path: root/lib/tools/doc/src/instrument.xml

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

<erlref>
  <header>
    <copyright>
      <year>1998</year><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>instrument</title>
    <prepared>Arndt Jonasson</prepared>
    <responsible>Torbj&ouml;rn Johnsson</responsible>
    <docno>1</docno>
    <approved>Bj&ouml;rn Gustavsson</approved>
    <checked></checked>
    <date>98-04-01</date>
    <rev>PA1</rev>
    <file>instrument.sgml</file>
  </header>
  <module since="">instrument</module>
  <modulesummary>Analysis and Utility Functions for Instrumentation</modulesummary>
  <description>
    <p>The module <c>instrument</c> contains support for studying the resource
      usage in an Erlang runtime system. Currently, only the allocation of memory can
      be studied.</p>
    <note>
      <p>Note that this whole module is experimental, and the representations
        used as well as the functionality is likely to change in the future.</p>
    </note>
  </description>
  <datatypes>
    <datatype>
      <name name="block_histogram"/>
      <desc>
        <p>A histogram of block sizes where each interval's upper bound is 
          twice as high as the one before it.</p>
        <p>The upper bound of the first interval is provided by the function
           that returned the histogram, and the last interval has no upper
           bound.</p>
      </desc>
    </datatype>
    <datatype>
      <name name="allocation_summary"/>
      <desc>
        <p>A summary of allocated block sizes (including their headers) grouped
          by their <c><anno>Origin</anno></c> and <c><anno>Type</anno></c>.</p>
        <p><c><anno>Origin</anno></c> is generally which NIF or driver that
          allocated the blocks, or 'system' if it could not be determined.</p>
        <p><c><anno>Type</anno></c> is the allocation category that the blocks
          belong to, e.g. <c>db_term</c>, <c>message</c> or <c>binary</c>.</p>
        <p>If one or more carriers could not be scanned in full without harming
          the responsiveness of the system, <c><anno>UnscannedSize</anno></c>
          is the number of bytes that had to be skipped.</p>
       </desc>
    </datatype>
    <datatype>
      <name name="carrier_info_list"/>
      <desc>
        <p><c><anno>AllocatorType</anno></c> is the type of the allocator that
          employs this carrier.</p>
        <p><c><anno>TotalSize</anno></c> is the total size of the carrier,
          including its header.</p>
        <p><c><anno>AllocatedSize</anno></c> is the combined size of the
          carrier's allocated blocks, including their headers.</p>
        <p><c><anno>AllocatedCount</anno></c> is the number of allocated
          blocks in the carrier.</p>
        <p><c><anno>InPool</anno></c> is whether the carrier is in the
          migration pool.</p>
        <p><c><anno>FreeBlocks</anno></c> is a histogram of the free block
          sizes in the carrier.</p>
        <p>If the carrier could not be scanned in full without harming the
          responsiveness of the system, <c><anno>UnscannedSize</anno></c> is
          the number of bytes that had to be skipped.</p>
       </desc>
    </datatype>
  </datatypes>
  <funcs>

    <func>
      <name name="allocations" arity="0" since="OTP 21.0"/>
      <fsummary>Return a summary of all allocations in the system.</fsummary>
      <desc>
        <p>Shorthand for
          <seealso marker="#allocations/1"><c>allocations(#{})</c>.</seealso></p>
      </desc>
    </func>

    <func>
      <name name="allocations" arity="1" since="OTP 21.0"/>
      <fsummary>Return a summary of all allocations filtered by allocator type
        and scheduler id.</fsummary>
      <desc>
        <p>Returns a summary of all tagged allocations in the system,
          optionally filtered by allocator type and scheduler id.</p>
        <p>Only binaries and allocations made by NIFs and drivers are tagged by
          default, but this can be configured an a per-allocator basis with the
          <seealso marker="erts:erts_alloc#M_atags"><c>+M&lt;S&gt;atags</c>
          </seealso> emulator option.</p>
        <p>If tagged allocations are not enabled on any of the specified
          allocator types, the call will fail with
          <c>{error, not_enabled}</c>.</p>
        <p>The following options can be used:</p>
        <taglist>
          <tag><c>allocator_types</c></tag>
          <item>
            <p>The allocator types that will be searched. Defaults to all
              <c>alloc_util</c> allocators.</p>
          </item>
          <tag><c>scheduler_ids</c></tag>
          <item>
            <p>The scheduler ids whose allocator instances will be searched. A
              scheduler id of 0 will refer to the global instance that is not
              tied to any particular scheduler. Defaults to all schedulers and
              the global instance.</p>
          </item>
          <tag><c>histogram_start</c></tag>
          <item>
            <p>The upper bound of the first interval in the allocated block
              size histograms. Defaults to 128.</p>
          </item>
          <tag><c>histogram_width</c></tag>
          <item>
            <p>The number of intervals in the allocated block size histograms.
              Defaults to 18.</p>
          </item>
        </taglist>
        <p><em>Example:</em></p>
        <code type="none"><![CDATA[
> instrument:allocations(#{ histogram_start => 128, histogram_width => 15 }).
{ok,{128,0,
     #{udp_inet =>
           #{driver_event_state => {0,0,0,0,0,0,0,0,0,1,0,0,0,0,0}},
       system =>
           #{heap => {0,0,0,0,20,4,2,2,2,3,0,1,0,0,1},
             db_term => {271,3,1,52,80,1,0,0,0,0,0,0,0,0,0},
             code => {0,0,0,5,3,6,11,22,19,20,10,2,1,0,0},
             binary => {18,0,0,0,7,0,0,1,0,0,0,0,0,0,0},
             message => {0,40,78,2,2,0,0,0,0,0,0,0,0,0,0},
             ... }
       spawn_forker =>
           #{driver_select_data_state =>
                 {1,0,0,0,0,0,0,0,0,0,0,0,0,0,0}},
       ram_file_drv => #{drv_binary => {0,0,0,0,0,0,1,0,0,0,0,0,0,0,0}},
       prim_file =>
           #{process_specific_data => {2,0,0,0,0,0,0,0,0,0,0,0,0,0,0},
             nif_trap_export_entry => {0,4,0,0,0,0,0,0,0,0,0,0,0,0,0},
             monitor_extended => {0,1,0,0,0,0,0,0,0,0,0,0,0,0,0},
             drv_binary => {0,0,0,0,0,0,1,0,3,5,0,0,0,1,0},
             binary => {0,4,0,0,0,0,0,0,0,0,0,0,0,0,0}},
       prim_buffer =>
           #{nif_internal => {0,4,0,0,0,0,0,0,0,0,0,0,0,0,0},
             binary => {0,4,0,0,0,0,0,0,0,0,0,0,0,0,0}}}}}
     ]]></code>
      </desc>
    </func>

    <func>
      <name name="carriers" arity="0" since="OTP 21.0"/>
      <fsummary>Return a list of all carriers in the system.</fsummary>
      <desc>
        <p>Shorthand for
          <seealso marker="#carriers/1"><c>carriers(#{})</c>.</seealso></p>
      </desc>
    </func>

    <func>
      <name name="carriers" arity="1" since="OTP 21.0"/>
      <fsummary>Return a list of all carriers filtered by allocator type and
        scheduler id.</fsummary>
      <desc>
        <p>Returns a summary of all carriers in the system, optionally filtered
          by allocator type and scheduler id.</p>
        <p>If the specified allocator types are not enabled, the call will fail
          with <c>{error, not_enabled}</c>.</p>
        <p>The following options can be used:</p>
        <taglist>
          <tag><c>allocator_types</c></tag>
          <item>
            <p>The allocator types that will be searched. Defaults to all
              <c>alloc_util</c> allocators.</p>
          </item>
          <tag><c>scheduler_ids</c></tag>
          <item>
            <p>The scheduler ids whose allocator instances will be searched. A
              scheduler id of 0 will refer to the global instance that is not
              tied to any particular scheduler. Defaults to all schedulers and
              the global instance.</p>
          </item>
          <tag><c>histogram_start</c></tag>
          <item>
            <p>The upper bound of the first interval in the free block size
              histograms. Defaults to 512.</p>
          </item>
          <tag><c>histogram_width</c></tag>
          <item>
            <p>The number of intervals in the free block size histograms.
              Defaults to 14.</p>
          </item>
        </taglist>
        <p><em>Example:</em></p>
        <code type="none"><![CDATA[
> instrument:carriers(#{ histogram_start => 512, histogram_width => 8 }).
{ok,{512,
     [{ll_alloc,1048576,0,1048344,71,false,{0,0,0,0,0,0,0,0}},
      {binary_alloc,1048576,0,324640,13,false,{3,0,0,1,0,0,0,2}},
      {eheap_alloc,2097152,0,1037200,45,false,{2,1,1,3,4,3,2,2}},
      {fix_alloc,32768,0,29544,82,false,{22,0,0,0,0,0,0,0}},
      {...}|...]}}
     ]]></code>
      </desc>
    </func>

  </funcs>

  <section>
    <title>See Also</title>
    <p><seealso marker="erts:erts_alloc">erts_alloc(3)</seealso>,
      <seealso marker="erts:erl">erl(1)</seealso></p>
  </section>
</erlref>