path: root/erts/doc/src/atomics.xml

<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">

<erlref>
  <header>
    <copyright>
      <year>2018</year>
      <holder>Ericsson AB. All Rights Reserved.</holder>
    </copyright>
    <legalnotice>
      Licensed under the Apache License, Version 2.0 (the "License");
      you may not use this file except in compliance with the License.
      You may obtain a copy of the License at

          http://www.apache.org/licenses/LICENSE-2.0

      Unless required by applicable law or agreed to in writing, software
      distributed under the License is distributed on an "AS IS" BASIS,
      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
      See the License for the specific language governing permissions and
      limitations under the License.
    </legalnotice>

    <title>atomics</title>
  </header>
  <module since="OTP 21.2">atomics</module>
  <modulesummary>Atomic Functions</modulesummary>
  <description>
    <p>This module provides a set of functions to do atomic operations towards
    mutable atomic variables. The implementation utilizes only
    atomic hardware instructions without any software level locking, which makes
    it very efficient for concurrent access. The atomics are organized into
    arrays with the following semantics:</p>
    <list type="bulleted">
      <item>
	<p>Atomics are 64 bit integers.</p>
      </item>
      <item>
	<p>Atomics can be represented as either signed or unsigned.</p>
      </item>
      <item>
	<p>Atomics wrap around at overflow and underflow operations.</p>
      </item>
      <item>
	<p>All operations guarantee atomicity. No intermediate results can be
	seen. The result of one mutation can only be the input to one
	following mutation.</p>
      </item>
      <item>
	<p>All atomic operations are mutually ordered. If atomic B is updated
	<em>after</em> atomic A, then that is how it will appear to any
	concurrent readers. No one can read the new value of B and then read the
	old value of A.</p>
      </item>
      <item>
	<p>Indexes into atomic arrays are one-based. An atomic array of
	arity N contains N atomics with index from 1 to N.</p>
      </item>
    </list>
  </description>

  <datatypes>
    <datatype>
      <name name="atomics_ref"/>
      <desc><p>Identifies an atomic array returned from
        <seealso marker="#new/2"><c>new/2</c></seealso>.</p>
      </desc>
    </datatype>
  </datatypes>

  <funcs>
    <func>
      <name name="new" arity="2" since="OTP 21.2"/>
      <fsummary>Create atomic array</fsummary>
      <desc>
        <p>Create a new atomic array of <c><anno>Arity</anno></c> atomics.</p>
	<p>Argument <c><anno>Opts</anno></c> is a list of the following possible
	options:</p>
	<taglist>
	  <tag><c>{signed, boolean()}</c></tag>
	  <item><p>Indicate if the elements of the array will be treated
	  as signed or unsigned integers. Default is <c>true</c> (signed).</p>
	  <p>The integer interval for signed atomics are from <c>-(1 bsl 63)</c>
	  to <c>(1 bsl 63)-1</c> and for unsigned atomics from <c>0</c> to <c>(1
	  bsl 64)-1</c>.</p>
	  </item>
	</taglist>
        <p>Atomics are not tied to the current process and are automatically
        garbage collected when they are no longer referenced.</p>
      </desc>
    </func>

    <func>
      <name name="put" arity="3" since="OTP 21.2"/>
      <fsummary>Set atomic value</fsummary>
      <desc>
        <p>Set atomic to <c><anno>Value</anno></c>.</p>
      </desc>
    </func>

    <func>
      <name name="get" arity="2" since="OTP 21.2"/>
      <fsummary>Read atomic value</fsummary>
      <desc>
        <p>Read atomic value.</p>
      </desc>
    </func>

    <func>
      <name name="add" arity="3" since="OTP 21.2"/>
      <fsummary>Add to atomic</fsummary>
      <desc>
        <p>Add <c><anno>Incr</anno></c> to atomic.</p>
      </desc>
    </func>

    <func>
      <name name="add_get" arity="3" since="OTP 21.2"/>
      <fsummary>Atomic add and get</fsummary>
      <desc>
        <p>Atomic addition and return of the result.</p>
      </desc>
    </func>

    <func>
      <name name="sub" arity="3" since="OTP 21.2"/>
      <fsummary>Subtract from atomic</fsummary>
      <desc>
        <p>Subtract <c><anno>Decr</anno></c> from atomic.</p>
      </desc>
    </func>

    <func>
      <name name="sub_get" arity="3" since="OTP 21.2"/>
      <fsummary>Atomic sub and get</fsummary>
      <desc>
        <p>Atomic subtraction and return of the result.</p>
      </desc>
    </func>

    <func>
      <name name="exchange" arity="3" since="OTP 21.2"/>
      <fsummary>Atomic exchange.</fsummary>
      <desc>
        <p>Atomically replaces the value of the atomic with
	<c><anno>Desired</anno></c> and returns the value it held
	previously.</p>
      </desc>
    </func>
    
    <func>
    <name name="compare_exchange" arity="4" since="OTP 21.2"/>
       <fsummary>Atomic compare and exchange.</fsummary>
       <desc>
         <p>Atomically compares the atomic with <c><anno>Expected</anno></c>,
 	and if those are equal, set atomic to <c><anno>Desired</anno></c>.
 	Returns <c>ok</c> if <c><anno>Desired</anno></c> was written. Returns
 	the actual atomic value if not equal to <c><anno>Expected</anno></c>.</p>
       </desc>
    </func>
    
    <func>
      <name name="info" arity="1" since="OTP 21.2"/>
      <fsummary>Get information about atomic array.</fsummary>
      <desc>
        <p>Return information about an atomic array in a map. The map
	has the following keys:</p>
	<taglist>
	  <tag><c>size</c></tag>
	  <item><p>The number of atomics in the array.</p></item>
	  <tag><c>max</c></tag>
	  <item><p>The highest possible value an atomic in this array can
	  hold.</p></item>
	  <tag><c>min</c></tag>
	  <item><p>The lowest possible value an atomic in this array can
	  hold.</p></item>
	  <tag><c>memory</c></tag>
	  <item><p>Approximate memory consumption for the array in
	  bytes.</p></item>
	</taglist>
      </desc>
    </func>

 </funcs>
</erlref>