<?xml version="1.0" encoding="utf-8" ?>
<!DOCTYPE erlref SYSTEM "erlref.dtd">
<erlref>
<header>
<copyright>
<year>2009</year>
<year>2017</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.
The Initial Developer of the Original Code is Ericsson AB.
</legalnotice>
<title>lcnt</title>
<prepared>Björn-Egil Dahlberg</prepared>
<responsible>nobody</responsible>
<docno></docno>
<approved>nobody</approved>
<checked></checked>
<date>2009-11-26</date>
<rev>PA1</rev>
<file>lcnt.xml</file>
</header>
<module>lcnt</module>
<modulesummary>A runtime system Lock Profiling tool.</modulesummary>
<description>
<p>The <c>lcnt</c> module is used to profile the internal ethread locks in the
Erlang Runtime System. With <c>lcnt</c> enabled, internal counters in the
runtime system are updated each time a lock is taken. The counters stores
information about the number of acquisition tries and the number of collisions
that has occurred during the acquisition tries. The counters also record the
waiting time a lock has caused for a blocked thread when a collision has occurred.
</p>
<p>
The data produced by the lock counters will give an estimate on how well
the runtime system will behave from a parallelizable view point for the
scenarios tested. This tool was mainly developed to help Erlang runtime
developers iron out potential and generic bottlenecks.
</p>
<p>Locks in the emulator are named after what type of resource they protect and where
in the emulator they are initialized, those are lock 'classes'. Most of those
locks are also instantiated several times, and given unique identifiers, to increase
locking granularity. Typically an instantiated lock protects a disjunct set of
the resource, for example ets tables, processes or ports. In other cases it protects a
specific range of a resource, for example <c>pix_lock</c> which protects index to process
mappings, and is given a unique number within the class. A unique lock in <c>lcnt</c>
is referenced by a name (class) and an identifier: <c>{Name, Id}</c>.
</p>
<p>Some locks in the system are static and protects global resources, for example
<c>bif_timers</c> and the <c>run_queue</c> locks. Other locks are dynamic and not
necessarily long lived, for example process locks and ets-table locks. The
statistics data from short lived locks can be stored separately when the locks
are deleted. This behavior is by default turned off to save memory but can be
turned on via <c>lcnt:rt_opt({copy_save, true})</c>. The <c>lcnt:apply/1,2,3</c>
functions enables this behavior during profiling.
</p>
</description>
<funcs>
<func>
<name>start() -> {ok, Pid} | {error, {already_started, Pid}} </name>
<fsummary>Starts the lock profiler server.</fsummary>
<type>
<v>Pid = pid()</v>
</type>
<desc>
<p>Starts the lock profiler server. The server only act as a medium for the
user and performs filtering and printing of data collected by <c>lcnt:collect/1</c>.
</p>
</desc>
</func>
<func>
<name>stop() -> ok</name>
<fsummary>Stops the lock profiler server.</fsummary>
<desc>
<p>Stops the lock profiler server.</p>
</desc>
</func>
<func>
<name>collect() -> ok</name>
<fsummary>Same as <c>collect(node())</c>.</fsummary>
<desc><p>Same as <c>collect(node())</c>.</p></desc>
</func>
<func>
<name>collect(Node) -> ok</name>
<fsummary>Collects lock statistics from the runtime system.</fsummary>
<type>
<v>Node = node()</v>
</type>
<desc>
<p>Collects lock statistics from the runtime system. The function starts a
server if it is not already started. It then populates the server with lock
statistics. If the server held any lock statistics data before the collect then
that data is lost.
</p>
</desc>
</func>
<func>
<name>clear() -> ok</name>
<fsummary>Same as <c>clear(node())</c>.</fsummary>
<desc><p>Same as <c>clear(node())</c>.</p></desc>
</func>
<func>
<name>clear(Node) -> ok</name>
<fsummary>Clears the internal lock statistics from runtime system.</fsummary>
<type>
<v>Node = node()</v>
</type>
<desc>
<p>Clears the internal lock statistics from the runtime system. This does not clear the
data on the server only on runtime system. All counters for static locks are zeroed,
all dynamic locks currently alive are zeroed and all saved locks now destroyed are removed.
It also resets the duration timer.
</p>
</desc>
</func>
<func>
<name>conflicts() -> ok</name>
<fsummary>Same as <c>conflicts([])</c>.</fsummary>
<desc><p>Same as <c>conflicts([])</c>.</p></desc>
</func>
<func>
<name>conflicts([Option]) -> ok</name>
<fsummary>Prints a list of internal lock counters.</fsummary>
<type>
<v>Option = {sort, Sort} | {reverse, bool()} | {thresholds, [Thresholds]} | {print, [Print | {Print, integer()}]} | {max_locks, MaxLocks} | {combine, bool()}</v>
<v>Sort = name | id | type | tries | colls | ratio | time | entry</v>
<v>Thresholds = {tries, integer()} | {colls, integer()} | {time, integer()}</v>
<v>Print = name | id | type | entry | tries | colls | ratio | time | duration</v>
<v>MaxLocks = integer() | none</v>
</type>
<desc>
<p>Prints a list of internal locks and its statistics.</p>
<p>For option description, see <seealso marker="#inspect/2">lcnt:inspect/2</seealso>.</p>
</desc>
</func>
<func>
<name>locations() -> ok</name>
<fsummary>Same as <c>locations([])</c>.</fsummary>
<desc>
<p>Same as <c>locations([])</c>.</p>
</desc>
</func>
<func>
<name>locations([Option]) -> ok</name>
<fsummary>Prints a list of internal lock counters by source code locations.</fsummary>
<type>
<v>Option = {sort, Sort} | {thresholds, [Thresholds]} | {print, [Print | {Print, integer()}]} | {max_locks, MaxLocks} | {combine, bool()}</v>
<v>Sort = name | id | type | tries | colls | ratio | time | entry</v>
<v>Thresholds = {tries, integer()} | {colls, integer()} | {time, integer()}</v>
<v>Print = name | id | type | entry | tries | colls | ratio | time | duration</v>
<v>MaxLocks = integer() | none</v>
</type>
<desc>
<p>Prints a list of internal lock counters by source code locations.</p>
<p>For option description, see <seealso marker="#inspect/2">lcnt:inspect/2</seealso>.</p>
</desc>
</func>
<func>
<name>inspect(Lock) -> ok</name>
<fsummary>Same as <c>inspect(Lock, [])</c>.</fsummary>
<desc><p>Same as <c>inspect(Lock, [])</c>.</p></desc>
</func>
<func>
<name>inspect(Lock, [Option]) -> ok</name>
<fsummary>Prints a list of internal lock counters for a specific lock.</fsummary>
<type>
<v>Lock = Name | {Name, Id | [Id]}</v>
<v>Name = atom() | pid() | port()</v>
<v>Id = atom() | integer() | pid() | port()</v>
<v>Option = {sort, Sort} | {thresholds, [Thresholds]} | {print, [Print | {Print, integer()}]} | {max_locks, MaxLocks} | {combine, bool()} | {locations, bool()}</v>
<v>Sort = name | id | type | tries | colls | ratio | time</v>
<v>Thresholds = {tries, integer()} | {colls, integer()} | {time, integer()}</v>
<v>Print = name | id | type | entry | tries | colls | ratio | time | duration</v>
<v>MaxLocks = integer() | none</v>
</type>
<desc>
<p>Prints a list of internal lock counters for a specific lock.</p>
<p>Lock <c>Name</c> and <c>Id</c> for ports and processes are interchangeable with the use of <c>lcnt:swap_pid_keys/0</c> and is the reason why <c>pid()</c> and <c>port()</c> options can be used in both <c>Name</c> and <c>Id</c> space. Both pids and ports are special identifiers with stripped creation and can be recreated with <seealso marker="#pid/3">lcnt:pid/2,3</seealso> and <seealso marker="#port/2">lcnt:port/1,2</seealso>. </p>
<p>Option description:</p>
<taglist>
<tag><c>{combine, bool()}</c></tag>
<item>Combine the statistics from different instances of a lock class.
<br/>Default: <c>true</c>
</item>
<tag><c>{locations, bool()}</c></tag>
<item>Print the statistics by source file and line numbers.
<br/>Default: <c>false</c>
</item>
<tag><c>{max_locks, MaxLocks}</c></tag>
<item>Maximum number of locks printed or no limit with <c>none</c>.
<br/>Default: <c>20</c>
</item>
<tag><c>{print, PrintOptions}</c></tag>
<item>Printing options:
<taglist>
<tag><c>name</c></tag>
<item>Named lock or named set of locks (classes). The same name used for initializing the lock in the VM.</item>
<tag><c>id</c></tag>
<item>Internal id for set of locks, not always unique. This could be table name for ets tables (db_tab), port id for ports, integer identifiers for allocators, etc.</item>
<tag><c>type</c></tag>
<item>Type of lock: <c>rw_mutex</c>, <c>mutex</c>, <c>spinlock</c>, <c>rw_spinlock</c> or <c>proclock</c>.</item>
<tag><c>entry</c></tag>
<item>In combination with <c>{locations, true}</c> this option prints the lock operations source file and line number entry-points along with statistics for each entry. </item>
<tag><c>tries</c></tag>
<item>Number of acquisitions of this lock.</item>
<tag><c>colls</c></tag>
<item>Number of collisions when a thread tried to acquire this lock. This is when a trylock is EBUSY, a write try on read held rw_lock, a try read on write held rw_lock, a thread tries to lock an already locked lock. (Internal states supervises this).</item>
<tag><c>ratio</c></tag>
<item>The ratio between the number of collisions and the number of tries (acquisitions) in percentage.</item>
<tag><c>time</c></tag>
<item>Accumulated waiting time for this lock. This could be greater than actual wall clock time, it is accumulated for all threads. Trylock conflicts does not accumulate time.</item>
<tag><c>duration</c></tag>
<item>Percentage of accumulated waiting time of wall clock time. This percentage can be higher than 100% since accumulated time is from all threads.</item>
</taglist>
<br/>Default: <c>[name,id,tries,colls,ratio,time,duration]</c>
</item>
<tag><c>{reverse, bool()}</c></tag>
<item>Reverses the order of sorting.
<br/>Default: <c>false</c>
</item>
<tag><c>{sort, Sort}</c></tag>
<item>Column sorting orders.
<br/>Default: <c>time</c>
</item>
<tag><c>{thresholds, Thresholds}</c></tag>
<item>Filtering thresholds. Anything values above the threshold value are passed through.
<br/>Default: <c>[{tries, 0}, {colls, 0}, {time, 0}]</c>
</item>
</taglist>
</desc>
</func>
<func>
<name>information() -> ok</name>
<fsummary>Prints lcnt server state and generic information about collected lock statistics.</fsummary>
<desc>
<p>Prints lcnt server state and generic information about collected lock statistics.</p>
</desc>
</func>
<func>
<name>swap_pid_keys() -> ok</name>
<fsummary>Swaps places on <c>Name</c> and <c>Id</c> space for ports and processes.</fsummary>
<desc>
<p>Swaps places on <c>Name</c> and <c>Id</c> space for ports and processes.</p>
</desc>
</func>
<func>
<name>load(Filename) -> ok</name>
<fsummary>Restores previously saved data to the server.</fsummary>
<type>
<v>Filename = filename()</v>
</type>
<desc>
<p>Restores previously saved data to the server.</p>
</desc>
</func>
<func>
<name>save(Filename) -> ok</name>
<fsummary>Saves the collected data to file.</fsummary>
<type>
<v>Filename = filename()</v>
</type>
<desc>
<p>Saves the collected data to file.</p>
</desc>
</func>
</funcs>
<section>
<title>Convenience functions</title>
<p>The following functions are used for convenience.</p>
</section>
<funcs>
<func>
<name>apply(Fun) -> term()</name>
<fsummary>Same as <c>apply(Fun, [])</c>.</fsummary>
<type>
<v>Fun = fun()</v>
</type>
<desc>
<p>Same as <c>apply(Fun, [])</c>.</p>
</desc>
</func>
<func>
<name>apply(Fun, Args) -> term()</name>
<fsummary>Same as <c>apply(Module, Function, Args)</c>.</fsummary>
<type>
<v>Fun = fun()</v>
<v>Args = [term()]</v>
</type>
<desc>
<p>Same as <c>apply(Module, Function, Args)</c>.</p>
</desc>
</func>
<func>
<name>apply(Module, Function, Args) -> term()</name>
<fsummary>Clears counters, applies function and collects the profiling results.</fsummary>
<type>
<v>Module = atom()</v>
<v>Function = atom()</v>
<v>Args = [term()]</v>
</type>
<desc>
<p> Clears the lock counters and then setups the instrumentation to save all destroyed locks.
After setup the function is called, passing the elements in <c>Args</c> as arguments.
When the function returns the statistics are immediately collected to the server. After the
collection the instrumentation is returned to its previous behavior.
The result of the applied function is returned.
</p>
<warning>
<p>
This function should only be used for micro-benchmarks; it sets <c>copy_save</c>
to <c>true</c> for the duration of the call, which can quickly lead to running
out of memory.
</p>
</warning>
</desc>
</func>
<func>
<name>pid(Id, Serial) -> pid()</name>
<fsummary>Same as <c>pid(node(), Id, Serial)</c>.</fsummary>
<desc><p>Same as <c>pid(node(), Id, Serial)</c>.</p></desc>
</func>
<func>
<name>pid(Node, Id, Serial) -> pid()</name>
<fsummary>Creates a process id with creation 0.</fsummary>
<type>
<v>Node = node()</v>
<v>Id = integer()</v>
<v>Serial = integer()</v>
</type>
<desc>
<p>Creates a process id with creation 0. Example:</p>
</desc>
</func>
<func>
<name>port(Id) -> port()</name>
<fsummary>Same as <c>port(node(), Id)</c>.</fsummary>
<desc><p>Same as <c>port(node(), Id)</c>.</p></desc>
</func>
<func>
<name>port(Node, Id) -> port()</name>
<fsummary>Creates a port id with creation 0.</fsummary>
<type>
<v>Node = node()</v>
<v>Id = integer()</v>
</type>
<desc><p>Creates a port id with creation 0.</p></desc>
</func>
</funcs>
<section>
<title>Internal runtime lock counter controllers</title>
<p> The following functions control the behavior of the internal counters. </p>
</section>
<funcs>
<func>
<name>rt_collect() -> [lock_counter_data()]</name>
<fsummary>Same as <c>rt_collect(node())</c>.</fsummary>
<desc> <p>Same as <c>rt_collect(node())</c>.</p> </desc>
</func>
<func>
<name>rt_collect(Node) -> [lock_counter_data()]</name>
<fsummary>Returns a list of raw lock counter data.</fsummary>
<type>
<v>Node = node()</v>
</type>
<desc> <p>Returns a list of raw lock counter data.</p> </desc>
</func>
<func>
<name>rt_clear() -> ok</name>
<fsummary>Same as <c>rt_clear(node())</c>.</fsummary>
<desc> <p>Same as <c>rt_clear(node())</c>.</p> </desc>
</func>
<func>
<name>rt_clear(Node) -> ok</name>
<fsummary>Clears the internal counters.</fsummary>
<type>
<v>Node = node()</v>
</type>
<desc> <p>Clear the internal counters. Same as <c>lcnt:clear(Node)</c>.</p></desc>
</func>
<func>
<name>rt_mask() -> [category_atom()]</name>
<fsummary>Same as <c>rt_mask(node())</c>.</fsummary>
<desc><p>Same as <c>rt_mask(node())</c>.</p></desc>
</func>
<func>
<name>rt_mask(Node) -> [category_atom()]</name>
<fsummary>Returns the current lock category mask.</fsummary>
<type>
<v>Node = node()</v>
</type>
<desc>
<p>
Refer to <c>rt_mask/2</c> for a list of valid categories. All
categories are enabled by default.
</p>
</desc>
</func>
<func>
<name>rt_mask(Categories) -> ok | {error, copy_save_enabled}</name>
<fsummary>Same as <c>rt_mask(node(), Categories)</c>.</fsummary>
<type>
<v>Categories = [atom()]</v>
</type>
<desc><p>Same as <c>rt_mask(node(), Categories)</c>.</p></desc>
</func>
<func>
<name>rt_mask(Node, Categories) -> ok | {error, copy_save_enabled}</name>
<fsummary>Changes the lock category mask.</fsummary>
<type>
<v>Node = node()</v>
<v>Categories = [atom()]</v>
</type>
<desc>
<p>
Sets the lock category mask to the given categories.
</p>
<p>
This will fail if the <c>copy_save</c> option is enabled; see
<c>lcnt:rt_opt/2</c>.
</p>
<p>Valid categories are:</p>
<taglist>
<item><c>allocator</c></item>
<item><c>db</c> (ETS tables)</item>
<item><c>debug</c></item>
<item><c>distribution</c></item>
<item><c>generic</c></item>
<item><c>io</c></item>
<item><c>process</c></item>
<item><c>scheduler</c></item>
</taglist>
<p>
This list is subject to change at any time, as is the category any given lock
may belong to.
</p>
</desc>
</func>
<func>
<name>rt_opt({Type, bool()}) -> bool()</name>
<fsummary>Same as <c>rt_opt(node(), {Type, Opt})</c>.</fsummary>
<desc> <p>Same as <c>rt_opt(node(), {Type, Opt})</c>.</p> </desc>
</func>
<func>
<name>rt_opt(Node, {Type, bool()}) -> bool()</name>
<fsummary>Changes the lock counter behavior and returns the previous behaviour.</fsummary>
<type>
<v>Node = node()</v>
<v>Type = copy_save | process_locks</v>
</type>
<desc>
<p>Option description:</p>
<taglist>
<tag><c>{copy_save, bool()}</c></tag>
<item>Retains the statistics of destroyed locks.
<br/>Default: <c>false</c>
<warning>
<p>
This option will use a lot of memory when enabled, which must be
reclaimed with <c>lcnt:rt_clear</c>. Note that it makes no distinction
between locks that were destroyed and locks for which counting was
disabled, so enabling this option will disable changes to the lock
category mask.
</p>
</warning>
</item>
<tag><c>{process_locks, bool()}</c></tag>
<item>Profile process locks, equal to adding <c>process</c> to the lock category mask;
see <c>lcnt:rt_mask/2</c>
<br/>Default: <c>true</c>
</item>
</taglist>
</desc>
</func>
</funcs>
<section>
<title>See Also</title>
<p> <seealso marker="lcnt_chapter">LCNT User's Guide</seealso></p>
</section>
</erlref>
