blob: 9996063d41764b5ae6045084e2f4b53651d54d9e [file] [log] [blame]
vsomeip
=======
// This enables a nice TOC as a sidebar
:toc2:
// Show all headings in TOC
:toclevels: 4
// Show icons if e.g. TIP: or IMPORTANT is used
:icons:
// Set the directory where the default icons can be found
:iconsdir: {asciidoc-confdir}/{iconsdir}
// number all headings
:numbered:
// this embeds images (e.g. the icons for TIP: $TEXT) into the html file
:data-uri:
Copyright
+++++++++
Copyright (C) 2015-2017, Bayerische Motoren Werke Aktiengesellschaft (BMW AG)
License
+++++++
This Source Code Form is subject to the terms of the Mozilla Public
License, v. 2.0. If a copy of the MPL was not distributed with this
file, You can obtain one at http://mozilla.org/MPL/2.0/.
Version
+++++++
// set the version to the one we get from cmake
// or pass it via -a version=$VSOMEIP_VERSION to asciidoc
This documentation was generated for version {version} of vsomeip.
vsomeip Overview
----------------
The vsomeip stack implements the http://some-ip.com/[Scalable service-Oriented
MiddlewarE over IP (SOME/IP)] protocol. The stack consists out of:
* a shared library for SOME/IP (`libvsomeip.so`)
* a second shared library for SOME/IP's service discovery (`libvsomeip-sd.so`)
which is loaded during runtime if the service discovery is enabled.
Build Instructions
------------------
Dependencies
~~~~~~~~~~~~
* A C++11 enabled compiler like gcc >= 4.8 is needed.
* vsomeip uses cmake as buildsystem.
* vsomeip uses Boost >= 1.55:
** Ubuntu 14.04:
*** `sudo apt-get install libboost-system1.55-dev libboost-thread1.55-dev
libboost-log1.55-dev`
** Ubuntu 12.04: a PPA is necessary to use version 1.54 of Boost:
*** URL: https://launchpad.net/~boost-latest/+archive/ubuntu/ppa
*** `sudo add-apt-repository ppa:boost-latest/ppa`
*** `sudo apt-get install libboost-system1.55-dev libboost-thread1.55-dev
libboost-log1.55-dev`
* For the tests Google's test framework
https://code.google.com/p/googletest/[gtest] in version 1.7.0 is needed
** URL: https://googletest.googlecode.com/files/gtest-1.7.0.zip[direct link,
version 1.7.0]
* To build the documentation asciidoc, source-highlight, doxygen and graphviz is needed:
** `sudo apt-get install asciidoc source-highlight doxygen graphviz`
Compilation
~~~~~~~~~~~
anchor:Compilation[]
For compilation call:
[source, bash]
----
mkdir build
cd build
cmake ..
make
----
To specify a installation directory (like `--prefix=` if you're used to
autotools) call cmake like:
[source, bash]
----
cmake -DCMAKE_INSTALL_PREFIX:PATH=$YOUR_PATH ..
make
make install
----
Compilation with predefined unicast and/or diagnosis address
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
To predefine the unicast address, call cmake like:
[source,bash]
----
cmake -DUNICAST_ADDRESS=<YOUR IP ADDRESS> ..
----
To predefine the diagnosis address, call cmake like:
[source,bash]
----
cmake -DDIAGNOSIS_ADDRESS=<YOUR DIAGNOSIS ADDRESS> ..
----
The diagnosis address is a single byte value.
Compilation with signal handling
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
To compile vsomeip with signal handling (SIGINT/SIGTERM) enabled,
call cmake like:
[source,bash]
----
cmake -DENABLE_SIGNAL_HANDLING=1 ..
----
In the default setting, the application has to take care of shutting
down vsomeip in case these signals are received.
Compilation with user defined "READY" message
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
To compile vsomeip with a user defined message signal the IP routing
to be ready to send/receive messages, call cmake like:
[source,bash]
----
cmake -DROUTING_READY_MESSAGE=<YOUR MESSAGE> ..
----
Compilation of examples
^^^^^^^^^^^^^^^^^^^^^^^
For compilation of the examples call:
[source, bash]
----
mkdir build
cd build
cmake ..
make examples
----
Compilation of tests
^^^^^^^^^^^^^^^^^^^^
To compile the tests, first unzip gtest to location of your desire.
Some of the tests require a second node on the same network. There are two cmake
variables which are used to automatically adapt the json files to the used
network setup:
* `TEST_IP_MASTER`: The IP address of the interface which will act as test
master.
* `TEST_IP_SLAVE`: The IP address of the interface of the second node which will
act as test slave.
If one of this variables isn't specified, only the tests using local
communication exclusively will be runnable.
Additionally the unit tests require enabled signal handling which can be enabled
via the `ENABLE_SIGNAL_HANDLING` cmake variable.
Example, compilation of tests:
[source, bash]
----
mkdir build
cd build
export GTEST_ROOT=$PATH_TO_GTEST/gtest-1.7.0/
cmake -DENABLE_SIGNAL_HANDLING=1 -DTEST_IP_MASTER=10.0.3.1 -DTEST_IP_SLAVE=10.0.3.125 ..
make check
----
Additional make targets for the tests:
* Call `make build_tests` to only compile the tests
* Call `ctest` in the build directory to execute the tests without a verbose
output
* To run single tests call `ctest --verbose --tests-regex $TESTNAME` short
form: `ctest -V -R $TESTNAME`
* To list all available tests run `ctest -N`.
* For further information about the tests please have a look at the
`readme.txt` in the `test` subdirectory.
For development purposes two cmake variables exist which control if the
json files and test scripts are copied (default) or symlinked into the build
directory. These settings are ignored on Windows.
* `TEST_SYMLINK_CONFIG_FILES`: Controls if the json and scripts needed
to run the tests are copied or symlinked into the build directory. (Default:
OFF, ignored on Windows)
* `TEST_SYMLINK_CONFIG_FILES_RELATIVE`: Controls if the json and scripts needed
to run the tests are symlinked relatively into the build directory.
(Default: OFF, ignored on Windows)
Example cmake call:
[source, bash]
----
cmake -DTEST_SYMLINK_CONFIG_FILES=ON -DTEST_SYMLINK_CONFIG_FILES_RELATIVE=ON ..
----
For compilation of only a subset of tests (for a quick
functionality check) the cmake variable `TESTS_BAT` has
to be set:
Example cmake call:
[source, bash]
----
cmake -DTESTS_BAT=ON ..
----
Compilation of vsomeip_ctrl
^^^^^^^^^^^^^^^^^^^^^^^^^^^
For compilation of the <<vsomeip_ctrl>> utility call:
[source, bash]
----
mkdir build
cd build
cmake ..
make vsomeip_ctrl
----
Generating the documentation
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
To generate the documentation call cmake as described in <<Compilation>> and
then call `make doc`.
This will generate:
* The README file in html: `$BUILDDIR/documentation/README.html`
* A doxygen documentation in `$BUILDDIR/documentation/html/index.html`
Starting vsomeip Applications / Used environment variables
----------------------------------------------------------
On startup the following environment variables are read out:
* `VSOMEIP_APPLICATION_NAME`: This environment variable is used to specify the
name of the application. This name is later used to map a client id to the
application in the configuration file. It is independent from the
application's binary name.
* `VSOMEIP_CONFIGURATION`: vsomeip uses the default configuration file `/etc/vsomeip.json`
and/or the default configuration folder `/etc/vsomeip`. This can be overridden by a
local configuration file `./vsomeip.json` and/or a local configuration folder `./vsomeip`.
If `VSOMEIP_CONFIGURATION` is set to a valid file or directory path, this is used instead
of the standard configuration (thus neither default nor local file/folder will be parsed).
* `VSOMEIP_MANDATORY_CONFIGURATION_FILES`: vsomeip allows to specify mandatory configuration
files to speed-up application startup. While mandatory configuration files are read by all
applications, all other configuration files are only read by the application that is
responsible for connections to external devices. If this configuration variable is not set,
the default mandatory files vsomeip_std.json, vsomeip_app.json and vsomeip_plc.json are used.
* `VSOMEIP_CLIENTSIDELOGGING`: Set this variable to an empty string to enable logging of
any received messages to DLT in all applications acting as routing manager proxies. For
example add the following line to the application's systemd service file:
`Environment=VSOMEIP_CLIENTSIDELOGGING=""`
To enable service-specific logs, provide a space- or colon-separated list of ServiceIDs (using
4-digit hexadecimal notation, optionally followed by dot-separted InstanceID). For example:
`Environment=VSOMEIP_CLIENTSIDELOGGING="b003.0001 f013.000a 1001 1002"`
`Environment=VSOMEIP_CLIENTSIDELOGGING="b003.0001:f013.000a:1001:1002"`
NOTE: If the file/folder that is configured by `VSOMEIP_CONFIGURATION` does _not_ exist,
the default configuration locations will be used.
NOTE: vsomeip will parse and use the configuration from all files in a configuration folder
but will _not_ consider directories within the configuration folder.
In the following example the application `my_vsomeip_application` is started.
The settings are read from the file `my_settings.json` in the current working
directory. The client id for the application can be found under the name
`my_vsomeip_client` in the configuration file.
[source, bash]
----
#!/bin/bash
export VSOMEIP_APPLICATION_NAME=my_vsomeip_client
export VSOMEIP_CONFIGURATION=my_settings.json
./my_vsomeip_application
----
Configuration File Structure
----------------------------
The configuration files for vsomeip are http://www.json.org/[JSON]-Files and are
composed out of multiple key value pairs and arrays.
[quote, , json.org]
____
* An object is an unordered set of name/value pairs. An object begins with `{
(left brace)` and ends with `} (right brace)`. Each name is followed by `:
(colon)` and the name/value pairs are separated by `, (comma)`.
* An array is an ordered collection of values. An array begins with `[ (left
bracket)` and ends with `] (right bracket)`. Values are separated by `,
(comma)`.
* A value can be a _string_ in double quotes, or a _number_, or `true` or `false`
or `null`, or an _object_ or an _array_. These structures can be nested.
____
Configuration file element explanation:
* 'unicast'
+
The IP address of the host system.
+
* 'netmask'
+
The netmask to specify the subnet of the host system.
+
* 'diagnosis'
+
The diagnosis address (byte) that will be used to build client identifiers. The
diagnosis address is assigned to the most significant byte in all client
identifiers if not specified otherwise (for example through a predefined client
ID).
+
* 'diagnosis_mask'
+
The diagnosis mask (2 byte) is used to control the amount of bits used for the
diagnosis address in client identifiers. The default value is `0xFF00` meaning
the most significant byte of the client ID is reserved for the diagnosis
address. Setting the mask to `0xFE00` allows to only use the 7 most
significant bits of the client ID as a diagnosis address. This can be used to
increase the maximum amount of concurrent active clients on an ECU.
* 'network'
+
Network identifier used to support multiple routing managers on one host. This
setting changes the name of the shared memory segment in `/dev/shm` and the name
of the unix domain sockets in `/tmp/`. Defaults to `vsomeip` meaning the shared
memory will be named `/dev/shm/vsomeip` and the unix domain sockets will be
named `/tmp/vsomeip-$CLIENTID`
+
//Logging
* 'logging'
+
** 'level'
+
Specifies the log level (valid values: _trace_, _debug_, _info_, _warning_,
_error_, _fatal_).
+
** 'console'
+
Specifies whether logging via console is enabled (valid values: _true, false_).
+
** 'file'
+
*** 'enable'
+
Specifies whether a log file should be created (valid values: _true, false_).
+
*** 'path'
+
The absolute path of the log file.
+
** 'dlt'
+
Specifies whether Diagnostic Log and Trace (DLT) is enabled (valid values:
_true, false_).
+
** 'version'
+
Configures logging of the vsomeip version
+
*** 'enable'
+
Enable or disable cyclic logging of vsomeip version, defaults to true (valid
values: _true, false_)
+
*** 'interval'
+
Configures interval in seconds to log the vsomeip version. Default value is 10.
+
** 'memory_log_interval'
+
Configures interval in seconds in which the routing manager logs its used
memory. Setting a value greater than zero enables the logging.
+
** 'status_log_interval'
+
Configures interval in seconds in which the routing manager logs its internal
status. Setting a value greater than zero enables the logging.
+
//Tracing
* anchor:config-tracing[]'tracing' (optional)
+
** 'enable'
+
Specifies whether the tracing of the SOME/IP messages is enabled
(valid values: _true, false_). Default value is _false_.
If tracing is enabled, the messages will be forwarded to DLT by
the <<traceconnector, Trace Connector>>
+
** 'sd_enable'
+
Specifies whether the tracing of the SOME/IP service discovery messages is
enabled (valid values: _true, false_). Default value is _false_.
+
** 'channels (array)' (optional)
+
Contains the channels to DLT.
+
NOTE: You can set up multiple channels to DLT over that you can forward the
messages.
+
*** 'name'
+
The name of the channel.
+
*** 'id'
+
The id of the channel.
+
** 'filters (array)' (optional)
+
Contains the filters that are applied on the messages.
+
NOTE: You can apply filters respectively filter rules on the messages with
specific criterias and expressions. So only the filtered messages are forwarded
to DLT.
+
*** 'channel' (optional)
+
The id of the channel over that the filtered messages are forwarded to DLT. If
no channel is specified the default channel is used.
+
IMPORTANT: If multiple filters are used, the channel MUST always be
specified and accordingly defined! Each filter needs its own channel!
+
*** 'services (array)' (optional)
+
Contains the service ids. The messages that relates to the specified services will be filtered.
+
*** 'methods (array)' (optional)
+
Contains the method ids. The messages that relates to the specified methods will be filtered.
+
*** 'clients (array)' (optional)
+
Contains the client ids. The messages that relates to the specified clients will
be filtered.
+
*** 'type' (optional)
+
Specifies the filter type (valid values: "positive", "negative"). When a positive
filter is used and a message matches one of the filter rules, the message will be
traced/forwarded to DLT. With a negative filter messages can be excluded. So when a
message matches one of the filter rules, the message will not be traced/forwarded to
DLT. Default value is "positive".
+
//Applications
* 'applications (array)'
+
Contains the applications of the host system that use this config file.
+
** 'name'
+
The name of the application.
+
** 'id'
+
The id of the application. Usually its high byte is equal to the diagnosis address. In this
case the low byte must be different from zero. Thus, if the diagnosis address is 0x63, valid
values range from 0x6301 until 0x63FF. It is also possible to use id values with a high byte
different from the diagnosis address.
+
** 'max_dispatchers'
+
The maximum number of threads that shall be used to execute the application callbacks.
+
** 'max_dispatch_time'
+
The maximum time that an application callback may consume before the callback is
considered to be blocked (and an additional thread is used to execute pending
callbacks if max_dispatchers is configured greater than 0).
+
** 'threads' (optional)
+
The number of internal threads to process messages and events within an application.
Valid values are 1-255. Default is 2.
+
** 'request_debounce_time' (optional)
+
Specifies a debounce-time interval in ms in which request-service messages are sent to
the routing manager. If an application requests many services in short same time
the load of sent messages to the routing manager and furthermore the replies from the
routing manager (which contains the routing info for the requested service if available)
can be heavily reduced. The default value if not specified is 10ms.
+
* `services` (array)
+
Contains the services of the service provider.
** `service`
+
The id of the service.
** `instance`
+
The id of the service instance.
** `protocol` (optional)
+
The protocol that is used to implement the service instance. The default setting
is _someip_. If a different setting is provided, vsomeip does not open the specified
port (server side) or does not connect to the specified port (client side). Thus,
this option can be used to let the service discovery announce a service that is
externally implemented.
** `unicast` (optional)
+
The unicast that hosts the service instance.
+
NOTE: The unicast address is needed if external service instances shall be used,
but service discovery is disabled. In this case, the provided unicast address
is used to access the service instance.
** `reliable`
+
Specifies that the communication with the service is reliable respectively the
TCP protocol is used for communication.
*** `port`
+
The port of the TCP endpoint.
*** `enable-magic-cookies`
+
Specifies whether magic cookies are enabled (valid values: _true_, _false_).
** `unreliable`
+
Specifies that the communication with the service is unreliable respectively the
UDP protocol is used for communication (valid values: the _port_ of the UDP
endpoint).
** `events` (array)
+
Contains the events of the service.
*** `event`
+
The id of the event.
*** `is_field`
+
Specifies whether the event is of type field.
+
NOTE: A field is a combination of getter, setter and notification event. It
contains at least a getter, a setter, or a notifier. The notifier sends an event
message that transports the current value of a field on change.
*** `is_reliable`
+
Specifies whether the communication is reliable respectively whether the event
is sent with the TCP protocol (valid values: _true_,_false_).
+
If the value is _false_ the UDP protocol will be used.
** `eventgroups` (array)
+
Events can be grouped together into on event group. For a client it is thus
possible to subscribe for an event group and to receive the appropriate events
within the group.
*** `eventgroup`
+
The id of the event group.
*** `events` (array)
+
Contains the ids of the appropriate events.
*** `multicast`
+
Specifies the multicast that is used to publish the eventgroup.
**** `address`
+
The multicast address.
**** `port`
+
The multicast port.
*** `threshold`
+
Specifies when to use multicast and when to use unicast to send a notification event.
Must be set to a non-negative number. If it is set to zero, all events of the eventgroup
will be sent by unicast. Otherwise, the events will be sent by unicast as long as the
number of subscribers is lower than the threshold and by multicast if the number
of subscribers is greater or equal. This means, a threshold of 1 will lead to all events
being sent by multicast. The default value is _0_.
* `clients` (array)
+
The client-side ports that shall be used to connect to a specific service.
For each service, an array of ports to be used for reliable / unreliable
communication can be specified. vsomeip will take the first free port of
the list. If no free port can be found, the connection will fail. If
vsomeip is asked to connect to a service instance without specified port(s),
the port will be selected by the system. This implies that the user has
to ensure that the ports configured here do not overlap with the ports
automatically selected by the IP stack.
** `service`
** `instance`
+
Together they specify the service instance the port configuration shall be applied to.
** `reliable` (array)
+
The list of client ports to be used for reliable (TCP) communication to the given
service instance.
** `unreliable` (array)
+
The list of client ports to be used for unreliable (UDP) communication to the given
service instance.
+
Additionally there is the possibility to configure mappings between ranges of client
ports and ranges of remote service ports.
(If a client port is configured for a specific service / instance, the port range mapping is ignored)
** `reliable_remote_ports`
+
Specifies a range of reliable remote service ports
** `unreliable_remote_ports`
+
Specifies a range of unreliable remote service ports
** `reliable_client_ports`
+
Specifies the range of reliable client ports to be mapped to the reliable_remote_ports range
** `unreliable_client_ports`
+
Specifies the range of unreliable client ports to be mapped to the unreliable_remote_ports range
** `first`
+
Specifies the lower bound of a port range
** `last`
+
Specifies the upper bound of a port range
* `payload-sizes` (array)
+
Array to limit the maximum allowed payload sizes per IP and port. If not
specified otherwise the allowed payload sizes are unlimited. The settings in
this array only affect communication over TCP. To limit the local payload size
`max-payload-size-local` can be used.
** `unicast`
+
On client side: the IP of the remote service for which the payload size should
be limited.
+
On service side: the IP of the offered service for which the payload size for
receiving and sending should be limited.
** `ports` (array)
+
Array which holds pairs of port and payload size statements.
*** `port`
+
On client side: the port of the remote service for which the payload size should
be limited.
+
On service side: the port of the offered service for which the payload size for
receiving and sending should be limited.
*** `max-payload-size`
+
On client side: the payload size limit in bytes of a message sent to the
remote service hosted on beforehand specified IP and port.
+
On service side: the payload size limit in bytes of messages received and sent
by the service offered on previously specified IP and port.
+
If multiple services are hosted on the same port they all share the limit
specified.
* `max-payload-size-local`
+
The maximum allowed payload size for node internal communication in bytes. By
default the payload size for node internal communication is unlimited. It can be
limited via this setting.
* `max-payload-size-reliable`
+
The maximum allowed payload size for TCP communication in
bytes. By default the payload size for TCP communication is
unlimited. It can be limited via this setting.
* `endpoint-queue-limits` (array)
+
Array to limit the maximum allowed size in bytes of cached outgoing messages per
IP and port (message queue size per endpoint). If not specified otherwise the
allowed queue size is unlimited. The settings in this array only affect external
communication. To limit the local queue size `endpoint-queue-limit-local` can
be used.
** `unicast`
+
On client side: the IP of the remote service for which the queue size of sent
requests should be limited.
+
On service side: the IP of the offered service for which the queue size for
sent responses should be limited. This IP address is therefore
identical to the IP address specified via `unicast` setting on top level of the
json file.
** `ports` (array)
+
Array which holds pairs of port and queue size statements.
*** `port`
+
On client side: the port of the remote service for which the queue size of sent
requests should be limited.
+
On service side: the port of the offered service for which the queue size for
send responses should be limited.
*** `queue-size-limit`
+
On client side: the queue size limit in bytes of messages sent to the
remote service hosted on beforehand specified IP and port.
+
On service side: the queue size limit in bytes for responses sent by the service
offered on previously specified IP and port.
+
If multiple services are hosted on the same port they all share the limit
specified.
* `endpoint-queue-limit-external`
+
Setting to limit the maximum allowed size in bytes of cached outgoing messages
for external communication (message queue size per endpoint). By default the
queue size for external communication is unlimited. It can be limited via this
setting. Settings done in the `endpoint-queue-limits` array override this
setting.
* `endpoint-queue-limit-local`
+
Setting to limit the maximum allowed size in bytes of cached outgoing messages
for local communication (message queue size per endpoint). By default the queue
size for node internal communication is unlimited. It can be limited via this
setting.
* `buffer-shrink-threshold`
+
The number of processed messages which are half the size or smaller than the
allocated buffer used to process them before the memory for the buffer is
released and starts to grow dynamically again. This setting can be useful in
scenarios where only a small number of the overall messages are a lot bigger
then the rest and the memory allocated to process them should be released in a
timely manner. If not specified via this setting the buffer sizes by default
aren't reseted and are as big as the biggest processed message.
+
Example: `buffer-shrink-threshold` is set to 50. A message with 500 bytes has to
be processed and the buffers grow accordingly. After this message 50 consecutive
messages smaller than 250 bytes have to be processed before the buffer size is
reduced and starts to grow dynamically again.
* `internal_services` (optional array)
+
Specifies service/instance ranges for pure internal service-instances.
This information is used by vsomeip to avoid sending Find-Service messages
via the Service-Discovery when a client is requesting a not available service-
instance. Its can either be done on service/instance level or on service level
only which then includes all instance from 0x0000-0xffff.
** `first`
+
The lowest entry of the internal service range.
*** `service`
+
The lowest Service-ID in hex of the internal service range.
*** `instance` (optional)
+
The lowest Instance-ID in hex of a internal service-instance range.
If not specified the lowest Instance-ID is 0x0000.
** `last`
+
The highest entry of the internal service range.
*** `service`
+
The highest Service-ID in hex of a internal service range.
*** `instance` (optional)
+
The highest Instance-ID in hex of a internal service-instance range.
If not specified the highest Instance-ID is 0xFFFF.
* `debounce` (optional array)
+
Events/fields sent by external devices will be forwarded to the
applications only if a configurable function evaluates to true. The
function checks whether the event/field payload has changed and whether
a specified interval has been elapsed since the last forwarding.
** `service`
+
Service ID which hosts the events to be debounced.
** `instance`
+
Instance ID which hosts the events to be debounced.
** `events`
+
Array of events which shall be debounced based on the following
configuration options.
*** `event`
+
Event ID.
*** `on_change`
+
Specifies whether the event is only forwared on
paylaod change or not. (valid values: _true_, _false_).
*** `ignore`
+
Array of payload indexes with given bit mask (optional)
to be ignored in payload change evaluation.
Instead of specifying an index / bitmask pair, one can only define the paylaod index
which shall be ignored in the evaluation.
**** `index`
+
Payload index to be checked with given bitmask.
**** `mask`
+
1Byte bitmask applied to byte at given payload index.
Example mask: 0x0f ignores payload changes in low nibble of the byte at given index.
*** `interval`
+
Specifies if the event shall be debounced based on elapsed time interval.
(valid values: _time in ms_, _never_).
*** `on_change_resets_interval_` (optional)
Specifies if interval timer is reset when payload change was detected.
(valid values: _false_, _true_).
* `routing`
+
The name of the application that is responsible for the routing.
* `service-discovery`
+
Contains settings related to the Service Discovery of the host application.
** `enable`
+
Specifies whether the Service Discovery is enabled (valid values: _true_,
_false_). The default value is _true_.
** `multicast`
+
The multicast address which the messages of the Service Discovery will be sent
to. The default value is _224.0.0.1_.
** `port`
+
The port of the Service Discovery. The default setting is _30490_.
** `protocol`
+
The protocol that is used for sending the Service Discovery messages (valid
values: _tcp_, _udp_). The default setting is _udp_.
** `initial_delay_min`
+
Minimum delay before first offer message.
** `initial_delay_max`
+
Maximum delay before first offer message.
** `repetitions_base_delay`
+
Base delay sending offer messages within the repetition phase.
** `repetitions_max`
+
Maximum number of repetitions for provided services within the
repetition phase.
** `ttl`
+
Lifetime of entries for provided services as well as consumed services and eventgroups.
** `ttl_factor_offers` (optional array)
+
Array which holds correction factors for incoming remote offers. If a value
greater than one is specified for a service instance, the TTL field of the
corresponding service entry will be multiplied with the specified factor. +
Example: An offer of a service is received with a TTL of 3 sec and the TTL
factor is set to 5. The remote node stops offering the service w/o sending a
StopOffer message. The service will then expire (marked as unavailable) 15 seconds
after the last offer has been received.
*** `service`
+
The id of the service.
*** `instance`
+
The id of the service instance.
*** `ttl_factor`
+
TTL correction factor
** `ttl_factor_subscriptions` (optional array)
+
Array which holds correction factors for incoming remote subscriptions. If a
value greater than one is specified for a service instance, the TTL field of the
corresponding eventgroup entry will be multiplied with the specified factor. +
Example: A remote subscription to an offered service is received with a TTL of 3
sec and the TTL factor is set to 5. The remote node stops resubscribing to the
service w/o sending a StopSubscribeEventgroup message. The subscription will
then expire 15 seconds after the last resubscription has been received.
*** `service`
+
The id of the service.
*** `instance`
+
The id of the service instance.
*** `ttl_factor`
+
TTL correction factor
** `cyclic_offer_delay`
+
Cycle of the OfferService messages in the main phase.
** `request_response_delay`
+
Minimum delay of a unicast message to a multicast message for
provided services and eventgroups.
+
** `offer_debounce_time`
+
Time which the stack collects new service offers before they enter the
repetition phase. This can be used to reduce the number of
sent messages during startup. The default setting is _500ms_.
+
//Watchdog
* anchor:config-watchdog[]'watchdog' (optional)
+
The Watchdog sends periodically pings to all known local clients.
If a client isn't responding within a configurred time/amount of pongs
the watchdog deregisters this application/client.
If not configured the watchdog isn't activated.
+
** 'enable'
+
Specifies whether the watchdog is enabled or disabled.
(valid values: _true, false_), (default is _false_).
+
** 'timeout'
+
Specifies the timeout in ms the watchdog gets activated if a ping
isn't answered with a pong by a local client within that time.
(valid values: _2 - 2^32_), (default is _5000_ ms).
+
** 'allowed_missing_pongs'
+
Specifies the amount of allowed missing pongs.
(valid values: _1 - 2^32_), (default is _3_ pongs).
+
//CAPI-Selective Broadcasts support
* anchor:config-supports_selective_broadcasts[]'supports_selective_broadcasts' (optional)
+
This nodes allow to add a list of IP addresses on which CAPI-Selective-Broadcasts feature is supported.
If not specified the feature can't be used and the subscription behavior of the stack is same as with
normal events.
+
** 'address'
+
Specifies an IP-Address (in IPv4 or IPv6 notation) on which the "selective"-feature is supported.
Multiple addresses can be configuered.
Security
--------
vsomeip has a security implementation based on UNIX credentials.
If activated every local connection is authenticated during connect using the standard UNIX credential passing mechanism.
During authentification a client transfers its client identifier together with its credentials (UID / GID) to the server which is then matched against the configuration.
If received credentials don't match the policy the socket will be immediately closed by the server and an message is logged.
If accepted the client identifier is bound to the receiving socket and can therefore be used to do further security checks on incoming messages (vsomeip messages as well as internal commands).
In general clients can be configured to be allowed/denied to request (means communicate with) and offer different service instances.
Every incoming vsomeip message (request/response/notifcation) as well as offer service requests or local subscriptions are then checked against the policy.
If an incoming vsomeip message or another operation (e.g. offer/subscribe) violates the configured policies it is skipped and a message is logged.
Furthermore if an application receives informations about other clients/services in the system, it must be received from the authenticated routing manager.
This is to avoid malicious applications faking the routing manager and therefore being able to wrongly inform other clients about services running on the system.
Therefore, whenever the "security" tag is specified, the routing manager (e.g. vsomeipd) must be a configured application with a fixed client identifier.
See chapter "Configuration File Structure" on how to configure an application to use a specific client identifier.
Credential passing is only possible via Unix-Domain-Sockets and therefore only available for local communication.
However if security is activated method calls from remote clients to local services are checked as well which means remote clients needs to be explicitly allowed.
Such a policy looks same in case for local clients except the _credentials_ tag can be skipped.
Security configuration
~~~~~~~~~~~~~~~~~~~~~~
The available configuration switches for the security feature are:
// Security
* anchor:config-policy[]'security' (optional)
+
If specified the credential passing mechanism is activated. However no credential or security checks are done as long as _check_credentials_ isn't set to _true_, but the routing manager client ID must be configured if security tag is specified and shall not be set to 0x6300.
** 'check_credentials (optional)'
+
Specifies whether security checks are active or not. This includes credentials checks on connect as well as all policies checks configured in follow.
(valid values: _true, false_), (default is _false_).
** 'policies' (array)
+
Specifies the security policies. Each policy at least needs to specify _allow_ or _deny_.
*** 'client' (optional)
+
Specifies a client for which a security policy will be applied (valid value: A valid client identifier in hex: e.g. _0x1234_).
It is also possible to specify a client identifier range to easily apply a policy to a set of clients.
A usecase is e.g. to allow a set of remote clients communicate with local services offered remote.
+
No client specification equals to any client (_0xFFFF_). Such policies are applied if a client has no specific policy.
**** 'first'
+
Specifies the first client of a range (first is included).
(valid value: A valid client identifier in hex: e.g. _0x1234_)
**** 'last'
+
Specifies the last client id of a range (last is included).
(valid value: A valid client identifier in hex: e.g. _0x1234_)
*** 'credentials' (optional)
+
Specifies the credentials of the above client(s).
If _check_credentials_ is set to _true_ the credentials for the above client(s) (if they running locally) needs to be specified correctly to ensure local socket authentification can succeed.
This entry is optional due to the fact that remote clients needs to be configured as well to allow to communicate with local services as already mentioned above.
For remote clients this entry should be skipped.
**** 'uid'
+
Specifies the LINUX user id of the above client(s) as decimal number.
As a wildcard "any" can be used.
**** 'gid'
+
Specifies the LINUX group id of the above client(s) as decimal number.
As a wildcard "any" can be used.
**** 'allow/deny' (optional)
+
Specifies whether the LINUX user and group ids are allowed or denied for the policy.
***** 'uid' (array)
+
Specifies a list of LINUX user ids. These may either be specified as decimal numbers or as ranges. Ranges
are specified by the first and the last valid id (see example below).
***** 'gid' (array)
+
Specifies a list of LINUX group ids. These may either be specified as decimal numbers or as ranges. Ranges
are specified by the first and the last valid id (see example below).
*** 'allow/deny'
+
This tag specifies either _allow_ or _deny_ depending on white- or blacklisting is needed. Specifing _allow_ and _deny_ entries in one policy is therefore not allowed.
With _allow_ a whitelisting of what is allowed can be done which means an empty _allow_ tag implies everything is denied.
With _deny_ a blacklisting of what is allowed can be done which means an empty _deny_ tag implies everything is allowed.
**** 'requests' (array)
+
Specifies a set of serivce instance pairs which the above client(s) is allowed/denied to communicate with.
***** 'service'
+
Specifies a service for the _requests_.
***** 'instance'
+
Specifies a instance for the _requests_
**** 'offers' (array)
+
Specifies a set of service instance pairs which are allowed/denied to be offered by the above client(s).
***** 'service'
+
Specifies a service for the _offers_.
***** 'instance'
+
Specifies a instance for the _offers_
Security configuration example
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[source, bash]
----
...
"security" :
{
...
"policies" :
[
{
...
"credentials" :
{
"uid" : "44",
"gid" : "any"
},
"allow" :
[
"requests" :
[
{
"service" : "0x6731",
"instance" : "0x0001"
}
]
]
},
{
"credentials" :
{
"deny" :
[
{
"uid" : [ "1000", { "first" : "1002", "last" : "max" }],
"gid" : [ "0", { "first" : "100", "last" : "243" }, "300"]
},
{
"uid" : "55",
"gid" : "55"
}
]
},
"allow" :
[
"requests" :
[
{
"service" : "0x6732",
"instance" : "0x0001"
}
]
]
}
]
}
----
The config/ folder contains some addition vsomeip configuration files to run the vsomeip
examples with activated security checks.
Additionally there's a security test in the `test/` subfolder which can be used
for further reference. +
They give a basic overview how to use the security related configuration tags described
in this chapter to run a simple request/response or subscribe/notify example locally or
remotely.
Audit Mode
~~~~~~~~~~
vsomeip's security implementation can be put in a so called 'Audit Mode' where
all security violations will be logged but allowed. This mode can be used to
build a security configuration.
To activate the 'Audit Mode' the 'security' object has to be included in the
json file but the 'check_credentials' switch has to be set to false. For
example:
[source, bash]
----
[...]
"services" :
[
[...]
],
"security" :
{
"check_credentials" : "false"
},
"routing" : "service-sample",
[...]
----
Autoconfiguration
-----------------
vsomeip supports the automatic configuration of client identifiers and the routing.
The first application that starts using vsomeip will automatically become the
routing manager if it is _not_ explicitly configured. The client identifiers
are generated from the diagnosis address that can be specified by defining
DIAGNOSIS_ADDRESS when compiling vsomeip. vsomeip will use the diagnosis address
as the high byte and enumerate the connecting applications within the low byte
of the client identifier.
Autoconfiguration of client identifiers isn't meant to be used together with vsomeip Security.
Every client running locally needs to have at least its own credentials configured when security is activated to ensure the credential checks can pass.
Practically that means if a client requests its identifier over the autoconfiguration for which no credentials are configured (at least it isn't known which client identifier is used beforehand) it is impossible for that client to establish a connection to a server endpoint.
However if the credentials for all clients are same it's possible to configure them for the overall (or DIAGNOSIS_ADDRESS) client identifier range to mix autoconfiguration together with activated security.
vsomeipd
--------
The vsomeipd is a minimal vsomeip application intended to offer routing manager
functionality on a node where one system wide configuration file is present.
The vsomeipd uses the application name `vsomeipd` by default. This name can be
overridden by specifying `-DROUTING=$DESIRED_NAME` during the cmake call.
Example: Starting the daemon on a system where the system wide configuration is
stored under `/etc/vsomeip.json`:
[source, bash]
----
VSOMEIP_CONFIGURATION=/etc/vsomeip.json ./vsomeipd
----
When using the daemon it should be ensured that:
* In the system wide configuration file the vsomeipd is defined as
routing manager, meaning it contains the line `"routing" : "vsomeipd"`.
If the default name is overridden the entry has to be adapted accordingly.
The system wide configuration file should contain the information about all
other offered services on the system as well.
* There's no other vsomeip configuration file used on the system which contains
a `"routing"` entry. As there can only be one routing manager per system.
vsomeip Hello World
-------------------
In this paragraph a Hello World program consisting out of a client and a service
is developed. The client sends a message containing a string to the service.
The service appends the received string to the string `Hello` and sends it back
to the client.
Upon receiving a response from the service the client prints the payload of the
response ("Hello World").
This example is intended to be run on the same host.
All files listed here are contained in the `examples\hello_world` subdirectory.
Build instructions
~~~~~~~~~~~~~~~~~~
The example can build with its own CMakeFile, please compile the vsomeip stack
before hand as described in <<Compilation>>. Then compile the example starting
from the repository root directory as followed:
[source, bash]
----
cd examples/hello_world
mkdir build
cd build
cmake ..
make
----
Starting and expected output
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Starting and expected output of service
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[source, bash]
----
$ VSOMEIP_CONFIGURATION=../helloworld-local.json \
VSOMEIP_APPLICATION_NAME=hello_world_service \
./hello_world_service
2015-04-01 11:31:13.248437 [info] Using configuration file: ../helloworld-local.json
2015-04-01 11:31:13.248766 [debug] Routing endpoint at /tmp/vsomeip-0
2015-04-01 11:31:13.248913 [info] Service Discovery disabled. Using static routing information.
2015-04-01 11:31:13.248979 [debug] Application(hello_world_service, 4444) is initialized.
2015-04-01 11:31:22.705010 [debug] Application/Client 5555 got registered!
----
Starting and expected output of client
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[source, bash]
----
$ VSOMEIP_CONFIGURATION=../helloworld-local.json \
VSOMEIP_APPLICATION_NAME=hello_world_client \
./hello_world_client
2015-04-01 11:31:22.704166 [info] Using configuration file: ../helloworld-local.json
2015-04-01 11:31:22.704417 [debug] Connecting to [0] at /tmp/vsomeip-0
2015-04-01 11:31:22.704630 [debug] Listening at /tmp/vsomeip-5555
2015-04-01 11:31:22.704680 [debug] Application(hello_world_client, 5555) is initialized.
Sending: World
Received: Hello World
----
CMakeFile
~~~~~~~~~
[source, bash]
----
include::../examples/hello_world/CMakeLists.txt[]
----
Configuration File For Client and Service
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
[source, bash]
----
include::../examples/hello_world/helloworld-local.json[]
----
Service
~~~~~~~
[source, bash]
----
include::../examples/hello_world/hello_world_service.cpp[]
----
The service example results in the following program execution:
:numbered!:
[float]
Main
^^^^^
. __main()__
+
First the application is initialized. After the initialization is
finished the application is started.
[float]
Initialization
^^^^^^^^^^^^^^
[start=2]
. __init()__
+
The initialization contains the registration of a message
handler and an event handler.
+
The message handler declares a callback (__on_message_cbk__) for messages that
are sent to the specific service (specifying the service id, the service
instance id and the service method id).
+
The event handler declares a callback (__on_event_cbk__) for events that occur.
One event can be the successful registration of the application at the runtime.
[float]
Start
^^^^^
[start=3]
. __start()__
+
The application will be started. This function only returns when the application
will be stopped.
[float]
Callbacks
^^^^^^^^^
[start=4]
. __on_state_cbk()__
+
This function is called by the application when an state change occurred. If
the application was successfully registered at the runtime then the specific
service is offered.
. __on_message_cbk()__
+
This function is called when a message/request from a client for the specified
service was received.
+
First a response based upon the request is created.
Afterwards the string 'Hello' will be concatenated with the payload of the
client's request.
After that the payload of the response is created. The payload data is set with
the previously concatenated string.
Finally the response is sent back to the client and the application is stopped.
[float]
Stop
^^^^
[start=6]
. __stop()__
+
This function stops offering the service, unregister the message and the event
handler and shuts down the application.
:numbered:
Client
~~~~~~
[source, bash]
----
include::../examples/hello_world/hello_world_client.cpp[]
----
The client example results in the following program execution:
:numbered!:
[float]
Main
^^^^^
. __main()__
+
First the application is initialized. After the initialization is finished the
application is started.
[float]
Initialization
^^^^^^^^^^^^^^
[start=2]
. __init()__
+
The initialization contains the registration of a message handler, an event
handler and an availability handler.
+
The event handler declares again a callback (__on_state_cbk__) for state changes
that occur.
+
The message handler declares a callback (__on_message_cbk__) for messages that
are received from any service, any service instance and any method.
+
The availability handler declares a callback (__on_availability_cbk__) which is
called when the specific service is available (specifying the service id and the
service instance id).
[float]
Start
^^^^^
[start=3]
. __start()__
+
The application will be started. This function only returns when the application
will be stopped.
[float]
Callbacks
^^^^^^^^^
[start=4]
. __on_state_cbk()__
+
This function is called by the application when an state change occurred. If the
application was successfully registered at the runtime then the specific service
is requested.
. __on_availability_cbk()__
+
This function is called when the requested service is available or no longer
available.
+
First there is a check if the change of the availability is related to the
'hello world service' and the availability changed to true.
If the check is successful a service request is created and the appropriate
service information are set (service id, service instance id, service method
id).
After that the payload of the request is created. The data of the payload is
'World' and will be set afterwards.
Finally the request is sent to the service.
. __on_message_cbk()__
+
This function is called when a message/response was received.
If the response is from the requested service, of type 'RESPONSE' and the return
code is 'OK' then the payload of the response is printed. Finally the
application is stopped.
[float]
Stop
^^^^
[start=7]
. __stop()__
+
This function unregister the event and the message handler and shuts down the
application.
:numbered:
Trace Connector
---------------
anchor:traceconnector[]
Overview/Prerequisites
~~~~~~~~~~~~~~~~~~~~~~
The Trace Connector is used to forward the internal messages that are sent over
the Unix Domain Sockets to DLT. +
So a prerequisite is that DLT is installed and the module can be found in
context of CMake.
Configuration
~~~~~~~~~~~~~
Static Configuration
^^^^^^^^^^^^^^^^^^^^
The Trace Connector can be configured statically over the
<<config-tracing,JSON configuration file>>. +
[float]
Example 1 (Minimal Configuration)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
[source, bash]
----
{
...
"tracing" :
{
"enable" : "true"
},
...
----
This is the minimal configuration of the Trace Connector. This just enables the
tracing and all of the sent internal messages will be traced/forwarded to DLT.
[float]
Example 2 (Using Filters)
^^^^^^^^^^^^^^^^^^^^^^^^^
[source, bash]
----
{
...
"tracing" :
{
"enable" : "true",
"channels" :
[
{
"name" : "My channel",
"id" : "MC"
}
],
"filters" : [
{
"channel" : "MC",
"services" : [ "0x1234" ],
"methods" : [ "0x80e8" ],
"clients" : [ "0x1343" ],
"type" : "positive"
}
]
},
...
----
You can apply filters to the messages. In this example only the messages that
* are addressed to the service with the id _0x1234_
+
* relates to the method with the id _0x80e8_
+
* relates to the client with the id _0x1234_
will be traced/forwarded to DLT. This is a kind of positive filtering. To
exclude messages from tracing/forwarding to DLT, the filter type can be
set to "negative". This would suppress the messages that relate to the
specified service, method and client ids. The default filter type is
psoitive. +
The messages will be forwarded over the channel with the id _MC_. If
just one filter is used, then the definition of a channel is optional.
But if multiple filters are used, each filter needs an own channel! +
In this example each criteria has only one expression/value but it's also possible
to define multiple values to get a more fine-grained filter. +
The ids of the filter criterias can be found in the appropriate _.fdepl_ files
and in the rest of the configuration file. +
Dynamic Configuration
^^^^^^^^^^^^^^^^^^^^^
The Trace Connector can also be configured dynamically over its interfaces.
[float]
Example:
^^^^^^^^
[source, bash]
----
// get trace connector
std::shared_ptr<tc::trace_connector> its_trace_connector = tc::trace_connector::get();
// add channel
its_trace_connector->add_channel("MC", "My channel");
//add filter rule
tc::trace_connector::filter_rule_t its_filter_rule;
its_filter_rule[tc::filter_criteria_e::SERVICES] = { 4660 };
its_filter_rule[tc::filter_criteria_e::METHODS] = { 33000 };
its_filter_rule[tc::filter_criteria_e::CLIENTS] = { 4931 };
its_trace_connector->add_filter_rule("MC", its_filter_rule);
// init trace connector
its_trace_connector->init();
// enable trace connector
its_trace_connector->set_enabled(true);
----
Tools
-----
vsomeip_ctrl
~~~~~~~~~~~~
anchor:vsomeip_ctrl[]
`vsomeip_ctrl` is a small utility which can be used to send SOME/IP messages
from the commandline. If a response arrives within 5 seconds the response will
be printed.
* It can be build via `vsomeip_ctrl` make target (`make vsomeip_ctrl`).
* The instance id of the target service has to be passed in hexadecimal
notation.
* The complete message has to be passed in hexadecimal notation.
* See the `--help` parameter for available options.
* If `vsomeip_ctrl` is used to send messages to a remote service and no
`vsomeipd` is running on the local machine, make sure to pass a json
configuration file where `vsomeip_ctrl` is set as routing manager via
environment variable.
* If `vsomeip_ctrl` is used to send messages to a local service and no
`vsomeipd` is running on the local machine, make sure to use the same json
configuration file as the local service.
Example: Calling method with method id 0x80e8 on service with service id 0x1234,
instance id 0x5678:
[source, bash]
----
./vsomeip_ctrl --instance 5678 --message 123480e800000015134300030100000000000009efbbbf576f726c6400
----
Example: Sending a message to service with service id 0x1234, instance id
0x5678 and method id 0x0bb8 via TCP
[source, bash]
----
./vsomeip_ctrl --tcp --instance 5678 --message 12340bb8000000081344000101010000
----