<?xml version="1.0" encoding="latin1" ?> <!DOCTYPE cref SYSTEM "cref.dtd"> <cref> <header> <copyright> <year>2002</year><year>2011</year> <holder>Ericsson AB. All Rights Reserved.</holder> </copyright> <legalnotice> The contents of this file are subject to the Erlang Public License, Version 1.1, (the "License"); you may not use this file except in compliance with the License. You should have received a copy of the Erlang Public License along with this software. If not, it can be retrieved online at http://www.erlang.org/. Software distributed under the License is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License for the specific language governing rights and limitations under the License. </legalnotice> <title>erts_alloc</title> <prepared>Rickard Green</prepared> <docno>1</docno> <date>03-06-11</date> <rev>1</rev> <file>erts_alloc.xml</file> </header> <lib>erts_alloc</lib> <libsummary>An Erlang Run-Time System internal memory allocator library.</libsummary> <description> <p><c>erts_alloc</c> is an Erlang Run-Time System internal memory allocator library. <c>erts_alloc</c> provides the Erlang Run-Time System with a number of memory allocators.</p> </description> <section> <title>Allocators</title> <marker id="allocators"></marker> <p>Currently the following allocators are present:</p> <taglist> <tag><c>temp_alloc</c></tag> <item>Allocator used for temporary allocations.</item> <tag><c>eheap_alloc</c></tag> <item>Allocator used for Erlang heap data, such as Erlang process heaps.</item> <tag><c>binary_alloc</c></tag> <item>Allocator used for Erlang binary data.</item> <tag><c>ets_alloc</c></tag> <item>Allocator used for ETS data.</item> <tag><c>driver_alloc</c></tag> <item>Allocator used for driver data.</item> <tag><c>sl_alloc</c></tag> <item>Allocator used for memory blocks that are expected to be short-lived.</item> <tag><c>ll_alloc</c></tag> <item>Allocator used for memory blocks that are expected to be long-lived, for example Erlang code.</item> <tag><c>fix_alloc</c></tag> <item>A fast allocator used for some frequently used fixed size data types.</item> <tag><c>std_alloc</c></tag> <item>Allocator used for most memory blocks not allocated via any of the other allocators described above.</item> <tag><c>sys_alloc</c></tag> <item>This is normally the default <c>malloc</c> implementation used on the specific OS.</item> <tag><c>mseg_alloc</c></tag> <item>A memory segment allocator. <c>mseg_alloc</c> is used by other allocators for allocating memory segments and is currently only available on systems that have the <c>mmap</c> system call. Memory segments that are deallocated are kept for a while in a segment cache before they are destroyed. When segments are allocated, cached segments are used if possible instead of creating new segments. This in order to reduce the number of system calls made.</item> <tag><c>sbmbc_alloc</c></tag> <item>Allocator used by other allocators for allocation of carriers where only small blocks are placed. Currently this allocator is disabled by default.</item> </taglist> <p><c>sys_alloc</c> is always enabled and cannot be disabled. <c>mseg_alloc</c> is always enabled if it is available and an allocator that uses it is enabled. All other allocators can be <seealso marker="#M_e">enabled or disabled</seealso>. By default all allocators are enabled. When an allocator is disabled, <c>sys_alloc</c> is used instead of the disabled allocator. <c>sbmbc_alloc</c> is an exception. If <c>sbmbc_alloc</c> is disabled, other allocators will not handle small blocks in separate carriers.</p> <p>The main idea with the <c>erts_alloc</c> library is to separate memory blocks that are used differently into different memory areas, and by this achieving less memory fragmentation. By putting less effort in finding a good fit for memory blocks that are frequently allocated than for those less frequently allocated, a performance gain can be achieved.</p> </section> <section> <marker id="alloc_util"></marker> <title>The alloc_util framework</title> <p>Internally a framework called <c>alloc_util</c> is used for implementing allocators. <c>sys_alloc</c>, and <c>mseg_alloc</c> do not use this framework; hence, the following does <em>not</em> apply to them.</p> <p>An allocator manages multiple areas, called carriers, in which memory blocks are placed. A carrier is either placed in a separate memory segment (allocated via <c>mseg_alloc</c>), in the heap segment (allocated via <c>sys_alloc</c>), or inside another carrier (in case it is a carrier created by <c>sbmbc_alloc</c>). Multiblock carriers are used for storage of several blocks. Singleblock carriers are used for storage of one block. Blocks that are larger than the value of the singleblock carrier threshold (<seealso marker="#M_sbct">sbct</seealso>) parameter are placed in singleblock carriers. Blocks that are smaller than the value of the <c>sbct</c> parameter are placed in multiblock carriers. Blocks that are smaller than the small block multiblock carrier threshold (<seealso marker="#M_sbmbct">sbmbct</seealso>) will be placed in multiblock carriers only used for small blocks. Normally an allocator creates a "main multiblock carrier". Main multiblock carriers are never deallocated. The size of the main multiblock carrier is determined by the value of the <seealso marker="#M_mmbcs">mmbcs</seealso> parameter.</p> <p><marker id="mseg_mbc_sizes"></marker>Sizes of multiblock carriers allocated via <c>mseg_alloc</c> are decided based on the values of the largest multiblock carrier size (<seealso marker="#M_lmbcs">lmbcs</seealso>), the smallest multiblock carrier size (<seealso marker="#M_smbcs">smbcs</seealso>), and the multiblock carrier growth stages (<seealso marker="#M_mbcgs">mbcgs</seealso>) parameters. If <c>nc</c> is the current number of multiblock carriers (the main multiblock carrier excluded) managed by an allocator, the size of the next <c>mseg_alloc</c> multiblock carrier allocated by this allocator will roughly be <c><![CDATA[smbcs+nc*(lmbcs-smbcs)/mbcgs]]></c> when <c><![CDATA[nc <= mbcgs]]></c>, and <c>lmbcs</c> when <c><![CDATA[nc > mbcgs]]></c>. If the value of the <c>sbct</c> parameter should be larger than the value of the <c>lmbcs</c> parameter, the allocator may have to create multiblock carriers that are larger than the value of the <c>lmbcs</c> parameter, though. The size of multiblock carriers for small blocks is determined by the small block multiblock carrier size (<seealso marker="#M_sbmbcs">sbmbcs</seealso>). Singleblock carriers allocated via <c>mseg_alloc</c> are sized to whole pages.</p> <p>Sizes of carriers allocated via <c>sys_alloc</c> are decided based on the value of the <c>sys_alloc</c> carrier size (<seealso marker="#Muycs">ycs</seealso>) parameter. The size of a carrier is the least number of multiples of the value of the <c>ycs</c> parameter that satisfies the request.</p> <p>Coalescing of free blocks are always performed immediately. Boundary tags (headers and footers) in free blocks are used which makes the time complexity for coalescing constant.</p> <p><marker id="strategy"></marker>The memory allocation strategy used for multiblock carriers by an allocator is configurable via the <seealso marker="#M_as">as</seealso> parameter. Currently the following strategies are available:</p> <taglist> <tag>Best fit</tag> <item> <p>Strategy: Find the smallest block that satisfies the requested block size.</p> <p>Implementation: A balanced binary search tree is used. The time complexity is proportional to log N, where N is the number of sizes of free blocks.</p> </item> <tag>Address order best fit</tag> <item> <p>Strategy: Find the smallest block that satisfies the requested block size. If multiple blocks are found, choose the one with the lowest address.</p> <p>Implementation: A balanced binary search tree is used. The time complexity is proportional to log N, where N is the number of free blocks.</p> </item> <tag>Address order first fit</tag> <item> <p>Strategy: Find the block with the lowest address that satisfies the requested block size.</p> <p>Implementation: A balanced binary search tree is used. The time complexity is proportional to log N, where N is the number of free blocks.</p> </item> <tag>Good fit</tag> <item> <p>Strategy: Try to find the best fit, but settle for the best fit found during a limited search.</p> <p>Implementation: The implementation uses segregated free lists with a maximum block search depth (in each list) in order to find a good fit fast. When the maximum block search depth is small (by default 3) this implementation has a time complexity that is constant. The maximum block search depth is configurable via the <seealso marker="#M_mbsd">mbsd</seealso> parameter.</p> </item> <tag>A fit</tag> <item> <p>Strategy: Do not search for a fit, inspect only one free block to see if it satisfies the request. This strategy is only intended to be used for temporary allocations.</p> <p>Implementation: Inspect the first block in a free-list. If it satisfies the request, it is used; otherwise, a new carrier is created. The implementation has a time complexity that is constant.</p> <p>As of erts version 5.6.1 the emulator will refuse to use this strategy on other allocators than <c>temp_alloc</c>. This since it will only cause problems for other allocators.</p> </item> </taglist> <p>Apart from the ordinary allocators described above a number of pre-allocators are used for some specific data types. These pre-allocators pre-allocate a fixed amount of memory for certain data types when the run-time system starts. As long as pre-allocated memory is available, it will be used. When no pre-allocated memory is available, memory will be allocated in ordinary allocators. These pre-allocators are typically much faster than the ordinary allocators, but can only satisfy a limited amount of requests.</p> </section> <note><p> Currently only allocators using the best fit and the address order best fit strategies are able to use "small block multi block carriers". </p></note> <section> <marker id="flags"></marker> <title>System Flags Effecting erts_alloc</title> <warning> <p>Only use these flags if you are absolutely sure what you are doing. Unsuitable settings may cause serious performance degradation and even a system crash at any time during operation.</p> </warning> <p>Memory allocator system flags have the following syntax: <c><![CDATA[+M<S><P> <V>]]></c> where <c><![CDATA[<S>]]></c> is a letter identifying a subsystem, <c><![CDATA[<P>]]></c> is a parameter, and <c><![CDATA[<V>]]></c> is the value to use. The flags can be passed to the Erlang emulator (<seealso marker="erl">erl</seealso>) as command line arguments.</p> <p>System flags effecting specific allocators have an upper-case letter as <c><![CDATA[<S>]]></c>. The following letters are used for the currently present allocators:</p> <list type="bulleted"> <item><c>B: binary_alloc</c></item> <item><c>C: sbmbc_alloc</c></item> <item><c>D: std_alloc</c></item> <item><c>E: ets_alloc</c></item> <item><c>F: fix_alloc</c></item> <item><c>H: eheap_alloc</c></item> <item><c>L: ll_alloc</c></item> <item><c>M: mseg_alloc</c></item> <item><c>R: driver_alloc</c></item> <item><c>S: sl_alloc</c></item> <item><c>T: temp_alloc</c></item> <item><c>Y: sys_alloc</c></item> </list> <p>The following flags are available for configuration of <c>mseg_alloc</c>:</p> <taglist> <tag><marker id="MMamcbf"><c><![CDATA[+MMamcbf <size>]]></c></marker></tag> <item> Absolute max cache bad fit (in kilobytes). A segment in the memory segment cache is not reused if its size exceeds the requested size with more than the value of this parameter. Default value is 4096. </item> <tag><marker id="MMrmcbf"><c><![CDATA[+MMrmcbf <ratio>]]></c></marker></tag> <item> Relative max cache bad fit (in percent). A segment in the memory segment cache is not reused if its size exceeds the requested size with more than relative max cache bad fit percent of the requested size. Default value is 20.</item> <tag><marker id="MMmcs"><c><![CDATA[+MMmcs <amount>]]></c></marker></tag> <item> Max cached segments. The maximum number of memory segments stored in the memory segment cache. Valid range is 0-30. Default value is 5.</item> </taglist> <p>The following flags are available for configuration of <c>sys_alloc</c>:</p> <taglist> <tag><marker id="MYe"><c>+MYe true</c></marker></tag> <item> Enable <c>sys_alloc</c>. Note: <c>sys_alloc</c> cannot be disabled.</item> <tag><marker id="MYm"><c>+MYm libc</c></marker></tag> <item> <c>malloc</c> library to use. Currently only <c>libc</c> is available. <c>libc</c> enables the standard <c>libc</c> malloc implementation. By default <c>libc</c> is used.</item> <tag><marker id="MYtt"><c><![CDATA[+MYtt <size>]]></c></marker></tag> <item> Trim threshold size (in kilobytes). This is the maximum amount of free memory at the top of the heap (allocated by <c>sbrk</c>) that will be kept by <c>malloc</c> (not released to the operating system). When the amount of free memory at the top of the heap exceeds the trim threshold, <c>malloc</c> will release it (by calling <c>sbrk</c>). Trim threshold is given in kilobytes. Default trim threshold is 128. <em>Note:</em> This flag will only have any effect when the emulator has been linked with the GNU C library, and uses its <c>malloc</c> implementation.</item> <tag><marker id="MYtp"><c><![CDATA[+MYtp <size>]]></c></marker></tag> <item> Top pad size (in kilobytes). This is the amount of extra memory that will be allocated by <c>malloc</c> when <c>sbrk</c> is called to get more memory from the operating system. Default top pad size is 0. <em>Note:</em> This flag will only have any effect when the emulator has been linked with the GNU C library, and uses its <c>malloc</c> implementation.</item> </taglist> <p>The following flags are available for configuration of allocators based on <c>alloc_util</c>. If <c>u</c> is used as subsystem identifier (i.e., <c><![CDATA[<S> = u]]></c>) all allocators based on <c>alloc_util</c> will be effected. If <c>B</c>, <c>D</c>, <c>E</c>, <c>F</c>, <c>H</c>, <c>L</c>, <c>R</c>, <c>S</c>, or <c>T</c> is used as subsystem identifier, only the specific allocator identified will be effected:</p> <taglist> <tag><marker id="M_as"><c><![CDATA[+M<S>as bf|aobf|aoff|gf|af]]></c></marker></tag> <item> Allocation strategy. Valid strategies are <c>bf</c> (best fit), <c>aobf</c> (address order best fit), <c>aoff</c> (address order first fit), <c>gf</c> (good fit), and <c>af</c> (a fit). See <seealso marker="#strategy">the description of allocation strategies</seealso> in "the <c>alloc_util</c> framework" section.</item> <tag><marker id="M_asbcst"><c><![CDATA[+M<S>asbcst <size>]]></c></marker></tag> <item> Absolute singleblock carrier shrink threshold (in kilobytes). When a block located in an <c>mseg_alloc</c> singleblock carrier is shrunk, the carrier will be left unchanged if the amount of unused memory is less than this threshold; otherwise, the carrier will be shrunk. See also <seealso marker="#M_rsbcst">rsbcst</seealso>.</item> <tag><marker id="M_e"><c><![CDATA[+M<S>e true|false]]></c></marker></tag> <item> Enable allocator <c><![CDATA[<S>]]></c>.</item> <tag><marker id="M_lmbcs"><c><![CDATA[+M<S>lmbcs <size>]]></c></marker></tag> <item> Largest (<c>mseg_alloc</c>) multiblock carrier size (in kilobytes). See <seealso marker="#mseg_mbc_sizes">the description on how sizes for mseg_alloc multiblock carriers are decided</seealso> in "the <c>alloc_util</c> framework" section.</item> <tag><marker id="M_mbcgs"><c><![CDATA[+M<S>mbcgs <ratio>]]></c></marker></tag> <item> (<c>mseg_alloc</c>) multiblock carrier growth stages. See <seealso marker="#mseg_mbc_sizes">the description on how sizes for mseg_alloc multiblock carriers are decided</seealso> in "the <c>alloc_util</c> framework" section.</item> <tag><marker id="M_mbsd"><c><![CDATA[+M<S>mbsd <depth>]]></c></marker></tag> <item> Max block search depth. This flag has effect only if the good fit strategy has been selected for allocator <c><![CDATA[<S>]]></c>. When the good fit strategy is used, free blocks are placed in segregated free-lists. Each free list contains blocks of sizes in a specific range. The max block search depth sets a limit on the maximum number of blocks to inspect in a free list during a search for suitable block satisfying the request.</item> <tag><marker id="M_mmbcs"><c><![CDATA[+M<S>mmbcs <size>]]></c></marker></tag> <item> Main multiblock carrier size. Sets the size of the main multiblock carrier for allocator <c><![CDATA[<S>]]></c>. The main multiblock carrier is allocated via <c><![CDATA[sys_alloc]]></c> and is never deallocated.</item> <tag><marker id="M_mmmbc"><c><![CDATA[+M<S>mmmbc <amount>]]></c></marker></tag> <item> Max <c>mseg_alloc</c> multiblock carriers. Maximum number of multiblock carriers allocated via <c>mseg_alloc</c> by allocator <c><![CDATA[<S>]]></c>. When this limit has been reached, new multiblock carriers will be allocated via <c>sys_alloc</c>.</item> <tag><marker id="M_mmsbc"><c><![CDATA[+M<S>mmsbc <amount>]]></c></marker></tag> <item> Max <c>mseg_alloc</c> singleblock carriers. Maximum number of singleblock carriers allocated via <c>mseg_alloc</c> by allocator <c><![CDATA[<S>]]></c>. When this limit has been reached, new singleblock carriers will be allocated via <c>sys_alloc</c>.</item> <tag><marker id="M_ramv"><c><![CDATA[+M<S>ramv <bool>]]></c></marker></tag> <item> Realloc always moves. When enabled, reallocate operations will more or less be translated into an allocate, copy, free sequence. This often reduce memory fragmentation, but costs performance. </item> <tag><marker id="M_rmbcmt"><c><![CDATA[+M<S>rmbcmt <ratio>]]></c></marker></tag> <item> Relative multiblock carrier move threshold (in percent). When a block located in a multiblock carrier is shrunk, the block will be moved if the ratio of the size of the returned memory compared to the previous size is more than this threshold; otherwise, the block will be shrunk at current location.</item> <tag><marker id="M_rsbcmt"><c><![CDATA[+M<S>rsbcmt <ratio>]]></c></marker></tag> <item> Relative singleblock carrier move threshold (in percent). When a block located in a singleblock carrier is shrunk to a size smaller than the value of the <seealso marker="#M_sbct">sbct</seealso> parameter, the block will be left unchanged in the singleblock carrier if the ratio of unused memory is less than this threshold; otherwise, it will be moved into a multiblock carrier. </item> <tag><marker id="M_rsbcst"><c><![CDATA[+M<S>rsbcst <ratio>]]></c></marker></tag> <item> Relative singleblock carrier shrink threshold (in percent). When a block located in an <c>mseg_alloc</c> singleblock carrier is shrunk, the carrier will be left unchanged if the ratio of unused memory is less than this threshold; otherwise, the carrier will be shrunk. See also <seealso marker="#M_asbcst">asbcst</seealso>.</item> <tag><marker id="M_sbct"><c><![CDATA[+M<S>sbct <size>]]></c></marker></tag> <item> Singleblock carrier threshold. Blocks larger than this threshold will be placed in singleblock carriers. Blocks smaller than this threshold will be placed in multiblock carriers.</item> <tag><marker id="M_sbmbcs"><c><![CDATA[+M<S>sbmbcs <size>]]></c></marker></tag> <item> Small block multiblock carrier size (in bytes). Memory blocks smaller than the small block multiblock carrier threshold (<seealso marker="#M_sbmbct">sbmbct</seealso>) will be placed in multiblock carriers used for small blocks only. This parameter determines the size of such carriers. </item> <tag><marker id="M_sbmbct"><c><![CDATA[+M<S>sbmbct <size>]]></c></marker></tag> <item> Small block multiblock carrier threshold (in bytes). Memory blocks smaller than this threshold will be placed in multiblock carriers used for small blocks only. </item> <tag><marker id="M_smbcs"><c><![CDATA[+M<S>smbcs <size>]]></c></marker></tag> <item> Smallest (<c>mseg_alloc</c>) multiblock carrier size (in kilobytes). See <seealso marker="#mseg_mbc_sizes">the description on how sizes for mseg_alloc multiblock carriers are decided</seealso> in "the <c>alloc_util</c> framework" section.</item> <tag><marker id="M_t"><c><![CDATA[+M<S>t true|false]]></c></marker></tag> <item> <p>Multiple, thread specific instances of the allocator. This option will only have any effect on the runtime system with SMP support. Default behaviour on the runtime system with SMP support:</p> <taglist> <tag><c>ll_alloc</c></tag> <item><c>1</c> instance.</item> <tag>Other allocators</tag> <item><c>NoSchedulers+1</c> instances. Each scheduler will use a lock-free instance of its own and other threads will use a common instance.</item> </taglist> <p>It was previously (before ERTS version 5.9) possible to configure a smaller amount of thread specific instances than schedulers. This is, however, not possible any more.</p> </item> </taglist> <p>Currently the following flags are available for configuration of <c>alloc_util</c>, i.e. all allocators based on <c>alloc_util</c> will be effected:</p> <taglist> <tag><marker id="Muycs"><c><![CDATA[+Muycs <size>]]></c></marker></tag> <item> <c>sys_alloc</c> carrier size. Carriers allocated via <c>sys_alloc</c> will be allocated in sizes which are multiples of the <c>sys_alloc</c> carrier size. This is not true for main multiblock carriers and carriers allocated during a memory shortage, though.</item> <tag><marker id="Mummc"><c><![CDATA[+Mummc <amount>]]></c></marker></tag> <item> Max <c>mseg_alloc</c> carriers. Maximum number of carriers placed in separate memory segments. When this limit has been reached, new carriers will be placed in memory retrieved from <c>sys_alloc</c>.</item> </taglist> <p>Instrumentation flags:</p> <taglist> <tag><marker id="Mim"><c>+Mim true|false</c></marker></tag> <item> A map over current allocations is kept by the emulator. The allocation map can be retrieved via the <c>instrument</c> module. <c>+Mim true</c> implies <c>+Mis true</c>. <c>+Mim true</c> is the same as <seealso marker="erl#instr">-instr</seealso>.</item> <tag><marker id="Mis"><c>+Mis true|false</c></marker></tag> <item> Status over allocated memory is kept by the emulator. The allocation status can be retrieved via the <c>instrument</c> module.</item> <tag><marker id="Mit"><c>+Mit X</c></marker></tag> <item> Reserved for future use. Do <em>not</em> use this flag.</item> </taglist> <note> <p>When instrumentation of the emulator is enabled, the emulator uses more memory and runs slower.</p> </note> <p>Other flags:</p> <taglist> <tag><marker id="Mea"><c>+Mea min|max|r9c|r10b|r11b|config</c></marker></tag> <item> <taglist> <tag><c>min</c></tag> <item> Disables all allocators that can be disabled. </item> <tag><c>max</c></tag> <item> Enables all allocators (currently default). </item> <tag><c>r9c|r10b|r11b</c></tag> <item> Configures all allocators as they were configured in respective OTP release. These will eventually be removed. </item> <tag><c>config</c></tag> <item> Disables features that cannot be enabled while creating an allocator configuration with <seealso marker="runtime_tools:erts_alloc_config">erts_alloc_config(3)</seealso>. Note, this option should only be used while running <c>erts_alloc_config</c>, <em>not</em> when using the created configuration. </item> </taglist> </item> </taglist> <p>Only some default values have been presented here. <seealso marker="erts:erlang#system_info_allocator">erlang:system_info(allocator)</seealso>, and <seealso marker="erts:erlang#system_info_allocator_tuple">erlang:system_info({allocator, Alloc})</seealso> can be used in order to obtain currently used settings and current status of the allocators.</p> <note> <p>Most of these flags are highly implementation dependent, and they may be changed or removed without prior notice.</p> <p><c>erts_alloc</c> is not obliged to strictly use the settings that have been passed to it (it may even ignore them).</p> </note> <p><seealso marker="runtime_tools:erts_alloc_config">erts_alloc_config(3)</seealso> is a tool that can be used to aid creation of an <c>erts_alloc</c> configuration that is suitable for a limited number of runtime scenarios.</p> </section> <section> <title>SEE ALSO</title> <p><seealso marker="runtime_tools:erts_alloc_config">erts_alloc_config(3)</seealso>, <seealso marker="erl">erl(1)</seealso>, <seealso marker="tools:instrument">instrument(3)</seealso>, <seealso marker="erts:erlang">erlang(3)</seealso></p> </section> </cref>