diff options
Diffstat (limited to 'erts/doc/src/erlang.xml')
| -rw-r--r-- | erts/doc/src/erlang.xml | 479 |
1 files changed, 450 insertions, 29 deletions
diff --git a/erts/doc/src/erlang.xml b/erts/doc/src/erlang.xml index 7926157fa5..1f8e89768c 100644 --- a/erts/doc/src/erlang.xml +++ b/erts/doc/src/erlang.xml @@ -59,6 +59,12 @@ </datatype> <datatype> + <name name="message_queue_data"></name> + <desc><p>See <seealso marker="#process_flag_message_queue_data"><c>erlang:process_flag(message_queue_data, MQD)</c></seealso>.</p> + </desc> + </datatype> + + <datatype> <name name="timestamp"></name> <desc><p>See <seealso marker="#timestamp/0">erlang:timestamp/0</seealso>.</p> </desc> @@ -125,6 +131,17 @@ </note> </item> + <tag><c>perf_counter</c></tag> + <item><p>Symbolic representation of the performance counter + time unit used by the Erlang runtime system.</p> + + <p>The <c>perf_counter</c> time unit behaves much in the same way + as the <c>native</c> time unit. That is it might differ inbetween + run-time restarts. You get values of this type by calling + <seealso marker="kernel:os#perf_counter/0"><c>os:perf_counter()</c></seealso> + </p> + </item> + </taglist> <p>The <c>time_unit/0</c> type may be extended. Use @@ -1105,7 +1122,7 @@ <fsummary>Prints a term on standard output.</fsummary> <desc> <p>Prints a text representation of <c><anno>Term</anno></c> on the - standard output. On OSE, the term is printed to the ramlog.</p> + standard output.</p> <warning> <p>This BIF is intended for debugging only.</p> </warning> @@ -4281,9 +4298,59 @@ os_prompt% </pre> <p>Returns the old value of the flag.</p> </desc> </func> - + <marker id="process_flag_message_queue_data"/> <func> <name name="process_flag" arity="2" clause_i="5"/> + <fsummary>Set process flag <c>message_queue_data</c> for the calling process</fsummary> + <type name="message_queue_data"/> + <desc> + <p>This flag determines how messages in the message queue + are stored. When the flag is:</p> + <taglist> + <tag><c>off_heap</c></tag> + <item><p> + <em>All</em> messages in the message queue will be stored + outside of the process heap. This implies that <em>no</em> + messages in the message queue will be part of a garbage + collection of the process. + </p></item> + <tag><c>on_heap</c></tag> + <item><p> + All messages in the message queue will eventually be + placed on heap. They may however temporarily be stored + off heap. This is how messages always have been stored + up until ERTS version 8.0. + </p></item> + <tag><c>mixed</c></tag> + <item><p> + Messages may be placed either on the heap or outside + of the heap. + </p></item> + </taglist> + <p> + The default <c>message_queue_data</c> process flag is determined + by the <seealso marker="erl#+xmqd"><c>+xmqd</c></seealso> + <c>erl</c> command line argument. + </p> + <p> + If the process potentially may get a hugh amount of messages, + you are recommended to set the flag to <c>off_heap</c>. This + since a garbage collection with lots of messages placed on + the heap may become extremly expensive and the process may + consume large amounts of memory. Performance of the + actual message passing is however generally better when not + using the <c>off_heap</c> flag. + </p> + <p> + When changing this flag messages will be moved. This work + has been initiated but not completed when this function + call returns. + </p> + <p>Returns the old value of the flag.</p> + </desc> + </func> + <func> + <name name="process_flag" arity="2" clause_i="6"/> <fsummary>Sets process flag <c>priority</c> for the calling process.</fsummary> <type name="priority_level"/> <desc> @@ -4357,7 +4424,7 @@ os_prompt% </pre> </func> <func> - <name name="process_flag" arity="2" clause_i="6"/> + <name name="process_flag" arity="2" clause_i="7"/> <fsummary>Sets process flag <c>save_calls</c> for the calling process.</fsummary> <desc> <p><c><anno>N</anno></c> must be an integer in the interval 0..10000. @@ -4388,7 +4455,7 @@ os_prompt% </pre> </func> <func> - <name name="process_flag" arity="2" clause_i="7"/> + <name name="process_flag" arity="2" clause_i="8"/> <fsummary>Sets process flag <c>sensitive</c> for the calling process.</fsummary> <desc> <p>Sets or clears flag <c>sensitive</c> for the current process. @@ -4442,6 +4509,7 @@ os_prompt% </pre> <type name="process_info_result_item"/> <type name="priority_level"/> <type name="stack_item"/> + <type name="message_queue_data" /> <desc> <p>Returns a list containing <c><anno>InfoTuple</anno></c>s with miscellaneous information about the process identified by @@ -4494,6 +4562,7 @@ os_prompt% </pre> <type name="process_info_result_item"/> <type name="stack_item"/> <type name="priority_level"/> + <type name="message_queue_data" /> <desc> <p>Returns information about the process identified by <c><anno>Pid</anno></c>, as specified by @@ -4585,6 +4654,17 @@ os_prompt% </pre> The content of <c><anno>GCInfo</anno></c> can be changed without prior notice.</p> </item> + <tag><c>{garbage_collection_info, <anno>GCInfo</anno>}</c></tag> + <item> + <p><c><anno>GCInfo</anno></c> is a list containing miscellaneous + detailed information about garbage collection for this process. + The content of <c><anno>GCInfo</anno></c> can be changed without + prior notice. + See <seealso marker="#gc_start">gc_start</seealso> in + <seealso marker="#trace/3">erlang:trace/3</seealso> for details about + what each item means. + </p> + </item> <tag><c>{group_leader, <anno>GroupLeader</anno>}</c></tag> <item> <p><c><anno>GroupLeader</anno></c> is group leader for the I/O of @@ -4662,6 +4742,15 @@ os_prompt% </pre> monitor by name, the list item is <c>{process, {<anno>RegName</anno>, <anno>Node</anno>}}</c>.</p> </item> + <tag><c>{message_queue_data, <anno>MQD</anno>}</c></tag> + <item> + <p>Returns the current state of the <c>message_queue_data</c> + process flag. <c><anno>MQD</anno></c> is either <c>off_heap</c>, + <c>on_heap</c>, or <c>mixed</c>. For more information, see the + documentation of + <seealso marker="#process_flag_message_queue_data"><c>process_flag(message_queue_data, + MQD)</c></seealso>.</p> + </item> <tag><c>{priority, <anno>Level</anno>}</c></tag> <item> <p><c><anno>Level</anno></c> is the current priority level for @@ -4804,6 +4893,12 @@ os_prompt% </pre> <seealso marker="kernel:code">code(3)</seealso>) and is not to be used elsewhere.</p> </warning> + <note> + <p>As from <c>ERTS</c> 8.0 (OTP 19), any lingering processes + that still execute the old code will be killed by this function. + In earlier versions, such incorrect use could cause much + more fatal failures, like emulator crash.</p> + </note> <p>Failure: <c>badarg</c> if there is no old code for <c><anno>Module</anno></c>.</p> </desc> @@ -5430,6 +5525,8 @@ true</pre> <name name="spawn_opt" arity="2"/> <fsummary>Creates a new process with a fun as entry point.</fsummary> <type name="priority_level"/> + <type name="message_queue_data" /> + <type name="spawn_opt_option" /> <desc> <p>Returns the process identifier (pid) of a new process started by the application of <c><anno>Fun</anno></c> @@ -5445,6 +5542,8 @@ true</pre> <name name="spawn_opt" arity="3"/> <fsummary>Creates a new process with a fun as entry point on a given node.</fsummary> <type name="priority_level"/> + <type name="message_queue_data" /> + <type name="spawn_opt_option" /> <desc> <p>Returns the process identifier (pid) of a new process started by the application of <c><anno>Fun</anno></c> to the @@ -5459,6 +5558,8 @@ true</pre> <name name="spawn_opt" arity="4"/> <fsummary>Creates a new process with a function as entry point.</fsummary> <type name="priority_level"/> + <type name="message_queue_data" /> + <type name="spawn_opt_option" /> <desc> <p>Works as <seealso marker="#spawn/3">spawn/3</seealso>, except that an @@ -5560,6 +5661,18 @@ true</pre> fine-tuning an application and to measure the execution time with various <c><anno>VSize</anno></c> values.</p> </item> + <tag><c>{message_queue_data, <anno>MQD</anno>}</c></tag> + <item> + <p>Sets the state of the <c>message_queue_data</c> process + flag. <c><anno>MQD</anno></c> should be either <c>off_heap</c>, + <c>on_heap</c>, or <c>mixed</c>. The default + <c>message_queue_data</c> process flag is determined by the + <seealso marker="erl#+xmqd"><c>+xmqd</c></seealso> <c>erl</c> + command line argument. For more information, see the + documentation of + <seealso marker="#process_flag_message_queue_data"><c>process_flag(message_queue_data, + <anno>MQD</anno>)</c></seealso>.</p> + </item> </taglist> </desc> </func> @@ -5568,6 +5681,8 @@ true</pre> <name name="spawn_opt" arity="5"/> <fsummary>Creates a new process with a function as entry point on a given node.</fsummary> <type name="priority_level"/> + <type name="message_queue_data" /> + <type name="spawn_opt_option" /> <desc> <p>Returns the process identifier (pid) of a new process started by the application @@ -5685,8 +5800,31 @@ true</pre> <anno>Dest</anno>, <anno>Msg</anno>, [])</c></seealso>.</p> </desc> </func> + <func> <name name="statistics" arity="1" clause_i="1"/> + <fsummary>Information about active processes and ports.</fsummary> + <desc><marker id="statistics_active_tasks"></marker> + <p> + Returns a list where each element represents the amount + of active processes and ports on each run queue and its + associated scheduler. That is, the number of processes and + ports that are ready to run, or are currently running. The + element location in the list corresponds to the scheduler + and its run queue. The first element corresponds to scheduler + number 1 and so on. The information is <em>not</em> gathered + atomically. That is, the result is not necessarily a + consistent snapshot of the state, but instead quite + efficiently gathered. See also, + <seealso marker="#statistics_total_active_tasks"><c>statistics(total_active_tasks)</c></seealso>, + <seealso marker="#statistics_run_queue_lengths"><c>statistics(run_queue_lengths)</c></seealso>, and + <seealso marker="#statistics_total_run_queue_lengths"><c>statistics(total_run_queue_lengths)</c></seealso>. + </p> + </desc> + </func> + + <func> + <name name="statistics" arity="1" clause_i="2"/> <fsummary>Information about context switches.</fsummary> <desc> <p>Returns the total number of context switches since the @@ -5695,7 +5833,7 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="2"/> + <name name="statistics" arity="1" clause_i="3"/> <fsummary>Information about exact reductions.</fsummary> <desc> <marker id="statistics_exact_reductions"></marker> @@ -5709,7 +5847,7 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="3"/> + <name name="statistics" arity="1" clause_i="4"/> <fsummary>Information about garbage collection.</fsummary> <desc> <p>Returns information about garbage collection, for example:</p> @@ -5721,7 +5859,7 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="4"/> + <name name="statistics" arity="1" clause_i="5"/> <fsummary>Information about I/O.</fsummary> <desc> <p>Returns <c><anno>Input</anno></c>, @@ -5732,7 +5870,147 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="5"/> + <name name="statistics" arity="1" clause_i="6"/> + <fsummary>Information about microstate accounting.</fsummary> + <desc> + <marker id="statistics_microstate_accounting"></marker> + <p> + Microstate accounting can be used to measure how much time the Erlang + runtime system spends doing various tasks. It is designed to be as + lightweight as possible, but there will be some overhead when this + is enabled. Microstate accounting is meant to be a profiling tool + to help figure out performance bottlenecks. + To <c>start</c>/<c>stop</c>/<c>reset</c> microstate_accounting you use + the system_flag + <seealso marker="#system_flag_microstate_accounting"> + <c>microstate_accounting</c></seealso>. + </p> + <p> + <c>erlang:statistics(microstate_accounting)</c> returns a list of maps + representing some of the OS threads within ERTS. Each map contains + <c>type</c> and <c>id</c> fields that can be used to identify what + thread it is, and also a counters field that contains data about how + much time has been spent in the various states.</p> + <pre> +> <input>erlang:statistics(microstate_accounting).</input> +[#{counters => #{aux => 1899182914, + check_io => 2605863602, + emulator => 45731880463, + gc => 1512206910, + other => 5421338456, + port => 221631, + sleep => 5150294100}, + id => 1, + type => scheduler}|...] + </pre> + <p>The time unit is the same as returned by + <seealso marker="kernel:os#perf_counter/0"> + <c>os:perf_counter/0</c></seealso>. + So to convert it to milliseconds you could do something like this:</p> + <pre> +lists:map( + fun(#{ counters := Cnt } = M) -> + MsCnt = maps:map(fun(_K, PerfCount) -> + erlang:convert_time_unit(PerfCount, perf_counter, 1000) + end, Cnt), + M#{ counters := MsCnt } + end, erlang:statistics(microstate_accounting)). + </pre> + <p> + It is important to note that these values are not guaranteed to be + the exact time spent in each state. This is because of various + optimisation done in order to keep the overhead as small as possible. + </p> + + <p>Currently the following <c><anno>MSAcc_Thread_Type</anno></c> are available:</p> + <taglist> + <tag><c>scheduler</c></tag> + <item>The main execution threads that do most of the work.</item> + <tag><c>async</c></tag><item>Async threads are used by various + linked-in drivers (mainly the file drivers) do offload non-cpu + intensive work.</item> + <tag><c>aux</c></tag><item>Takes care of any work that is not + specifically assigned to a scheduler.</item> + </taglist> + <p>Currently the following <c><anno>MSAcc_Thread_State</anno></c>s are available. + All states are exclusive, meaning that a thread cannot be in two states + at once. So if you add the numbers of all counters in a thread + you will get the total run-time for that thread.</p> + <taglist> + <tag><c>aux</c></tag> + <item>Time spent handling auxiliary jobs.</item> + <tag><c>check_io</c></tag> + <item>Time spent checking for new I/O events.</item> + <tag><c>emulator</c></tag> + <item>Time spent executing erlang processes.</item> + <tag><c>gc</c></tag> + <item>Time spent doing garbage collection. When extra states are + enabled this is the time spent doing non-fullsweep garbage + collections.</item> + <tag><c>other</c></tag> + <item>Time spent doing unaccounted things.</item> + <tag><c>port</c></tag> + <item>Time spent executing ports.</item> + <tag><c>sleep</c></tag> + <item>Time spent sleeping.</item> + </taglist> + <p>It is possible to add more fine grained <c><anno>MSAcc_Thread_State</anno></c>s + through configure. + (e.g. <c>./configure --with-microstate-accounting=extra</c>). + Enabling these states will cause a performance degradation when + microstate accounting is turned off and increase the overhead when + it is turned on.</p> + <taglist> + <tag><c>alloc</c></tag> + <item>Time spent managing memory. Without extra states this time is + spread out over all other states.</item> + <tag><c>bif</c></tag> + <item>Time spent in bifs. Without extra states this time is part of + the <c>emulator</c> state.</item> + <tag><c>busy_wait</c></tag> + <item>Time spent busy waiting. This is also the state where a + scheduler no longer reports that it is active when using + <seealso marker="#statistics_scheduler_wall_time"> + <c>erlang:statistics(scheduler_wall_time)</c></seealso>. + So if you add all other states but this and sleep and then divide that + by all time in the thread you should get something very similar to the + scheduler_wall_time fraction. Without extra states this time is part + of the <c>other</c> state.</item> + <tag><c>ets</c></tag> + <item>Time spent executing ETS bifs. Without extra states this time is + part of the <c>emulator</c> state.</item> + <tag><c>gc_full</c></tag> + <item>Time spent doing fullsweep garbage collection. Without extra + states this time is part of the <c>gc</c> state.</item> + <tag><c>nif</c></tag> + <item>Time spent in nifs. Without extra states this time is part of + the <c>emulator</c> state.</item> + <tag><c>send</c></tag> + <item>Time spent sending messages (processes only). Without extra + states this time is part of the <c>emulator</c> state.</item> + <tag><c>timers</c></tag> + <item>Time spent managing timers. Without extra states this time is + part of the <c>other</c> state.</item> + </taglist> + <p>There is a utility module called + <seealso marker="runtime_tools:msacc"><c>msacc</c></seealso> in + runtime_tools that can be used to more easily analyse these + statistics.</p> + + <p> + Returns <c>undefined</c> if the system flag + <seealso marker="#system_flag_microstate_accounting"> + <c>microstate_accounting</c></seealso> + is turned off. + </p> + <p>The list of thread information is unsorted and may appear in + different order between calls.</p> + <note><p>The threads and states are subject to change without any + prior notice.</p></note> + </desc> + </func> + <func> + <name name="statistics" arity="1" clause_i="7"/> <fsummary>Information about reductions.</fsummary> <desc> <marker id="statistics_reductions"></marker> @@ -5750,16 +6028,43 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="6"/> - <fsummary>Information about the run-queue.</fsummary> - <desc> - <p>Returns the total length of run-queues, that is, the number - of processes that are ready to run on all available run-queues.</p> + <name name="statistics" arity="1" clause_i="8"/> + <fsummary>Information about the run-queues.</fsummary> + <desc><marker id="statistics_run_queue"></marker> + <p> + Returns the total length of the run-queues. That is, the number + of processes and ports that are ready to run on all available + run-queues. The information is gathered atomically. That + is, the result is a consistent snapshot of the state, but + this operation is much more expensive compared to + <seealso marker="#statistics_total_run_queue_lengths"><c>statistics(total_run_queue_lengths)</c></seealso>. + This especially when a large amount of schedulers is used. + </p> </desc> </func> <func> - <name name="statistics" arity="1" clause_i="7"/> + <name name="statistics" arity="1" clause_i="9"/> + <fsummary>Information about the run-queue lengths.</fsummary> + <desc><marker id="statistics_run_queue_lengths"></marker> + <p> + Returns a list where each element represents the amount + of processes and ports ready to run for each run queue. The + element location in the list corresponds to the run queue + of a scheduler. The first element corresponds to the run + queue of scheduler number 1 and so on. The information is + <em>not</em> gathered atomically. That is, the result is + not necessarily a consistent snapshot of the state, but + instead quite efficiently gathered. See also, + <seealso marker="#statistics_total_run_queue_lengths"><c>statistics(total_run_queue_lengths)</c></seealso>, + <seealso marker="#statistics_active_tasks"><c>statistics(active_tasks)</c></seealso>, and + <seealso marker="#statistics_total_active_tasks"><c>statistics(total_active_tasks)</c></seealso>. + </p> + </desc> + </func> + + <func> + <name name="statistics" arity="1" clause_i="10"/> <fsummary>Information about runtime.</fsummary> <desc> <p>Returns information about runtime, in milliseconds.</p> @@ -5774,7 +6079,7 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="8"/> + <name name="statistics" arity="1" clause_i="11"/> <fsummary>Information about each schedulers work time.</fsummary> <desc> <marker id="statistics_scheduler_wall_time"></marker> @@ -5845,7 +6150,44 @@ ok </func> <func> - <name name="statistics" arity="1" clause_i="9"/> + <name name="statistics" arity="1" clause_i="12"/> + <fsummary>Information about active processes and ports.</fsummary> + <desc><marker id="statistics_total_active_tasks"></marker> + <p> + Returns the total amount of active processes and ports in + the system. That is, the number of processes and ports that + are ready to run, or are currently running. The information + is <em>not</em> gathered atomically. That is, the result + is not necessarily a consistent snapshot of the state, but + instead quite efficiently gathered. See also, + <seealso marker="#statistics_active_tasks"><c>statistics(active_tasks)</c></seealso>, + <seealso marker="#statistics_run_queue_lengths"><c>statistics(run_queue_lengths)</c></seealso>, and + <seealso marker="#statistics_total_run_queue_lengths"><c>statistics(total_run_queue_lengths)</c></seealso>. + </p> + </desc> + </func> + + <func> + <name name="statistics" arity="1" clause_i="13"/> + <fsummary>Information about the run-queue lengths.</fsummary> + <desc><marker id="statistics_total_run_queue_lengths"></marker> + <p> + Returns the total length of the run-queues. That is, the number + of processes and ports that are ready to run on all available + run-queues. The information is <em>not</em> gathered atomically. + That is, the result is not necessarily a consistent snapshot of + the state, but much more efficiently gathered compared to + <seealso marker="#statistics_run_queue"><c>statistics(run_queue)</c></seealso>. + See also, + <seealso marker="#statistics_run_queue_lengths"><c>statistics(run_queue_lengths)</c></seealso>, + <seealso marker="#statistics_total_active_tasks"><c>statistics(total_active_tasks)</c></seealso>, and + <seealso marker="#statistics_active_tasks"><c>statistics(active_tasks)</c></seealso>. + </p> + </desc> + </func> + + <func> + <name name="statistics" arity="1" clause_i="14"/> <fsummary>Information about wall clock.</fsummary> <desc> <p>Returns information about wall clock. <c>wall_clock</c> can @@ -6079,6 +6421,17 @@ ok <func> <name name="system_flag" arity="2" clause_i="5"/> + <fsummary>Set system flag microstate_accounting</fsummary> + <desc><p><marker id="system_flag_microstate_accounting"></marker> + Turns on/off microstate accounting measurements. By passing reset it is possible to reset + all counters to 0.</p> + <p>For more information see, + <seealso marker="#statistics_microstate_accounting">erlang:statistics(microstate_accounting)</seealso>. + </p> + </desc> + </func> + <func> + <name name="system_flag" arity="2" clause_i="6"/> <fsummary>Sets system flag <c>min_heap_size</c>.</fsummary> <desc> <p>Sets the default minimum heap size for processes. The size @@ -6093,7 +6446,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="6"/> + <name name="system_flag" arity="2" clause_i="7"/> <fsummary>Sets system flag <c>min_bin_vheap_size</c>.</fsummary> <desc> <p>Sets the default minimum binary virtual heap size for @@ -6110,7 +6463,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="7"/> + <name name="system_flag" arity="2" clause_i="8"/> <fsummary>Sets system flag <c>multi_scheduling</c>.</fsummary> <desc> <p><marker id="system_flag_multi_scheduling"></marker> @@ -6148,7 +6501,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="8"/> + <name name="system_flag" arity="2" clause_i="9"/> <fsummary>Sets system flag <c>scheduler_bind_type</c>.</fsummary> <type name="scheduler_bind_type"/> <desc> @@ -6266,7 +6619,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="9"/> + <name name="system_flag" arity="2" clause_i="10"/> <fsummary>Sets system flag <c>scheduler_wall_time</c>.</fsummary> <desc><p><marker id="system_flag_scheduler_wall_time"></marker> Turns on or off scheduler wall time measurements.</p> @@ -6276,7 +6629,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="10"/> + <name name="system_flag" arity="2" clause_i="11"/> <fsummary>Sets system flag <c>schedulers_online</c>.</fsummary> <desc> <p><marker id="system_flag_schedulers_online"></marker> @@ -6301,7 +6654,7 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="11"/> + <name name="system_flag" arity="2" clause_i="12"/> <fsummary>Sets system flag <c>trace_control_word</c>.</fsummary> <desc> <p>Sets the value of the node trace control word to @@ -6640,6 +6993,7 @@ ok <name name="system_info" arity="1" clause_i="65"/> <name name="system_info" arity="1" clause_i="66"/> <name name="system_info" arity="1" clause_i="67"/> + <name name="system_info" arity="1" clause_i="68"/> <fsummary>Information about the system.</fsummary> <desc> <p>Returns various information about the current system @@ -7045,7 +7399,17 @@ ok used by the runtime system. It is on the form "<major ver>.<minor ver>".</p> </item> - <tag><c>otp_release</c></tag> + <tag><marker id="system_info_message_queue_data"><c>message_queue_data</c></marker></tag> + <item> + <p>Returns the default value of the <c>message_queue_data</c> + process flag which is either <c>off_heap</c>, <c>on_heap</c>, or <c>mixed</c>. + This default is set by the <c>erl</c> command line argument + <seealso marker="erl#+xmqd"><c>+xmqd</c></seealso>. For more information on the + <c>message_queue_data</c> process flag, see documentation of + <seealso marker="#process_flag_message_queue_data"><c>process_flag(message_queue_data, + MQD)</c></seealso>.</p> + </item> + <tag><marker id="system_info_otp_release"><c>otp_release</c></marker></tag> <item> <marker id="system_info_otp_release"></marker> <p>Returns a string containing the OTP release number of the @@ -7653,6 +8017,14 @@ ok <c>inactive</c>, and later <c>active</c> when the port callback returns.</p> </item> + <tag><c>monotonic_timestamp</c></tag> + <item> + <p>Timestamps in profile messages will use + <seealso marker="time_correction#Erlang_Monotonic_Time">Erlang + monotonic time</seealso>. The time-stamp (Ts) has the same + format and value as produced by + <c>erlang:monotonic_time(nano_seconds)</c>.</p> + </item> <tag><c>runnable_procs</c></tag> <item> <p>If a process is put into or removed from the run queue, a @@ -7673,6 +8045,25 @@ ok <c>{profile, scheduler, Id, State, NoScheds, Ts}</c>, is sent to <c><anno>ProfilerPid</anno></c>.</p> </item> + <tag><c>strict_monotonic_timestamp</c></tag> + <item> + <p>Timestamps in profile messages will consisting of + <seealso marker="time_correction#Erlang_Monotonic_Time">Erlang + monotonic time</seealso> and a monotonically increasing + integer. The time-stamp (Ts) has the same format and value + as produced by <c>{erlang:monotonic_time(nano_seconds), + erlang:unique_integer([monotonic])}</c>.</p> + </item> + <tag><c>timestamp</c></tag> + <item> + <p>Timestamps in profile messages will include a + time-stamp (Ts) that has the same form as returned by + <c>erlang:now()</c>. This is also the default if no + timestamp flag is given. If <c>cpu_timestamp</c> has + been enabled via <c>erlang:trace/3</c>, this will also + effect the timestamp produced in profiling messages + when <c>timestamp</c> flag is enabled.</p> + </item> </taglist> <note><p><c>erlang:system_profile</c> is considered experimental and its behavior can change in a future release.</p> @@ -8032,7 +8423,10 @@ timestamp() -> <tag><c>cpu_timestamp</c></tag> <item> <p>A global trace flag for the Erlang node that makes all - trace time-stamps to be in CPU time, not wall clock time. + trace time-stamps using the <c>timestamp</c> flag to be + in CPU time, not wall clock time. That is, <c>cpu_timestamp</c> + will not be used if <c>monotonic_timestamp</c>, or + <c>strict_monotonic_timestamp</c> is enabled. Only allowed with <c>PidSpec==all</c>. If the host machine OS does not support high-resolution CPU time measurements, <c>trace/3</c> exits with @@ -8040,6 +8434,26 @@ timestamp() -> not synchronize this value across cores, so be prepared that time might seem to go backwards when using this option.</p> </item> + <tag><c>monotonic_timestamp</c></tag> + <item> + <p>Includes an + <seealso marker="time_correction#Erlang_Monotonic_Time">Erlang + monotonic time</seealso> time-stamp in all trace messages. The + time-stamp (Ts) has the same format and value as produced by + <c>erlang:monotonic_time(nano_seconds)</c>. This flag overrides + the <c>cpu_timestamp</c> flag.</p> + </item> + <tag><c>strict_monotonic_timestamp</c></tag> + <item> + <p>Includes an timestamp consisting of + <seealso marker="time_correction#Erlang_Monotonic_Time">Erlang + monotonic time</seealso> and a monotonically increasing + integer in all trace messages. The time-stamp (Ts) has the + same format and value as produced by + <c>{erlang:monotonic_time(nano_seconds), + erlang:unique_integer([monotonic])}</c>. This flag overrides + the <c>cpu_timestamp</c> flag.</p> + </item> <tag><c>arity</c></tag> <item> <p>Used with the <c>call</c> trace flag. @@ -8086,9 +8500,16 @@ timestamp() -> in the following list. <c>Pid</c> is the process identifier of the traced process in which the traced event has occurred. The third tuple element is the message tag.</p> - <p>If flag <c>timestamp</c> is given, the first tuple - element is <c>trace_ts</c> instead, and the time-stamp - is added last in the message tuple.</p> + <p>If flag <c>timestamp</c>, <c>strict_monotonic_timestamp</c>, or + <c>monotonic_timestamp</c> is given, the first tuple + element is <c>trace_ts</c> instead, and the time-stamp + is added as an extra element last in the message tuple. If + multiple timestamp flags are passed, <c>timestamp</c> has + precedence over <c>strict_monotonic_timestamp</c> which + in turn has precedence over <c>monotonic_timestamp</c>. All + timestamp flags are remembered, so if two are passed + and the one with highest precedence later is disabled + the other one will become active.</p> <marker id="trace_3_trace_messages"></marker> <taglist> <tag><c>{trace, Pid, 'receive', Msg}</c></tag> @@ -8183,14 +8604,14 @@ timestamp() -> <p>When <c>Pid</c> is scheduled to run. The process runs in function <c>{M, F, Arity}</c>. On some rare occasions, the current function cannot be determined, - then the last element <c>Arity</c> is <c>0</c>.</p> + then the last element is <c>0</c>.</p> </item> <tag><c>{trace, Pid, out, {M, F, Arity} | 0}</c></tag> <item> <p>When <c>Pid</c> is scheduled out. The process was running in function {M, F, Arity}. On some rare occasions, the current function cannot be determined, then the last - element <c>Arity</c> is <c>0</c>.</p> + element is <c>0</c>.</p> </item> <tag><c>{trace, Pid, gc_start, Info}</c></tag> <item> |