
+  }
+#endif
 
-      if (next != 0)
-      {
-        /* into top */
-        if (next == top(ar_ptr))
-        {
-          if ((long)(nextsize + prevsize + newsize) >= (long)(nb + MINSIZE))
-          {
-            unlink(prev, bck, fwd);
-            newp = prev;
-            newsize += prevsize + nextsize;
-            MALLOC_COPY(BOUNDED_N(chunk2mem(newp), oldsize), oldmem, oldsize,
-                        1);
-            top(ar_ptr) = chunk_at_offset(newp, nb);
-            set_head(top(ar_ptr), (newsize - nb) | PREV_INUSE);
-            set_head_size(newp, nb);
-            return newp;
-          }
-        }
+  /* realloc of null is supposed to be same as malloc */
+  if (oldmem == 0) return _int_malloc(av, bytes);
 
-        /* into next chunk */
-        else if (((long)(nextsize + prevsize + newsize) >= (long)(nb)))
-        {
-          unlink(next, bck, fwd);
-          unlink(prev, bck, fwd);
-          newp = prev;
-          newsize += nextsize + prevsize;
-          MALLOC_COPY(BOUNDED_N(chunk2mem(newp), oldsize), oldmem, oldsize, 1);
-          goto split;
-        }
-      }
+  checked_request2size(bytes, nb);
 
-      /* backward only */
-      if (prev != 0 && (long)(prevsize + newsize) >= (long)nb)
-      {
-        unlink(prev, bck, fwd);
-        newp = prev;
-        newsize += prevsize;
-        MALLOC_COPY(BOUNDED_N(chunk2mem(newp), oldsize), oldmem, oldsize, 1);
-        goto split;
-      }
-    }
+  oldp    = mem2chunk(oldmem);
+  oldsize = chunksize(oldp);
 
-    /* Must allocate */
+  check_inuse_chunk(av, oldp);
 
-    newp = chunk_alloc (ar_ptr, nb);
+  if (!chunk_is_mmapped(oldp)) {
 
-    if (newp == 0) {
-      /* Maybe the failure is due to running out of mmapped areas. */
-      if (ar_ptr != &main_arena) {
-        (void)mutex_lock(&main_arena.mutex);
-        newp = chunk_alloc(&main_arena, nb);
-        (void)mutex_unlock(&main_arena.mutex);
-      } else {
-#if USE_ARENAS
-        /* ... or sbrk() has failed and there is still a chance to mmap() */
-        arena* ar_ptr2 = arena_get2(ar_ptr->next ? ar_ptr : 0, nb);
-        if(ar_ptr2) {
-          newp = chunk_alloc(ar_ptr2, nb);
-          (void)mutex_unlock(&ar_ptr2->mutex);
+    if ((unsigned long)(oldsize) >= (unsigned long)(nb)) {
+      /* already big enough; split below */
+      newp = oldp;
+      newsize = oldsize;
+    }
+
+    else {
+      next = chunk_at_offset(oldp, oldsize);
+
+      /* Try to expand forward into top */
+      if (next == av->top &&
+          (unsigned long)(newsize = oldsize + chunksize(next)) >=
+          (unsigned long)(nb + MINSIZE)) {
+        set_head_size(oldp, nb | (av != &main_arena ? NON_MAIN_ARENA : 0));
+        av->top = chunk_at_offset(oldp, nb);
+        set_head(av->top, (newsize - nb) | PREV_INUSE);
+	check_inuse_chunk(av, oldp);
+        return chunk2mem(oldp);
+      }
+      
+      /* Try to expand forward into next chunk;  split off remainder below */
+      else if (next != av->top && 
+               !inuse(next) &&
+               (unsigned long)(newsize = oldsize + chunksize(next)) >=
+               (unsigned long)(nb)) {
+        newp = oldp;
+        unlink(next, bck, fwd);
+      }
+
+      /* allocate, copy, free */
+      else {
+        newmem = _int_malloc(av, nb - MALLOC_ALIGN_MASK);
+        if (newmem == 0)
+          return 0; /* propagate failure */
+      
+        newp = mem2chunk(newmem);
+        newsize = chunksize(newp);
+        
+        /*
+          Avoid copy if newp is next chunk after oldp.
+        */
+        if (newp == next) {
+          newsize += oldsize;
+          newp = oldp;
+        }
+        else {
+          /*
+            Unroll copy of <= 36 bytes (72 if 8byte sizes)
+            We know that contents have an odd number of
+            INTERNAL_SIZE_T-sized words; minimally 3.
+          */
+          
+          copysize = oldsize - SIZE_SZ;
+          s = (INTERNAL_SIZE_T*)(oldmem);
+          d = (INTERNAL_SIZE_T*)(newmem);
+          ncopies = copysize / sizeof(INTERNAL_SIZE_T);
+          assert(ncopies >= 3);
+          
+          if (ncopies > 9)
+            MALLOC_COPY(d, s, copysize);
+          
+          else {
+            *(d+0) = *(s+0);
+            *(d+1) = *(s+1);
+            *(d+2) = *(s+2);
+            if (ncopies > 4) {
+              *(d+3) = *(s+3);
+              *(d+4) = *(s+4);
+              if (ncopies > 6) {
+                *(d+5) = *(s+5);
+                *(d+6) = *(s+6);
+                if (ncopies > 8) {
+                  *(d+7) = *(s+7);
+                  *(d+8) = *(s+8);
+                }
+              }
+            }
+          }
+          
+          _int_free(av, oldmem);
+          check_inuse_chunk(av, newp);
+          return chunk2mem(newp);
         }
-#endif
       }
-      if (newp == 0) /* propagate failure */
-        return 0;
     }
 
-    /* Avoid copy if newp is next chunk after oldp. */
-    /* (This can only happen when new chunk is sbrk'ed.) */
+    /* If possible, free extra space in old or extended chunk */
 
-    if ( newp == next_chunk(oldp))
-    {
-      newsize += chunksize(newp);
-      newp = oldp;
-      goto split;
+    assert((unsigned long)(newsize) >= (unsigned long)(nb));
+
+    remainder_size = newsize - nb;
+
+    if (remainder_size < MINSIZE) { /* not enough extra to split off */
+      set_head_size(newp, newsize | (av != &main_arena ? NON_MAIN_ARENA : 0));
+      set_inuse_bit_at_offset(newp, newsize);
+    }
+    else { /* split remainder */
+      remainder = chunk_at_offset(newp, nb);
+      set_head_size(newp, nb | (av != &main_arena ? NON_MAIN_ARENA : 0));
+      set_head(remainder, remainder_size | PREV_INUSE |
+	       (av != &main_arena ? NON_MAIN_ARENA : 0));
+      /* Mark remainder as inuse so free() won't complain */
+      set_inuse_bit_at_offset(remainder, remainder_size);
+      _int_free(av, chunk2mem(remainder)); 
     }
 
-    /* Otherwise copy, free, and exit */
-    MALLOC_COPY(BOUNDED_N(chunk2mem(newp), oldsize), oldmem, oldsize, 0);
-    chunk_free(ar_ptr, oldp);
-    return newp;
+    check_inuse_chunk(av, newp);
+    return chunk2mem(newp);
   }
 
+  /*
+    Handle mmap cases
+  */
 
- split:  /* split off extra room in old or expanded chunk */
-
-  if (newsize - nb >= MINSIZE) /* split off remainder */
-  {
-    remainder = chunk_at_offset(newp, nb);
-    remainder_size = newsize - nb;
-    set_head_size(newp, nb);
-    set_head(remainder, remainder_size | PREV_INUSE);
-    set_inuse_bit_at_offset(remainder, remainder_size);
-    chunk_free(ar_ptr, remainder);
-  }
-  else
-  {
-    set_head_size(newp, newsize);
-    set_inuse_bit_at_offset(newp, newsize);
-  }
+  else {
+#if HAVE_MMAP
 
-  check_inuse_chunk(ar_ptr, newp);
-  return newp;
-}
+#if HAVE_MREMAP
+    INTERNAL_SIZE_T offset = oldp->prev_size;
+    size_t pagemask = mp_.pagesize - 1;
+    char *cp;
+    unsigned long sum;
+    
+    /* Note the extra SIZE_SZ overhead */
+    newsize = (nb + offset + SIZE_SZ + pagemask) & ~pagemask;
+
+    /* don't need to remap if still within same page */
+    if (oldsize == newsize - offset) 
+      return oldmem;
+
+    cp = (char*)mremap((char*)oldp - offset, oldsize + offset, newsize, 1);
+    
+    if (cp != MAP_FAILED) {
+
+      newp = (mchunkptr)(cp + offset);
+      set_head(newp, (newsize - offset)|IS_MMAPPED);
+      
+      assert(aligned_OK(chunk2mem(newp)));
+      assert((newp->prev_size == offset));
+      
+      /* update statistics */
+      sum = mp_.mmapped_mem += newsize - oldsize;
+      if (sum > (unsigned long)(mp_.max_mmapped_mem)) 
+        mp_.max_mmapped_mem = sum;
+#ifdef NO_THREADS
+      sum += main_arena.system_mem;
+      if (sum > (unsigned long)(mp_.max_total_mem)) 
+        mp_.max_total_mem = sum;
+#endif
+      
+      return chunk2mem(newp);
+    }
+#endif
 
+    /* Note the extra SIZE_SZ overhead. */
+    if ((unsigned long)(oldsize) >= (unsigned long)(nb + SIZE_SZ)) 
+      newmem = oldmem; /* do nothing */
+    else {
+      /* Must alloc, copy, free. */
+      newmem = _int_malloc(av, nb - MALLOC_ALIGN_MASK);
+      if (newmem != 0) {
+        MALLOC_COPY(newmem, oldmem, oldsize - 2*SIZE_SZ);
+        _int_free(av, oldmem);
+      }
+    }
+    return newmem;
 
-
+#else 
+    /* If !HAVE_MMAP, but chunk_is_mmapped, user must have overwritten mem */
+    check_malloc_state(av);
+    MALLOC_FAILURE_ACTION;
+    return 0;
+#endif
+  }
+}
 
 /*
+  ------------------------------ memalign ------------------------------
+*/
 
-  memalign algorithm:
+Void_t*
+_int_memalign(mstate av, size_t alignment, size_t bytes)
+{
+  INTERNAL_SIZE_T nb;             /* padded  request size */
+  char*           m;              /* memory returned by malloc call */
+  mchunkptr       p;              /* corresponding chunk */
+  char*           brk;            /* alignment point within p */
+  mchunkptr       newp;           /* chunk to return */
+  INTERNAL_SIZE_T newsize;        /* its size */
+  INTERNAL_SIZE_T leadsize;       /* leading space before alignment point */
+  mchunkptr       remainder;      /* spare room at end to split off */
+  unsigned long   remainder_size; /* its size */
+  INTERNAL_SIZE_T size;
 
-    memalign requests more than enough space from malloc, finds a spot
-    within that chunk that meets the alignment request, and then
-    possibly frees the leading and trailing space.
+  /* If need less alignment than we give anyway, just relay to malloc */
 
-    The alignment argument must be a power of two. This property is not
-    checked by memalign, so misuse may result in random runtime errors.
+  if (alignment <= MALLOC_ALIGNMENT) return _int_malloc(av, bytes);
 
-    8-byte alignment is guaranteed by normal malloc calls, so don't
-    bother calling memalign with an argument of 8 or less.
+  /* Otherwise, ensure that it is at least a minimum chunk size */
 
-    Overreliance on memalign is a sure way to fragment space.
+  if (alignment <  MINSIZE) alignment = MINSIZE;
 
-*/
+  /* Make sure alignment is power of 2 (in case MINSIZE is not).  */
+  if ((alignment & (alignment - 1)) != 0) {
+    size_t a = MALLOC_ALIGNMENT * 2;
+    while ((unsigned long)a < (unsigned long)alignment) a <<= 1;
+    alignment = a;
+  }
 
+  checked_request2size(bytes, nb);
 
-#if __STD_C
-Void_t* mEMALIGn(size_t alignment, size_t bytes)
-#else
-Void_t* mEMALIGn(alignment, bytes) size_t alignment; size_t bytes;
-#endif
-{
-  arena *ar_ptr;
-  INTERNAL_SIZE_T    nb;      /* padded  request size */
-  mchunkptr p;
-
-#if defined _LIBC || defined MALLOC_HOOKS
-  __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, size_t,
-                                        __const __malloc_ptr_t)) =
-    __memalign_hook;
-  if (hook != NULL) {
-    Void_t* result;
+  /*
+    Strategy: find a spot within that chunk that meets the alignment
+    request, and then possibly free the leading and trailing space.
+  */
 
-#if defined __GNUC__ && __GNUC__ >= 2
-    result = (*hook)(alignment, bytes, RETURN_ADDRESS (0));
-#else
-    result = (*hook)(alignment, bytes, NULL);
-#endif
-    return result;
-  }
-#endif
 
-  /* If need less alignment than we give anyway, just relay to malloc */
+  /* Call malloc with worst case padding to hit alignment. */
 
-  if (alignment <= MALLOC_ALIGNMENT) return mALLOc(bytes);
+  m  = (char*)(_int_malloc(av, nb + alignment + MINSIZE));
 
-  /* Otherwise, ensure that it is at least a minimum chunk size */
+  if (m == 0) return 0; /* propagate failure */
 
-  if (alignment <  MINSIZE) alignment = MINSIZE;
+  p = mem2chunk(m);
 
-  if(request2size(bytes, nb))
-    return 0;
-  arena_get(ar_ptr, nb + alignment + MINSIZE);
-  if(!ar_ptr)
-    return 0;
-  p = chunk_align(ar_ptr, nb, alignment);
-  (void)mutex_unlock(&ar_ptr->mutex);
-  if(!p) {
-    /* Maybe the failure is due to running out of mmapped areas. */
-    if(ar_ptr != &main_arena) {
-      (void)mutex_lock(&main_arena.mutex);
-      p = chunk_align(&main_arena, nb, alignment);
-      (void)mutex_unlock(&main_arena.mutex);
-    } else {
-#if USE_ARENAS
-      /* ... or sbrk() has failed and there is still a chance to mmap() */
-      ar_ptr = arena_get2(ar_ptr->next ? ar_ptr : 0, nb);
-      if(ar_ptr) {
-        p = chunk_align(ar_ptr, nb, alignment);
-        (void)mutex_unlock(&ar_ptr->mutex);
-      }
-#endif
-    }
-    if(!p) return 0;
-  }
-  return BOUNDED_N(chunk2mem(p), bytes);
-}
+  if ((((unsigned long)(m)) % alignment) != 0) { /* misaligned */
 
-static mchunkptr
-internal_function
-#if __STD_C
-chunk_align(arena* ar_ptr, INTERNAL_SIZE_T nb, size_t alignment)
-#else
-chunk_align(ar_ptr, nb, alignment)
-arena* ar_ptr; INTERNAL_SIZE_T nb; size_t alignment;
-#endif
-{
-  unsigned long m;            /* memory returned by malloc call */
-  mchunkptr p;                /* corresponding chunk */
-  char*     brk;              /* alignment point within p */
-  mchunkptr newp;             /* chunk to return */
-  INTERNAL_SIZE_T  newsize;   /* its size */
-  INTERNAL_SIZE_T  leadsize;  /* leading space befor alignment point */
-  mchunkptr remainder;        /* spare room at end to split off */
-  long      remainder_size;   /* its size */
-
-  /* Call chunk_alloc with worst case padding to hit alignment. */
-  p = chunk_alloc(ar_ptr, nb + alignment + MINSIZE);
-  if (p == 0)
-    return 0; /* propagate failure */
-
-  m = (unsigned long)chunk2mem(p);
-
-  if ((m % alignment) == 0) /* aligned */
-  {
-#if HAVE_MMAP
-    if(chunk_is_mmapped(p)) {
-      return p; /* nothing more to do */
-    }
-#endif
-  }
-  else /* misaligned */
-  {
     /*
-      Find an aligned spot inside chunk.
-      Since we need to give back leading space in a chunk of at
-      least MINSIZE, if the first calculation places us at
-      a spot with less than MINSIZE leader, we can move to the
-      next aligned spot -- we've allocated enough total room so that
-      this is always possible.
+      Find an aligned spot inside chunk.  Since we need to give back
+      leading space in a chunk of at least MINSIZE, if the first
+      calculation places us at a spot with less than MINSIZE leader,
+      we can move to the next aligned spot -- we've allocated enough
+      total room so that this is always possible.
     */
 
-    brk = (char*)mem2chunk(((m + alignment - 1)) & -(long)alignment);
-    if ((long)(brk - (char*)(p)) < (long)MINSIZE) brk += alignment;
+    brk = (char*)mem2chunk(((unsigned long)(m + alignment - 1)) &
+                           -((signed long) alignment));
+    if ((unsigned long)(brk - (char*)(p)) < MINSIZE)
+      brk += alignment;
 
-    newp = chunk_at_offset(brk, 0);
+    newp = (mchunkptr)brk;
     leadsize = brk - (char*)(p);
     newsize = chunksize(p) - leadsize;
 
-#if HAVE_MMAP
-    if(chunk_is_mmapped(p))
-    {
+    /* For mmapped chunks, just adjust offset */
+    if (chunk_is_mmapped(p)) {
       newp->prev_size = p->prev_size + leadsize;
       set_head(newp, newsize|IS_MMAPPED);
-      return newp;
+      return chunk2mem(newp);
     }
-#endif
 
-    /* give back leader, use the rest */
-
-    set_head(newp, newsize | PREV_INUSE);
+    /* Otherwise, give back leader, use the rest */
+    set_head(newp, newsize | PREV_INUSE |
+	     (av != &main_arena ? NON_MAIN_ARENA : 0));
     set_inuse_bit_at_offset(newp, newsize);
-    set_head_size(p, leadsize);
-    chunk_free(ar_ptr, p);
+    set_head_size(p, leadsize | (av != &main_arena ? NON_MAIN_ARENA : 0));
+    _int_free(av, chunk2mem(p));
     p = newp;
 
-    assert (newsize>=nb && (((unsigned long)(chunk2mem(p))) % alignment) == 0);
+    assert (newsize >= nb &&
+            (((unsigned long)(chunk2mem(p))) % alignment) == 0);
   }
 
   /* Also give back spare room at the end */
-
-  remainder_size = chunksize(p) - nb;
-
-  if (remainder_size >= (long)MINSIZE)
-  {
-    remainder = chunk_at_offset(p, nb);
-    set_head(remainder, remainder_size | PREV_INUSE);
-    set_head_size(p, nb);
-    chunk_free(ar_ptr, remainder);
+  if (!chunk_is_mmapped(p)) {
+    size = chunksize(p);
+    if ((unsigned long)(size) > (unsigned long)(nb + MINSIZE)) {
+      remainder_size = size - nb;
+      remainder = chunk_at_offset(p, nb);
+      set_head(remainder, remainder_size | PREV_INUSE |
+	       (av != &main_arena ? NON_MAIN_ARENA : 0));
+      set_head_size(p, nb);
+      _int_free(av, chunk2mem(remainder));
+    }
   }
 
-  check_inuse_chunk(ar_ptr, p);
-  return p;
+  check_inuse_chunk(av, p);
+  return chunk2mem(p);
 }
 
-
+#if 0
+/*
+  ------------------------------ calloc ------------------------------
+*/
+
+#if __STD_C
+Void_t* cALLOc(size_t n_elements, size_t elem_size)
+#else
+Void_t* cALLOc(n_elements, elem_size) size_t n_elements; size_t elem_size;
+#endif
+{
+  mchunkptr p;
+  unsigned long clearsize;
+  unsigned long nclears;
+  INTERNAL_SIZE_T* d;
 
+  Void_t* mem = mALLOc(n_elements * elem_size);
+
+  if (mem != 0) {
+    p = mem2chunk(mem);
+
+#if MMAP_CLEARS
+    if (!chunk_is_mmapped(p)) /* don't need to clear mmapped space */
+#endif
+    {  
+      /*
+        Unroll clear of <= 36 bytes (72 if 8byte sizes)
+        We know that contents have an odd number of
+        INTERNAL_SIZE_T-sized words; minimally 3.
+      */
+
+      d = (INTERNAL_SIZE_T*)mem;
+      clearsize = chunksize(p) - SIZE_SZ;
+      nclears = clearsize / sizeof(INTERNAL_SIZE_T);
+      assert(nclears >= 3);
+
+      if (nclears > 9)
+        MALLOC_ZERO(d, clearsize);
+
+      else {
+        *(d+0) = 0;
+        *(d+1) = 0;
+        *(d+2) = 0;
+        if (nclears > 4) {
+          *(d+3) = 0;
+          *(d+4) = 0;
+          if (nclears > 6) {
+            *(d+5) = 0;
+            *(d+6) = 0;
+            if (nclears > 8) {
+              *(d+7) = 0;
+              *(d+8) = 0;
+            }
+          }
+        }
+      }
+    }
+  }
+  return mem;
+}
+#endif /* 0 */
 
 /*
-    valloc just invokes memalign with alignment argument equal
-    to the page size of the system (or as near to this as can
-    be figured out from all the includes/defines above.)
+  ------------------------- independent_calloc -------------------------
 */
 
+Void_t**
 #if __STD_C
-Void_t* vALLOc(size_t bytes)
+_int_icalloc(mstate av, size_t n_elements, size_t elem_size, Void_t* chunks[])
 #else
-Void_t* vALLOc(bytes) size_t bytes;
+_int_icalloc(av, n_elements, elem_size, chunks)
+mstate av; size_t n_elements; size_t elem_size; Void_t* chunks[];
 #endif
 {
-  if(__malloc_initialized < 0)
-    ptmalloc_init ();
-  return mEMALIGn (malloc_getpagesize, bytes);
+  size_t sz = elem_size; /* serves as 1-element array */
+  /* opts arg of 3 means all elements are same size, and should be cleared */
+  return iALLOc(av, n_elements, &sz, 3, chunks);
 }
 
 /*
-  pvalloc just invokes valloc for the nearest pagesize
-  that will accommodate request
+  ------------------------- independent_comalloc -------------------------
 */
 
-
+Void_t**
 #if __STD_C
-Void_t* pvALLOc(size_t bytes)
+_int_icomalloc(mstate av, size_t n_elements, size_t sizes[], Void_t* chunks[])
 #else
-Void_t* pvALLOc(bytes) size_t bytes;
+_int_icomalloc(av, n_elements, sizes, chunks)
+mstate av; size_t n_elements; size_t sizes[]; Void_t* chunks[];
 #endif
 {
-  size_t pagesize;
-  if(__malloc_initialized < 0)
-    ptmalloc_init ();
-  pagesize = malloc_getpagesize;
-  return mEMALIGn (pagesize, (bytes + pagesize - 1) & ~(pagesize - 1));
+  return iALLOc(av, n_elements, sizes, 0, chunks);
 }
 
-/*
 
-  calloc calls chunk_alloc, then zeroes out the allocated chunk.
+/*
+  ------------------------------ ialloc ------------------------------
+  ialloc 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
 */
 
+
+static Void_t**
 #if __STD_C
-Void_t* cALLOc(size_t n, size_t elem_size)
+iALLOc(mstate av, size_t n_elements, size_t* sizes, int opts, Void_t* chunks[])
 #else
-Void_t* cALLOc(n, elem_size) size_t n; size_t elem_size;
+iALLOc(av, n_elements, sizes, opts, chunks)
+mstate av; size_t n_elements; size_t* sizes; int opts; Void_t* chunks[];
 #endif
 {
-  arena *ar_ptr;
-  mchunkptr p, oldtop;
-  INTERNAL_SIZE_T sz, csz, oldtopsize;
-  Void_t* mem;
+  INTERNAL_SIZE_T element_size;   /* chunksize of each element, if all same */
+  INTERNAL_SIZE_T contents_size;  /* total size of elements */
+  INTERNAL_SIZE_T array_size;     /* request size of pointer array */
+  Void_t*         mem;            /* malloced aggregate space */
+  mchunkptr       p;              /* corresponding chunk */
+  INTERNAL_SIZE_T remainder_size; /* remaining bytes while splitting */
+  Void_t**        marray;         /* either "chunks" or malloced ptr array */
+  mchunkptr       array_chunk;    /* chunk for malloced ptr array */
+  int             mmx;            /* to disable mmap */
+  INTERNAL_SIZE_T size;           
+  INTERNAL_SIZE_T size_flags;
+  size_t          i;
+
+  /* Ensure initialization/consolidation */
+  if (have_fastchunks(av)) malloc_consolidate(av);
+
+  /* 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_t**) _int_malloc(av, 0);
+    marray = 0;
+    array_size = request2size(n_elements * (sizeof(Void_t*)));
+  }
 
-#if defined _LIBC || defined MALLOC_HOOKS
-  __malloc_ptr_t (*hook) __MALLOC_PMT ((size_t, __const __malloc_ptr_t)) =
-    __malloc_hook;
-  if (hook != NULL) {
-    sz = n * elem_size;
-#if defined __GNUC__ && __GNUC__ >= 2
-    mem = (*hook)(sz, RETURN_ADDRESS (0));
-#else
-    mem = (*hook)(sz, NULL);
-#endif
-    if(mem == 0)
-      return 0;
-#ifdef HAVE_MEMSET
-    return memset(mem, 0, sz);
-#else
-    while(sz > 0) ((char*)mem)[--sz] = 0; /* rather inefficient */
-    return mem;
-#endif
+  /* 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]);     
   }
-#endif
 
-  if(request2size(n * elem_size, sz))
-    return 0;
-  arena_get(ar_ptr, sz);
-  if(!ar_ptr)
+  /* subtract out alignment bytes from total to minimize overallocation */
+  size = contents_size + array_size - MALLOC_ALIGN_MASK;
+  
+  /* 
+     Allocate the aggregate chunk.
+     But first disable mmap so malloc won't use it, since
+     we would not be able to later free/realloc space internal
+     to a segregated mmap region.
+  */
+  mmx = mp_.n_mmaps_max;   /* disable mmap */
+  mp_.n_mmaps_max = 0;
+  mem = _int_malloc(av, size);
+  mp_.n_mmaps_max = mmx;   /* reset mmap */
+  if (mem == 0) 
     return 0;
 
-  /* Check if expand_top called, in which case there may be
-     no need to clear. */
-#if MORECORE_CLEARS
-  oldtop = top(ar_ptr);
-  oldtopsize = chunksize(top(ar_ptr));
-#if MORECORE_CLEARS < 2
-  /* Only newly allocated memory is guaranteed to be cleared.  */
-  if (ar_ptr == &main_arena &&
-      oldtopsize < sbrk_base + max_sbrked_mem - (char *)oldtop)
-    oldtopsize = (sbrk_base + max_sbrked_mem - (char *)oldtop);
-#endif
-#endif
-  p = chunk_alloc (ar_ptr, sz);
-
-  /* Only clearing follows, so we can unlock early. */
-  (void)mutex_unlock(&ar_ptr->mutex);
+  p = mem2chunk(mem);
+  assert(!chunk_is_mmapped(p)); 
+  remainder_size = chunksize(p);
 
-  if (p == 0) {
-    /* Maybe the failure is due to running out of mmapped areas. */
-    if(ar_ptr != &main_arena) {
-      (void)mutex_lock(&main_arena.mutex);
-      p = chunk_alloc(&main_arena, sz);
-      (void)mutex_unlock(&main_arena.mutex);
-    } else {
-#if USE_ARENAS
-      /* ... or sbrk() has failed and there is still a chance to mmap() */
-      (void)mutex_lock(&main_arena.mutex);
-      ar_ptr = arena_get2(ar_ptr->next ? ar_ptr : 0, sz);
-      (void)mutex_unlock(&main_arena.mutex);
-      if(ar_ptr) {
-        p = chunk_alloc(ar_ptr, sz);
-        (void)mutex_unlock(&ar_ptr->mutex);
-      }
-#endif
-    }
-    if (p == 0) return 0;
+  if (opts & 0x2) {       /* optionally clear the elements */
+    MALLOC_ZERO(mem, remainder_size - SIZE_SZ - array_size);
   }
-  mem = BOUNDED_N(chunk2mem(p), n * elem_size);
 
-  /* Two optional cases in which clearing not necessary */
+  size_flags = PREV_INUSE | (av != &main_arena ? NON_MAIN_ARENA : 0);
 
-#if HAVE_MMAP
-  if (chunk_is_mmapped(p)) return mem;
-#endif
+  /* If not provided, allocate the pointer array as final part of chunk */
+  if (marray == 0) {
+    array_chunk = chunk_at_offset(p, contents_size);
+    marray = (Void_t**) (chunk2mem(array_chunk));
+    set_head(array_chunk, (remainder_size - contents_size) | size_flags);
+    remainder_size = contents_size;
+  }
 
-  csz = chunksize(p);
+  /* 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_head(p, size | size_flags);
+      p = chunk_at_offset(p, size);
+    }
+    else { /* the final element absorbs any overallocation slop */
+      set_head(p, remainder_size | size_flags);
+      break;
+    }
+  }
 
-#if MORECORE_CLEARS
-  if (p == oldtop && csz > oldtopsize) {
-    /* clear only the bytes from non-freshly-sbrked memory */
-    csz = oldtopsize;
+#if MALLOC_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(av, mem2chunk(marray));
   }
+
+  for (i = 0; i != n_elements; ++i)
+    check_inuse_chunk(av, mem2chunk(marray[i]));
 #endif
 
-  csz -= SIZE_SZ;
-  MALLOC_ZERO(BOUNDED_N(chunk2mem(p), csz), csz);
-  return mem;
+  return marray;
 }
 
-/*
-
-  cfree just calls free. It is needed/defined on some systems
-  that pair it with calloc, presumably for odd historical reasons.
 
+/*
+  ------------------------------ valloc ------------------------------
 */
 
-#if !defined(_LIBC)
+Void_t*
 #if __STD_C
-void cfree(Void_t *mem)
+_int_valloc(mstate av, size_t bytes)
 #else
-void cfree(mem) Void_t *mem;
+_int_valloc(av, bytes) mstate av; size_t bytes;
 #endif
 {
-  fREe(mem);
+  /* Ensure initialization/consolidation */
+  if (have_fastchunks(av)) malloc_consolidate(av);
+  return _int_memalign(av, mp_.pagesize, bytes);
 }
-#endif
-
-
 
 /*
-
-    Malloc_trim gives memory back to the system (via negative
-    arguments to sbrk) if there is unused memory at the `high' end of
-    the malloc pool. You can call this after freeing large blocks of
-    memory to potentially reduce the system-level memory requirements
-    of a program. However, it cannot guarantee to reduce memory. Under
-    some allocation patterns, some large free blocks of memory will be
-    locked between two used chunks, so they cannot be given back to
-    the system.
-
-    The `pad' argument to malloc_trim represents the amount of free
-    trailing space to leave untrimmed. If this argument is zero,
-    only the minimum amount of memory to maintain internal data
-    structures will be left (one page or less). Non-zero arguments
-    can be supplied to maintain enough trailing space to service
-    future expected allocations without having to re-obtain memory
-    from the system.
-
-    Malloc_trim returns 1 if it actually released any memory, else 0.
-
+  ------------------------------ pvalloc ------------------------------
 */
 
+
+Void_t*
 #if __STD_C
-int mALLOC_TRIm(size_t pad)
+_int_pvalloc(mstate av, size_t bytes)
 #else
-int mALLOC_TRIm(pad) size_t pad;
+_int_pvalloc(av, bytes) mstate av, size_t bytes;
 #endif
 {
-  int res;
+  size_t pagesz;
 
-  (void)mutex_lock(&main_arena.mutex);
-  res = main_trim(pad);
-  (void)mutex_unlock(&main_arena.mutex);
-  return res;
+  /* Ensure initialization/consolidation */
+  if (have_fastchunks(av)) malloc_consolidate(av);
+  pagesz = mp_.pagesize;
+  return _int_memalign(av, pagesz, (bytes + pagesz - 1) & ~(pagesz - 1));
 }
+   
 
-/* Trim the main arena. */
+/*
+  ------------------------------ malloc_trim ------------------------------
+*/
 
-static int
-internal_function
 #if __STD_C
-main_trim(size_t pad)
+int mTRIm(size_t pad)
 #else
-main_trim(pad) size_t pad;
+int mTRIm(pad) size_t pad;
 #endif
 {
-  mchunkptr top_chunk;   /* The current top chunk */
-  long  top_size;        /* Amount of top-most memory */
-  long  extra;           /* Amount to release */
-  char* current_brk;     /* address returned by pre-check sbrk call */
-  char* new_brk;         /* address returned by negative sbrk call */
-
-  unsigned long pagesz = malloc_getpagesize;
-
-  top_chunk = top(&main_arena);
-  top_size = chunksize(top_chunk);
-  extra = ((top_size - pad - MINSIZE + (pagesz-1)) / pagesz - 1) * pagesz;
+  mstate av = &main_arena; /* already locked */
 
-  if (extra < (long)pagesz) /* Not enough memory to release */
-    return 0;
-
-  /* Test to make sure no one else called sbrk */
-  current_brk = (char*)(MORECORE (0));
-  if (current_brk != (char*)(top_chunk) + top_size)
-    return 0;     /* Apparently we don't own memory; must fail */
-
-  new_brk = (char*)(MORECORE (-extra));
-
-#if defined _LIBC || defined MALLOC_HOOKS
-  /* Call the `morecore' hook if necessary.  */
-  if (__after_morecore_hook)
-    (*__after_morecore_hook) ();
-#endif
+  /* Ensure initialization/consolidation */
+  malloc_consolidate(av);
 
-  if (new_brk == (char*)(MORECORE_FAILURE)) { /* sbrk failed? */
-    /* Try to figure out what we have */
-    current_brk = (char*)(MORECORE (0));
-    top_size = current_brk - (char*)top_chunk;
-    if (top_size >= (long)MINSIZE) /* if not, we are very very dead! */
-    {
-      sbrked_mem = current_brk - sbrk_base;
-      set_head(top_chunk, top_size | PREV_INUSE);
-    }
-    check_chunk(&main_arena, top_chunk);
-    return 0;
-  }
-  sbrked_mem -= extra;
-
-  /* Success. Adjust top accordingly. */
-  set_head(top_chunk, (top_size - extra) | PREV_INUSE);
-  check_chunk(&main_arena, top_chunk);
-  return 1;
-}
-
-#if USE_ARENAS
-
-static int
-internal_function
-#if __STD_C
-heap_trim(heap_info *heap, size_t pad)
+#ifndef MORECORE_CANNOT_TRIM        
+  return sYSTRIm(pad, av);
 #else
-heap_trim(heap, pad) heap_info *heap; size_t pad;
+  return 0;
 #endif
-{
-  unsigned long pagesz = malloc_getpagesize;
-  arena *ar_ptr = heap->ar_ptr;
-  mchunkptr top_chunk = top(ar_ptr), p, bck, fwd;
-  heap_info *prev_heap;
-  long new_size, top_size, extra;
-
-  /* Can this heap go away completely ? */
-  while(top_chunk == chunk_at_offset(heap, sizeof(*heap))) {
-    prev_heap = heap->prev;
-    p = chunk_at_offset(prev_heap, prev_heap->size - (MINSIZE-2*SIZE_SZ));
-    assert(p->size == (0|PREV_INUSE)); /* must be fencepost */
-    p = prev_chunk(p);
-    new_size = chunksize(p) + (MINSIZE-2*SIZE_SZ);
-    assert(new_size>0 && new_size<(long)(2*MINSIZE));
-    if(!prev_inuse(p))
-      new_size += p->prev_size;
-    assert(new_size>0 && new_size<HEAP_MAX_SIZE);
-    if(new_size + (HEAP_MAX_SIZE - prev_heap->size) < pad + MINSIZE + pagesz)
-      break;
-    ar_ptr->size -= heap->size;
-    arena_mem -= heap->size;
-    delete_heap(heap);
-    heap = prev_heap;
-    if(!prev_inuse(p)) { /* consolidate backward */
-      p = prev_chunk(p);
-      unlink(p, bck, fwd);
-    }
-    assert(((unsigned long)((char*)p + new_size) & (pagesz-1)) == 0);
-    assert( ((char*)p + new_size) == ((char*)heap + heap->size) );
-    top(ar_ptr) = top_chunk = p;
-    set_head(top_chunk, new_size | PREV_INUSE);
-    check_chunk(ar_ptr, top_chunk);
-  }
-  top_size = chunksize(top_chunk);
-  extra = ((top_size - pad - MINSIZE + (pagesz-1))/pagesz - 1) * pagesz;
-  if(extra < (long)pagesz)
-    return 0;
-  /* Try to shrink. */
-  if(grow_heap(heap, -extra) != 0)
-    return 0;
-  ar_ptr->size -= extra;
-  arena_mem -= extra;
-
-  /* Success. Adjust top accordingly. */
-  set_head(top_chunk, (top_size - extra) | PREV_INUSE);
-  check_chunk(ar_ptr, top_chunk);
-  return 1;
 }
 
-#endif /* USE_ARENAS */
-
-
 
 /*
-  malloc_usable_size:
-
-    This routine tells you how many bytes you can actually use in an
-    allocated chunk, which may be more than you requested (although
-    often not). You can use this many bytes without worrying about
-    overwriting other allocated objects. Not a particularly great
-    programming practice, but still sometimes useful.
-
+  ------------------------- malloc_usable_size -------------------------
 */
 
 #if __STD_C
-size_t mALLOC_USABLE_SIZe(Void_t* mem)
+size_t mUSABLe(Void_t* mem)
 #else
-size_t mALLOC_USABLE_SIZe(mem) Void_t* mem;
+size_t mUSABLe(mem) Void_t* mem;
 #endif
 {
   mchunkptr p;
-
-  if (mem == 0)
-    return 0;
-  else
-  {
+  if (mem != 0) {
     p = mem2chunk(mem);
-    if(!chunk_is_mmapped(p))
-    {
-      if (!inuse(p)) return 0;
-      check_inuse_chunk(arena_for_ptr(mem), p);
+    if (chunk_is_mmapped(p))
+      return chunksize(p) - 2*SIZE_SZ;
+    else if (inuse(p))
       return chunksize(p) - SIZE_SZ;
-    }
-    return chunksize(p) - 2*SIZE_SZ;
   }
+  return 0;
 }
 
+/*
+  ------------------------------ mallinfo ------------------------------
+*/
 
-
-
-/* Utility to update mallinfo for malloc_stats() and mallinfo() */
-
-static void
-#if __STD_C
-malloc_update_mallinfo(arena *ar_ptr, struct mallinfo *mi)
-#else
-malloc_update_mallinfo(ar_ptr, mi) arena *ar_ptr; struct mallinfo *mi;
-#endif
+struct mallinfo mALLINFo(mstate av)
 {
-  int i, navail;
+  struct mallinfo mi;
+  int i;
   mbinptr b;
   mchunkptr p;
-#if MALLOC_DEBUG
-  mchunkptr q;
-#endif
   INTERNAL_SIZE_T avail;
+  INTERNAL_SIZE_T fastavail;
+  int nblocks;
+  int nfastblocks;
 
-  (void)mutex_lock(&ar_ptr->mutex);
-  avail = chunksize(top(ar_ptr));
-  navail = ((long)(avail) >= (long)MINSIZE)? 1 : 0;
+  /* Ensure initialization */
+  if (av->top == 0)  malloc_consolidate(av);
 
-  for (i = 1; i < NAV; ++i)
-  {
-    b = bin_at(ar_ptr, i);
-    for (p = last(b); p != b; p = p->bk)
-    {
-#if MALLOC_DEBUG
-      check_free_chunk(ar_ptr, p);
-      for (q = next_chunk(p);
-           q != top(ar_ptr) && inuse(q) && (long)chunksize(q) > 0;
-           q = next_chunk(q))
-        check_inuse_chunk(ar_ptr, q);
-#endif
-      avail += chunksize(p);
-      navail++;
-    }
-  }
+  check_malloc_state(av);
 
-  mi->arena = ar_ptr->size;
-  mi->ordblks = navail;
-  mi->smblks = mi->usmblks = mi->fsmblks = 0; /* clear unused fields */
-  mi->uordblks = ar_ptr->size - avail;
-  mi->fordblks = avail;
-  mi->hblks = n_mmaps;
-  mi->hblkhd = mmapped_mem;
-  mi->keepcost = chunksize(top(ar_ptr));
+  /* Account for top */
+  avail = chunksize(av->top);
+  nblocks = 1;  /* top always exists */
 
-  (void)mutex_unlock(&ar_ptr->mutex);
-}
+  /* traverse fastbins */
+  nfastblocks = 0;
+  fastavail = 0;
 
-#if USE_ARENAS && MALLOC_DEBUG > 1
-
-/* Print the complete contents of a single heap to stderr. */
+  for (i = 0; i < NFASTBINS; ++i) {
+    for (p = av->fastbins[i]; p != 0; p = p->fd) {
+      ++nfastblocks;
+      fastavail += chunksize(p);
+    }
+  }
 
-static void
-#if __STD_C
-dump_heap(heap_info *heap)
-#else
-dump_heap(heap) heap_info *heap;
-#endif
-{
-  char *ptr;
-  mchunkptr p;
+  avail += fastavail;
 
-  fprintf(stderr, "Heap %p, size %10lx:\n", heap, (long)heap->size);
-  ptr = (heap->ar_ptr != (arena*)(heap+1)) ?
-    (char*)(heap + 1) : (char*)(heap + 1) + sizeof(arena);
-  p = (mchunkptr)(((unsigned long)ptr + MALLOC_ALIGN_MASK) &
-                  ~MALLOC_ALIGN_MASK);
-  for(;;) {
-    fprintf(stderr, "chunk %p size %10lx", p, (long)p->size);
-    if(p == top(heap->ar_ptr)) {
-      fprintf(stderr, " (top)\n");
-      break;
-    } else if(p->size == (0|PREV_INUSE)) {
-      fprintf(stderr, " (fence)\n");
-      break;
+  /* traverse regular bins */
+  for (i = 1; i < NBINS; ++i) {
+    b = bin_at(av, i);
+    for (p = last(b); p != b; p = p->bk) {
+      ++nblocks;
+      avail += chunksize(p);
     }
-    fprintf(stderr, "\n");
-    p = next_chunk(p);
   }
-}
-
-#endif
 
-
+  mi.smblks = nfastblocks;
+  mi.ordblks = nblocks;
+  mi.fordblks = avail;
+  mi.uordblks = av->system_mem - avail;
+  mi.arena = av->system_mem;
+  mi.hblks = mp_.n_mmaps;
+  mi.hblkhd = mp_.mmapped_mem;
+  mi.fsmblks = fastavail;
+  mi.keepcost = chunksize(av->top);
+  mi.usmblks = mp_.max_total_mem;
+  return mi;
+}
 
 /*
-
-  malloc_stats:
-
-    For all arenas separately and in total, prints on stderr the
-    amount of space obtained from the system, and the current number
-    of bytes allocated via malloc (or realloc, etc) but not yet
-    freed. (Note that this is the number of bytes allocated, not the
-    number requested. It will be larger than the number requested
-    because of alignment and bookkeeping overhead.)  When not compiled
-    for multiple threads, the maximum amount of allocated memory
-    (which may be more than current if malloc_trim and/or munmap got
-    called) is also reported.  When using mmap(), prints the maximum
-    number of simultaneous mmap regions used, too.
-
+  ------------------------------ malloc_stats ------------------------------
 */
 
-void mALLOC_STATs()
+void mSTATs()
 {
   int i;
-  arena *ar_ptr;
+  mstate ar_ptr;
   struct mallinfo mi;
-  unsigned int in_use_b = mmapped_mem, system_b = in_use_b;
+  unsigned int in_use_b = mp_.mmapped_mem, system_b = in_use_b;
 #if THREAD_STATS
   long stat_lock_direct = 0, stat_lock_loop = 0, stat_lock_wait = 0;
 #endif
 
-  for(i=0, ar_ptr = &main_arena;; i++) {
-    malloc_update_mallinfo(ar_ptr, &mi);
+  for (i=0, ar_ptr = &main_arena;; i++) {
+    (void)mutex_lock(&ar_ptr->mutex);
+    mi = mALLINFo(ar_ptr);
     fprintf(stderr, "Arena %d:\n", i);
     fprintf(stderr, "system bytes     = %10u\n", (unsigned int)mi.arena);
     fprintf(stderr, "in use bytes     = %10u\n", (unsigned int)mi.uordblks);
+#if MALLOC_DEBUG > 1
+    if (i > 0)
+      dump_heap(heap_for_ptr(top(ar_ptr)));
+#endif
     system_b += mi.arena;
     in_use_b += mi.uordblks;
 #if THREAD_STATS
@@ -4225,15 +5005,7 @@ void mALLOC_STATs()
     stat_lock_loop += ar_ptr->stat_lock_loop;
     stat_lock_wait += ar_ptr->stat_lock_wait;
 #endif
-#if USE_ARENAS && MALLOC_DEBUG > 1
-    if(ar_ptr != &main_arena) {
-      heap_info *heap;
-      (void)mutex_lock(&ar_ptr->mutex);
-      heap = heap_for_ptr(top(ar_ptr));
-      while(heap) { dump_heap(heap); heap = heap->prev; }
-      (void)mutex_unlock(&ar_ptr->mutex);
-    }
-#endif
+    (void)mutex_unlock(&ar_ptr->mutex);
     ar_ptr = ar_ptr->next;
     if(ar_ptr == &main_arena) break;
   }
@@ -4245,11 +5017,12 @@ void mALLOC_STATs()
   fprintf(stderr, "system bytes     = %10u\n", system_b);
   fprintf(stderr, "in use bytes     = %10u\n", in_use_b);
 #ifdef NO_THREADS
-  fprintf(stderr, "max system bytes = %10u\n", (unsigned int)max_total_mem);
+  fprintf(stderr, "max system bytes = %10u\n", (unsigned int)mp_.max_total_mem);
 #endif
 #if HAVE_MMAP
-  fprintf(stderr, "max mmap regions = %10u\n", (unsigned int)max_n_mmaps);
-  fprintf(stderr, "max mmap bytes   = %10lu\n", max_mmapped_mem);
+  fprintf(stderr, "max mmap regions = %10u\n", (unsigned int)mp_.max_n_mmaps);
+  fprintf(stderr, "max mmap bytes   = %10lu\n",
+	  (unsigned long)mp_.max_mmapped_mem);
 #endif
 #if THREAD_STATS
   fprintf(stderr, "heaps created    = %10d\n",  stat_n_heaps);
@@ -4261,39 +5034,9 @@ void mALLOC_STATs()
 #endif
 }
 
-/*
-  mallinfo returns a copy of updated current mallinfo.
-  The information reported is for the arena last used by the thread.
-*/
-
-struct mallinfo mALLINFo()
-{
-  struct mallinfo mi;
-  Void_t *vptr = NULL;
-
-#ifndef NO_THREADS
-  tsd_getspecific(arena_key, vptr);
-  if(vptr == ATFORK_ARENA_PTR)
-    vptr = (Void_t*)&main_arena;
-#endif
-  malloc_update_mallinfo((vptr ? (arena*)vptr : &main_arena), &mi);
-  return mi;
-}
-
-
-
 
 /*
-  mallopt:
-
-    mallopt is the general SVID/XPG interface to tunable parameters.
-    The format is to provide a (parameter-number, parameter-value) pair.
-    mallopt then sets the corresponding parameter to the argument
-    value if it can (i.e., so long as the value is meaningful),
-    and returns 1 if successful else 0.
-
-    See descriptions of tunable parameters above.
-
+  ------------------------------ mallopt ------------------------------
 */
 
 #if __STD_C
@@ -4302,597 +5045,199 @@ int mALLOPt(int param_number, int value)
 int mALLOPt(param_number, value) int param_number; int value;
 #endif
 {
-  switch(param_number)
-  {
-    case M_TRIM_THRESHOLD:
-      trim_threshold = value; return 1;
-    case M_TOP_PAD:
-      top_pad = value; return 1;
-    case M_MMAP_THRESHOLD:
-#if USE_ARENAS
-      /* Forbid setting the threshold too high. */
-      if((unsigned long)value > HEAP_MAX_SIZE/2) return 0;
-#endif
-      mmap_threshold = value; return 1;
-    case M_MMAP_MAX:
-#if HAVE_MMAP
-      n_mmaps_max = value; return 1;
-#else
-      if (value != 0) return 0; else  n_mmaps_max = value; return 1;
-#endif
-    case M_CHECK_ACTION:
-      check_action = value; return 1;
+  mstate av = &main_arena;
+  int res = 1;
 
-    default:
-      return 0;
-  }
-}
+  (void)mutex_lock(&av->mutex);
+  /* Ensure initialization/consolidation */
+  malloc_consolidate(av);
 
-
-
-/* Get/set state: malloc_get_state() records the current state of all
-   malloc variables (_except_ for the actual heap contents and `hook'
-   function pointers) in a system dependent, opaque data structure.
-   This data structure is dynamically allocated and can be free()d
-   after use.  malloc_set_state() restores the state of all malloc
-   variables to the previously obtained state.  This is especially
-   useful when using this malloc as part of a shared library, and when
-   the heap contents are saved/restored via some other method.  The
-   primary example for this is GNU Emacs with its `dumping' procedure.
-   `Hook' function pointers are never saved or restored by these
-   functions, with two exceptions: If malloc checking was in use when
-   malloc_get_state() was called, then malloc_set_state() calls
-   __malloc_check_init() if possible; if malloc checking was not in
-   use in the recorded state but the user requested malloc checking,
-   then the hooks are reset to 0.  */
-
-#define MALLOC_STATE_MAGIC   0x444c4541l
-#define MALLOC_STATE_VERSION (0*0x100l + 1l) /* major*0x100 + minor */
+  switch(param_number) {
+  case M_MXFAST:
+    if (value >= 0 && value <= MAX_FAST_SIZE) {
+      set_max_fast(av, value);
+    }
+    else
+      res = 0;
+    break;
 
-struct malloc_state {
-  long          magic;
-  long          version;
-  mbinptr       av[NAV * 2 + 2];
-  char*         sbrk_base;
-  int           sbrked_mem_bytes;
-  unsigned long trim_threshold;
-  unsigned long top_pad;
-  unsigned int  n_mmaps_max;
-  unsigned long mmap_threshold;
-  int           check_action;
-  unsigned long max_sbrked_mem;
-  unsigned long max_total_mem;
-  unsigned int  n_mmaps;
-  unsigned int  max_n_mmaps;
-  unsigned long mmapped_mem;
-  unsigned long max_mmapped_mem;
-  int           using_malloc_checking;
-};
+  case M_TRIM_THRESHOLD:
+    mp_.trim_threshold = value;
+    break;
 
-Void_t*
-mALLOC_GET_STATe()
-{
-  struct malloc_state* ms;
-  int i;
-  mbinptr b;
+  case M_TOP_PAD:
+    mp_.top_pad = value;
+    break;
 
-  ms = (struct malloc_state*)mALLOc(sizeof(*ms));
-  if (!ms)
-    return 0;
-  (void)mutex_lock(&main_arena.mutex);
-  ms->magic = MALLOC_STATE_MAGIC;
-  ms->version = MALLOC_STATE_VERSION;
-  ms->av[0] = main_arena.av[0];
-  ms->av[1] = main_arena.av[1];
-  for(i=0; i<NAV; i++) {
-    b = bin_at(&main_arena, i);
-    if(first(b) == b)
-      ms->av[2*i+2] = ms->av[2*i+3] = 0; /* empty bin (or initial top) */
-    else {
-      ms->av[2*i+2] = first(b);
-      ms->av[2*i+3] = last(b);
-    }
-  }
-  ms->sbrk_base = sbrk_base;
-  ms->sbrked_mem_bytes = sbrked_mem;
-  ms->trim_threshold = trim_threshold;
-  ms->top_pad = top_pad;
-  ms->n_mmaps_max = n_mmaps_max;
-  ms->mmap_threshold = mmap_threshold;
-  ms->check_action = check_action;
-  ms->max_sbrked_mem = max_sbrked_mem;
-#ifdef NO_THREADS
-  ms->max_total_mem = max_total_mem;
-#else
-  ms->max_total_mem = 0;
-#endif
-  ms->n_mmaps = n_mmaps;
-  ms->max_n_mmaps = max_n_mmaps;
-  ms->mmapped_mem = mmapped_mem;
-  ms->max_mmapped_mem = max_mmapped_mem;
-#if defined _LIBC || defined MALLOC_HOOKS
-  ms->using_malloc_checking = using_malloc_checking;
-#else
-  ms->using_malloc_checking = 0;
+  case M_MMAP_THRESHOLD:
+#if USE_ARENAS
+    /* Forbid setting the threshold too high. */
+    if((unsigned long)value > HEAP_MAX_SIZE/2)
+      res = 0;
+    else
 #endif
-  (void)mutex_unlock(&main_arena.mutex);
-  return (Void_t*)ms;
-}
+      mp_.mmap_threshold = value;
+    break;
 
-int
-#if __STD_C
-mALLOC_SET_STATe(Void_t* msptr)
-#else
-mALLOC_SET_STATe(msptr) Void_t* msptr;
+  case M_MMAP_MAX:
+#if !HAVE_MMAP
+    if (value != 0)
+      res = 0;
+    else
 #endif
-{
-  struct malloc_state* ms = (struct malloc_state*)msptr;
-  int i;
-  mbinptr b;
+      mp_.n_mmaps_max = value;
+    break;
 
-#if defined _LIBC || defined MALLOC_HOOKS
-  disallow_malloc_check = 1;
-#endif
-  ptmalloc_init();
-  if(ms->magic != MALLOC_STATE_MAGIC) return -1;
-  /* Must fail if the major version is too high. */
-  if((ms->version & ~0xffl) > (MALLOC_STATE_VERSION & ~0xffl)) return -2;
-  (void)mutex_lock(&main_arena.mutex);
-  main_arena.av[0] = ms->av[0];
-  main_arena.av[1] = ms->av[1];
-  for(i=0; i<NAV; i++) {
-    b = bin_at(&main_arena, i);
-    if(ms->av[2*i+2] == 0)
-      first(b) = last(b) = b;
-    else {
-      first(b) = ms->av[2*i+2];
-      last(b) = ms->av[2*i+3];
-      if(i > 0) {
-        /* Make sure the links to the `av'-bins in the heap are correct. */
-        first(b)->bk = b;
-        last(b)->fd = b;
-      }
-    }
-  }
-  sbrk_base = ms->sbrk_base;
-  sbrked_mem = ms->sbrked_mem_bytes;
-  trim_threshold = ms->trim_threshold;
-  top_pad = ms->top_pad;
-  n_mmaps_max = ms->n_mmaps_max;
-  mmap_threshold = ms->mmap_threshold;
-  check_action = ms->check_action;
-  max_sbrked_mem = ms->max_sbrked_mem;
-#ifdef NO_THREADS
-  max_total_mem = ms->max_total_mem;
-#endif
-  n_mmaps = ms->n_mmaps;
-  max_n_mmaps = ms->max_n_mmaps;
-  mmapped_mem = ms->mmapped_mem;
-  max_mmapped_mem = ms->max_mmapped_mem;
-  /* add version-dependent code here */
-  if (ms->version >= 1) {
-#if defined _LIBC || defined MALLOC_HOOKS
-    /* Check whether it is safe to enable malloc checking, or whether
-       it is necessary to disable it.  */
-    if (ms->using_malloc_checking && !using_malloc_checking &&
-        !disallow_malloc_check)
-      __malloc_check_init ();
-    else if (!ms->using_malloc_checking && using_malloc_checking) {
-      __malloc_hook = 0;
-      __free_hook = 0;
-      __realloc_hook = 0;
-      __memalign_hook = 0;
-      using_malloc_checking = 0;
-    }
-#endif
+  case M_CHECK_ACTION:
+    check_action = value;
+    break;
   }
-
-  (void)mutex_unlock(&main_arena.mutex);
-  return 0;
+  (void)mutex_unlock(&av->mutex);
+  return res;
 }
 
-
 
-#if defined _LIBC || defined MALLOC_HOOKS
+/* 
+  -------------------- Alternative MORECORE functions --------------------
+*/
 
-/* A simple, standard set of debugging hooks.  Overhead is `only' one
-   byte per chunk; still this will catch most cases of double frees or
-   overruns.  The goal here is to avoid obscure crashes due to invalid
-   usage, unlike in the MALLOC_DEBUG code. */
 
-#define MAGICBYTE(p) ( ( ((size_t)p >> 3) ^ ((size_t)p >> 11)) & 0xFF )
+/*
+  General Requirements for MORECORE.
 
-/* Instrument a chunk with overrun detector byte(s) and convert it
-   into a user pointer with requested size sz. */
+  The MORECORE function must have the following properties:
 
-static Void_t*
-internal_function
-#if __STD_C
-chunk2mem_check(mchunkptr p, size_t sz)
-#else
-chunk2mem_check(p, sz) mchunkptr p; size_t sz;
-#endif
-{
-  unsigned char* m_ptr = (unsigned char*)BOUNDED_N(chunk2mem(p), sz);
-  size_t i;
-
-  for(i = chunksize(p) - (chunk_is_mmapped(p) ? 2*SIZE_SZ+1 : SIZE_SZ+1);
-      i > sz;
-      i -= 0xFF) {
-    if(i-sz < 0x100) {
-      m_ptr[i] = (unsigned char)(i-sz);
-      break;
-    }
-    m_ptr[i] = 0xFF;
-  }
-  m_ptr[sz] = MAGICBYTE(p);
-  return (Void_t*)m_ptr;
-}
+  If MORECORE_CONTIGUOUS is false:
 
-/* Convert a pointer to be free()d or realloc()ed to a valid chunk
-   pointer.  If the provided pointer is not valid, return NULL. */
+    * MORECORE must allocate in multiples of pagesize. It will
+      only be called with arguments that are multiples of pagesize.
 
-static mchunkptr
-internal_function
-#if __STD_C
-mem2chunk_check(Void_t* mem)
-#else
-mem2chunk_check(mem) Void_t* mem;
-#endif
-{
-  mchunkptr p;
-  INTERNAL_SIZE_T sz, c;
-  unsigned char magic;
+    * MORECORE(0) must return an address that is at least 
+      MALLOC_ALIGNMENT aligned. (Page-aligning always suffices.)
 
-  p = mem2chunk(mem);
-  if(!aligned_OK(p)) return NULL;
-  if( (char*)p>=sbrk_base && (char*)p<(sbrk_base+sbrked_mem) ) {
-    /* Must be a chunk in conventional heap memory. */
-    if(chunk_is_mmapped(p) ||
-       ( (sz = chunksize(p)), ((char*)p + sz)>=(sbrk_base+sbrked_mem) ) ||
-       sz<MINSIZE || sz&MALLOC_ALIGN_MASK || !inuse(p) ||
-       ( !prev_inuse(p) && (p->prev_size&MALLOC_ALIGN_MASK ||
-                            (long)prev_chunk(p)<(long)sbrk_base ||
-                            next_chunk(prev_chunk(p))!=p) ))
-      return NULL;
-    magic = MAGICBYTE(p);
-    for(sz += SIZE_SZ-1; (c = ((unsigned char*)p)[sz]) != magic; sz -= c) {
-      if(c<=0 || sz<(c+2*SIZE_SZ)) return NULL;
-    }
-    ((unsigned char*)p)[sz] ^= 0xFF;
-  } else {
-    unsigned long offset, page_mask = malloc_getpagesize-1;
-
-    /* mmap()ed chunks have MALLOC_ALIGNMENT or higher power-of-two
-       alignment relative to the beginning of a page.  Check this
-       first. */
-    offset = (unsigned long)mem & page_mask;
-    if((offset!=MALLOC_ALIGNMENT && offset!=0 && offset!=0x10 &&
-        offset!=0x20 && offset!=0x40 && offset!=0x80 && offset!=0x100 &&
-        offset!=0x200 && offset!=0x400 && offset!=0x800 && offset!=0x1000 &&
-        offset<0x2000) ||
-       !chunk_is_mmapped(p) || (p->size & PREV_INUSE) ||
-       ( (((unsigned long)p - p->prev_size) & page_mask) != 0 ) ||
-       ( (sz = chunksize(p)), ((p->prev_size + sz) & page_mask) != 0 ) )
-      return NULL;
-    magic = MAGICBYTE(p);
-    for(sz -= 1; (c = ((unsigned char*)p)[sz]) != magic; sz -= c) {
-      if(c<=0 || sz<(c+2*SIZE_SZ)) return NULL;
-    }
-    ((unsigned char*)p)[sz] ^= 0xFF;
-  }
-  return p;
-}
+  else (i.e. If MORECORE_CONTIGUOUS is true):
 
-/* Check for corruption of the top chunk, and try to recover if
-   necessary. */
+    * Consecutive calls to MORECORE with positive arguments
+      return increasing addresses, indicating that space has been
+      contiguously extended.
 
-static int
-internal_function
-#if __STD_C
-top_check(void)
-#else
-top_check()
-#endif
-{
-  mchunkptr t = top(&main_arena);
-  char* brk, * new_brk;
-  INTERNAL_SIZE_T front_misalign, sbrk_size;
-  unsigned long pagesz = malloc_getpagesize;
-
-  if((char*)t + chunksize(t) == sbrk_base + sbrked_mem ||
-     t == initial_top(&main_arena)) return 0;
-
-  if(check_action & 1)
-    fprintf(stderr, "malloc: top chunk is corrupt\n");
-  if(check_action & 2)
-    abort();
-
-  /* Try to set up a new top chunk. */
-  brk = MORECORE(0);
-  front_misalign = (unsigned long)chunk2mem(brk) & MALLOC_ALIGN_MASK;
-  if (front_misalign > 0)
-    front_misalign = MALLOC_ALIGNMENT - front_misalign;
-  sbrk_size = front_misalign + top_pad + MINSIZE;
-  sbrk_size += pagesz - ((unsigned long)(brk + sbrk_size) & (pagesz - 1));
-  new_brk = (char*)(MORECORE (sbrk_size));
-  if (new_brk == (char*)(MORECORE_FAILURE)) return -1;
-  sbrked_mem = (new_brk - sbrk_base) + sbrk_size;
-
-  top(&main_arena) = (mchunkptr)(brk + front_misalign);
-  set_head(top(&main_arena), (sbrk_size - front_misalign) | PREV_INUSE);
+    * MORECORE need not allocate in multiples of pagesize.
+      Calls to MORECORE need not have args of multiples of pagesize.
 
-  return 0;
-}
+    * MORECORE need not page-align.
 
-static Void_t*
-#if __STD_C
-malloc_check(size_t sz, const Void_t *caller)
-#else
-malloc_check(sz, caller) size_t sz; const Void_t *caller;
-#endif
-{
-  mchunkptr victim;
-  INTERNAL_SIZE_T nb;
+  In either case:
 
-  if(request2size(sz+1, nb))
-    return 0;
-  (void)mutex_lock(&main_arena.mutex);
-  victim = (top_check() >= 0) ? chunk_alloc(&main_arena, nb) : NULL;
-  (void)mutex_unlock(&main_arena.mutex);
-  if(!victim) return NULL;
-  return chunk2mem_check(victim, sz);
-}
-
-static void
-#if __STD_C
-free_check(Void_t* mem, const Void_t *caller)
-#else
-free_check(mem, caller) Void_t* mem; const Void_t *caller;
-#endif
-{
-  mchunkptr p;
-
-  if(!mem) return;
-  (void)mutex_lock(&main_arena.mutex);
-  p = mem2chunk_check(mem);
-  if(!p) {
-    (void)mutex_unlock(&main_arena.mutex);
-    if(check_action & 1)
-      fprintf(stderr, "free(): invalid pointer %p!\n", mem);
-    if(check_action & 2)
-      abort();
-    return;
-  }
-#if HAVE_MMAP
-  if (chunk_is_mmapped(p)) {
-    (void)mutex_unlock(&main_arena.mutex);
-    munmap_chunk(p);
-    return;
-  }
-#endif
-#if 0 /* Erase freed memory. */
-  memset(mem, 0, chunksize(p) - (SIZE_SZ+1));
-#endif
-  chunk_free(&main_arena, p);
-  (void)mutex_unlock(&main_arena.mutex);
-}
+    * MORECORE may allocate more memory than requested. (Or even less,
+      but this will generally result in a malloc failure.)
 
-static Void_t*
-#if __STD_C
-realloc_check(Void_t* oldmem, size_t bytes, const Void_t *caller)
-#else
-realloc_check(oldmem, bytes, caller)
-     Void_t* oldmem; size_t bytes; const Void_t *caller;
-#endif
-{
-  mchunkptr oldp, newp;
-  INTERNAL_SIZE_T nb, oldsize;
+    * MORECORE must not allocate memory when given argument zero, but
+      instead return one past the end address of memory from previous
+      nonzero call. This malloc does NOT call MORECORE(0)
+      until at least one call with positive arguments is made, so
+      the initial value returned is not important.
 
-  if (oldmem == 0) return malloc_check(bytes, NULL);
-  (void)mutex_lock(&main_arena.mutex);
-  oldp = mem2chunk_check(oldmem);
-  if(!oldp) {
-    (void)mutex_unlock(&main_arena.mutex);
-    if(check_action & 1)
-      fprintf(stderr, "realloc(): invalid pointer %p!\n", oldmem);
-    if(check_action & 2)
-      abort();
-    return malloc_check(bytes, NULL);
-  }
-  oldsize = chunksize(oldp);
-
-  if(request2size(bytes+1, nb)) {
-    (void)mutex_unlock(&main_arena.mutex);
-    return 0;
-  }
-
-#if HAVE_MMAP
-  if (chunk_is_mmapped(oldp)) {
-#if HAVE_MREMAP
-    newp = mremap_chunk(oldp, nb);
-    if(!newp) {
-#endif
-      /* Note the extra SIZE_SZ overhead. */
-      if(oldsize - SIZE_SZ >= nb) newp = oldp; /* do nothing */
-      else {
-        /* Must alloc, copy, free. */
-        newp = (top_check() >= 0) ? chunk_alloc(&main_arena, nb) : NULL;
-        if (newp) {
-          MALLOC_COPY(BOUNDED_N(chunk2mem(newp), nb),
-		      oldmem, oldsize - 2*SIZE_SZ, 0);
-          munmap_chunk(oldp);
-        }
-      }
-#if HAVE_MREMAP
-    }
-#endif
-  } else {
-#endif /* HAVE_MMAP */
-    newp = (top_check() >= 0) ?
-      chunk_realloc(&main_arena, oldp, oldsize, nb) : NULL;
-#if 0 /* Erase freed memory. */
-    nb = chunksize(newp);
-    if(oldp<newp || oldp>=chunk_at_offset(newp, nb)) {
-      memset((char*)oldmem + 2*sizeof(mbinptr), 0,
-             oldsize - (2*sizeof(mbinptr)+2*SIZE_SZ+1));
-    } else if(nb > oldsize+SIZE_SZ) {
-      memset((char*)BOUNDED_N(chunk2mem(newp), bytes) + oldsize,
-	     0, nb - (oldsize+SIZE_SZ));
-    }
-#endif
-#if HAVE_MMAP
-  }
-#endif
-  (void)mutex_unlock(&main_arena.mutex);
+    * 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.
 
-  if(!newp) return NULL;
-  return chunk2mem_check(newp, bytes);
-}
+    * MORECORE need not handle negative arguments -- it may instead
+      just return MORECORE_FAILURE 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,
 
-static Void_t*
-#if __STD_C
-memalign_check(size_t alignment, size_t bytes, const Void_t *caller)
-#else
-memalign_check(alignment, bytes, caller)
-     size_t alignment; size_t bytes; const Void_t *caller;
-#endif
-{
-  INTERNAL_SIZE_T nb;
-  mchunkptr p;
+  There is some variation across systems about the type of the
+  argument to sbrk/MORECORE. If size_t is unsigned, then it cannot
+  actually be size_t, because sbrk supports negative args, so it is
+  normally the signed type of the same width as size_t (sometimes
+  declared as "intptr_t", and sometimes "ptrdiff_t").  It doesn't much
+  matter though. Internally, we use "long" as arguments, which should
+  work across all reasonable possibilities.
 
-  if (alignment <= MALLOC_ALIGNMENT) return malloc_check(bytes, NULL);
-  if (alignment <  MINSIZE) alignment = MINSIZE;
+  Additionally, if MORECORE ever returns failure for a positive
+  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 map noncontiguous space.
 
-  if(request2size(bytes+1, nb))
-    return 0;
-  (void)mutex_lock(&main_arena.mutex);
-  p = (top_check() >= 0) ? chunk_align(&main_arena, nb, alignment) : NULL;
-  (void)mutex_unlock(&main_arena.mutex);
-  if(!p) return NULL;
-  return chunk2mem_check(p, bytes);
-}
+  If you'd like mmap to ALWAYS be used, you can define MORECORE to be
+  a function that always returns MORECORE_FAILURE.
 
-#ifndef NO_THREADS
+  If you are using this malloc with something other than sbrk (or its
+  emulation) to supply memory regions, you probably want to set
+  MORECORE_CONTIGUOUS as false.  As an example, 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:
 
-/* The following hooks are used when the global initialization in
-   ptmalloc_init() hasn't completed yet. */
+      #define MORECORE osMoreCore
+      #define MORECORE_CONTIGUOUS 0
 
-static Void_t*
-#if __STD_C
-malloc_starter(size_t sz, const Void_t *caller)
-#else
-malloc_starter(sz, caller) size_t sz; const Void_t *caller;
-#endif
-{
-  INTERNAL_SIZE_T nb;
-  mchunkptr victim;
+  There is also a shutdown routine that should somehow be called for
+  cleanup upon program exit.
 
-  if(request2size(sz, nb))
-    return 0;
-  victim = chunk_alloc(&main_arena, nb);
+  #define MAX_POOL_ENTRIES 100
+  #define MINIMUM_MORECORE_SIZE  (64 * 1024)
+  static int next_os_pool;
+  void *our_os_pools[MAX_POOL_ENTRIES];
 
-  return victim ? BOUNDED_N(chunk2mem(victim), sz) : 0;
-}
-
-static void
-#if __STD_C
-free_starter(Void_t* mem, const Void_t *caller)
-#else
-free_starter(mem, caller) Void_t* mem; const Void_t *caller;
-#endif
-{
-  mchunkptr p;
-
-  if(!mem) return;
-  p = mem2chunk(mem);
-#if HAVE_MMAP
-  if (chunk_is_mmapped(p)) {
-    munmap_chunk(p);
-    return;
-  }
-#endif
-  chunk_free(&main_arena, p);
-}
-
-/* The following hooks are used while the `atfork' handling mechanism
-   is active. */
+  void *osMoreCore(int size)
+  {
+    void *ptr = 0;
+    static void *sbrk_top = 0;
 
-static Void_t*
-#if __STD_C
-malloc_atfork (size_t sz, const Void_t *caller)
-#else
-malloc_atfork(sz, caller) size_t sz; const Void_t *caller;
-#endif
-{
-  Void_t *vptr = NULL;
-  INTERNAL_SIZE_T nb;
-  mchunkptr victim;
-
-  tsd_getspecific(arena_key, vptr);
-  if(vptr == ATFORK_ARENA_PTR) {
-    /* We are the only thread that may allocate at all.  */
-    if(save_malloc_hook != malloc_check) {
-      if(request2size(sz, nb))
-        return 0;
-      victim = chunk_alloc(&main_arena, nb);
-      return victim ? BOUNDED_N(chunk2mem(victim), sz) : 0;
-    } else {
-      if(top_check()<0 || request2size(sz+1, nb))
-        return 0;
-      victim = chunk_alloc(&main_arena, nb);
-      return victim ? chunk2mem_check(victim, sz) : 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 *) MORECORE_FAILURE;
+      }
+      // save ptrs so they can be freed during cleanup
+      our_os_pools[next_os_pool] = ptr;
+      next_os_pool++;
+      ptr = (void *) ((((unsigned long) 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 *) MORECORE_FAILURE;
+    }
+    else
+    {
+      return sbrk_top;
     }
-  } else {
-    /* Suspend the thread until the `atfork' handlers have completed.
-       By that time, the hooks will have been reset as well, so that
-       mALLOc() can be used again. */
-    (void)mutex_lock(&list_lock);
-    (void)mutex_unlock(&list_lock);
-    return mALLOc(sz);
   }
-}
 
-static void
-#if __STD_C
-free_atfork(Void_t* mem, const Void_t *caller)
-#else
-free_atfork(mem, caller) Void_t* mem; const Void_t *caller;
-#endif
-{
-  Void_t *vptr = NULL;
-  arena *ar_ptr;
-  mchunkptr p;                          /* chunk corresponding to mem */
-
-  if (mem == 0)                              /* free(0) has no effect */
-    return;
-
-  p = mem2chunk(mem);         /* do not bother to replicate free_check here */
+  // cleanup any allocated memory pools
+  // called as last thing before shutting down driver
 
-#if HAVE_MMAP
-  if (chunk_is_mmapped(p))                       /* release mmapped memory. */
+  void osCleanupMem(void)
   {
-    munmap_chunk(p);
-    return;
-  }
-#endif
+    void **ptr;
 
-  ar_ptr = arena_for_ptr(p);
-  tsd_getspecific(arena_key, vptr);
-  if(vptr != ATFORK_ARENA_PTR)
-    (void)mutex_lock(&ar_ptr->mutex);
-  chunk_free(ar_ptr, p);
-  if(vptr != ATFORK_ARENA_PTR)
-    (void)mutex_unlock(&ar_ptr->mutex);
-}
-
-#endif /* !defined NO_THREADS */
+    for (ptr = our_os_pools; ptr < &our_os_pools[MAX_POOL_ENTRIES]; ptr++)
+      if (*ptr)
+      {
+         PoolDeallocate(*ptr);
+         *ptr = 0;
+      }
+  }
 
-#endif /* defined _LIBC || defined MALLOC_HOOKS */
+*/
 
-
 
 #ifdef _LIBC
+
 /* We need a wrapper function for one of the additions of POSIX.  */
 int
 __posix_memalign (void **memptr, size_t alignment, size_t size)
@@ -4906,11 +5251,10 @@ __posix_memalign (void **memptr, size_t alignment, size_t size)
 
   mem = __libc_memalign (alignment, size);
 
-  if (mem != NULL)
-    {
-      *memptr = mem;
-      return 0;
-    }
+  if (mem != NULL) {
+    *memptr = mem;
+    return 0;
+  }
 
   return ENOMEM;
 }
@@ -4932,111 +5276,17 @@ weak_alias (__malloc_usable_size, malloc_usable_size)
 weak_alias (__malloc_trim, malloc_trim)
 weak_alias (__malloc_get_state, malloc_get_state)
 weak_alias (__malloc_set_state, malloc_set_state)
-#endif
 
-/*
+#endif /* _LIBC */
 
+/* ------------------------------------------------------------
 History:
 
-    V2.6.4-pt3 Thu Feb 20 1997 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
-      * Added malloc_get/set_state() (mainly for use in GNU emacs),
-        using interface from Marcus Daniels
-      * All parameters are now adjustable via environment variables
-
-    V2.6.4-pt2 Sat Dec 14 1996 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
-      * Added debugging hooks
-      * Fixed possible deadlock in realloc() when out of memory
-      * Don't pollute namespace in glibc: use __getpagesize, __mmap, etc.
-
-    V2.6.4-pt Wed Dec  4 1996 Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
-      * Very minor updates from the released 2.6.4 version.
-      * Trimmed include file down to exported data structures.
-      * Changes from H.J. Lu for glibc-2.0.
-
-    V2.6.3i-pt Sep 16 1996  Wolfram Gloger (wmglo@dent.med.uni-muenchen.de)
-      * Many changes for multiple threads
-      * Introduced arenas and heaps
-
-    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 threshold
-      * Eliminate block-local decls to simplify tracing and debugging.
-      * Support another case of realloc via move into top
-      * Fix error occurring 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.)
+[see ftp://g.oswego.edu/pub/misc/malloc.c for the history of dlmalloc]
 
 */
+/*
+ * Local variables:
+ * c-basic-offset: 2
+ * End:
+ */
diff --git a/malloc/malloc.h b/malloc/malloc.h
index 8accd64699..aaa95b3a2c 100644
--- a/malloc/malloc.h
+++ b/malloc/malloc.h
@@ -20,24 +20,26 @@
 #ifndef _MALLOC_H
 #define _MALLOC_H 1
 
+#ifdef _LIBC
 #include <features.h>
+#endif
 
 /*
-  `ptmalloc', a malloc implementation for multiple threads without
-  lock contention, by Wolfram Gloger <wmglo@dent.med.uni-muenchen.de>.
-  See the files `ptmalloc.c' or `COPYRIGHT' for copying conditions.
+  $Id$
+  `ptmalloc2', a malloc implementation for multiple threads without
+  lock contention, by Wolfram Gloger <wg@malloc.de>.
 
-  VERSION 2.6.4-pt Wed Dec  4 00:35:54 MET 1996
+  VERSION 2.7.0
 
-  This work is mainly derived from malloc-2.6.4 by Doug Lea
+  This work is mainly derived from malloc-2.7.0 by Doug Lea
   <dl@cs.oswego.edu>, which is available from:
 
-                 ftp://g.oswego.edu/pub/misc/malloc.c
+                 ftp://gee.cs.oswego.edu/pub/misc/malloc.c
 
   This trimmed-down header file only provides function prototypes and
   the exported data structures.  For more detailed function
   descriptions and compile-time options, see the source file
-  `ptmalloc.c'.
+  `malloc.c'.
 */
 
 #if defined(__STDC__) || defined (__cplusplus)
@@ -112,11 +114,6 @@ extern "C" {
 #endif
 extern int __malloc_initialized;
 
-/* Initialize global configuration.  Not needed with GNU libc. */
-#ifndef __GLIBC__
-extern void ptmalloc_init __MALLOC_P ((void));
-#endif
-
 /* Allocate SIZE bytes of memory.  */
 extern __malloc_ptr_t malloc __MALLOC_P ((size_t __size)) __attribute_malloc__;
 
@@ -156,16 +153,17 @@ extern __malloc_ptr_t __default_morecore __MALLOC_P ((ptrdiff_t __size))
        __attribute_malloc__;
 
 /* SVID2/XPG mallinfo structure */
+
 struct mallinfo {
-  int arena;    /* total space allocated from system */
-  int ordblks;  /* number of non-inuse chunks */
-  int smblks;   /* unused -- always zero */
+  int arena;    /* non-mmapped space allocated from system */
+  int ordblks;  /* number of free chunks */
+  int smblks;   /* number of fastbin blocks */
   int hblks;    /* number of mmapped regions */
-  int hblkhd;   /* total space in mmapped regions */
-  int usmblks;  /* unused -- always zero */
-  int fsmblks;  /* unused -- always zero */
+  int hblkhd;   /* space in mmapped regions */
+  int usmblks;  /* maximum total allocated space */
+  int fsmblks;  /* space available in freed fastbin blocks */
   int uordblks; /* total allocated space */
-  int fordblks; /* total non-inuse space */
+  int fordblks; /* total free space */
   int keepcost; /* top-most, releasable (via malloc_trim) space */
 };
 
@@ -174,7 +172,7 @@ extern struct mallinfo mallinfo __MALLOC_P ((void));
 
 /* SVID2/XPG mallopt options */
 #ifndef M_MXFAST
-# define M_MXFAST  1	/* UNUSED in this malloc */
+# define M_MXFAST  1	/* maximum request size for "fastbins" */
 #endif
 #ifndef M_NLBLKS
 # define M_NLBLKS  2	/* UNUSED in this malloc */
@@ -214,7 +212,6 @@ extern __malloc_ptr_t malloc_get_state __MALLOC_P ((void));
    malloc_get_state(). */
 extern int malloc_set_state __MALLOC_P ((__malloc_ptr_t __ptr));
 
-#if defined __GLIBC__ || defined MALLOC_HOOKS
 /* Called once when malloc is initialized; redefining this variable in
    the application provides the preferred way to set up the hook
    pointers. */
@@ -234,7 +231,19 @@ extern void (*__after_morecore_hook) __MALLOC_PMT ((void));
 
 /* Activate a standard set of debugging hooks. */
 extern void __malloc_check_init __MALLOC_P ((void));
-#endif
+
+/* Internal routines, operating on "arenas".  */
+struct malloc_state;
+typedef struct malloc_state *mstate;
+
+extern mstate         _int_new_arena __MALLOC_P ((size_t __ini_size));
+extern __malloc_ptr_t _int_malloc __MALLOC_P ((mstate __m, size_t __size));
+extern void           _int_free __MALLOC_P ((mstate __m, __malloc_ptr_t __ptr));
+extern __malloc_ptr_t _int_realloc __MALLOC_P ((mstate __m,
+						__malloc_ptr_t __ptr,
+						size_t __size));
+extern __malloc_ptr_t _int_memalign __MALLOC_P ((mstate __m, size_t __alignment,
+						 size_t __size));
 
 #ifdef __cplusplus
 }; /* end of extern "C" */
diff --git a/malloc/tst-mallocstate.c b/malloc/tst-mallocstate.c
new file mode 100644
index 0000000000..97a10586d3
--- /dev/null
+++ b/malloc/tst-mallocstate.c
@@ -0,0 +1,82 @@
+/* Copyright (C) 2001 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+   Contributed by Wolfram Gloger <wg@malloc.de>, 2001.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, write to the Free
+   Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
+   02111-1307 USA.  */
+
+#include <errno.h>
+#include <stdio.h>
+#include "malloc.h"
+
+static int errors = 0;
+
+static void
+merror (const char *msg)
+{
+  ++errors;
+  printf ("Error: %s\n", msg);
+}
+
+int
+main (void)
+{
+  void *p1, *p2;
+  void *save_state;
+  long i;
+
+  errno = 0;
+
+  p1 = malloc (10);
+  if (p1 == NULL)
+    merror ("malloc (10) failed.");
+
+  p2 = malloc (20);
+  if (p2 == NULL)
+    merror ("malloc (20) failed.");
+
+  free (malloc (10));
+
+  for (i=0; i<100; ++i)
+    {
+      save_state = malloc_get_state ();
+      if (save_state == NULL)
+	{
+	  merror ("malloc_get_state () failed.");
+	  break;
+	}
+      /*free (malloc (10)); This could change the top chunk! */
+      malloc_set_state (save_state);
+      p1 = realloc (p1, i*4 + 4);
+      if (p1 == NULL)
+	merror ("realloc (i*4) failed.");
+      free (save_state);
+    }
+
+  p1 = realloc (p1, 40);
+  free (p2);
+  p2 = malloc (10);
+  if (p2 == NULL)
+    merror ("malloc (10) failed.");
+  free (p1);
+
+  return errors != 0;
+}
+
+/*
+ * Local variables:
+ * c-basic-offset: 2
+ * End:
+ */
diff --git a/stdlib/msort.c b/stdlib/msort.c
index 7d21c10fc9..3668370cd5 100644
--- a/stdlib/msort.c
+++ b/stdlib/msort.c
@@ -1,9 +1,7 @@
 /* An alternative to qsort, with an identical interface.
    This file is part of the GNU C Library.
-   Copyright (C) 1992, 1995-1997, 1999, 2000, 2001, 2002
-   Free Software Foundation, Inc.
-   Original Implementation by Mike Haertel, September 1988.
-   Towers of Hanoi Mergesort by Roger Sayle, January 2002.
+   Copyright (C) 1992, 1995-1997, 1999, 2000, 2001 Free Software Foundation, Inc.
+   Written by Mike Haertel, September 1988.
 
    The GNU C Library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Lesser General Public
@@ -21,372 +19,70 @@
    02111-1307 USA.  */
 
 #include <alloca.h>
-#include <limits.h>
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
 #include <memcopy.h>
 #include <errno.h>
 
-
-/* Check whether pointer P is aligned for access by type T. */
-#define TYPE_ALIGNED(P,T)  (((char *) (P) - (char *) 0) % __alignof__ (T) == 0)
-
-
-static int hanoi_sort (char *b, size_t n, size_t s,
-                        __compar_fn_t cmp, char *t);
-static int hanoi_sort_int (int *b, size_t n,
-                           __compar_fn_t cmp, int *t);
-#if INT_MAX != LONG_MAX
-static int hanoi_sort_long (long int *b, size_t n,
-                            __compar_fn_t cmp, long int *t);
-#endif
 static void msort_with_tmp (void *b, size_t n, size_t s,
-			    __compar_fn_t cmp, void *t);
-
-
-/* This routine implements "Towers of Hanoi Mergesort".  The algorithm
-   sorts the n elements of size s pointed to by array b using comparison
-   function cmp.  The argument t points to a suitable temporary buffer.
-   If the return value is zero, the sorted array is returned in b, and
-   for non-zero return values the sorted array is returned in t.  */
-static int
-hanoi_sort (char *b, size_t n, size_t s, __compar_fn_t cmp, char *t)
-{
-  size_t n1, n2;
-  char *b1,*b2;
-  char *t1,*t2;
-  char *s1,*s2;
-  size_t size;
-  int result;
-  char *ptr;
-
-  if (n <= 1)
-    return 0;
-
-  if (n == 2)
-    {
-      b2 = b + s;
-      if ((*cmp) (b, b2) <= 0)
-	return 0;
-      memcpy (__mempcpy (t, b2, s), b, s);
-      return 1;
-    }
-
-  n1 = n/2;
-  n2 = n - n1;
-  /* n1 < n2!  */
-
-  size = n1 * s;
-  b1 = b;
-  b2 = b + size;
-
-  t1 = t;
-  t2 = t + size;
-
-  /* Recursively call hanoi_sort to sort the two halves of the array.
-     Depending upon the return values, determine the values s1 and s2
-     the locations of the two sorted subarrays, ptr, the location to
-     contain the sorted array and result, the return value for this
-     function.  Note that "ptr = result? t : b".  */
-  if (hanoi_sort (b1, n1, s, cmp, t1))
-    {
-      if (hanoi_sort (b2, n2, s, cmp, t2))
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = b2;
-	}
-    }
-  else
-    {
-      if (hanoi_sort (b2, n2, s, cmp, t2))
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = b2;
-	}
-    }
+			    __compar_fn_t cmp, char *t);
 
-  /*  Merge the two sorted arrays s1 and s2 of n1 and n2 elements
-      respectively, placing the result in ptr.  On entry, n1 > 0
-      && n2 > 0, and with each iteration either n1 or n2 is decreased
-      until either reaches zero, and the loop terminates via return.  */
-  for (;;)
-    {
-      if ((*cmp) (s1, s2) <= 0)
-	{
-	  ptr = (char *) __mempcpy (ptr, s1, s);
-	  s1 += s;
-	  --n1;
-	  if (n1 == 0)
-            {
-              if (ptr != s2)
-                memcpy (ptr, s2, n2 * s);
-              return result;
-            }
-	}
-      else
-	{
-	  ptr = (char *) __mempcpy (ptr, s2, s);
-	  s2 += s;
-	  --n2;
-	  if (n2 == 0)
-	    {
-	      memcpy (ptr, s1, n1 * s);
-	      return result;
-	    }
-        }
-    }
-}
-
-
-/* This routine is a variant of hanoi_sort that is optimized for the
-   case where items to be sorted are the size of ints, and both b and
-   t are suitably aligned.  The parameter s in not needed as it is
-   known to be sizeof(int).  */
-static int
-hanoi_sort_int (int *b, size_t n, __compar_fn_t cmp, int *t)
-{
-  size_t n1, n2;
-  int *b1,*b2;
-  int *t1,*t2;
-  int *s1,*s2;
-  int result;
-  int *ptr;
-
-  if (n <= 1)
-    return 0;
-
-  if (n == 2)
-    {
-      if ((*cmp) (b, b + 1) <= 0)
-	return 0;
-      t[0] = b[1];
-      t[1] = b[0];
-      return 1;
-    }
-
-  n1 = n/2;
-  n2 = n - n1;
-  /* n1 < n2!  */
-
-  b1 = b;
-  b2 = b + n1;
-
-  t1 = t;
-  t2 = t + n1;
-
-  /* Recursively call hanoi_sort_int to sort the two halves.  */
-  if (hanoi_sort_int (b1, n1, cmp, t1))
-    {
-      if (hanoi_sort_int (b2, n2, cmp, t2))
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = b2;
-	}
-    }
-  else
-    {
-      if (hanoi_sort_int (b2, n2, cmp, t2))
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = b2;
-	}
-    }
-
-  /*  Merge n1 elements from s1 and n2 elements from s2 into ptr.  */
-  for (;;)
-    {
-      if ((*cmp) (s1, s2) <= 0)
-	{
-	  *ptr++ = *s1++;
-	  --n1;
-	  if (n1 == 0)
-            {
-              if (ptr != s2)
-                memcpy (ptr, s2, n2 * sizeof (int));
-              return result;
-            }
-	}
-      else
-	{
-	  *ptr++ = *s2++;
-	  --n2;
-	  if (n2 == 0)
-	    {
-	      memcpy (ptr, s1, n1 * sizeof (int));
-	      return result;
-	    }
-	}
-    }
-}
-
-
-#if INT_MAX != LONG_MAX
-/* This routine is a variant of hanoi_sort that is optimized for the
-   case where items to be sorted are the size of longs, and both b and
-   t are suitably aligned.  The parameter s in not needed as it is
-   known to be sizeof(long).  In case sizeof(int)== sizeof(long) we
-   do not need this code since it would be the same as hanoi_sort_int.  */
-static int
-hanoi_sort_long (long int *b, size_t n, __compar_fn_t cmp, long int *t)
+static void
+msort_with_tmp (void *b, size_t n, size_t s, __compar_fn_t cmp,
+		char *t)
 {
+  char *tmp;
+  char *b1, *b2;
   size_t n1, n2;
-  long int *b1,*b2;
-  long int *t1,*t2;
-  long int *s1,*s2;
-  int result;
-  long int *ptr;
 
   if (n <= 1)
-    return 0;
-
-  if (n == 2)
-    {
-      if ((*cmp) (b, b + 1) <= 0)
-	return 0;
-      t[0] = b[1];
-      t[1] = b[0];
-      return 1;
-    }
+    return;
 
-  n1 = n/2;
+  n1 = n / 2;
   n2 = n - n1;
-  /* n1 < n2!  */
-
   b1 = b;
-  b2 = b + n1;
-
-  t1 = t;
-  t2 = t + n1;
-
-  /* Recursively call hanoi_sort_long to sort the two halves.  */
-  if (hanoi_sort_long (b1, n1, cmp, t1))
-    {
-      if (hanoi_sort_long (b2, n2, cmp, t2))
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 0;
-	  ptr = b;
-	  s1 = t1;
-	  s2 = b2;
-	}
-    }
+  b2 = (char *) b + (n1 * s);
+
+  msort_with_tmp (b1, n1, s, cmp, t);
+  msort_with_tmp (b2, n2, s, cmp, t);
+
+  tmp = t;
+
+  if (s == OPSIZ && (b1 - (char *) 0) % OPSIZ == 0)
+    /* We are operating on aligned words.  Use direct word stores.  */
+    while (n1 > 0 && n2 > 0)
+      {
+	if ((*cmp) (b1, b2) <= 0)
+	  {
+	    --n1;
+	    *((op_t *) tmp)++ = *((op_t *) b1)++;
+	  }
+	else
+	  {
+	    --n2;
+	    *((op_t *) tmp)++ = *((op_t *) b2)++;
+	  }
+      }
   else
-    {
-      if (hanoi_sort_long (b2, n2, cmp, t2))
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = t2;
-	}
-      else
-	{
-	  result = 1;
-	  ptr = t;
-	  s1 = b1;
-	  s2 = b2;
-	}
-    }
-
-  /*  Merge n1 elements from s1 and n2 elements from s2 into ptr.  */
-  for (;;)
-    {
-      if ((*cmp) (s1, s2) <= 0)
-	{
-	  *ptr++ = *s1++;
-	  --n1;
-	  if (n1 == 0)
-            {
-              if (ptr != s2)
-                memcpy (ptr, s2, n2 * sizeof (long));
-              return result;
-            }
-	}
-      else
-	{
-	  *ptr++ = *s2++;
-	  --n2;
-	  if (n2 == 0)
-	    {
-	      memcpy (ptr, s1, n1 * sizeof (long));
-	      return result;
-	    }
-        }
-    }
-}
-#endif
-
-
-/* This routine preserves the original interface to msort_with_tmp and
-   determines which variant of hanoi_sort to call, based upon item size
-   and alignment.  */
-
-static void
-msort_with_tmp (void *b, size_t n, size_t s, __compar_fn_t cmp, void *t)
-{
-  const size_t size = n * s;
-
-  if (s == sizeof (int) && TYPE_ALIGNED (b, int))
-    {
-      if (hanoi_sort_int (b, n, cmp, t))
-        memcpy (b, t, size);
-    }
-#if INT_MAX != LONG_MAX
-  else if (s == sizeof (long int) && TYPE_ALIGNED (b, long int))
-    {
-      if (hanoi_sort_long (b, n, cmp, t))
-        memcpy (b, t, size);
-    }
-#endif
-  else
-    {
-      /* Call the generic implementation.  */
-      if (hanoi_sort (b, n, s, cmp, t))
-        memcpy (b, t, size);
-    }
+    while (n1 > 0 && n2 > 0)
+      {
+	if ((*cmp) (b1, b2) <= 0)
+	  {
+	    tmp = (char *) __mempcpy (tmp, b1, s);
+	    b1 += s;
+	    --n1;
+	  }
+	else
+	  {
+	    tmp = (char *) __mempcpy (tmp, b2, s);
+	    b2 += s;
+	    --n2;
+	  }
+      }
+  if (n1 > 0)
+    memcpy (tmp, b1, n1 * s);
+  memcpy (b, t, (n - n2) * s);
 }
 
 void
@@ -397,7 +93,7 @@ qsort (void *b, size_t n, size_t s, __compar_fn_t cmp)
   if (size < 1024)
     {
       void *buf = __alloca (size);
-
+      
       /* The temporary array is small, so put it on the stack.  */
       msort_with_tmp (b, n, s, cmp, buf);
     }
@@ -434,7 +130,7 @@ qsort (void *b, size_t n, size_t s, __compar_fn_t cmp)
 	   measured in bytes.  */
 
       /* If the memory requirements are too high don't allocate memory.  */
-      if ((long int) (size / pagesize) > phys_pages)
+      if (size / pagesize > phys_pages)
 	_quicksort (b, n, s, cmp);
       else
 	{
diff --git a/stdlib/qsort.c b/stdlib/qsort.c
index 512277ad5f..1ac268bec9 100644
--- a/stdlib/qsort.c
+++ b/stdlib/qsort.c
@@ -92,9 +92,6 @@ _quicksort (void *const pbase, size_t total_elems, size_t size,
 {
   register char *base_ptr = (char *) pbase;
 
-  /* Allocating SIZE bytes for a pivot buffer facilitates a better
-     algorithm below since we can do comparisons directly on the pivot. */
-  char *pivot_buffer = (char *) __alloca (size);
   const size_t max_thresh = MAX_THRESH * size;
 
   if (total_elems == 0)
@@ -113,8 +110,6 @@ _quicksort (void *const pbase, size_t total_elems, size_t size,
           char *left_ptr;
           char *right_ptr;
 
-	  char *pivot = pivot_buffer;
-
 	  /* Select median value from among LO, MID, and HI. Rearrange
 	     LO and HI so the three values are sorted. This lowers the
 	     probability of picking a pathological pivot value and
@@ -132,8 +127,6 @@ _quicksort (void *const pbase, size_t total_elems, size_t size,
 	  if ((*cmp) ((void *) mid, (void *) lo) < 0)
 	    SWAP (mid, lo, size);
 	jump_over:;
-	  memcpy (pivot, mid, size);
-	  pivot = pivot_buffer;
 
 	  left_ptr  = lo + size;
 	  right_ptr = hi - size;
@@ -143,15 +136,19 @@ _quicksort (void *const pbase, size_t total_elems, size_t size,
 	     that this algorithm runs much faster than others. */
 	  do
 	    {
-	      while ((*cmp) ((void *) left_ptr, (void *) pivot) < 0)
+	      while ((*cmp) ((void *) left_ptr, (void *) mid) < 0)
 		left_ptr += size;
 
-	      while ((*cmp) ((void *) pivot, (void *) right_ptr) < 0)
+	      while ((*cmp) ((void *) mid, (void *) right_ptr) < 0)
 		right_ptr -= size;
 
 	      if (left_ptr < right_ptr)
 		{
 		  SWAP (left_ptr, right_ptr, size);
+		  if (mid == left_ptr)
+		    mid = right_ptr;
+		  else if (mid == right_ptr)
+		    mid = left_ptr;
 		  left_ptr += size;
 		  right_ptr -= size;
 		}