diff options
Diffstat (limited to 'erts/doc/src')
| -rw-r--r-- | erts/doc/src/Makefile | 11 | ||||
| -rw-r--r-- | erts/doc/src/atomics.xml | 183 | ||||
| -rw-r--r-- | erts/doc/src/counters.xml | 170 | ||||
| -rw-r--r-- | erts/doc/src/erl_nif.xml | 67 | ||||
| -rw-r--r-- | erts/doc/src/notes.xml | 66 | ||||
| -rw-r--r-- | erts/doc/src/persistent_term.xml | 290 | ||||
| -rw-r--r-- | erts/doc/src/ref_man.xml | 3 | ||||
| -rw-r--r-- | erts/doc/src/specs.xml | 3 |
8 files changed, 763 insertions, 30 deletions
diff --git a/erts/doc/src/Makefile b/erts/doc/src/Makefile index 21aa3db864..40f74b78ff 100644 --- a/erts/doc/src/Makefile +++ b/erts/doc/src/Makefile @@ -52,18 +52,17 @@ XML_REF3_EFILES = \ erlang.xml \ erl_tracer.xml \ init.xml \ + persistent_term.xml \ + atomics.xml \ + counters.xml \ zlib.xml XML_REF3_FILES = \ + $(XML_REF3_EFILES) \ driver_entry.xml \ erl_nif.xml \ - erl_tracer.xml \ erl_driver.xml \ - erl_prim_loader.xml \ - erlang.xml \ - erts_alloc.xml \ - init.xml \ - zlib.xml + erts_alloc.xml XML_PART_FILES = \ part.xml diff --git a/erts/doc/src/atomics.xml b/erts/doc/src/atomics.xml new file mode 100644 index 0000000000..3fca92fb97 --- /dev/null +++ b/erts/doc/src/atomics.xml @@ -0,0 +1,183 @@ +<?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>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 follwing 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"/> + <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> + </desc> + </func> + + <func> + <name name="put" arity="3"/> + <fsummary>Set atomic value</fsummary> + <desc> + <p>Set atomic to <c><anno>Value</anno></c>.</p> + </desc> + </func> + + <func> + <name name="get" arity="2"/> + <fsummary>Read atomic value</fsummary> + <desc> + <p>Read atomic value.</p> + </desc> + </func> + + <func> + <name name="add" arity="3"/> + <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"/> + <fsummary>Atomic add and get</fsummary> + <desc> + <p>Atomic addition and return of the result.</p> + </desc> + </func> + + <func> + <name name="sub" arity="3"/> + <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"/> + <fsummary>Atomic sub and get</fsummary> + <desc> + <p>Atomic subtraction and return of the result.</p> + </desc> + </func> + + <func> + <name name="exchange" arity="3"/> + <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"/> + <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"/> + <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> diff --git a/erts/doc/src/counters.xml b/erts/doc/src/counters.xml new file mode 100644 index 0000000000..ba4a22759f --- /dev/null +++ b/erts/doc/src/counters.xml @@ -0,0 +1,170 @@ +<?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>counters</title> + </header> + <module>counters</module> + <modulesummary>Counter Functions</modulesummary> + <description> + <p>This module provides a set of functions to do operations towards + shared mutable counter variables. The implementation does not utilize any + software level locking, which makes it very efficient for concurrent + access. The counters are organized into arrays with the follwing + semantics:</p> + <list type="bulleted"> + <item> + <p>Counters are 64 bit signed integers.</p> + </item> + <item> + <p>Counters wrap around at overflow and underflow operations.</p> + </item> + <item><p>Counters are initialized to zero and can then only be written to + by adding or subtracting.</p> + </item> + <item> + <p>Write operations guarantee atomicity. No intermediate results can be + seen from a single write operation.</p> + </item> + <item> + <p>Two types of counter arrays can be created with options <c>atomics</c> or + <c>write_concurrency</c>. The <c>atomics</c> counters have good allround + performance with nice consistent semantics while + <c>write_concurrency</c> counters offers even better concurrent + write performance at the expense of some potential read + inconsistencies. See <seealso marker="#new/2"><c>new/2</c></seealso>.</p> + </item> + <item> + <p>Indexes into counter arrays are one-based. A counter array of + size N contains N counters with index from 1 to N.</p> + </item> + </list> + </description> + + <datatypes> + <datatype> + <name name="counters_ref"/> + <desc><p>Identifies a counter array returned from + <seealso marker="#new/2"><c>new/2</c></seealso>.</p> + </desc> + </datatype> + </datatypes> + + <funcs> + <func> + <name name="new" arity="2"/> + <fsummary>Create counter array</fsummary> + <desc> + <p>Create a new counter array of <c><anno>Size</anno></c> counters.</p> + <p>Argument <c><anno>Opts</anno></c> is a list of the following possible + options:</p> + <taglist> + <tag><c>atomics</c> (Default)</tag> + <item><p>Counters will be sequentially consistent. If write + operation A is done sequencially before write operation B, then a concurrent reader + may see none of them, only A, or both A and B. It cannot see only B.</p> + </item> + <tag><c>write_concurrency</c></tag> + <item><p>This is an optimization to achieve very efficient concurrent + <seealso marker="#add/3"><c>add</c></seealso> and <seealso + marker="#sub/3"><c>sub</c></seealso> operations at the expense of potential read + inconsistency and memory consumption per counter.</p> + <p>Read operations may see sequentially inconsistent results with + regard to concurrent write operations. Even if write operation A is done + sequencially before write operation B, a concurrent reader may see any + combination of A and B, including only B. A read operation is only + guaranteed to see all writes done sequentially before the read. No writes + are ever lost, but will eventually all be seen.</p> + <p>The typical use case for <c>write_concurrency</c> is when + concurrent calls to <seealso marker="#add/3"><c>add</c></seealso> and + <seealso marker="#sub/3"><c>sub</c></seealso> toward the same counters + are very frequent, while calls to <seealso marker="#get/2"><c>get</c> + </seealso> and <seealso marker="#put/3"><c>put</c></seealso> are much + less frequent. The lack of absolute read consistency must also be + acceptable.</p> + </item> + </taglist> + </desc> + </func> + + <func> + <name name="get" arity="2"/> + <fsummary>Read counter value</fsummary> + <desc> + <p>Read counter value.</p> + </desc> + </func> + + <func> + <name name="add" arity="3"/> + <fsummary>Add to counter</fsummary> + <desc> + <p>Add <c><anno>Incr</anno></c> to counter at index + <c><anno>Ix</anno></c>.</p> + </desc> + </func> + + <func> + <name name="sub" arity="3"/> + <fsummary>Subtract from counter</fsummary> + <desc> + <p>Subtract <c><anno>Decr</anno></c> from counter at index + <c><anno>Ix</anno></c>.</p> + </desc> + </func> + + <func> + <name name="put" arity="3"/> + <fsummary>Set counter to value</fsummary> + <desc> + <p>Write <c><anno>Value</anno></c> to counter at index + <c><anno>Ix</anno></c>.</p> + <note> + <p>Despite its name, the <c>write_concurrency</c> optimization does not + improve <c>put</c>. A call to <c>put</c> is a relative heavy + operation compared to the very lightweight and scalable <seealso + marker="#add/3"><c>add</c></seealso> and <seealso marker="#sub/3"> + <c>sub</c></seealso>. The cost for a <c>put</c> with + <c>write_concurrency</c> is lika a <seealso marker="#get/2"><c>get</c> + </seealso> plus a <c>put</c> without <c>write_concurrency</c>.</p> + </note> + </desc> + </func> + + <func> + <name name="info" arity="1"/> + <fsummary>Get information about counter array.</fsummary> + <desc> + <p>Return information about a counter array in a map. The map + has the following keys (at least):</p> + <taglist> + <tag><c>size</c></tag> + <item><p>The number of counters in the array.</p></item> + <tag><c>memory</c></tag> + <item><p>Approximate memory consumption for the array in + bytes.</p></item> + </taglist> + </desc> + </func> + + </funcs> +</erlref> diff --git a/erts/doc/src/erl_nif.xml b/erts/doc/src/erl_nif.xml index 190ec12d0e..095fc79bdf 100644 --- a/erts/doc/src/erl_nif.xml +++ b/erts/doc/src/erl_nif.xml @@ -293,7 +293,7 @@ return term;</code> arguments. When you write to a shared state either through static variables or <seealso marker="#enif_priv_data"> <c>enif_priv_data</c></seealso>, you need to supply your own explicit - synchronization. This includes terms in process-independent + synchronization. This includes terms in process independent environments that are shared between threads. Resource objects also require synchronization if you treat them as mutable.</p> <p>The library initialization callbacks <c>load</c> and @@ -596,7 +596,7 @@ int writeiovec(ErlNifEnv *env, ERL_NIF_TERM term, ERL_NIF_TERM *tail, <c>--enable-static-nifs</c>, you must define <c>STATIC_ERLANG_NIF</c> before the <c>ERL_NIF_INIT</c> declaration.</p> </item> - <tag><marker id="load"/><c>int (*load)(ErlNifEnv* env, void** priv_data, + <tag><marker id="load"/><c>int (*load)(ErlNifEnv* caller_env, void** priv_data, ERL_NIF_TERM load_info)</c></tag> <item> <p><c>load</c> is called when the NIF library is loaded @@ -612,7 +612,7 @@ int writeiovec(ErlNifEnv *env, ERL_NIF_TERM term, ERL_NIF_TERM *tail, anything other than <c>0</c>. <c>load</c> can be <c>NULL</c> if initialization is not needed.</p> </item> - <tag><marker id="upgrade"/><c>int (*upgrade)(ErlNifEnv* env, void** + <tag><marker id="upgrade"/><c>int (*upgrade)(ErlNifEnv* caller_env, void** priv_data, void** old_priv_data, ERL_NIF_TERM load_info)</c></tag> <item> <p><c>upgrade</c> is called when the NIF library is loaded @@ -626,7 +626,7 @@ int writeiovec(ErlNifEnv *env, ERL_NIF_TERM term, ERL_NIF_TERM *tail, <p>The library fails to load if <c>upgrade</c> returns anything other than <c>0</c> or if <c>upgrade</c> is <c>NULL</c>.</p> </item> - <tag><marker id="unload"/><c>void (*unload)(ErlNifEnv* env, void* + <tag><marker id="unload"/><c>void (*unload)(ErlNifEnv* caller_env, void* priv_data)</c></tag> <item> <p><c>unload</c> is called when the module code that @@ -654,27 +654,41 @@ int writeiovec(ErlNifEnv *env, ERL_NIF_TERM term, ERL_NIF_TERM *tail, <p><c>ErlNifEnv</c> represents an environment that can host Erlang terms. All terms in an environment are valid as long as the environment is valid. <c>ErlNifEnv</c> is an opaque type; pointers to - it can only be passed on to API functions. Two types of environments + it can only be passed on to API functions. Three types of environments exist:</p> <taglist> - <tag>Process-bound environment</tag> + <tag>Process bound environment</tag> <item> <p>Passed as the first argument to all NIFs. All function arguments passed to a NIF belong to that environment. The return value from a NIF must also be a term belonging to the same environment.</p> - <p>A process-bound environment contains transient information + <p>A process bound environment contains transient information about the calling Erlang process. The environment is only valid in the thread where it was supplied as argument until the NIF returns. It is thus useless and dangerous to store pointers to - process-bound environments between NIF calls.</p> + process bound environments between NIF calls.</p> </item> - <tag>Process-independent environment</tag> + <tag>Callback environment</tag> + <item> + <p>Passed as the first argument to all the non-NIF callback functions + (<seealso marker="#load"><c>load</c></seealso>, + <seealso marker="#upgrade"><c>upgrade</c></seealso>, + <seealso marker="#unload"><c>unload</c></seealso>, + <seealso marker="#ErlNifResourceDtor"><c>dtor</c></seealso>, + <seealso marker="#ErlNifResourceDown"><c>down</c></seealso> and + <seealso marker="#ErlNifResourceStop"><c>stop</c></seealso>). + Works like a process bound environment but with a temporary + pseudo process that "terminates" when the callback has + returned. Terms may be created in this environment but they will + only be accessible during the callback.</p> + </item> + <tag>Process independent environment</tag> <item> <p>Created by calling <seealso marker="#enif_alloc_env"> <c>enif_alloc_env</c></seealso>. This environment can be used to store terms between NIF calls and to send terms with <seealso marker="#enif_send"><c>enif_send</c></seealso>. A - process-independent environment with all its terms is valid until + process independent environment with all its terms is valid until you explicitly invalidate it with <seealso marker="#enif_free_env"><c>enif_free_env</c></seealso> or <c>enif_send</c>.</p> @@ -799,7 +813,7 @@ typedef struct { <tag><marker id="ErlNifResourceDtor"/><c>ErlNifResourceDtor</c></tag> <item> <code type="none"> -typedef void ErlNifResourceDtor(ErlNifEnv* env, void* obj);</code> +typedef void ErlNifResourceDtor(ErlNifEnv* caller_env, void* obj);</code> <p>The function prototype of a resource destructor function.</p> <p>The <c>obj</c> argument is a pointer to the resource. The only allowed use for the resource in the destructor is to access its @@ -809,7 +823,7 @@ typedef void ErlNifResourceDtor(ErlNifEnv* env, void* obj);</code> <tag><marker id="ErlNifResourceDown"/><c>ErlNifResourceDown</c></tag> <item> <code type="none"> -typedef void ErlNifResourceDown(ErlNifEnv* env, void* obj, ErlNifPid* pid, ErlNifMonitor* mon);</code> +typedef void ErlNifResourceDown(ErlNifEnv* caller_env, void* obj, ErlNifPid* pid, ErlNifMonitor* mon);</code> <p>The function prototype of a resource down function, called on the behalf of <seealso marker="#enif_monitor_process"> enif_monitor_process</seealso>. <c>obj</c> is the resource, <c>pid</c> @@ -820,7 +834,7 @@ typedef void ErlNifResourceDown(ErlNifEnv* env, void* obj, ErlNifPid* pid, ErlNi <tag><marker id="ErlNifResourceStop"/><c>ErlNifResourceStop</c></tag> <item> <code type="none"> -typedef void ErlNifResourceStop(ErlNifEnv* env, void* obj, ErlNifEvent event, int is_direct_call);</code> +typedef void ErlNifResourceStop(ErlNifEnv* caller_env, void* obj, ErlNifEvent event, int is_direct_call);</code> <p>The function prototype of a resource stop function, called on the behalf of <seealso marker="#enif_select"> enif_select</seealso>. <c>obj</c> is the resource, <c>event</c> is OS event, @@ -987,7 +1001,7 @@ typedef struct { <name><ret>ErlNifEnv *</ret><nametext>enif_alloc_env()</nametext></name> <fsummary>Create a new environment.</fsummary> <desc> - <p>Allocates a new process-independent environment. The environment can + <p>Allocates a new process independent environment. The environment can be used to hold terms that are not bound to any process. Such terms can later be copied to a process environment with <seealso marker="#enif_make_copy"><c>enif_make_copy</c></seealso> or @@ -1211,14 +1225,17 @@ typedef struct { </func> <func> - <name><ret>int</ret><nametext>enif_demonitor_process(ErlNifEnv* env, void* obj, + <name><ret>int</ret><nametext>enif_demonitor_process(ErlNifEnv* caller_env, void* obj, const ErlNifMonitor* mon)</nametext></name> <fsummary>Cancel a process monitor.</fsummary> <desc> <marker id="enif_demonitor_process"></marker> <p>Cancels a monitor created earlier with <seealso marker="#enif_monitor_process"> <c>enif_monitor_process</c></seealso>. Argument <c>obj</c> is a pointer - to the resource holding the monitor and <c>*mon</c> identifies the monitor.</p> + to the resource holding the monitor and <c>*mon</c> identifies the + monitor.</p> + <p>Argument <c>caller_env</c> is the environment of the calling process + or callback. Must only be NULL if calling from a custom thread.</p> <p>Returns <c>0</c> if the monitor was successfully identified and removed. Returns a non-zero value if the monitor could not be identified, which means it was either</p> @@ -2572,7 +2589,7 @@ enif_map_iterator_destroy(env, &iter);</code> </func> <func> - <name><ret>int</ret><nametext>enif_monitor_process(ErlNifEnv* env, void* obj, + <name><ret>int</ret><nametext>enif_monitor_process(ErlNifEnv* caller_env, void* obj, const ErlNifPid* target_pid, ErlNifMonitor* mon)</nametext></name> <fsummary>Monitor a process from a resource.</fsummary> <desc> @@ -2593,6 +2610,8 @@ enif_map_iterator_destroy(env, &iter);</code> <seealso marker="#enif_compare_monitors"><c>enif_compare_monitors</c></seealso>. A monitor is automatically removed when it triggers or when the resource is deallocated.</p> + <p>Argument <c>caller_env</c> is the environment of the calling process + or callback. Must only be NULL if calling from a custom thread.</p> <p>Returns <c>0</c> on success, < 0 if no <c>down</c> callback is provided, and > 0 if the process is no longer alive.</p> <p>This function is only thread-safe when the emulator with SMP support @@ -2768,7 +2787,7 @@ enif_map_iterator_destroy(env, &iter);</code> <item>The port ID of the receiving port. The port ID is to refer to a port on the local node.</item> <tag><c>msg_env</c></tag> - <item>The environment of the message term. Can be a process-independent + <item>The environment of the message term. Can be a process independent environment allocated with <seealso marker="#enif_alloc_env"> <c>enif_alloc_env</c></seealso> or <c>NULL</c>.</item> <tag><c>msg</c></tag> @@ -3124,26 +3143,26 @@ if (retval & ERL_NIF_SELECT_STOP_CALLED) { <p>Initializes the <seealso marker="#ErlNifPid"><c>ErlNifPid</c></seealso> variable at <c>*pid</c> to represent the calling process.</p> <p>Returns <c>pid</c> if successful, or NULL if <c>caller_env</c> is not - a <seealso marker="#ErlNifEnv">process-bound environment</seealso>.</p> + a <seealso marker="#ErlNifEnv">process bound environment</seealso>.</p> </desc> </func> <func> - <name><ret>int</ret><nametext>enif_send(ErlNifEnv* env, ErlNifPid* to_pid, + <name><ret>int</ret><nametext>enif_send(ErlNifEnv* caller_env, ErlNifPid* to_pid, ErlNifEnv* msg_env, ERL_NIF_TERM msg)</nametext></name> <fsummary>Send a message to a process.</fsummary> <desc> <p>Sends a message to a process.</p> <taglist> - <tag><c>env</c></tag> - <item>The environment of the calling process. Must be <c>NULL</c> - only if calling from a created thread.</item> + <tag><c>caller_env</c></tag> + <item>The environment of the calling process or callback. Must be <c>NULL</c> + only if calling from a custom thread not spawned by ERTS.</item> <tag><c>*to_pid</c></tag> <item>The pid of the receiving process. The pid is to refer to a process on the local node.</item> <tag><c>msg_env</c></tag> <item>The environment of the message term. Must be a - process-independent environment allocated with + process independent environment allocated with <seealso marker="#enif_alloc_env"><c>enif_alloc_env</c></seealso> or NULL.</item> <tag><c>msg</c></tag> diff --git a/erts/doc/src/notes.xml b/erts/doc/src/notes.xml index 533e1d5eb7..2a823d9fe7 100644 --- a/erts/doc/src/notes.xml +++ b/erts/doc/src/notes.xml @@ -31,6 +31,42 @@ </header> <p>This document describes the changes made to the ERTS application.</p> +<section><title>Erts 10.1.3</title> + + <section><title>Improvements and New Features</title> + <list> + <item> + <p>Added an optional <c>./configure</c> flag to compile + the emulator with spectre mitigation: + <c>--with-spectre-mitigation</c></p> + <p>Note that this requires a recent version of GCC with + support for spectre mitigation and the + <c>--mindirect-branch=thunk</c> flag, such as + <c>8.1</c>.</p> + <p> + Own Id: OTP-15430 Aux Id: ERIERL-237 </p> + </item> + </list> + </section> + +</section> + +<section><title>Erts 10.1.2</title> + + <section><title>Fixed Bugs and Malfunctions</title> + <list> + <item> + <p>Fixed a rare bug where files could be closed on a + normal instead of an IO scheduler, resulting in system + instability if the operation blocked.</p> + <p> + Own Id: OTP-15421</p> + </item> + </list> + </section> + +</section> + <section><title>Erts 10.1.1</title> <section><title>Fixed Bugs and Malfunctions</title> @@ -1355,6 +1391,21 @@ </section> +<section><title>Erts 9.3.3.6</title> + + <section><title>Improvements and New Features</title> + <list> + <item> + <p>List subtraction (The <c>--</c> operator) will now + yield properly on large inputs.</p> + <p> + Own Id: OTP-15371</p> + </item> + </list> + </section> + +</section> + <section><title>Erts 9.3.3.5</title> <section><title>Fixed Bugs and Malfunctions</title> @@ -4799,6 +4850,21 @@ </section> +<section><title>Erts 7.3.1.6</title> + + <section><title>Improvements and New Features</title> + <list> + <item> + <p>List subtraction (The <c>--</c> operator) will now + yield properly on large inputs.</p> + <p> + Own Id: OTP-15371</p> + </item> + </list> + </section> + +</section> + <section><title>Erts 7.3.1.5</title> <section><title>Fixed Bugs and Malfunctions</title> diff --git a/erts/doc/src/persistent_term.xml b/erts/doc/src/persistent_term.xml new file mode 100644 index 0000000000..29a6c67051 --- /dev/null +++ b/erts/doc/src/persistent_term.xml @@ -0,0 +1,290 @@ +<?xml version="1.0" encoding="utf-8" ?> +<!DOCTYPE erlref SYSTEM "erlref.dtd"> + +<erlref> + <header> + <copyright> + <year>2018</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>persistent_term</title> + <prepared></prepared> + <docno></docno> + <date></date> + <rev></rev> + <file>persistent_term.xml</file> + </header> + <module>persistent_term</module> + <modulesummary>Persistent terms.</modulesummary> + <description> + <p>This module is similar to <seealso + marker="stdlib:ets"><c>ets</c></seealso> in that it provides a + storage for Erlang terms that can be accessed in constant time, + but with the difference that <c>persistent_term</c> has been + highly optimized for reading terms at the expense of writing and + updating terms. When a persistent term is updated or deleted, a + global garbage collection pass is run to scan all processes for + the deleted term, and to copy it into each process that still uses + it. Therefore, <c>persistent_term</c> is suitable for storing + Erlang terms that are frequently accessed but never or + infrequently updated.</p> + + <warning><p>Persistent terms is an advanced feature and is not a + general replacement for ETS tables. Before using persistent terms, + make sure to fully understand the consequence to system + performance when updating or deleting persistent terms.</p></warning> + + <p>Term lookup (using <seealso + marker="#get/1"><c>get/1</c></seealso>), is done in constant time + and without taking any locks, and the term is <strong>not</strong> + copied to the heap (as is the case with terms stored in ETS + tables).</p> + + <p>Storing or updating a term (using <seealso + marker="#put/2"><c>put/2</c></seealso>) is proportional to the + number of already created persistent terms because the hash table + holding the keys will be copied. In addition, the term itself will + be copied.</p> + + <p>When a (complex) term is deleted (using <seealso + marker="#erase/1"><c>erase/1</c></seealso>) or replaced by another + (using <seealso marker="#put/2"><c>put/2</c></seealso>), a global + garbage collection is initiated. It works like this:</p> + + <list> + <item><p>All processes in the system will be scheduled to run a + scan of their heaps for the term that has been deleted. While + such scan is relatively light-weight, if there are many + processes, the system can become less responsive until all + process have scanned their heaps.</p></item> + + <item><p>If the deleted term (or any part of it) is still used + by a process, that process will do a major (fullsweep) garbage + collection and copy the term into the process. However, at most + two processes at a time will be scheduled to do that kind of + garbage collection.</p></item> + </list> + + <p>Deletion of atoms and other terms that fit in one machine word + is specially optimized to avoid doing a global GC. It is still not + recommended to update persistent terms with such values too + frequently because the hash table holding the keys is copied every + time a persistent term is updated.</p> + + <p>Some examples are suitable uses for persistent terms are:</p> + + <list> + <item><p>Storing of configuration data that must be easily + accessible by all processes.</p></item> + + <item><p>Storing of references for NIF resources.</p></item> + + <item><p>Storing of references for efficient counters.</p></item> + + <item><p>Storing an atom to indicate a logging level or whether debugging + is turned on.</p></item> + </list> + + </description> + + <section> + <title>Storing Huge Persistent Terms</title> + <p>The current implementation of persistent terms uses the literal + <seealso marker="erts_alloc">allocator</seealso> also used for + literals (constant terms) in BEAM code. By default, 1 GB of + virtual address space is reserved for literals in BEAM code and + persistent terms. The amount of virtual address space reserved for + literals can be changed by using the <seealso + marker="erts_alloc#MIscs"><c>+MIscs option</c></seealso> when + starting the emulator.</p> + + <p>Here is an example how the reserved virtual address space for literals + can be raised to 2 GB (2048 MB):</p> + + <pre> + erl +MIscs 2048</pre> + </section> + + <section> + <title>Warning For Many Persistent Terms</title> + <p>The runtime system will send a warning report to the + error logger if more than 20000 persistent terms have been + created. It will look like this:</p> + +<pre> +More than 20000 persistent terms have been created. +It is recommended to avoid creating an excessive number of +persistent terms, as creation and deletion of persistent terms +will be slower as the number of persistent terms increases.</pre> + </section> + + <section> + <title>Best Practices for Using Persistent Terms</title> + + <p>It is recommended to use keys like <c>?MODULE</c> or + <c>{?MODULE,SubKey}</c> to avoid name collisions.</p> + + <p>Prefer creating a few large persistent terms to creating many + small persistent terms. The execution time for storing a + persistent term is proportional to the number of already existing + terms.</p> + + <p>Updating a persistent term with the same value as it already + has is specially optimized to do nothing quickly; thus, there is + no need compare the old and new values and avoid calling + <seealso marker="#put/2"><c>put/2</c></seealso> if the values + are equal.</p> + + <p>When atoms or other terms that fit in one machine word are + deleted, no global GC is needed. Therefore, persistent terms that + have atoms as their values can be updated more frequently, but + note that updating such persistent terms is still much more + expensive than reading them.</p> + + <p>Updating or deleting a persistent term will trigger a global GC + if the term does not fit in one machine word. Processes will be + scheduled as usual, but all processes will be made runnable at + once, which will make the system less responsive until all process + have run and scanned their heaps for the deleted terms. One way to + minimize the effects on responsiveness could be to minimize the + number of processes on the node before updating or deleting a + persistent term. It would also be wise to avoid updating terms + when the system is at peak load.</p> + + <p>Avoid storing a retrieved persistent term in a process if that + persistent term could be deleted or updated in the future. If a + process holds a reference to a persistent term when the term is + deleted, the process will be garbage collected and the term copied + to process.</p> + + <p>Avoid updating or deleting more than one persistent term at a + time. Each deleted term will trigger its own global GC. That + means that deleting N terms will make the system less responsive N + times longer than deleting a single persistent term. Therefore, + terms that are to be updated at the same time should be collected + into a larger term, for example, a map or a tuple.</p> + </section> + + <section> + <title>Example</title> + + <p>The following example shows how lock contention for ETS tables + can be minimized by having one ETS table for each scheduler. The + table identifiers for the ETS tables are stored as a single + persistent term:</p> + +<pre> + %% There is one ETS table for each scheduler. + Sid = erlang:system_info(scheduler_id), + Tid = element(Sid, persistent_term:get(?MODULE)), + ets:update_counter(Tid, Key, 1).</pre> + + </section> + + <datatypes> + <datatype> + <name name="key"/> + <desc> + <p>Any Erlang term.</p> + </desc> + </datatype> + <datatype> + <name name="value"/> + <desc> + <p>Any Erlang term.</p> + </desc> + </datatype> + </datatypes> + + <funcs> + <func> + <name name="erase" arity="1"/> + <fsummary>Erase the name for a persistent term.</fsummary> + <desc> + <p>Erase the name for the persistent term with key + <c><anno>Key</anno></c>. The return value will be <c>true</c> + if there was a persistent term with the key + <c><anno>Key</anno></c>, and <c>false</c> if there was no + persistent term associated with the key.</p> + <p>If there existed a previous persistent term associated with + key <c><anno>Key</anno></c>, a global GC has been initiated + when <c>erase/1</c> returns. See <seealso + marker="#description">Description</seealso>.</p> + </desc> + </func> + + <func> + <name name="get" arity="0"/> + <fsummary>Get all persistent terms.</fsummary> + <desc> + <p>Retrieve the keys and values for all persistent terms. + The keys will be copied to the heap for the process calling + <c>get/0</c>, but the values will not.</p> + </desc> + </func> + + <func> + <name name="get" arity="1"/> + <fsummary>Get the value for a persistent term.</fsummary> + <desc> + <p>Retrieve the value for the persistent term associated with + the key <c><anno>Key</anno></c>. The lookup will be made in + constant time and the value will not be copied to the heap + of the calling process.</p> + <p>This function fails with a <c>badarg</c> exception if no + term has been stored with the key + <c><anno>Key</anno></c>.</p> + <p>If the calling process holds on to the value of the + persistent term and the persistent term is deleted in the future, + the term will be copied to the process.</p> + </desc> + </func> + + <func> + <name name="info" arity="0"/> + <fsummary>Get information about persistent terms.</fsummary> + <desc> + <p>Return information about persistent terms in a map. The map + has the following keys:</p> + <taglist> + <tag><c>count</c></tag> + <item><p>The number of persistent terms.</p></item> + <tag><c>memory</c></tag> + <item><p>The total amount of memory (measured in bytes) + used by all persistent terms.</p></item> + </taglist> + </desc> + </func> + + <func> + <name name="put" arity="2"/> + <fsummary>Store a term.</fsummary> + <desc> + <p>Store the value <c><anno>Value</anno></c> as a persistent term and + associate it with the key <c><anno>Key</anno></c>.</p> + <p>If the value <c><anno>Value</anno></c> is equal to the value + previously stored for the key, <c>put/2</c> will do nothing and return + quickly.</p> + <p>If there existed a previous persistent term associated with + key <c><anno>Key</anno></c>, a global GC has been initiated + when <c>put/2</c> returns. See <seealso + marker="#description">Description</seealso>.</p> + </desc> + </func> + </funcs> +</erlref> diff --git a/erts/doc/src/ref_man.xml b/erts/doc/src/ref_man.xml index 0617463a7b..a78aaa449e 100644 --- a/erts/doc/src/ref_man.xml +++ b/erts/doc/src/ref_man.xml @@ -34,6 +34,7 @@ <xi:include href="erl_prim_loader.xml"/> <xi:include href="erlang.xml"/> <xi:include href="init.xml"/> + <xi:include href="persistent_term.xml"/> <xi:include href="zlib.xml"/> <xi:include href="epmd.xml"/> <xi:include href="erl.xml"/> @@ -49,5 +50,7 @@ <xi:include href="erts_alloc.xml"/> <xi:include href="erl_nif.xml"/> <xi:include href="erl_tracer.xml"/> + <xi:include href="atomics.xml"/> + <xi:include href="counters.xml"/> </application> diff --git a/erts/doc/src/specs.xml b/erts/doc/src/specs.xml index ed6be650e5..0b943e6295 100644 --- a/erts/doc/src/specs.xml +++ b/erts/doc/src/specs.xml @@ -4,5 +4,8 @@ <xi:include href="../specs/specs_erlang.xml"/> <xi:include href="../specs/specs_erl_tracer.xml"/> <xi:include href="../specs/specs_init.xml"/> + <xi:include href="../specs/specs_persistent_term.xml"/> <xi:include href="../specs/specs_zlib.xml"/> + <xi:include href="../specs/specs_atomics.xml"/> + <xi:include href="../specs/specs_counters.xml"/> </specs> |