xref: /redis-3.2.3/deps/jemalloc/doc/jemalloc.html (revision 5268379e)
1<html><head><meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1"><title>JEMALLOC</title><meta name="generator" content="DocBook XSL Stylesheets V1.78.1"></head><body bgcolor="white" text="black" link="#0000FF" vlink="#840084" alink="#0000FF"><div class="refentry"><a name="idp45223136"></a><div class="titlepage"></div><div class="refnamediv"><h2>Name</h2><p>jemalloc &#8212; general purpose memory allocation functions</p></div><div class="refsect1"><a name="library"></a><h2>LIBRARY</h2><p>This manual describes jemalloc 4.0.3-0-ge9192eacf8935e29fc62fddc2701f7942b1cc02c.  More information
2    can be found at the <a class="ulink" href="http://www.canonware.com/jemalloc/" target="_top">jemalloc website</a>.</p></div><div class="refsynopsisdiv"><h2>SYNOPSIS</h2><div class="funcsynopsis"><pre class="funcsynopsisinfo">#include &lt;<code class="filename">jemalloc/jemalloc.h</code>&gt;</pre><div class="refsect2"><a name="idp44244480"></a><h3>Standard API</h3><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">malloc</b>(</code></td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">calloc</b>(</code></td><td>size_t <var class="pdparam">number</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">int <b class="fsfunc">posix_memalign</b>(</code></td><td>void **<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">alignment</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">aligned_alloc</b>(</code></td><td>size_t <var class="pdparam">alignment</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">realloc</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void <b class="fsfunc">free</b>(</code></td><td>void *<var class="pdparam">ptr</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="refsect2"><a name="idp46062768"></a><h3>Non-standard API</h3><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">mallocx</b>(</code></td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void *<b class="fsfunc">rallocx</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">size_t <b class="fsfunc">xallocx</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">extra</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">size_t <b class="fsfunc">sallocx</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void <b class="fsfunc">dallocx</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void <b class="fsfunc">sdallocx</b>(</code></td><td>void *<var class="pdparam">ptr</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">size_t <b class="fsfunc">nallocx</b>(</code></td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>int <var class="pdparam">flags</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">int <b class="fsfunc">mallctl</b>(</code></td><td>const char *<var class="pdparam">name</var>, </td></tr><tr><td>�</td><td>void *<var class="pdparam">oldp</var>, </td></tr><tr><td>�</td><td>size_t *<var class="pdparam">oldlenp</var>, </td></tr><tr><td>�</td><td>void *<var class="pdparam">newp</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">newlen</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">int <b class="fsfunc">mallctlnametomib</b>(</code></td><td>const char *<var class="pdparam">name</var>, </td></tr><tr><td>�</td><td>size_t *<var class="pdparam">mibp</var>, </td></tr><tr><td>�</td><td>size_t *<var class="pdparam">miblenp</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">int <b class="fsfunc">mallctlbymib</b>(</code></td><td>const size_t *<var class="pdparam">mib</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">miblen</var>, </td></tr><tr><td>�</td><td>void *<var class="pdparam">oldp</var>, </td></tr><tr><td>�</td><td>size_t *<var class="pdparam">oldlenp</var>, </td></tr><tr><td>�</td><td>void *<var class="pdparam">newp</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">newlen</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void <b class="fsfunc">malloc_stats_print</b>(</code></td><td>void <var class="pdparam">(*write_cb)</var>
3            <code>(</code>void *, const char *<code>)</code>
4          , </td></tr><tr><td>�</td><td>void *<var class="pdparam">cbopaque</var>, </td></tr><tr><td>�</td><td>const char *<var class="pdparam">opts</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">size_t <b class="fsfunc">malloc_usable_size</b>(</code></td><td>const void *<var class="pdparam">ptr</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">void <b class="fsfunc">(*malloc_message)</b>(</code></td><td>void *<var class="pdparam">cbopaque</var>, </td></tr><tr><td>�</td><td>const char *<var class="pdparam">s</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div><p><span class="type">const char *</span><code class="varname">malloc_conf</code>;</p></div></div></div><div class="refsect1"><a name="description"></a><h2>DESCRIPTION</h2><div class="refsect2"><a name="idp46115952"></a><h3>Standard API</h3><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>) function allocates
5      <em class="parameter"><code>size</code></em> bytes of uninitialized memory.  The allocated
6      space is suitably aligned (after possible pointer coercion) for storage
7      of any type of object.</p><p>The <code class="function">calloc</code>(<em class="parameter"><code></code></em>) function allocates
8      space for <em class="parameter"><code>number</code></em> objects, each
9      <em class="parameter"><code>size</code></em> bytes in length.  The result is identical to
10      calling <code class="function">malloc</code>(<em class="parameter"><code></code></em>) with an argument of
11      <em class="parameter"><code>number</code></em> * <em class="parameter"><code>size</code></em>, with the
12      exception that the allocated memory is explicitly initialized to zero
13      bytes.</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function
14      allocates <em class="parameter"><code>size</code></em> bytes of memory such that the
15      allocation's base address is a multiple of
16      <em class="parameter"><code>alignment</code></em>, and returns the allocation in the value
17      pointed to by <em class="parameter"><code>ptr</code></em>.  The requested
18      <em class="parameter"><code>alignment</code></em> must be a power of 2 at least as large as
19      <code class="code">sizeof(<span class="type">void *</span>)</code>.</p><p>The <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function
20      allocates <em class="parameter"><code>size</code></em> bytes of memory such that the
21      allocation's base address is a multiple of
22      <em class="parameter"><code>alignment</code></em>.  The requested
23      <em class="parameter"><code>alignment</code></em> must be a power of 2.  Behavior is
24      undefined if <em class="parameter"><code>size</code></em> is not an integral multiple of
25      <em class="parameter"><code>alignment</code></em>.</p><p>The <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function changes the
26      size of the previously allocated memory referenced by
27      <em class="parameter"><code>ptr</code></em> to <em class="parameter"><code>size</code></em> bytes.  The
28      contents of the memory are unchanged up to the lesser of the new and old
29      sizes.  If the new size is larger, the contents of the newly allocated
30      portion of the memory are undefined.  Upon success, the memory referenced
31      by <em class="parameter"><code>ptr</code></em> is freed and a pointer to the newly
32      allocated memory is returned.  Note that
33      <code class="function">realloc</code>(<em class="parameter"><code></code></em>) may move the memory allocation,
34      resulting in a different return value than <em class="parameter"><code>ptr</code></em>.
35      If <em class="parameter"><code>ptr</code></em> is <code class="constant">NULL</code>, the
36      <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function behaves identically to
37      <code class="function">malloc</code>(<em class="parameter"><code></code></em>) for the specified size.</p><p>The <code class="function">free</code>(<em class="parameter"><code></code></em>) function causes the
38      allocated memory referenced by <em class="parameter"><code>ptr</code></em> to be made
39      available for future allocations.  If <em class="parameter"><code>ptr</code></em> is
40      <code class="constant">NULL</code>, no action occurs.</p></div><div class="refsect2"><a name="idp46144704"></a><h3>Non-standard API</h3><p>The <code class="function">mallocx</code>(<em class="parameter"><code></code></em>),
41      <code class="function">rallocx</code>(<em class="parameter"><code></code></em>),
42      <code class="function">xallocx</code>(<em class="parameter"><code></code></em>),
43      <code class="function">sallocx</code>(<em class="parameter"><code></code></em>),
44      <code class="function">dallocx</code>(<em class="parameter"><code></code></em>),
45      <code class="function">sdallocx</code>(<em class="parameter"><code></code></em>), and
46      <code class="function">nallocx</code>(<em class="parameter"><code></code></em>) functions all have a
47      <em class="parameter"><code>flags</code></em> argument that can be used to specify
48      options.  The functions only check the options that are contextually
49      relevant.  Use bitwise or (<code class="code">|</code>) operations to
50      specify one or more of the following:
51        </p><div class="variablelist"><dl class="variablelist"><dt><a name="MALLOCX_LG_ALIGN"></a><span class="term"><code class="constant">MALLOCX_LG_ALIGN(<em class="parameter"><code>la</code></em>)
52            </code></span></dt><dd><p>Align the memory allocation to start at an address
53            that is a multiple of <code class="code">(1 &lt;&lt;
54            <em class="parameter"><code>la</code></em>)</code>.  This macro does not validate
55            that <em class="parameter"><code>la</code></em> is within the valid
56            range.</p></dd><dt><a name="MALLOCX_ALIGN"></a><span class="term"><code class="constant">MALLOCX_ALIGN(<em class="parameter"><code>a</code></em>)
57            </code></span></dt><dd><p>Align the memory allocation to start at an address
58            that is a multiple of <em class="parameter"><code>a</code></em>, where
59            <em class="parameter"><code>a</code></em> is a power of two.  This macro does not
60            validate that <em class="parameter"><code>a</code></em> is a power of 2.
61            </p></dd><dt><a name="MALLOCX_ZERO"></a><span class="term"><code class="constant">MALLOCX_ZERO</code></span></dt><dd><p>Initialize newly allocated memory to contain zero
62            bytes.  In the growing reallocation case, the real size prior to
63            reallocation defines the boundary between untouched bytes and those
64            that are initialized to contain zero bytes.  If this macro is
65            absent, newly allocated memory is uninitialized.</p></dd><dt><a name="MALLOCX_TCACHE"></a><span class="term"><code class="constant">MALLOCX_TCACHE(<em class="parameter"><code>tc</code></em>)
66            </code></span></dt><dd><p>Use the thread-specific cache (tcache) specified by
67            the identifier <em class="parameter"><code>tc</code></em>, which must have been
68            acquired via the <a class="link" href="#tcache.create">
69    "<code class="mallctl">tcache.create</code>"
70  </a>
71            mallctl.  This macro does not validate that
72            <em class="parameter"><code>tc</code></em> specifies a valid
73            identifier.</p></dd><dt><a name="MALLOC_TCACHE_NONE"></a><span class="term"><code class="constant">MALLOCX_TCACHE_NONE</code></span></dt><dd><p>Do not use a thread-specific cache (tcache).  Unless
74            <code class="constant">MALLOCX_TCACHE(<em class="parameter"><code>tc</code></em>)</code> or
75            <code class="constant">MALLOCX_TCACHE_NONE</code> is specified, an
76            automatically managed tcache will be used under many circumstances.
77            This macro cannot be used in the same <em class="parameter"><code>flags</code></em>
78            argument as
79            <code class="constant">MALLOCX_TCACHE(<em class="parameter"><code>tc</code></em>)</code>.</p></dd><dt><a name="MALLOCX_ARENA"></a><span class="term"><code class="constant">MALLOCX_ARENA(<em class="parameter"><code>a</code></em>)
80            </code></span></dt><dd><p>Use the arena specified by the index
81            <em class="parameter"><code>a</code></em>.  This macro has no effect for regions that
82            were allocated via an arena other than the one specified.  This
83            macro does not validate that <em class="parameter"><code>a</code></em> specifies an
84            arena index in the valid range.</p></dd></dl></div><p>
85      </p><p>The <code class="function">mallocx</code>(<em class="parameter"><code></code></em>) function allocates at
86      least <em class="parameter"><code>size</code></em> bytes of memory, and returns a pointer
87      to the base address of the allocation.  Behavior is undefined if
88      <em class="parameter"><code>size</code></em> is <code class="constant">0</code>, or if request size
89      overflows due to size class and/or alignment constraints.</p><p>The <code class="function">rallocx</code>(<em class="parameter"><code></code></em>) function resizes the
90      allocation at <em class="parameter"><code>ptr</code></em> to be at least
91      <em class="parameter"><code>size</code></em> bytes, and returns a pointer to the base
92      address of the resulting allocation, which may or may not have moved from
93      its original location.  Behavior is undefined if
94      <em class="parameter"><code>size</code></em> is <code class="constant">0</code>, or if request size
95      overflows due to size class and/or alignment constraints.</p><p>The <code class="function">xallocx</code>(<em class="parameter"><code></code></em>) function resizes the
96      allocation at <em class="parameter"><code>ptr</code></em> in place to be at least
97      <em class="parameter"><code>size</code></em> bytes, and returns the real size of the
98      allocation.  If <em class="parameter"><code>extra</code></em> is non-zero, an attempt is
99      made to resize the allocation to be at least <code class="code">(<em class="parameter"><code>size</code></em> +
100      <em class="parameter"><code>extra</code></em>)</code> bytes, though inability to allocate
101      the extra byte(s) will not by itself result in failure to resize.
102      Behavior is undefined if <em class="parameter"><code>size</code></em> is
103      <code class="constant">0</code>, or if <code class="code">(<em class="parameter"><code>size</code></em> + <em class="parameter"><code>extra</code></em>
104      &gt; <code class="constant">SIZE_T_MAX</code>)</code>.</p><p>The <code class="function">sallocx</code>(<em class="parameter"><code></code></em>) function returns the
105      real size of the allocation at <em class="parameter"><code>ptr</code></em>.</p><p>The <code class="function">dallocx</code>(<em class="parameter"><code></code></em>) function causes the
106      memory referenced by <em class="parameter"><code>ptr</code></em> to be made available for
107      future allocations.</p><p>The <code class="function">sdallocx</code>(<em class="parameter"><code></code></em>) function is an
108      extension of <code class="function">dallocx</code>(<em class="parameter"><code></code></em>) with a
109      <em class="parameter"><code>size</code></em> parameter to allow the caller to pass in the
110      allocation size as an optimization.  The minimum valid input size is the
111      original requested size of the allocation, and the maximum valid input
112      size is the corresponding value returned by
113      <code class="function">nallocx</code>(<em class="parameter"><code></code></em>) or
114      <code class="function">sallocx</code>(<em class="parameter"><code></code></em>).</p><p>The <code class="function">nallocx</code>(<em class="parameter"><code></code></em>) function allocates no
115      memory, but it performs the same size computation as the
116      <code class="function">mallocx</code>(<em class="parameter"><code></code></em>) function, and returns the real
117      size of the allocation that would result from the equivalent
118      <code class="function">mallocx</code>(<em class="parameter"><code></code></em>) function call.  Behavior is
119      undefined if <em class="parameter"><code>size</code></em> is <code class="constant">0</code>, or if
120      request size overflows due to size class and/or alignment
121      constraints.</p><p>The <code class="function">mallctl</code>(<em class="parameter"><code></code></em>) function provides a
122      general interface for introspecting the memory allocator, as well as
123      setting modifiable parameters and triggering actions.  The
124      period-separated <em class="parameter"><code>name</code></em> argument specifies a
125      location in a tree-structured namespace; see the <a class="xref" href="#mallctl_namespace" title="MALLCTL NAMESPACE">MALLCTL NAMESPACE</a> section for
126      documentation on the tree contents.  To read a value, pass a pointer via
127      <em class="parameter"><code>oldp</code></em> to adequate space to contain the value, and a
128      pointer to its length via <em class="parameter"><code>oldlenp</code></em>; otherwise pass
129      <code class="constant">NULL</code> and <code class="constant">NULL</code>.  Similarly, to
130      write a value, pass a pointer to the value via
131      <em class="parameter"><code>newp</code></em>, and its length via
132      <em class="parameter"><code>newlen</code></em>; otherwise pass <code class="constant">NULL</code>
133      and <code class="constant">0</code>.</p><p>The <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>) function
134      provides a way to avoid repeated name lookups for applications that
135      repeatedly query the same portion of the namespace, by translating a name
136      to a &#8220;Management Information Base&#8221; (MIB) that can be passed
137      repeatedly to <code class="function">mallctlbymib</code>(<em class="parameter"><code></code></em>).  Upon
138      successful return from <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>),
139      <em class="parameter"><code>mibp</code></em> contains an array of
140      <em class="parameter"><code>*miblenp</code></em> integers, where
141      <em class="parameter"><code>*miblenp</code></em> is the lesser of the number of components
142      in <em class="parameter"><code>name</code></em> and the input value of
143      <em class="parameter"><code>*miblenp</code></em>.  Thus it is possible to pass a
144      <em class="parameter"><code>*miblenp</code></em> that is smaller than the number of
145      period-separated name components, which results in a partial MIB that can
146      be used as the basis for constructing a complete MIB.  For name
147      components that are integers (e.g. the 2 in
148      <a class="link" href="#arenas.bin.i.size">
149    "<code class="mallctl">arenas.bin.2.size</code>"
150  </a>),
151      the corresponding MIB component will always be that integer.  Therefore,
152      it is legitimate to construct code like the following: </p><pre class="programlisting">
153unsigned nbins, i;
154size_t mib[4];
155size_t len, miblen;
156
157len = sizeof(nbins);
158mallctl("arenas.nbins", &amp;nbins, &amp;len, NULL, 0);
159
160miblen = 4;
161mallctlnametomib("arenas.bin.0.size", mib, &amp;miblen);
162for (i = 0; i &lt; nbins; i++) {
163	size_t bin_size;
164
165	mib[2] = i;
166	len = sizeof(bin_size);
167	mallctlbymib(mib, miblen, &amp;bin_size, &amp;len, NULL, 0);
168	/* Do something with bin_size... */
169}</pre><p>The <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>) function
170      writes human-readable summary statistics via the
171      <em class="parameter"><code>write_cb</code></em> callback function pointer and
172      <em class="parameter"><code>cbopaque</code></em> data passed to
173      <em class="parameter"><code>write_cb</code></em>, or
174      <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) if
175      <em class="parameter"><code>write_cb</code></em> is <code class="constant">NULL</code>.  This
176      function can be called repeatedly.  General information that never
177      changes during execution can be omitted by specifying "g" as a character
178      within the <em class="parameter"><code>opts</code></em> string.  Note that
179      <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) uses the
180      <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions internally, so
181      inconsistent statistics can be reported if multiple threads use these
182      functions simultaneously.  If <code class="option">--enable-stats</code> is
183      specified during configuration, &#8220;m&#8221; and &#8220;a&#8221; can
184      be specified to omit merged arena and per arena statistics, respectively;
185      &#8220;b&#8221;, &#8220;l&#8221;, and &#8220;h&#8221; can be specified to
186      omit per size class statistics for bins, large objects, and huge objects,
187      respectively.  Unrecognized characters are silently ignored.  Note that
188      thread caching may prevent some statistics from being completely up to
189      date, since extra locking would be required to merge counters that track
190      thread cache operations.
191      </p><p>The <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function
192      returns the usable size of the allocation pointed to by
193      <em class="parameter"><code>ptr</code></em>.  The return value may be larger than the size
194      that was requested during allocation.  The
195      <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function is not a
196      mechanism for in-place <code class="function">realloc</code>(<em class="parameter"><code></code></em>); rather
197      it is provided solely as a tool for introspection purposes.  Any
198      discrepancy between the requested allocation size and the size reported
199      by <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) should not be
200      depended on, since such behavior is entirely implementation-dependent.
201      </p></div></div><div class="refsect1"><a name="tuning"></a><h2>TUNING</h2><p>Once, when the first call is made to one of the memory allocation
202    routines, the allocator initializes its internals based in part on various
203    options that can be specified at compile- or run-time.</p><p>The string pointed to by the global variable
204    <code class="varname">malloc_conf</code>, the &#8220;name&#8221; of the file
205    referenced by the symbolic link named <code class="filename">/etc/malloc.conf</code>, and the value of the
206    environment variable <code class="envar">MALLOC_CONF</code>, will be interpreted, in
207    that order, from left to right as options.  Note that
208    <code class="varname">malloc_conf</code> may be read before
209    <code class="function">main</code>(<em class="parameter"><code></code></em>) is entered, so the declaration of
210    <code class="varname">malloc_conf</code> should specify an initializer that contains
211    the final value to be read by jemalloc.  <code class="varname">malloc_conf</code> is
212    a compile-time setting, whereas <code class="filename">/etc/malloc.conf</code> and <code class="envar">MALLOC_CONF</code>
213    can be safely set any time prior to program invocation.</p><p>An options string is a comma-separated list of option:value pairs.
214    There is one key corresponding to each <a class="link" href="#opt.abort">
215    "<code class="mallctl">opt.*</code>"
216  </a> mallctl (see the <a class="xref" href="#mallctl_namespace" title="MALLCTL NAMESPACE">MALLCTL NAMESPACE</a> section for options
217    documentation).  For example, <code class="literal">abort:true,narenas:1</code> sets
218    the <a class="link" href="#opt.abort">
219    "<code class="mallctl">opt.abort</code>"
220  </a> and <a class="link" href="#opt.narenas">
221    "<code class="mallctl">opt.narenas</code>"
222  </a> options.  Some
223    options have boolean values (true/false), others have integer values (base
224    8, 10, or 16, depending on prefix), and yet others have raw string
225    values.</p></div><div class="refsect1"><a name="implementation_notes"></a><h2>IMPLEMENTATION NOTES</h2><p>Traditionally, allocators have used
226    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span> to obtain memory, which is
227    suboptimal for several reasons, including race conditions, increased
228    fragmentation, and artificial limitations on maximum usable memory.  If
229    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span> is supported by the operating
230    system, this allocator uses both
231    <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> and
232    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>, in that order of preference;
233    otherwise only <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> is used.</p><p>This allocator uses multiple arenas in order to reduce lock
234    contention for threaded programs on multi-processor systems.  This works
235    well with regard to threading scalability, but incurs some costs.  There is
236    a small fixed per-arena overhead, and additionally, arenas manage memory
237    completely independently of each other, which means a small fixed increase
238    in overall memory fragmentation.  These overheads are not generally an
239    issue, given the number of arenas normally used.  Note that using
240    substantially more arenas than the default is not likely to improve
241    performance, mainly due to reduced cache performance.  However, it may make
242    sense to reduce the number of arenas if an application does not make much
243    use of the allocation functions.</p><p>In addition to multiple arenas, unless
244    <code class="option">--disable-tcache</code> is specified during configuration, this
245    allocator supports thread-specific caching for small and large objects, in
246    order to make it possible to completely avoid synchronization for most
247    allocation requests.  Such caching allows very fast allocation in the
248    common case, but it increases memory usage and fragmentation, since a
249    bounded number of objects can remain allocated in each thread cache.</p><p>Memory is conceptually broken into equal-sized chunks, where the
250    chunk size is a power of two that is greater than the page size.  Chunks
251    are always aligned to multiples of the chunk size.  This alignment makes it
252    possible to find metadata for user objects very quickly.</p><p>User objects are broken into three categories according to size:
253    small, large, and huge.  Small and large objects are managed entirely by
254    arenas; huge objects are additionally aggregated in a single data structure
255    that is shared by all threads.  Huge objects are typically used by
256    applications infrequently enough that this single data structure is not a
257    scalability issue.</p><p>Each chunk that is managed by an arena tracks its contents as runs of
258    contiguous pages (unused, backing a set of small objects, or backing one
259    large object).  The combination of chunk alignment and chunk page maps
260    makes it possible to determine all metadata regarding small and large
261    allocations in constant time.</p><p>Small objects are managed in groups by page runs.  Each run maintains
262    a bitmap to track which regions are in use.  Allocation requests that are no
263    more than half the quantum (8 or 16, depending on architecture) are rounded
264    up to the nearest power of two that is at least <code class="code">sizeof(<span class="type">double</span>)</code>.  All other object size
265    classes are multiples of the quantum, spaced such that there are four size
266    classes for each doubling in size, which limits internal fragmentation to
267    approximately 20% for all but the smallest size classes.  Small size classes
268    are smaller than four times the page size, large size classes are smaller
269    than the chunk size (see the <a class="link" href="#opt.lg_chunk">
270    "<code class="mallctl">opt.lg_chunk</code>"
271  </a> option), and
272    huge size classes extend from the chunk size up to one size class less than
273    the full address space size.</p><p>Allocations are packed tightly together, which can be an issue for
274    multi-threaded applications.  If you need to assure that allocations do not
275    suffer from cacheline sharing, round your allocation requests up to the
276    nearest multiple of the cacheline size, or specify cacheline alignment when
277    allocating.</p><p>The <code class="function">realloc</code>(<em class="parameter"><code></code></em>),
278    <code class="function">rallocx</code>(<em class="parameter"><code></code></em>), and
279    <code class="function">xallocx</code>(<em class="parameter"><code></code></em>) functions may resize allocations
280    without moving them under limited circumstances.  Unlike the
281    <code class="function">*allocx</code>(<em class="parameter"><code></code></em>) API, the standard API does not
282    officially round up the usable size of an allocation to the nearest size
283    class, so technically it is necessary to call
284    <code class="function">realloc</code>(<em class="parameter"><code></code></em>) to grow e.g. a 9-byte allocation to
285    16 bytes, or shrink a 16-byte allocation to 9 bytes.  Growth and shrinkage
286    trivially succeeds in place as long as the pre-size and post-size both round
287    up to the same size class.  No other API guarantees are made regarding
288    in-place resizing, but the current implementation also tries to resize large
289    and huge allocations in place, as long as the pre-size and post-size are
290    both large or both huge.  In such cases shrinkage always succeeds for large
291    size classes, but for huge size classes the chunk allocator must support
292    splitting (see <a class="link" href="#arena.i.chunk_hooks">
293    "<code class="mallctl">arena.&lt;i&gt;.chunk_hooks</code>"
294  </a>).
295    Growth only succeeds if the trailing memory is currently available, and
296    additionally for huge size classes the chunk allocator must support
297    merging.</p><p>Assuming 2 MiB chunks, 4 KiB pages, and a 16-byte quantum on a
298    64-bit system, the size classes in each category are as shown in <a class="xref" href="#size_classes" title="Table�1.�Size classes">Table 1</a>.</p><div class="table"><a name="size_classes"></a><p class="title"><b>Table�1.�Size classes</b></p><div class="table-contents"><table summary="Size classes" border="1"><colgroup><col align="left" class="c1"><col align="right" class="c2"><col align="left" class="c3"></colgroup><thead><tr><th align="left">Category</th><th align="right">Spacing</th><th align="left">Size</th></tr></thead><tbody><tr><td rowspan="9" align="left">Small</td><td align="right">lg</td><td align="left">[8]</td></tr><tr><td align="right">16</td><td align="left">[16, 32, 48, 64, 80, 96, 112, 128]</td></tr><tr><td align="right">32</td><td align="left">[160, 192, 224, 256]</td></tr><tr><td align="right">64</td><td align="left">[320, 384, 448, 512]</td></tr><tr><td align="right">128</td><td align="left">[640, 768, 896, 1024]</td></tr><tr><td align="right">256</td><td align="left">[1280, 1536, 1792, 2048]</td></tr><tr><td align="right">512</td><td align="left">[2560, 3072, 3584, 4096]</td></tr><tr><td align="right">1 KiB</td><td align="left">[5 KiB, 6 KiB, 7 KiB, 8 KiB]</td></tr><tr><td align="right">2 KiB</td><td align="left">[10 KiB, 12 KiB, 14 KiB]</td></tr><tr><td rowspan="8" align="left">Large</td><td align="right">2 KiB</td><td align="left">[16 KiB]</td></tr><tr><td align="right">4 KiB</td><td align="left">[20 KiB, 24 KiB, 28 KiB, 32 KiB]</td></tr><tr><td align="right">8 KiB</td><td align="left">[40 KiB, 48 KiB, 54 KiB, 64 KiB]</td></tr><tr><td align="right">16 KiB</td><td align="left">[80 KiB, 96 KiB, 112 KiB, 128 KiB]</td></tr><tr><td align="right">32 KiB</td><td align="left">[160 KiB, 192 KiB, 224 KiB, 256 KiB]</td></tr><tr><td align="right">64 KiB</td><td align="left">[320 KiB, 384 KiB, 448 KiB, 512 KiB]</td></tr><tr><td align="right">128 KiB</td><td align="left">[640 KiB, 768 KiB, 896 KiB, 1 MiB]</td></tr><tr><td align="right">256 KiB</td><td align="left">[1280 KiB, 1536 KiB, 1792 KiB]</td></tr><tr><td rowspan="7" align="left">Huge</td><td align="right">256 KiB</td><td align="left">[2 MiB]</td></tr><tr><td align="right">512 KiB</td><td align="left">[2560 KiB, 3 MiB, 3584 KiB, 4 MiB]</td></tr><tr><td align="right">1 MiB</td><td align="left">[5 MiB, 6 MiB, 7 MiB, 8 MiB]</td></tr><tr><td align="right">2 MiB</td><td align="left">[10 MiB, 12 MiB, 14 MiB, 16 MiB]</td></tr><tr><td align="right">4 MiB</td><td align="left">[20 MiB, 24 MiB, 28 MiB, 32 MiB]</td></tr><tr><td align="right">8 MiB</td><td align="left">[40 MiB, 48 MiB, 56 MiB, 64 MiB]</td></tr><tr><td align="right">...</td><td align="left">...</td></tr></tbody></table></div></div><br class="table-break"></div><div class="refsect1"><a name="mallctl_namespace"></a><h2>MALLCTL NAMESPACE</h2><p>The following names are defined in the namespace accessible via the
299    <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions.  Value types are
300    specified in parentheses, their readable/writable statuses are encoded as
301    <code class="literal">rw</code>, <code class="literal">r-</code>, <code class="literal">-w</code>, or
302    <code class="literal">--</code>, and required build configuration flags follow, if
303    any.  A name element encoded as <code class="literal">&lt;i&gt;</code> or
304    <code class="literal">&lt;j&gt;</code> indicates an integer component, where the
305    integer varies from 0 to some upper value that must be determined via
306    introspection.  In the case of
307    "<code class="mallctl">stats.arenas.&lt;i&gt;.*</code>"
308  ,
309    <code class="literal">&lt;i&gt;</code> equal to <a class="link" href="#arenas.narenas">
310    "<code class="mallctl">arenas.narenas</code>"
311  </a> can be
312    used to access the summation of statistics from all arenas.  Take special
313    note of the <a class="link" href="#epoch">
314    "<code class="mallctl">epoch</code>"
315  </a> mallctl,
316    which controls refreshing of cached dynamic statistics.</p><div class="variablelist"><dl class="variablelist"><dt><a name="version"></a><span class="term">
317
318    "<code class="mallctl">version</code>"
319
320          (<span class="type">const char *</span>)
321          <code class="literal">r-</code>
322        </span></dt><dd><p>Return the jemalloc version string.</p></dd><dt><a name="epoch"></a><span class="term">
323
324    "<code class="mallctl">epoch</code>"
325
326          (<span class="type">uint64_t</span>)
327          <code class="literal">rw</code>
328        </span></dt><dd><p>If a value is passed in, refresh the data from which
329        the <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) functions report values,
330        and increment the epoch.  Return the current epoch.  This is useful for
331        detecting whether another thread caused a refresh.</p></dd><dt><a name="config.cache_oblivious"></a><span class="term">
332
333    "<code class="mallctl">config.cache_oblivious</code>"
334
335          (<span class="type">bool</span>)
336          <code class="literal">r-</code>
337        </span></dt><dd><p><code class="option">--enable-cache-oblivious</code> was specified
338        during build configuration.</p></dd><dt><a name="config.debug"></a><span class="term">
339
340    "<code class="mallctl">config.debug</code>"
341
342          (<span class="type">bool</span>)
343          <code class="literal">r-</code>
344        </span></dt><dd><p><code class="option">--enable-debug</code> was specified during
345        build configuration.</p></dd><dt><a name="config.fill"></a><span class="term">
346
347    "<code class="mallctl">config.fill</code>"
348
349          (<span class="type">bool</span>)
350          <code class="literal">r-</code>
351        </span></dt><dd><p><code class="option">--enable-fill</code> was specified during
352        build configuration.</p></dd><dt><a name="config.lazy_lock"></a><span class="term">
353
354    "<code class="mallctl">config.lazy_lock</code>"
355
356          (<span class="type">bool</span>)
357          <code class="literal">r-</code>
358        </span></dt><dd><p><code class="option">--enable-lazy-lock</code> was specified
359        during build configuration.</p></dd><dt><a name="config.munmap"></a><span class="term">
360
361    "<code class="mallctl">config.munmap</code>"
362
363          (<span class="type">bool</span>)
364          <code class="literal">r-</code>
365        </span></dt><dd><p><code class="option">--enable-munmap</code> was specified during
366        build configuration.</p></dd><dt><a name="config.prof"></a><span class="term">
367
368    "<code class="mallctl">config.prof</code>"
369
370          (<span class="type">bool</span>)
371          <code class="literal">r-</code>
372        </span></dt><dd><p><code class="option">--enable-prof</code> was specified during
373        build configuration.</p></dd><dt><a name="config.prof_libgcc"></a><span class="term">
374
375    "<code class="mallctl">config.prof_libgcc</code>"
376
377          (<span class="type">bool</span>)
378          <code class="literal">r-</code>
379        </span></dt><dd><p><code class="option">--disable-prof-libgcc</code> was not
380        specified during build configuration.</p></dd><dt><a name="config.prof_libunwind"></a><span class="term">
381
382    "<code class="mallctl">config.prof_libunwind</code>"
383
384          (<span class="type">bool</span>)
385          <code class="literal">r-</code>
386        </span></dt><dd><p><code class="option">--enable-prof-libunwind</code> was specified
387        during build configuration.</p></dd><dt><a name="config.stats"></a><span class="term">
388
389    "<code class="mallctl">config.stats</code>"
390
391          (<span class="type">bool</span>)
392          <code class="literal">r-</code>
393        </span></dt><dd><p><code class="option">--enable-stats</code> was specified during
394        build configuration.</p></dd><dt><a name="config.tcache"></a><span class="term">
395
396    "<code class="mallctl">config.tcache</code>"
397
398          (<span class="type">bool</span>)
399          <code class="literal">r-</code>
400        </span></dt><dd><p><code class="option">--disable-tcache</code> was not specified
401        during build configuration.</p></dd><dt><a name="config.tls"></a><span class="term">
402
403    "<code class="mallctl">config.tls</code>"
404
405          (<span class="type">bool</span>)
406          <code class="literal">r-</code>
407        </span></dt><dd><p><code class="option">--disable-tls</code> was not specified during
408        build configuration.</p></dd><dt><a name="config.utrace"></a><span class="term">
409
410    "<code class="mallctl">config.utrace</code>"
411
412          (<span class="type">bool</span>)
413          <code class="literal">r-</code>
414        </span></dt><dd><p><code class="option">--enable-utrace</code> was specified during
415        build configuration.</p></dd><dt><a name="config.valgrind"></a><span class="term">
416
417    "<code class="mallctl">config.valgrind</code>"
418
419          (<span class="type">bool</span>)
420          <code class="literal">r-</code>
421        </span></dt><dd><p><code class="option">--enable-valgrind</code> was specified during
422        build configuration.</p></dd><dt><a name="config.xmalloc"></a><span class="term">
423
424    "<code class="mallctl">config.xmalloc</code>"
425
426          (<span class="type">bool</span>)
427          <code class="literal">r-</code>
428        </span></dt><dd><p><code class="option">--enable-xmalloc</code> was specified during
429        build configuration.</p></dd><dt><a name="opt.abort"></a><span class="term">
430
431    "<code class="mallctl">opt.abort</code>"
432
433          (<span class="type">bool</span>)
434          <code class="literal">r-</code>
435        </span></dt><dd><p>Abort-on-warning enabled/disabled.  If true, most
436        warnings are fatal.  The process will call
437        <span class="citerefentry"><span class="refentrytitle">abort</span>(3)</span> in these cases.  This option is
438        disabled by default unless <code class="option">--enable-debug</code> is
439        specified during configuration, in which case it is enabled by default.
440        </p></dd><dt><a name="opt.dss"></a><span class="term">
441
442    "<code class="mallctl">opt.dss</code>"
443
444          (<span class="type">const char *</span>)
445          <code class="literal">r-</code>
446        </span></dt><dd><p>dss (<span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>) allocation precedence as
447        related to <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> allocation.  The following
448        settings are supported if
449        <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span> is supported by the operating
450        system: &#8220;disabled&#8221;, &#8220;primary&#8221;, and
451        &#8220;secondary&#8221;; otherwise only &#8220;disabled&#8221; is
452        supported.  The default is &#8220;secondary&#8221; if
453        <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span> is supported by the operating
454        system; &#8220;disabled&#8221; otherwise.
455        </p></dd><dt><a name="opt.lg_chunk"></a><span class="term">
456
457    "<code class="mallctl">opt.lg_chunk</code>"
458
459          (<span class="type">size_t</span>)
460          <code class="literal">r-</code>
461        </span></dt><dd><p>Virtual memory chunk size (log base 2).  If a chunk
462        size outside the supported size range is specified, the size is
463        silently clipped to the minimum/maximum supported size.  The default
464        chunk size is 2 MiB (2^21).
465        </p></dd><dt><a name="opt.narenas"></a><span class="term">
466
467    "<code class="mallctl">opt.narenas</code>"
468
469          (<span class="type">size_t</span>)
470          <code class="literal">r-</code>
471        </span></dt><dd><p>Maximum number of arenas to use for automatic
472        multiplexing of threads and arenas.  The default is four times the
473        number of CPUs, or one if there is a single CPU.</p></dd><dt><a name="opt.lg_dirty_mult"></a><span class="term">
474
475    "<code class="mallctl">opt.lg_dirty_mult</code>"
476
477          (<span class="type">ssize_t</span>)
478          <code class="literal">r-</code>
479        </span></dt><dd><p>Per-arena minimum ratio (log base 2) of active to dirty
480        pages.  Some dirty unused pages may be allowed to accumulate, within
481        the limit set by the ratio (or one chunk worth of dirty pages,
482        whichever is greater), before informing the kernel about some of those
483        pages via <span class="citerefentry"><span class="refentrytitle">madvise</span>(2)</span> or a similar system call.  This
484        provides the kernel with sufficient information to recycle dirty pages
485        if physical memory becomes scarce and the pages remain unused.  The
486        default minimum ratio is 8:1 (2^3:1); an option value of -1 will
487        disable dirty page purging.  See <a class="link" href="#arenas.lg_dirty_mult">
488    "<code class="mallctl">arenas.lg_dirty_mult</code>"
489  </a>
490        and <a class="link" href="#arena.i.lg_dirty_mult">
491    "<code class="mallctl">arena.&lt;i&gt;.lg_dirty_mult</code>"
492  </a>
493        for related dynamic control options.</p></dd><dt><a name="opt.stats_print"></a><span class="term">
494
495    "<code class="mallctl">opt.stats_print</code>"
496
497          (<span class="type">bool</span>)
498          <code class="literal">r-</code>
499        </span></dt><dd><p>Enable/disable statistics printing at exit.  If
500        enabled, the <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>)
501        function is called at program exit via an
502        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function.  If
503        <code class="option">--enable-stats</code> is specified during configuration, this
504        has the potential to cause deadlock for a multi-threaded process that
505        exits while one or more threads are executing in the memory allocation
506        functions.  Furthermore, <code class="function">atexit</code>(<em class="parameter"><code></code></em>) may
507        allocate memory during application initialization and then deadlock
508        internally when jemalloc in turn calls
509        <code class="function">atexit</code>(<em class="parameter"><code></code></em>), so this option is not
510        univerally usable (though the application can register its own
511        <code class="function">atexit</code>(<em class="parameter"><code></code></em>) function with equivalent
512        functionality).  Therefore, this option should only be used with care;
513        it is primarily intended as a performance tuning aid during application
514        development.  This option is disabled by default.</p></dd><dt><a name="opt.junk"></a><span class="term">
515
516    "<code class="mallctl">opt.junk</code>"
517
518          (<span class="type">const char *</span>)
519          <code class="literal">r-</code>
520          [<code class="option">--enable-fill</code>]
521        </span></dt><dd><p>Junk filling.  If set to "alloc", each byte of
522        uninitialized allocated memory will be initialized to
523        <code class="literal">0xa5</code>.  If set to "free", all deallocated memory will
524        be initialized to <code class="literal">0x5a</code>.  If set to "true", both
525        allocated and deallocated memory will be initialized, and if set to
526        "false", junk filling be disabled entirely.  This is intended for
527        debugging and will impact performance negatively.  This option is
528        "false" by default unless <code class="option">--enable-debug</code> is specified
529        during configuration, in which case it is "true" by default unless
530        running inside <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>.</p></dd><dt><a name="opt.quarantine"></a><span class="term">
531
532    "<code class="mallctl">opt.quarantine</code>"
533
534          (<span class="type">size_t</span>)
535          <code class="literal">r-</code>
536          [<code class="option">--enable-fill</code>]
537        </span></dt><dd><p>Per thread quarantine size in bytes.  If non-zero, each
538        thread maintains a FIFO object quarantine that stores up to the
539        specified number of bytes of memory.  The quarantined memory is not
540        freed until it is released from quarantine, though it is immediately
541        junk-filled if the <a class="link" href="#opt.junk">
542    "<code class="mallctl">opt.junk</code>"
543  </a> option is
544        enabled.  This feature is of particular use in combination with <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>, which can detect attempts
545        to access quarantined objects.  This is intended for debugging and will
546        impact performance negatively.  The default quarantine size is 0 unless
547        running inside Valgrind, in which case the default is 16
548        MiB.</p></dd><dt><a name="opt.redzone"></a><span class="term">
549
550    "<code class="mallctl">opt.redzone</code>"
551
552          (<span class="type">bool</span>)
553          <code class="literal">r-</code>
554          [<code class="option">--enable-fill</code>]
555        </span></dt><dd><p>Redzones enabled/disabled.  If enabled, small
556        allocations have redzones before and after them.  Furthermore, if the
557        <a class="link" href="#opt.junk">
558    "<code class="mallctl">opt.junk</code>"
559  </a> option is
560        enabled, the redzones are checked for corruption during deallocation.
561        However, the primary intended purpose of this feature is to be used in
562        combination with <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>,
563        which needs redzones in order to do effective buffer overflow/underflow
564        detection.  This option is intended for debugging and will impact
565        performance negatively.  This option is disabled by
566        default unless running inside Valgrind.</p></dd><dt><a name="opt.zero"></a><span class="term">
567
568    "<code class="mallctl">opt.zero</code>"
569
570          (<span class="type">bool</span>)
571          <code class="literal">r-</code>
572          [<code class="option">--enable-fill</code>]
573        </span></dt><dd><p>Zero filling enabled/disabled.  If enabled, each byte
574        of uninitialized allocated memory will be initialized to 0.  Note that
575        this initialization only happens once for each byte, so
576        <code class="function">realloc</code>(<em class="parameter"><code></code></em>) and
577        <code class="function">rallocx</code>(<em class="parameter"><code></code></em>) calls do not zero memory that
578        was previously allocated.  This is intended for debugging and will
579        impact performance negatively.  This option is disabled by default.
580        </p></dd><dt><a name="opt.utrace"></a><span class="term">
581
582    "<code class="mallctl">opt.utrace</code>"
583
584          (<span class="type">bool</span>)
585          <code class="literal">r-</code>
586          [<code class="option">--enable-utrace</code>]
587        </span></dt><dd><p>Allocation tracing based on
588        <span class="citerefentry"><span class="refentrytitle">utrace</span>(2)</span> enabled/disabled.  This option
589        is disabled by default.</p></dd><dt><a name="opt.xmalloc"></a><span class="term">
590
591    "<code class="mallctl">opt.xmalloc</code>"
592
593          (<span class="type">bool</span>)
594          <code class="literal">r-</code>
595          [<code class="option">--enable-xmalloc</code>]
596        </span></dt><dd><p>Abort-on-out-of-memory enabled/disabled.  If enabled,
597        rather than returning failure for any allocation function, display a
598        diagnostic message on <code class="constant">STDERR_FILENO</code> and cause the
599        program to drop core (using
600        <span class="citerefentry"><span class="refentrytitle">abort</span>(3)</span>).  If an application is
601        designed to depend on this behavior, set the option at compile time by
602        including the following in the source code:
603        </p><pre class="programlisting">
604malloc_conf = "xmalloc:true";</pre><p>
605        This option is disabled by default.</p></dd><dt><a name="opt.tcache"></a><span class="term">
606
607    "<code class="mallctl">opt.tcache</code>"
608
609          (<span class="type">bool</span>)
610          <code class="literal">r-</code>
611          [<code class="option">--enable-tcache</code>]
612        </span></dt><dd><p>Thread-specific caching (tcache) enabled/disabled.  When
613        there are multiple threads, each thread uses a tcache for objects up to
614        a certain size.  Thread-specific caching allows many allocations to be
615        satisfied without performing any thread synchronization, at the cost of
616        increased memory use.  See the <a class="link" href="#opt.lg_tcache_max">
617    "<code class="mallctl">opt.lg_tcache_max</code>"
618  </a>
619        option for related tuning information.  This option is enabled by
620        default unless running inside <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a>, in which case it is
621        forcefully disabled.</p></dd><dt><a name="opt.lg_tcache_max"></a><span class="term">
622
623    "<code class="mallctl">opt.lg_tcache_max</code>"
624
625          (<span class="type">size_t</span>)
626          <code class="literal">r-</code>
627          [<code class="option">--enable-tcache</code>]
628        </span></dt><dd><p>Maximum size class (log base 2) to cache in the
629        thread-specific cache (tcache).  At a minimum, all small size classes
630        are cached, and at a maximum all large size classes are cached.  The
631        default maximum is 32 KiB (2^15).</p></dd><dt><a name="opt.prof"></a><span class="term">
632
633    "<code class="mallctl">opt.prof</code>"
634
635          (<span class="type">bool</span>)
636          <code class="literal">r-</code>
637          [<code class="option">--enable-prof</code>]
638        </span></dt><dd><p>Memory profiling enabled/disabled.  If enabled, profile
639        memory allocation activity.  See the <a class="link" href="#opt.prof_active">
640    "<code class="mallctl">opt.prof_active</code>"
641  </a>
642        option for on-the-fly activation/deactivation.  See the <a class="link" href="#opt.lg_prof_sample">
643    "<code class="mallctl">opt.lg_prof_sample</code>"
644  </a>
645        option for probabilistic sampling control.  See the <a class="link" href="#opt.prof_accum">
646    "<code class="mallctl">opt.prof_accum</code>"
647  </a>
648        option for control of cumulative sample reporting.  See the <a class="link" href="#opt.lg_prof_interval">
649    "<code class="mallctl">opt.lg_prof_interval</code>"
650  </a>
651        option for information on interval-triggered profile dumping, the <a class="link" href="#opt.prof_gdump">
652    "<code class="mallctl">opt.prof_gdump</code>"
653  </a>
654        option for information on high-water-triggered profile dumping, and the
655        <a class="link" href="#opt.prof_final">
656    "<code class="mallctl">opt.prof_final</code>"
657  </a>
658        option for final profile dumping.  Profile output is compatible with
659        the <span class="command"><strong>jeprof</strong></span> command, which is based on the
660        <span class="command"><strong>pprof</strong></span> that is developed as part of the <a class="ulink" href="http://code.google.com/p/gperftools/" target="_top">gperftools
661        package</a>.</p></dd><dt><a name="opt.prof_prefix"></a><span class="term">
662
663    "<code class="mallctl">opt.prof_prefix</code>"
664
665          (<span class="type">const char *</span>)
666          <code class="literal">r-</code>
667          [<code class="option">--enable-prof</code>]
668        </span></dt><dd><p>Filename prefix for profile dumps.  If the prefix is
669        set to the empty string, no automatic dumps will occur; this is
670        primarily useful for disabling the automatic final heap dump (which
671        also disables leak reporting, if enabled).  The default prefix is
672        <code class="filename">jeprof</code>.</p></dd><dt><a name="opt.prof_active"></a><span class="term">
673
674    "<code class="mallctl">opt.prof_active</code>"
675
676          (<span class="type">bool</span>)
677          <code class="literal">r-</code>
678          [<code class="option">--enable-prof</code>]
679        </span></dt><dd><p>Profiling activated/deactivated.  This is a secondary
680        control mechanism that makes it possible to start the application with
681        profiling enabled (see the <a class="link" href="#opt.prof">
682    "<code class="mallctl">opt.prof</code>"
683  </a> option) but
684        inactive, then toggle profiling at any time during program execution
685        with the <a class="link" href="#prof.active">
686    "<code class="mallctl">prof.active</code>"
687  </a> mallctl.
688        This option is enabled by default.</p></dd><dt><a name="opt.prof_thread_active_init"></a><span class="term">
689
690    "<code class="mallctl">opt.prof_thread_active_init</code>"
691
692          (<span class="type">bool</span>)
693          <code class="literal">r-</code>
694          [<code class="option">--enable-prof</code>]
695        </span></dt><dd><p>Initial setting for <a class="link" href="#thread.prof.active">
696    "<code class="mallctl">thread.prof.active</code>"
697  </a>
698        in newly created threads.  The initial setting for newly created threads
699        can also be changed during execution via the <a class="link" href="#prof.thread_active_init">
700    "<code class="mallctl">prof.thread_active_init</code>"
701  </a>
702        mallctl.  This option is enabled by default.</p></dd><dt><a name="opt.lg_prof_sample"></a><span class="term">
703
704    "<code class="mallctl">opt.lg_prof_sample</code>"
705
706          (<span class="type">size_t</span>)
707          <code class="literal">r-</code>
708          [<code class="option">--enable-prof</code>]
709        </span></dt><dd><p>Average interval (log base 2) between allocation
710        samples, as measured in bytes of allocation activity.  Increasing the
711        sampling interval decreases profile fidelity, but also decreases the
712        computational overhead.  The default sample interval is 512 KiB (2^19
713        B).</p></dd><dt><a name="opt.prof_accum"></a><span class="term">
714
715    "<code class="mallctl">opt.prof_accum</code>"
716
717          (<span class="type">bool</span>)
718          <code class="literal">r-</code>
719          [<code class="option">--enable-prof</code>]
720        </span></dt><dd><p>Reporting of cumulative object/byte counts in profile
721        dumps enabled/disabled.  If this option is enabled, every unique
722        backtrace must be stored for the duration of execution.  Depending on
723        the application, this can impose a large memory overhead, and the
724        cumulative counts are not always of interest.  This option is disabled
725        by default.</p></dd><dt><a name="opt.lg_prof_interval"></a><span class="term">
726
727    "<code class="mallctl">opt.lg_prof_interval</code>"
728
729          (<span class="type">ssize_t</span>)
730          <code class="literal">r-</code>
731          [<code class="option">--enable-prof</code>]
732        </span></dt><dd><p>Average interval (log base 2) between memory profile
733        dumps, as measured in bytes of allocation activity.  The actual
734        interval between dumps may be sporadic because decentralized allocation
735        counters are used to avoid synchronization bottlenecks.  Profiles are
736        dumped to files named according to the pattern
737        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.i&lt;iseq&gt;.heap</code>,
738        where <code class="literal">&lt;prefix&gt;</code> is controlled by the
739        <a class="link" href="#opt.prof_prefix">
740    "<code class="mallctl">opt.prof_prefix</code>"
741  </a>
742        option.  By default, interval-triggered profile dumping is disabled
743        (encoded as -1).
744        </p></dd><dt><a name="opt.prof_gdump"></a><span class="term">
745
746    "<code class="mallctl">opt.prof_gdump</code>"
747
748          (<span class="type">bool</span>)
749          <code class="literal">r-</code>
750          [<code class="option">--enable-prof</code>]
751        </span></dt><dd><p>Set the initial state of <a class="link" href="#prof.gdump">
752    "<code class="mallctl">prof.gdump</code>"
753  </a>, which when
754        enabled triggers a memory profile dump every time the total virtual
755        memory exceeds the previous maximum.  This option is disabled by
756        default.</p></dd><dt><a name="opt.prof_final"></a><span class="term">
757
758    "<code class="mallctl">opt.prof_final</code>"
759
760          (<span class="type">bool</span>)
761          <code class="literal">r-</code>
762          [<code class="option">--enable-prof</code>]
763        </span></dt><dd><p>Use an
764        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function to dump final memory
765        usage to a file named according to the pattern
766        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.f.heap</code>,
767        where <code class="literal">&lt;prefix&gt;</code> is controlled by the <a class="link" href="#opt.prof_prefix">
768    "<code class="mallctl">opt.prof_prefix</code>"
769  </a>
770        option.  Note that <code class="function">atexit</code>(<em class="parameter"><code></code></em>) may allocate
771        memory during application initialization and then deadlock internally
772        when jemalloc in turn calls <code class="function">atexit</code>(<em class="parameter"><code></code></em>), so
773        this option is not univerally usable (though the application can
774        register its own <code class="function">atexit</code>(<em class="parameter"><code></code></em>) function with
775        equivalent functionality).  This option is disabled by
776        default.</p></dd><dt><a name="opt.prof_leak"></a><span class="term">
777
778    "<code class="mallctl">opt.prof_leak</code>"
779
780          (<span class="type">bool</span>)
781          <code class="literal">r-</code>
782          [<code class="option">--enable-prof</code>]
783        </span></dt><dd><p>Leak reporting enabled/disabled.  If enabled, use an
784        <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span> function to report memory leaks
785        detected by allocation sampling.  See the
786        <a class="link" href="#opt.prof">
787    "<code class="mallctl">opt.prof</code>"
788  </a> option for
789        information on analyzing heap profile output.  This option is disabled
790        by default.</p></dd><dt><a name="thread.arena"></a><span class="term">
791
792    "<code class="mallctl">thread.arena</code>"
793
794          (<span class="type">unsigned</span>)
795          <code class="literal">rw</code>
796        </span></dt><dd><p>Get or set the arena associated with the calling
797        thread.  If the specified arena was not initialized beforehand (see the
798        <a class="link" href="#arenas.initialized">
799    "<code class="mallctl">arenas.initialized</code>"
800  </a>
801        mallctl), it will be automatically initialized as a side effect of
802        calling this interface.</p></dd><dt><a name="thread.allocated"></a><span class="term">
803
804    "<code class="mallctl">thread.allocated</code>"
805
806          (<span class="type">uint64_t</span>)
807          <code class="literal">r-</code>
808          [<code class="option">--enable-stats</code>]
809        </span></dt><dd><p>Get the total number of bytes ever allocated by the
810        calling thread.  This counter has the potential to wrap around; it is
811        up to the application to appropriately interpret the counter in such
812        cases.</p></dd><dt><a name="thread.allocatedp"></a><span class="term">
813
814    "<code class="mallctl">thread.allocatedp</code>"
815
816          (<span class="type">uint64_t *</span>)
817          <code class="literal">r-</code>
818          [<code class="option">--enable-stats</code>]
819        </span></dt><dd><p>Get a pointer to the the value that is returned by the
820        <a class="link" href="#thread.allocated">
821    "<code class="mallctl">thread.allocated</code>"
822  </a>
823        mallctl.  This is useful for avoiding the overhead of repeated
824        <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) calls.</p></dd><dt><a name="thread.deallocated"></a><span class="term">
825
826    "<code class="mallctl">thread.deallocated</code>"
827
828          (<span class="type">uint64_t</span>)
829          <code class="literal">r-</code>
830          [<code class="option">--enable-stats</code>]
831        </span></dt><dd><p>Get the total number of bytes ever deallocated by the
832        calling thread.  This counter has the potential to wrap around; it is
833        up to the application to appropriately interpret the counter in such
834        cases.</p></dd><dt><a name="thread.deallocatedp"></a><span class="term">
835
836    "<code class="mallctl">thread.deallocatedp</code>"
837
838          (<span class="type">uint64_t *</span>)
839          <code class="literal">r-</code>
840          [<code class="option">--enable-stats</code>]
841        </span></dt><dd><p>Get a pointer to the the value that is returned by the
842        <a class="link" href="#thread.deallocated">
843    "<code class="mallctl">thread.deallocated</code>"
844  </a>
845        mallctl.  This is useful for avoiding the overhead of repeated
846        <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>) calls.</p></dd><dt><a name="thread.tcache.enabled"></a><span class="term">
847
848    "<code class="mallctl">thread.tcache.enabled</code>"
849
850          (<span class="type">bool</span>)
851          <code class="literal">rw</code>
852          [<code class="option">--enable-tcache</code>]
853        </span></dt><dd><p>Enable/disable calling thread's tcache.  The tcache is
854        implicitly flushed as a side effect of becoming
855        disabled (see <a class="link" href="#thread.tcache.flush">
856    "<code class="mallctl">thread.tcache.flush</code>"
857  </a>).
858        </p></dd><dt><a name="thread.tcache.flush"></a><span class="term">
859
860    "<code class="mallctl">thread.tcache.flush</code>"
861
862          (<span class="type">void</span>)
863          <code class="literal">--</code>
864          [<code class="option">--enable-tcache</code>]
865        </span></dt><dd><p>Flush calling thread's thread-specific cache (tcache).
866        This interface releases all cached objects and internal data structures
867        associated with the calling thread's tcache.  Ordinarily, this interface
868        need not be called, since automatic periodic incremental garbage
869        collection occurs, and the thread cache is automatically discarded when
870        a thread exits.  However, garbage collection is triggered by allocation
871        activity, so it is possible for a thread that stops
872        allocating/deallocating to retain its cache indefinitely, in which case
873        the developer may find manual flushing useful.</p></dd><dt><a name="thread.prof.name"></a><span class="term">
874
875    "<code class="mallctl">thread.prof.name</code>"
876
877          (<span class="type">const char *</span>)
878          <code class="literal">r-</code> or
879          <code class="literal">-w</code>
880          [<code class="option">--enable-prof</code>]
881        </span></dt><dd><p>Get/set the descriptive name associated with the calling
882        thread in memory profile dumps.  An internal copy of the name string is
883        created, so the input string need not be maintained after this interface
884        completes execution.  The output string of this interface should be
885        copied for non-ephemeral uses, because multiple implementation details
886        can cause asynchronous string deallocation.  Furthermore, each
887        invocation of this interface can only read or write; simultaneous
888        read/write is not supported due to string lifetime limitations.  The
889        name string must nil-terminated and comprised only of characters in the
890        sets recognized
891        by <span class="citerefentry"><span class="refentrytitle">isgraph</span>(3)</span> and
892        <span class="citerefentry"><span class="refentrytitle">isblank</span>(3)</span>.</p></dd><dt><a name="thread.prof.active"></a><span class="term">
893
894    "<code class="mallctl">thread.prof.active</code>"
895
896          (<span class="type">bool</span>)
897          <code class="literal">rw</code>
898          [<code class="option">--enable-prof</code>]
899        </span></dt><dd><p>Control whether sampling is currently active for the
900        calling thread.  This is an activation mechanism in addition to <a class="link" href="#prof.active">
901    "<code class="mallctl">prof.active</code>"
902  </a>; both must
903        be active for the calling thread to sample.  This flag is enabled by
904        default.</p></dd><dt><a name="tcache.create"></a><span class="term">
905
906    "<code class="mallctl">tcache.create</code>"
907
908          (<span class="type">unsigned</span>)
909          <code class="literal">r-</code>
910          [<code class="option">--enable-tcache</code>]
911        </span></dt><dd><p>Create an explicit thread-specific cache (tcache) and
912        return an identifier that can be passed to the <a class="link" href="#MALLOCX_TCACHE"><code class="constant">MALLOCX_TCACHE(<em class="parameter"><code>tc</code></em>)</code></a>
913        macro to explicitly use the specified cache rather than the
914        automatically managed one that is used by default.  Each explicit cache
915        can be used by only one thread at a time; the application must assure
916        that this constraint holds.
917        </p></dd><dt><a name="tcache.flush"></a><span class="term">
918
919    "<code class="mallctl">tcache.flush</code>"
920
921          (<span class="type">unsigned</span>)
922          <code class="literal">-w</code>
923          [<code class="option">--enable-tcache</code>]
924        </span></dt><dd><p>Flush the specified thread-specific cache (tcache).  The
925        same considerations apply to this interface as to <a class="link" href="#thread.tcache.flush">
926    "<code class="mallctl">thread.tcache.flush</code>"
927  </a>,
928        except that the tcache will never be automatically be discarded.
929        </p></dd><dt><a name="tcache.destroy"></a><span class="term">
930
931    "<code class="mallctl">tcache.destroy</code>"
932
933          (<span class="type">unsigned</span>)
934          <code class="literal">-w</code>
935          [<code class="option">--enable-tcache</code>]
936        </span></dt><dd><p>Flush the specified thread-specific cache (tcache) and
937        make the identifier available for use during a future tcache creation.
938        </p></dd><dt><a name="arena.i.purge"></a><span class="term">
939
940    "<code class="mallctl">arena.&lt;i&gt;.purge</code>"
941
942          (<span class="type">void</span>)
943          <code class="literal">--</code>
944        </span></dt><dd><p>Purge unused dirty pages for arena &lt;i&gt;, or for
945        all arenas if &lt;i&gt; equals <a class="link" href="#arenas.narenas">
946    "<code class="mallctl">arenas.narenas</code>"
947  </a>.
948        </p></dd><dt><a name="arena.i.dss"></a><span class="term">
949
950    "<code class="mallctl">arena.&lt;i&gt;.dss</code>"
951
952          (<span class="type">const char *</span>)
953          <code class="literal">rw</code>
954        </span></dt><dd><p>Set the precedence of dss allocation as related to mmap
955        allocation for arena &lt;i&gt;, or for all arenas if &lt;i&gt; equals
956        <a class="link" href="#arenas.narenas">
957    "<code class="mallctl">arenas.narenas</code>"
958  </a>.  See
959        <a class="link" href="#opt.dss">
960    "<code class="mallctl">opt.dss</code>"
961  </a> for supported
962        settings.</p></dd><dt><a name="arena.i.lg_dirty_mult"></a><span class="term">
963
964    "<code class="mallctl">arena.&lt;i&gt;.lg_dirty_mult</code>"
965
966          (<span class="type">ssize_t</span>)
967          <code class="literal">rw</code>
968        </span></dt><dd><p>Current per-arena minimum ratio (log base 2) of active
969        to dirty pages for arena &lt;i&gt;.  Each time this interface is set and
970        the ratio is increased, pages are synchronously purged as necessary to
971        impose the new ratio.  See <a class="link" href="#opt.lg_dirty_mult">
972    "<code class="mallctl">opt.lg_dirty_mult</code>"
973  </a>
974        for additional information.</p></dd><dt><a name="arena.i.chunk_hooks"></a><span class="term">
975
976    "<code class="mallctl">arena.&lt;i&gt;.chunk_hooks</code>"
977
978          (<span class="type">chunk_hooks_t</span>)
979          <code class="literal">rw</code>
980        </span></dt><dd><p>Get or set the chunk management hook functions for arena
981        &lt;i&gt;.  The functions must be capable of operating on all extant
982        chunks associated with arena &lt;i&gt;, usually by passing unknown
983        chunks to the replaced functions.  In practice, it is feasible to
984        control allocation for arenas created via <a class="link" href="#arenas.extend">
985    "<code class="mallctl">arenas.extend</code>"
986  </a> such
987        that all chunks originate from an application-supplied chunk allocator
988        (by setting custom chunk hook functions just after arena creation), but
989        the automatically created arenas may have already created chunks prior
990        to the application having an opportunity to take over chunk
991        allocation.</p><pre class="programlisting">
992typedef struct {
993	chunk_alloc_t		*alloc;
994	chunk_dalloc_t		*dalloc;
995	chunk_commit_t		*commit;
996	chunk_decommit_t	*decommit;
997	chunk_purge_t		*purge;
998	chunk_split_t		*split;
999	chunk_merge_t		*merge;
1000} chunk_hooks_t;</pre><p>The <span class="type">chunk_hooks_t</span> structure comprises function
1001        pointers which are described individually below.  jemalloc uses these
1002        functions to manage chunk lifetime, which starts off with allocation of
1003        mapped committed memory, in the simplest case followed by deallocation.
1004        However, there are performance and platform reasons to retain chunks for
1005        later reuse.  Cleanup attempts cascade from deallocation to decommit to
1006        purging, which gives the chunk management functions opportunities to
1007        reject the most permanent cleanup operations in favor of less permanent
1008        (and often less costly) operations.  The chunk splitting and merging
1009        operations can also be opted out of, but this is mainly intended to
1010        support platforms on which virtual memory mappings provided by the
1011        operating system kernel do not automatically coalesce and split, e.g.
1012        Windows.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef void *<b class="fsfunc">(chunk_alloc_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">alignment</var>, </td></tr><tr><td>�</td><td>bool *<var class="pdparam">zero</var>, </td></tr><tr><td>�</td><td>bool *<var class="pdparam">commit</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk allocation function conforms to the
1013        <span class="type">chunk_alloc_t</span> type and upon success returns a pointer to
1014        <em class="parameter"><code>size</code></em> bytes of mapped memory on behalf of arena
1015        <em class="parameter"><code>arena_ind</code></em> such that the chunk's base address is a
1016        multiple of <em class="parameter"><code>alignment</code></em>, as well as setting
1017        <em class="parameter"><code>*zero</code></em> to indicate whether the chunk is zeroed and
1018        <em class="parameter"><code>*commit</code></em> to indicate whether the chunk is
1019        committed.  Upon error the function returns <code class="constant">NULL</code>
1020        and leaves <em class="parameter"><code>*zero</code></em> and
1021        <em class="parameter"><code>*commit</code></em> unmodified.  The
1022        <em class="parameter"><code>size</code></em> parameter is always a multiple of the chunk
1023        size.  The <em class="parameter"><code>alignment</code></em> parameter is always a power
1024        of two at least as large as the chunk size.  Zeroing is mandatory if
1025        <em class="parameter"><code>*zero</code></em> is true upon function entry.  Committing is
1026        mandatory if <em class="parameter"><code>*commit</code></em> is true upon function entry.
1027        If <em class="parameter"><code>chunk</code></em> is not <code class="constant">NULL</code>, the
1028        returned pointer must be <em class="parameter"><code>chunk</code></em> on success or
1029        <code class="constant">NULL</code> on error.  Committed memory may be committed
1030        in absolute terms as on a system that does not overcommit, or in
1031        implicit terms as on a system that overcommits and satisfies physical
1032        memory needs on demand via soft page faults.  Note that replacing the
1033        default chunk allocation function makes the arena's <a class="link" href="#arena.i.dss">
1034    "<code class="mallctl">arena.&lt;i&gt;.dss</code>"
1035  </a>
1036        setting irrelevant.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_dalloc_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>bool <var class="pdparam">committed</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>
1037        A chunk deallocation function conforms to the
1038        <span class="type">chunk_dalloc_t</span> type and deallocates a
1039        <em class="parameter"><code>chunk</code></em> of given <em class="parameter"><code>size</code></em> with
1040        <em class="parameter"><code>committed</code></em>/decommited memory as indicated, on
1041        behalf of arena <em class="parameter"><code>arena_ind</code></em>, returning false upon
1042        success.  If the function returns true, this indicates opt-out from
1043        deallocation; the virtual memory mapping associated with the chunk
1044        remains mapped, in the same commit state, and available for future use,
1045        in which case it will be automatically retained for later reuse.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_commit_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">offset</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">length</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk commit function conforms to the
1046        <span class="type">chunk_commit_t</span> type and commits zeroed physical memory to
1047        back pages within a <em class="parameter"><code>chunk</code></em> of given
1048        <em class="parameter"><code>size</code></em> at <em class="parameter"><code>offset</code></em> bytes,
1049        extending for <em class="parameter"><code>length</code></em> on behalf of arena
1050        <em class="parameter"><code>arena_ind</code></em>, returning false upon success.
1051        Committed memory may be committed in absolute terms as on a system that
1052        does not overcommit, or in implicit terms as on a system that
1053        overcommits and satisfies physical memory needs on demand via soft page
1054        faults. If the function returns true, this indicates insufficient
1055        physical memory to satisfy the request.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_decommit_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">offset</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">length</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk decommit function conforms to the
1056        <span class="type">chunk_decommit_t</span> type and decommits any physical memory
1057        that is backing pages within a <em class="parameter"><code>chunk</code></em> of given
1058        <em class="parameter"><code>size</code></em> at <em class="parameter"><code>offset</code></em> bytes,
1059        extending for <em class="parameter"><code>length</code></em> on behalf of arena
1060        <em class="parameter"><code>arena_ind</code></em>, returning false upon success, in which
1061        case the pages will be committed via the chunk commit function before
1062        being reused.  If the function returns true, this indicates opt-out from
1063        decommit; the memory remains committed and available for future use, in
1064        which case it will be automatically retained for later reuse.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_purge_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t<var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">offset</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">length</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk purge function conforms to the <span class="type">chunk_purge_t</span>
1065        type and optionally discards physical pages within the virtual memory
1066        mapping associated with <em class="parameter"><code>chunk</code></em> of given
1067        <em class="parameter"><code>size</code></em> at <em class="parameter"><code>offset</code></em> bytes,
1068        extending for <em class="parameter"><code>length</code></em> on behalf of arena
1069        <em class="parameter"><code>arena_ind</code></em>, returning false if pages within the
1070        purged virtual memory range will be zero-filled the next time they are
1071        accessed.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_split_t)</b>(</code></td><td>void *<var class="pdparam">chunk</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size_a</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size_b</var>, </td></tr><tr><td>�</td><td>bool <var class="pdparam">committed</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk split function conforms to the <span class="type">chunk_split_t</span>
1072        type and optionally splits <em class="parameter"><code>chunk</code></em> of given
1073        <em class="parameter"><code>size</code></em> into two adjacent chunks, the first of
1074        <em class="parameter"><code>size_a</code></em> bytes, and the second of
1075        <em class="parameter"><code>size_b</code></em> bytes, operating on
1076        <em class="parameter"><code>committed</code></em>/decommitted memory as indicated, on
1077        behalf of arena <em class="parameter"><code>arena_ind</code></em>, returning false upon
1078        success.  If the function returns true, this indicates that the chunk
1079        remains unsplit and therefore should continue to be operated on as a
1080        whole.</p><div class="funcsynopsis"><table border="0" class="funcprototype-table" summary="Function synopsis" style="cellspacing: 0; cellpadding: 0;"><tr><td><code class="funcdef">typedef bool <b class="fsfunc">(chunk_merge_t)</b>(</code></td><td>void *<var class="pdparam">chunk_a</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size_a</var>, </td></tr><tr><td>�</td><td>void *<var class="pdparam">chunk_b</var>, </td></tr><tr><td>�</td><td>size_t <var class="pdparam">size_b</var>, </td></tr><tr><td>�</td><td>bool <var class="pdparam">committed</var>, </td></tr><tr><td>�</td><td>unsigned <var class="pdparam">arena_ind</var><code>)</code>;</td></tr></table><div class="funcprototype-spacer">�</div></div><div class="literallayout"><p></p></div><p>A chunk merge function conforms to the <span class="type">chunk_merge_t</span>
1081        type and optionally merges adjacent chunks,
1082        <em class="parameter"><code>chunk_a</code></em> of given <em class="parameter"><code>size_a</code></em>
1083        and <em class="parameter"><code>chunk_b</code></em> of given
1084        <em class="parameter"><code>size_b</code></em> into one contiguous chunk, operating on
1085        <em class="parameter"><code>committed</code></em>/decommitted memory as indicated, on
1086        behalf of arena <em class="parameter"><code>arena_ind</code></em>, returning false upon
1087        success.  If the function returns true, this indicates that the chunks
1088        remain distinct mappings and therefore should continue to be operated on
1089        independently.</p></dd><dt><a name="arenas.narenas"></a><span class="term">
1090
1091    "<code class="mallctl">arenas.narenas</code>"
1092
1093          (<span class="type">unsigned</span>)
1094          <code class="literal">r-</code>
1095        </span></dt><dd><p>Current limit on number of arenas.</p></dd><dt><a name="arenas.initialized"></a><span class="term">
1096
1097    "<code class="mallctl">arenas.initialized</code>"
1098
1099          (<span class="type">bool *</span>)
1100          <code class="literal">r-</code>
1101        </span></dt><dd><p>An array of <a class="link" href="#arenas.narenas">
1102    "<code class="mallctl">arenas.narenas</code>"
1103  </a>
1104        booleans.  Each boolean indicates whether the corresponding arena is
1105        initialized.</p></dd><dt><a name="arenas.lg_dirty_mult"></a><span class="term">
1106
1107    "<code class="mallctl">arenas.lg_dirty_mult</code>"
1108
1109          (<span class="type">ssize_t</span>)
1110          <code class="literal">rw</code>
1111        </span></dt><dd><p>Current default per-arena minimum ratio (log base 2) of
1112        active to dirty pages, used to initialize <a class="link" href="#arena.i.lg_dirty_mult">
1113    "<code class="mallctl">arena.&lt;i&gt;.lg_dirty_mult</code>"
1114  </a>
1115        during arena creation.  See <a class="link" href="#opt.lg_dirty_mult">
1116    "<code class="mallctl">opt.lg_dirty_mult</code>"
1117  </a>
1118        for additional information.</p></dd><dt><a name="arenas.quantum"></a><span class="term">
1119
1120    "<code class="mallctl">arenas.quantum</code>"
1121
1122          (<span class="type">size_t</span>)
1123          <code class="literal">r-</code>
1124        </span></dt><dd><p>Quantum size.</p></dd><dt><a name="arenas.page"></a><span class="term">
1125
1126    "<code class="mallctl">arenas.page</code>"
1127
1128          (<span class="type">size_t</span>)
1129          <code class="literal">r-</code>
1130        </span></dt><dd><p>Page size.</p></dd><dt><a name="arenas.tcache_max"></a><span class="term">
1131
1132    "<code class="mallctl">arenas.tcache_max</code>"
1133
1134          (<span class="type">size_t</span>)
1135          <code class="literal">r-</code>
1136          [<code class="option">--enable-tcache</code>]
1137        </span></dt><dd><p>Maximum thread-cached size class.</p></dd><dt><a name="arenas.nbins"></a><span class="term">
1138
1139    "<code class="mallctl">arenas.nbins</code>"
1140
1141          (<span class="type">unsigned</span>)
1142          <code class="literal">r-</code>
1143        </span></dt><dd><p>Number of bin size classes.</p></dd><dt><a name="arenas.nhbins"></a><span class="term">
1144
1145    "<code class="mallctl">arenas.nhbins</code>"
1146
1147          (<span class="type">unsigned</span>)
1148          <code class="literal">r-</code>
1149          [<code class="option">--enable-tcache</code>]
1150        </span></dt><dd><p>Total number of thread cache bin size
1151        classes.</p></dd><dt><a name="arenas.bin.i.size"></a><span class="term">
1152
1153    "<code class="mallctl">arenas.bin.&lt;i&gt;.size</code>"
1154
1155          (<span class="type">size_t</span>)
1156          <code class="literal">r-</code>
1157        </span></dt><dd><p>Maximum size supported by size class.</p></dd><dt><a name="arenas.bin.i.nregs"></a><span class="term">
1158
1159    "<code class="mallctl">arenas.bin.&lt;i&gt;.nregs</code>"
1160
1161          (<span class="type">uint32_t</span>)
1162          <code class="literal">r-</code>
1163        </span></dt><dd><p>Number of regions per page run.</p></dd><dt><a name="arenas.bin.i.run_size"></a><span class="term">
1164
1165    "<code class="mallctl">arenas.bin.&lt;i&gt;.run_size</code>"
1166
1167          (<span class="type">size_t</span>)
1168          <code class="literal">r-</code>
1169        </span></dt><dd><p>Number of bytes per page run.</p></dd><dt><a name="arenas.nlruns"></a><span class="term">
1170
1171    "<code class="mallctl">arenas.nlruns</code>"
1172
1173          (<span class="type">unsigned</span>)
1174          <code class="literal">r-</code>
1175        </span></dt><dd><p>Total number of large size classes.</p></dd><dt><a name="arenas.lrun.i.size"></a><span class="term">
1176
1177    "<code class="mallctl">arenas.lrun.&lt;i&gt;.size</code>"
1178
1179          (<span class="type">size_t</span>)
1180          <code class="literal">r-</code>
1181        </span></dt><dd><p>Maximum size supported by this large size
1182        class.</p></dd><dt><a name="arenas.nhchunks"></a><span class="term">
1183
1184    "<code class="mallctl">arenas.nhchunks</code>"
1185
1186          (<span class="type">unsigned</span>)
1187          <code class="literal">r-</code>
1188        </span></dt><dd><p>Total number of huge size classes.</p></dd><dt><a name="arenas.hchunk.i.size"></a><span class="term">
1189
1190    "<code class="mallctl">arenas.hchunk.&lt;i&gt;.size</code>"
1191
1192          (<span class="type">size_t</span>)
1193          <code class="literal">r-</code>
1194        </span></dt><dd><p>Maximum size supported by this huge size
1195        class.</p></dd><dt><a name="arenas.extend"></a><span class="term">
1196
1197    "<code class="mallctl">arenas.extend</code>"
1198
1199          (<span class="type">unsigned</span>)
1200          <code class="literal">r-</code>
1201        </span></dt><dd><p>Extend the array of arenas by appending a new arena,
1202        and returning the new arena index.</p></dd><dt><a name="prof.thread_active_init"></a><span class="term">
1203
1204    "<code class="mallctl">prof.thread_active_init</code>"
1205
1206          (<span class="type">bool</span>)
1207          <code class="literal">rw</code>
1208          [<code class="option">--enable-prof</code>]
1209        </span></dt><dd><p>Control the initial setting for <a class="link" href="#thread.prof.active">
1210    "<code class="mallctl">thread.prof.active</code>"
1211  </a>
1212        in newly created threads.  See the <a class="link" href="#opt.prof_thread_active_init">
1213    "<code class="mallctl">opt.prof_thread_active_init</code>"
1214  </a>
1215        option for additional information.</p></dd><dt><a name="prof.active"></a><span class="term">
1216
1217    "<code class="mallctl">prof.active</code>"
1218
1219          (<span class="type">bool</span>)
1220          <code class="literal">rw</code>
1221          [<code class="option">--enable-prof</code>]
1222        </span></dt><dd><p>Control whether sampling is currently active.  See the
1223        <a class="link" href="#opt.prof_active">
1224    "<code class="mallctl">opt.prof_active</code>"
1225  </a>
1226        option for additional information, as well as the interrelated <a class="link" href="#thread.prof.active">
1227    "<code class="mallctl">thread.prof.active</code>"
1228  </a>
1229        mallctl.</p></dd><dt><a name="prof.dump"></a><span class="term">
1230
1231    "<code class="mallctl">prof.dump</code>"
1232
1233          (<span class="type">const char *</span>)
1234          <code class="literal">-w</code>
1235          [<code class="option">--enable-prof</code>]
1236        </span></dt><dd><p>Dump a memory profile to the specified file, or if NULL
1237        is specified, to a file according to the pattern
1238        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.m&lt;mseq&gt;.heap</code>,
1239        where <code class="literal">&lt;prefix&gt;</code> is controlled by the
1240        <a class="link" href="#opt.prof_prefix">
1241    "<code class="mallctl">opt.prof_prefix</code>"
1242  </a>
1243        option.</p></dd><dt><a name="prof.gdump"></a><span class="term">
1244
1245    "<code class="mallctl">prof.gdump</code>"
1246
1247          (<span class="type">bool</span>)
1248          <code class="literal">rw</code>
1249          [<code class="option">--enable-prof</code>]
1250        </span></dt><dd><p>When enabled, trigger a memory profile dump every time
1251        the total virtual memory exceeds the previous maximum.  Profiles are
1252        dumped to files named according to the pattern
1253        <code class="filename">&lt;prefix&gt;.&lt;pid&gt;.&lt;seq&gt;.u&lt;useq&gt;.heap</code>,
1254        where <code class="literal">&lt;prefix&gt;</code> is controlled by the <a class="link" href="#opt.prof_prefix">
1255    "<code class="mallctl">opt.prof_prefix</code>"
1256  </a>
1257        option.</p></dd><dt><a name="prof.reset"></a><span class="term">
1258
1259    "<code class="mallctl">prof.reset</code>"
1260
1261          (<span class="type">size_t</span>)
1262          <code class="literal">-w</code>
1263          [<code class="option">--enable-prof</code>]
1264        </span></dt><dd><p>Reset all memory profile statistics, and optionally
1265        update the sample rate (see <a class="link" href="#opt.lg_prof_sample">
1266    "<code class="mallctl">opt.lg_prof_sample</code>"
1267  </a>
1268        and <a class="link" href="#prof.lg_sample">
1269    "<code class="mallctl">prof.lg_sample</code>"
1270  </a>).
1271        </p></dd><dt><a name="prof.lg_sample"></a><span class="term">
1272
1273    "<code class="mallctl">prof.lg_sample</code>"
1274
1275          (<span class="type">size_t</span>)
1276          <code class="literal">r-</code>
1277          [<code class="option">--enable-prof</code>]
1278        </span></dt><dd><p>Get the current sample rate (see <a class="link" href="#opt.lg_prof_sample">
1279    "<code class="mallctl">opt.lg_prof_sample</code>"
1280  </a>).
1281        </p></dd><dt><a name="prof.interval"></a><span class="term">
1282
1283    "<code class="mallctl">prof.interval</code>"
1284
1285          (<span class="type">uint64_t</span>)
1286          <code class="literal">r-</code>
1287          [<code class="option">--enable-prof</code>]
1288        </span></dt><dd><p>Average number of bytes allocated between
1289        inverval-based profile dumps.  See the
1290        <a class="link" href="#opt.lg_prof_interval">
1291    "<code class="mallctl">opt.lg_prof_interval</code>"
1292  </a>
1293        option for additional information.</p></dd><dt><a name="stats.cactive"></a><span class="term">
1294
1295    "<code class="mallctl">stats.cactive</code>"
1296
1297          (<span class="type">size_t *</span>)
1298          <code class="literal">r-</code>
1299          [<code class="option">--enable-stats</code>]
1300        </span></dt><dd><p>Pointer to a counter that contains an approximate count
1301        of the current number of bytes in active pages.  The estimate may be
1302        high, but never low, because each arena rounds up when computing its
1303        contribution to the counter.  Note that the <a class="link" href="#epoch">
1304    "<code class="mallctl">epoch</code>"
1305  </a> mallctl has no bearing
1306        on this counter.  Furthermore, counter consistency is maintained via
1307        atomic operations, so it is necessary to use an atomic operation in
1308        order to guarantee a consistent read when dereferencing the pointer.
1309        </p></dd><dt><a name="stats.allocated"></a><span class="term">
1310
1311    "<code class="mallctl">stats.allocated</code>"
1312
1313          (<span class="type">size_t</span>)
1314          <code class="literal">r-</code>
1315          [<code class="option">--enable-stats</code>]
1316        </span></dt><dd><p>Total number of bytes allocated by the
1317        application.</p></dd><dt><a name="stats.active"></a><span class="term">
1318
1319    "<code class="mallctl">stats.active</code>"
1320
1321          (<span class="type">size_t</span>)
1322          <code class="literal">r-</code>
1323          [<code class="option">--enable-stats</code>]
1324        </span></dt><dd><p>Total number of bytes in active pages allocated by the
1325        application.  This is a multiple of the page size, and greater than or
1326        equal to <a class="link" href="#stats.allocated">
1327    "<code class="mallctl">stats.allocated</code>"
1328  </a>.
1329        This does not include <a class="link" href="#stats.arenas.i.pdirty">
1330
1331    "<code class="mallctl">stats.arenas.&lt;i&gt;.pdirty</code>"
1332  </a>, nor pages
1333        entirely devoted to allocator metadata.</p></dd><dt><a name="stats.metadata"></a><span class="term">
1334
1335    "<code class="mallctl">stats.metadata</code>"
1336
1337          (<span class="type">size_t</span>)
1338          <code class="literal">r-</code>
1339          [<code class="option">--enable-stats</code>]
1340        </span></dt><dd><p>Total number of bytes dedicated to metadata, which
1341        comprise base allocations used for bootstrap-sensitive internal
1342        allocator data structures, arena chunk headers (see <a class="link" href="#stats.arenas.i.metadata.mapped">
1343    "<code class="mallctl">stats.arenas.&lt;i&gt;.metadata.mapped</code>"
1344  </a>),
1345        and internal allocations (see <a class="link" href="#stats.arenas.i.metadata.allocated">
1346    "<code class="mallctl">stats.arenas.&lt;i&gt;.metadata.allocated</code>"
1347  </a>).</p></dd><dt><a name="stats.resident"></a><span class="term">
1348
1349    "<code class="mallctl">stats.resident</code>"
1350
1351          (<span class="type">size_t</span>)
1352          <code class="literal">r-</code>
1353          [<code class="option">--enable-stats</code>]
1354        </span></dt><dd><p>Maximum number of bytes in physically resident data
1355        pages mapped by the allocator, comprising all pages dedicated to
1356        allocator metadata, pages backing active allocations, and unused dirty
1357        pages.  This is a maximum rather than precise because pages may not
1358        actually be physically resident if they correspond to demand-zeroed
1359        virtual memory that has not yet been touched.  This is a multiple of the
1360        page size, and is larger than <a class="link" href="#stats.active">
1361    "<code class="mallctl">stats.active</code>"
1362  </a>.</p></dd><dt><a name="stats.mapped"></a><span class="term">
1363
1364    "<code class="mallctl">stats.mapped</code>"
1365
1366          (<span class="type">size_t</span>)
1367          <code class="literal">r-</code>
1368          [<code class="option">--enable-stats</code>]
1369        </span></dt><dd><p>Total number of bytes in active chunks mapped by the
1370        allocator.  This is a multiple of the chunk size, and is larger than
1371        <a class="link" href="#stats.active">
1372    "<code class="mallctl">stats.active</code>"
1373  </a>.
1374        This does not include inactive chunks, even those that contain unused
1375        dirty pages, which means that there is no strict ordering between this
1376        and <a class="link" href="#stats.resident">
1377    "<code class="mallctl">stats.resident</code>"
1378  </a>.</p></dd><dt><a name="stats.arenas.i.dss"></a><span class="term">
1379
1380    "<code class="mallctl">stats.arenas.&lt;i&gt;.dss</code>"
1381
1382          (<span class="type">const char *</span>)
1383          <code class="literal">r-</code>
1384        </span></dt><dd><p>dss (<span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>) allocation precedence as
1385        related to <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span> allocation.  See <a class="link" href="#opt.dss">
1386    "<code class="mallctl">opt.dss</code>"
1387  </a> for details.
1388        </p></dd><dt><a name="stats.arenas.i.lg_dirty_mult"></a><span class="term">
1389
1390    "<code class="mallctl">stats.arenas.&lt;i&gt;.lg_dirty_mult</code>"
1391
1392          (<span class="type">ssize_t</span>)
1393          <code class="literal">r-</code>
1394        </span></dt><dd><p>Minimum ratio (log base 2) of active to dirty pages.
1395        See <a class="link" href="#opt.lg_dirty_mult">
1396    "<code class="mallctl">opt.lg_dirty_mult</code>"
1397  </a>
1398        for details.</p></dd><dt><a name="stats.arenas.i.nthreads"></a><span class="term">
1399
1400    "<code class="mallctl">stats.arenas.&lt;i&gt;.nthreads</code>"
1401
1402          (<span class="type">unsigned</span>)
1403          <code class="literal">r-</code>
1404        </span></dt><dd><p>Number of threads currently assigned to
1405        arena.</p></dd><dt><a name="stats.arenas.i.pactive"></a><span class="term">
1406
1407    "<code class="mallctl">stats.arenas.&lt;i&gt;.pactive</code>"
1408
1409          (<span class="type">size_t</span>)
1410          <code class="literal">r-</code>
1411        </span></dt><dd><p>Number of pages in active runs.</p></dd><dt><a name="stats.arenas.i.pdirty"></a><span class="term">
1412
1413    "<code class="mallctl">stats.arenas.&lt;i&gt;.pdirty</code>"
1414
1415          (<span class="type">size_t</span>)
1416          <code class="literal">r-</code>
1417        </span></dt><dd><p>Number of pages within unused runs that are potentially
1418        dirty, and for which <code class="function">madvise</code>(<em class="parameter"><code>...</code></em>,
1419        <em class="parameter"><code><code class="constant">MADV_DONTNEED</code></code></em>) or
1420        similar has not been called.</p></dd><dt><a name="stats.arenas.i.mapped"></a><span class="term">
1421
1422    "<code class="mallctl">stats.arenas.&lt;i&gt;.mapped</code>"
1423
1424          (<span class="type">size_t</span>)
1425          <code class="literal">r-</code>
1426          [<code class="option">--enable-stats</code>]
1427        </span></dt><dd><p>Number of mapped bytes.</p></dd><dt><a name="stats.arenas.i.metadata.mapped"></a><span class="term">
1428
1429    "<code class="mallctl">stats.arenas.&lt;i&gt;.metadata.mapped</code>"
1430
1431          (<span class="type">size_t</span>)
1432          <code class="literal">r-</code>
1433          [<code class="option">--enable-stats</code>]
1434        </span></dt><dd><p>Number of mapped bytes in arena chunk headers, which
1435        track the states of the non-metadata pages.</p></dd><dt><a name="stats.arenas.i.metadata.allocated"></a><span class="term">
1436
1437    "<code class="mallctl">stats.arenas.&lt;i&gt;.metadata.allocated</code>"
1438
1439          (<span class="type">size_t</span>)
1440          <code class="literal">r-</code>
1441          [<code class="option">--enable-stats</code>]
1442        </span></dt><dd><p>Number of bytes dedicated to internal allocations.
1443        Internal allocations differ from application-originated allocations in
1444        that they are for internal use, and that they are omitted from heap
1445        profiles.  This statistic is reported separately from <a class="link" href="#stats.metadata">
1446    "<code class="mallctl">stats.metadata</code>"
1447  </a> and
1448        <a class="link" href="#stats.arenas.i.metadata.mapped">
1449    "<code class="mallctl">stats.arenas.&lt;i&gt;.metadata.mapped</code>"
1450  </a>
1451        because it overlaps with e.g. the <a class="link" href="#stats.allocated">
1452    "<code class="mallctl">stats.allocated</code>"
1453  </a> and
1454        <a class="link" href="#stats.active">
1455    "<code class="mallctl">stats.active</code>"
1456  </a>
1457        statistics, whereas the other metadata statistics do
1458        not.</p></dd><dt><a name="stats.arenas.i.npurge"></a><span class="term">
1459
1460    "<code class="mallctl">stats.arenas.&lt;i&gt;.npurge</code>"
1461
1462          (<span class="type">uint64_t</span>)
1463          <code class="literal">r-</code>
1464          [<code class="option">--enable-stats</code>]
1465        </span></dt><dd><p>Number of dirty page purge sweeps performed.
1466        </p></dd><dt><a name="stats.arenas.i.nmadvise"></a><span class="term">
1467
1468    "<code class="mallctl">stats.arenas.&lt;i&gt;.nmadvise</code>"
1469
1470          (<span class="type">uint64_t</span>)
1471          <code class="literal">r-</code>
1472          [<code class="option">--enable-stats</code>]
1473        </span></dt><dd><p>Number of <code class="function">madvise</code>(<em class="parameter"><code>...</code></em>,
1474        <em class="parameter"><code><code class="constant">MADV_DONTNEED</code></code></em>) or
1475        similar calls made to purge dirty pages.</p></dd><dt><a name="stats.arenas.i.purged"></a><span class="term">
1476
1477    "<code class="mallctl">stats.arenas.&lt;i&gt;.purged</code>"
1478
1479          (<span class="type">uint64_t</span>)
1480          <code class="literal">r-</code>
1481          [<code class="option">--enable-stats</code>]
1482        </span></dt><dd><p>Number of pages purged.</p></dd><dt><a name="stats.arenas.i.small.allocated"></a><span class="term">
1483
1484    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.allocated</code>"
1485
1486          (<span class="type">size_t</span>)
1487          <code class="literal">r-</code>
1488          [<code class="option">--enable-stats</code>]
1489        </span></dt><dd><p>Number of bytes currently allocated by small objects.
1490        </p></dd><dt><a name="stats.arenas.i.small.nmalloc"></a><span class="term">
1491
1492    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.nmalloc</code>"
1493
1494          (<span class="type">uint64_t</span>)
1495          <code class="literal">r-</code>
1496          [<code class="option">--enable-stats</code>]
1497        </span></dt><dd><p>Cumulative number of allocation requests served by
1498        small bins.</p></dd><dt><a name="stats.arenas.i.small.ndalloc"></a><span class="term">
1499
1500    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.ndalloc</code>"
1501
1502          (<span class="type">uint64_t</span>)
1503          <code class="literal">r-</code>
1504          [<code class="option">--enable-stats</code>]
1505        </span></dt><dd><p>Cumulative number of small objects returned to bins.
1506        </p></dd><dt><a name="stats.arenas.i.small.nrequests"></a><span class="term">
1507
1508    "<code class="mallctl">stats.arenas.&lt;i&gt;.small.nrequests</code>"
1509
1510          (<span class="type">uint64_t</span>)
1511          <code class="literal">r-</code>
1512          [<code class="option">--enable-stats</code>]
1513        </span></dt><dd><p>Cumulative number of small allocation requests.
1514        </p></dd><dt><a name="stats.arenas.i.large.allocated"></a><span class="term">
1515
1516    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.allocated</code>"
1517
1518          (<span class="type">size_t</span>)
1519          <code class="literal">r-</code>
1520          [<code class="option">--enable-stats</code>]
1521        </span></dt><dd><p>Number of bytes currently allocated by large objects.
1522        </p></dd><dt><a name="stats.arenas.i.large.nmalloc"></a><span class="term">
1523
1524    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.nmalloc</code>"
1525
1526          (<span class="type">uint64_t</span>)
1527          <code class="literal">r-</code>
1528          [<code class="option">--enable-stats</code>]
1529        </span></dt><dd><p>Cumulative number of large allocation requests served
1530        directly by the arena.</p></dd><dt><a name="stats.arenas.i.large.ndalloc"></a><span class="term">
1531
1532    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.ndalloc</code>"
1533
1534          (<span class="type">uint64_t</span>)
1535          <code class="literal">r-</code>
1536          [<code class="option">--enable-stats</code>]
1537        </span></dt><dd><p>Cumulative number of large deallocation requests served
1538        directly by the arena.</p></dd><dt><a name="stats.arenas.i.large.nrequests"></a><span class="term">
1539
1540    "<code class="mallctl">stats.arenas.&lt;i&gt;.large.nrequests</code>"
1541
1542          (<span class="type">uint64_t</span>)
1543          <code class="literal">r-</code>
1544          [<code class="option">--enable-stats</code>]
1545        </span></dt><dd><p>Cumulative number of large allocation requests.
1546        </p></dd><dt><a name="stats.arenas.i.huge.allocated"></a><span class="term">
1547
1548    "<code class="mallctl">stats.arenas.&lt;i&gt;.huge.allocated</code>"
1549
1550          (<span class="type">size_t</span>)
1551          <code class="literal">r-</code>
1552          [<code class="option">--enable-stats</code>]
1553        </span></dt><dd><p>Number of bytes currently allocated by huge objects.
1554        </p></dd><dt><a name="stats.arenas.i.huge.nmalloc"></a><span class="term">
1555
1556    "<code class="mallctl">stats.arenas.&lt;i&gt;.huge.nmalloc</code>"
1557
1558          (<span class="type">uint64_t</span>)
1559          <code class="literal">r-</code>
1560          [<code class="option">--enable-stats</code>]
1561        </span></dt><dd><p>Cumulative number of huge allocation requests served
1562        directly by the arena.</p></dd><dt><a name="stats.arenas.i.huge.ndalloc"></a><span class="term">
1563
1564    "<code class="mallctl">stats.arenas.&lt;i&gt;.huge.ndalloc</code>"
1565
1566          (<span class="type">uint64_t</span>)
1567          <code class="literal">r-</code>
1568          [<code class="option">--enable-stats</code>]
1569        </span></dt><dd><p>Cumulative number of huge deallocation requests served
1570        directly by the arena.</p></dd><dt><a name="stats.arenas.i.huge.nrequests"></a><span class="term">
1571
1572    "<code class="mallctl">stats.arenas.&lt;i&gt;.huge.nrequests</code>"
1573
1574          (<span class="type">uint64_t</span>)
1575          <code class="literal">r-</code>
1576          [<code class="option">--enable-stats</code>]
1577        </span></dt><dd><p>Cumulative number of huge allocation requests.
1578        </p></dd><dt><a name="stats.arenas.i.bins.j.nmalloc"></a><span class="term">
1579
1580    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nmalloc</code>"
1581
1582          (<span class="type">uint64_t</span>)
1583          <code class="literal">r-</code>
1584          [<code class="option">--enable-stats</code>]
1585        </span></dt><dd><p>Cumulative number of allocations served by bin.
1586        </p></dd><dt><a name="stats.arenas.i.bins.j.ndalloc"></a><span class="term">
1587
1588    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.ndalloc</code>"
1589
1590          (<span class="type">uint64_t</span>)
1591          <code class="literal">r-</code>
1592          [<code class="option">--enable-stats</code>]
1593        </span></dt><dd><p>Cumulative number of allocations returned to bin.
1594        </p></dd><dt><a name="stats.arenas.i.bins.j.nrequests"></a><span class="term">
1595
1596    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nrequests</code>"
1597
1598          (<span class="type">uint64_t</span>)
1599          <code class="literal">r-</code>
1600          [<code class="option">--enable-stats</code>]
1601        </span></dt><dd><p>Cumulative number of allocation
1602        requests.</p></dd><dt><a name="stats.arenas.i.bins.j.curregs"></a><span class="term">
1603
1604    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curregs</code>"
1605
1606          (<span class="type">size_t</span>)
1607          <code class="literal">r-</code>
1608          [<code class="option">--enable-stats</code>]
1609        </span></dt><dd><p>Current number of regions for this size
1610        class.</p></dd><dt><a name="stats.arenas.i.bins.j.nfills"></a><span class="term">
1611
1612    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nfills</code>"
1613
1614          (<span class="type">uint64_t</span>)
1615          <code class="literal">r-</code>
1616          [<code class="option">--enable-stats</code> <code class="option">--enable-tcache</code>]
1617        </span></dt><dd><p>Cumulative number of tcache fills.</p></dd><dt><a name="stats.arenas.i.bins.j.nflushes"></a><span class="term">
1618
1619    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nflushes</code>"
1620
1621          (<span class="type">uint64_t</span>)
1622          <code class="literal">r-</code>
1623          [<code class="option">--enable-stats</code> <code class="option">--enable-tcache</code>]
1624        </span></dt><dd><p>Cumulative number of tcache flushes.</p></dd><dt><a name="stats.arenas.i.bins.j.nruns"></a><span class="term">
1625
1626    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nruns</code>"
1627
1628          (<span class="type">uint64_t</span>)
1629          <code class="literal">r-</code>
1630          [<code class="option">--enable-stats</code>]
1631        </span></dt><dd><p>Cumulative number of runs created.</p></dd><dt><a name="stats.arenas.i.bins.j.nreruns"></a><span class="term">
1632
1633    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.nreruns</code>"
1634
1635          (<span class="type">uint64_t</span>)
1636          <code class="literal">r-</code>
1637          [<code class="option">--enable-stats</code>]
1638        </span></dt><dd><p>Cumulative number of times the current run from which
1639        to allocate changed.</p></dd><dt><a name="stats.arenas.i.bins.j.curruns"></a><span class="term">
1640
1641    "<code class="mallctl">stats.arenas.&lt;i&gt;.bins.&lt;j&gt;.curruns</code>"
1642
1643          (<span class="type">size_t</span>)
1644          <code class="literal">r-</code>
1645          [<code class="option">--enable-stats</code>]
1646        </span></dt><dd><p>Current number of runs.</p></dd><dt><a name="stats.arenas.i.lruns.j.nmalloc"></a><span class="term">
1647
1648    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nmalloc</code>"
1649
1650          (<span class="type">uint64_t</span>)
1651          <code class="literal">r-</code>
1652          [<code class="option">--enable-stats</code>]
1653        </span></dt><dd><p>Cumulative number of allocation requests for this size
1654        class served directly by the arena.</p></dd><dt><a name="stats.arenas.i.lruns.j.ndalloc"></a><span class="term">
1655
1656    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.ndalloc</code>"
1657
1658          (<span class="type">uint64_t</span>)
1659          <code class="literal">r-</code>
1660          [<code class="option">--enable-stats</code>]
1661        </span></dt><dd><p>Cumulative number of deallocation requests for this
1662        size class served directly by the arena.</p></dd><dt><a name="stats.arenas.i.lruns.j.nrequests"></a><span class="term">
1663
1664    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.nrequests</code>"
1665
1666          (<span class="type">uint64_t</span>)
1667          <code class="literal">r-</code>
1668          [<code class="option">--enable-stats</code>]
1669        </span></dt><dd><p>Cumulative number of allocation requests for this size
1670        class.</p></dd><dt><a name="stats.arenas.i.lruns.j.curruns"></a><span class="term">
1671
1672    "<code class="mallctl">stats.arenas.&lt;i&gt;.lruns.&lt;j&gt;.curruns</code>"
1673
1674          (<span class="type">size_t</span>)
1675          <code class="literal">r-</code>
1676          [<code class="option">--enable-stats</code>]
1677        </span></dt><dd><p>Current number of runs for this size class.
1678        </p></dd><dt><a name="stats.arenas.i.hchunks.j.nmalloc"></a><span class="term">
1679
1680    "<code class="mallctl">stats.arenas.&lt;i&gt;.hchunks.&lt;j&gt;.nmalloc</code>"
1681
1682          (<span class="type">uint64_t</span>)
1683          <code class="literal">r-</code>
1684          [<code class="option">--enable-stats</code>]
1685        </span></dt><dd><p>Cumulative number of allocation requests for this size
1686        class served directly by the arena.</p></dd><dt><a name="stats.arenas.i.hchunks.j.ndalloc"></a><span class="term">
1687
1688    "<code class="mallctl">stats.arenas.&lt;i&gt;.hchunks.&lt;j&gt;.ndalloc</code>"
1689
1690          (<span class="type">uint64_t</span>)
1691          <code class="literal">r-</code>
1692          [<code class="option">--enable-stats</code>]
1693        </span></dt><dd><p>Cumulative number of deallocation requests for this
1694        size class served directly by the arena.</p></dd><dt><a name="stats.arenas.i.hchunks.j.nrequests"></a><span class="term">
1695
1696    "<code class="mallctl">stats.arenas.&lt;i&gt;.hchunks.&lt;j&gt;.nrequests</code>"
1697
1698          (<span class="type">uint64_t</span>)
1699          <code class="literal">r-</code>
1700          [<code class="option">--enable-stats</code>]
1701        </span></dt><dd><p>Cumulative number of allocation requests for this size
1702        class.</p></dd><dt><a name="stats.arenas.i.hchunks.j.curhchunks"></a><span class="term">
1703
1704    "<code class="mallctl">stats.arenas.&lt;i&gt;.hchunks.&lt;j&gt;.curhchunks</code>"
1705
1706          (<span class="type">size_t</span>)
1707          <code class="literal">r-</code>
1708          [<code class="option">--enable-stats</code>]
1709        </span></dt><dd><p>Current number of huge allocations for this size class.
1710        </p></dd></dl></div></div><div class="refsect1"><a name="debugging_malloc_problems"></a><h2>DEBUGGING MALLOC PROBLEMS</h2><p>When debugging, it is a good idea to configure/build jemalloc with
1711    the <code class="option">--enable-debug</code> and <code class="option">--enable-fill</code>
1712    options, and recompile the program with suitable options and symbols for
1713    debugger support.  When so configured, jemalloc incorporates a wide variety
1714    of run-time assertions that catch application errors such as double-free,
1715    write-after-free, etc.</p><p>Programs often accidentally depend on &#8220;uninitialized&#8221;
1716    memory actually being filled with zero bytes.  Junk filling
1717    (see the <a class="link" href="#opt.junk">
1718    "<code class="mallctl">opt.junk</code>"
1719  </a>
1720    option) tends to expose such bugs in the form of obviously incorrect
1721    results and/or coredumps.  Conversely, zero
1722    filling (see the <a class="link" href="#opt.zero">
1723    "<code class="mallctl">opt.zero</code>"
1724  </a> option) eliminates
1725    the symptoms of such bugs.  Between these two options, it is usually
1726    possible to quickly detect, diagnose, and eliminate such bugs.</p><p>This implementation does not provide much detail about the problems
1727    it detects, because the performance impact for storing such information
1728    would be prohibitive.  However, jemalloc does integrate with the most
1729    excellent <a class="ulink" href="http://valgrind.org/" target="_top">Valgrind</a> tool if the
1730    <code class="option">--enable-valgrind</code> configuration option is enabled.</p></div><div class="refsect1"><a name="diagnostic_messages"></a><h2>DIAGNOSTIC MESSAGES</h2><p>If any of the memory allocation/deallocation functions detect an
1731    error or warning condition, a message will be printed to file descriptor
1732    <code class="constant">STDERR_FILENO</code>.  Errors will result in the process
1733    dumping core.  If the <a class="link" href="#opt.abort">
1734    "<code class="mallctl">opt.abort</code>"
1735  </a> option is set, most
1736    warnings are treated as errors.</p><p>The <code class="varname">malloc_message</code> variable allows the programmer
1737    to override the function which emits the text strings forming the errors
1738    and warnings if for some reason the <code class="constant">STDERR_FILENO</code> file
1739    descriptor is not suitable for this.
1740    <code class="function">malloc_message</code>(<em class="parameter"><code></code></em>) takes the
1741    <em class="parameter"><code>cbopaque</code></em> pointer argument that is
1742    <code class="constant">NULL</code> unless overridden by the arguments in a call to
1743    <code class="function">malloc_stats_print</code>(<em class="parameter"><code></code></em>), followed by a string
1744    pointer.  Please note that doing anything which tries to allocate memory in
1745    this function is likely to result in a crash or deadlock.</p><p>All messages are prefixed by
1746    &#8220;<code class="computeroutput">&lt;jemalloc&gt;: </code>&#8221;.</p></div><div class="refsect1"><a name="return_values"></a><h2>RETURN VALUES</h2><div class="refsect2"><a name="idp46949776"></a><h3>Standard API</h3><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>) and
1747      <code class="function">calloc</code>(<em class="parameter"><code></code></em>) functions return a pointer to the
1748      allocated memory if successful; otherwise a <code class="constant">NULL</code>
1749      pointer is returned and <code class="varname">errno</code> is set to
1750      <span class="errorname">ENOMEM</span>.</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function
1751      returns the value 0 if successful; otherwise it returns an error value.
1752      The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function will fail
1753      if:
1754        </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p>The <em class="parameter"><code>alignment</code></em> parameter is
1755            not a power of 2 at least as large as
1756            <code class="code">sizeof(<span class="type">void *</span>)</code>.
1757            </p></dd><dt><span class="term"><span class="errorname">ENOMEM</span></span></dt><dd><p>Memory allocation error.</p></dd></dl></div><p>
1758      </p><p>The <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function returns
1759      a pointer to the allocated memory if successful; otherwise a
1760      <code class="constant">NULL</code> pointer is returned and
1761      <code class="varname">errno</code> is set.  The
1762      <code class="function">aligned_alloc</code>(<em class="parameter"><code></code></em>) function will fail if:
1763        </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p>The <em class="parameter"><code>alignment</code></em> parameter is
1764            not a power of 2.
1765            </p></dd><dt><span class="term"><span class="errorname">ENOMEM</span></span></dt><dd><p>Memory allocation error.</p></dd></dl></div><p>
1766      </p><p>The <code class="function">realloc</code>(<em class="parameter"><code></code></em>) function returns a
1767      pointer, possibly identical to <em class="parameter"><code>ptr</code></em>, to the
1768      allocated memory if successful; otherwise a <code class="constant">NULL</code>
1769      pointer is returned, and <code class="varname">errno</code> is set to
1770      <span class="errorname">ENOMEM</span> if the error was the result of an
1771      allocation failure.  The <code class="function">realloc</code>(<em class="parameter"><code></code></em>)
1772      function always leaves the original buffer intact when an error occurs.
1773      </p><p>The <code class="function">free</code>(<em class="parameter"><code></code></em>) function returns no
1774      value.</p></div><div class="refsect2"><a name="idp46974576"></a><h3>Non-standard API</h3><p>The <code class="function">mallocx</code>(<em class="parameter"><code></code></em>) and
1775      <code class="function">rallocx</code>(<em class="parameter"><code></code></em>) functions return a pointer to
1776      the allocated memory if successful; otherwise a <code class="constant">NULL</code>
1777      pointer is returned to indicate insufficient contiguous memory was
1778      available to service the allocation request.  </p><p>The <code class="function">xallocx</code>(<em class="parameter"><code></code></em>) function returns the
1779      real size of the resulting resized allocation pointed to by
1780      <em class="parameter"><code>ptr</code></em>, which is a value less than
1781      <em class="parameter"><code>size</code></em> if the allocation could not be adequately
1782      grown in place.  </p><p>The <code class="function">sallocx</code>(<em class="parameter"><code></code></em>) function returns the
1783      real size of the allocation pointed to by <em class="parameter"><code>ptr</code></em>.
1784      </p><p>The <code class="function">nallocx</code>(<em class="parameter"><code></code></em>) returns the real size
1785      that would result from a successful equivalent
1786      <code class="function">mallocx</code>(<em class="parameter"><code></code></em>) function call, or zero if
1787      insufficient memory is available to perform the size computation.  </p><p>The <code class="function">mallctl</code>(<em class="parameter"><code></code></em>),
1788      <code class="function">mallctlnametomib</code>(<em class="parameter"><code></code></em>), and
1789      <code class="function">mallctlbymib</code>(<em class="parameter"><code></code></em>) functions return 0 on
1790      success; otherwise they return an error value.  The functions will fail
1791      if:
1792        </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><span class="errorname">EINVAL</span></span></dt><dd><p><em class="parameter"><code>newp</code></em> is not
1793            <code class="constant">NULL</code>, and <em class="parameter"><code>newlen</code></em> is too
1794            large or too small.  Alternatively, <em class="parameter"><code>*oldlenp</code></em>
1795            is too large or too small; in this case as much data as possible
1796            are read despite the error.</p></dd><dt><span class="term"><span class="errorname">ENOENT</span></span></dt><dd><p><em class="parameter"><code>name</code></em> or
1797            <em class="parameter"><code>mib</code></em> specifies an unknown/invalid
1798            value.</p></dd><dt><span class="term"><span class="errorname">EPERM</span></span></dt><dd><p>Attempt to read or write void value, or attempt to
1799            write read-only value.</p></dd><dt><span class="term"><span class="errorname">EAGAIN</span></span></dt><dd><p>A memory allocation failure
1800            occurred.</p></dd><dt><span class="term"><span class="errorname">EFAULT</span></span></dt><dd><p>An interface with side effects failed in some way
1801            not directly related to <code class="function">mallctl*</code>(<em class="parameter"><code></code></em>)
1802            read/write processing.</p></dd></dl></div><p>
1803      </p><p>The <code class="function">malloc_usable_size</code>(<em class="parameter"><code></code></em>) function
1804      returns the usable size of the allocation pointed to by
1805      <em class="parameter"><code>ptr</code></em>.  </p></div></div><div class="refsect1"><a name="environment"></a><h2>ENVIRONMENT</h2><p>The following environment variable affects the execution of the
1806    allocation functions:
1807      </p><div class="variablelist"><dl class="variablelist"><dt><span class="term"><code class="envar">MALLOC_CONF</code></span></dt><dd><p>If the environment variable
1808          <code class="envar">MALLOC_CONF</code> is set, the characters it contains
1809          will be interpreted as options.</p></dd></dl></div><p>
1810    </p></div><div class="refsect1"><a name="examples"></a><h2>EXAMPLES</h2><p>To dump core whenever a problem occurs:
1811      </p><pre class="screen">ln -s 'abort:true' /etc/malloc.conf</pre><p>
1812    </p><p>To specify in the source a chunk size that is 16 MiB:
1813      </p><pre class="programlisting">
1814malloc_conf = "lg_chunk:24";</pre></div><div class="refsect1"><a name="see_also"></a><h2>SEE ALSO</h2><p><span class="citerefentry"><span class="refentrytitle">madvise</span>(2)</span>,
1815    <span class="citerefentry"><span class="refentrytitle">mmap</span>(2)</span>,
1816    <span class="citerefentry"><span class="refentrytitle">sbrk</span>(2)</span>,
1817    <span class="citerefentry"><span class="refentrytitle">utrace</span>(2)</span>,
1818    <span class="citerefentry"><span class="refentrytitle">alloca</span>(3)</span>,
1819    <span class="citerefentry"><span class="refentrytitle">atexit</span>(3)</span>,
1820    <span class="citerefentry"><span class="refentrytitle">getpagesize</span>(3)</span></p></div><div class="refsect1"><a name="standards"></a><h2>STANDARDS</h2><p>The <code class="function">malloc</code>(<em class="parameter"><code></code></em>),
1821    <code class="function">calloc</code>(<em class="parameter"><code></code></em>),
1822    <code class="function">realloc</code>(<em class="parameter"><code></code></em>), and
1823    <code class="function">free</code>(<em class="parameter"><code></code></em>) functions conform to ISO/IEC
1824    9899:1990 (&#8220;ISO C90&#8221;).</p><p>The <code class="function">posix_memalign</code>(<em class="parameter"><code></code></em>) function conforms
1825    to IEEE Std 1003.1-2001 (&#8220;POSIX.1&#8221;).</p></div></div></body></html>
1826