Calling SOME/IP Methods
SOME/IP Methods
Methods implement a request-response communication in the SOME/IP protocol. A server offers a method. A client calls the method by sending a request message to the server. The server answers the request message with a response message. In contrast to SOME/IP events, no subscription of the offered service has to be setup in order to make method request call.
Request and response messages can carry a serialized payload. Typically, the payload is either the argument to the method or the return value of the method. For defining the service interface data types which can be used in methods and serializing structured data into bytes
follow the article SOME/IP Service Interface Data Types.
In SOME/IP there is also the possibility of fire&forget communication. The client sends a request message and does not expect a response from the server. The fire&forget communication is not implemented in someipy yet.
Step 1: Define the Data Types for Request and Response
In this example, we will call a SOME/IP method that calculates the sum of two signed integers and returns the result back to the client. When we are calling the method, we will pass the two addends with the request message to the server. The server will send back a response message containing the sum. For request and response, we need to define two data types Addends
and Sum
. The data types are the same types as defined in the article Offering SOME/IP Methods.
@dataclass
class Addends(SomeIpPayload):
addend1: Sint16
addend2: Sint16
def __init__(self, addend1: int = 0, addend2: int = 0):
self.addend1 = Sint16(addend1)
self.addend2 = Sint16(addend2)
@dataclass
class Sum(SomeIpPayload):
value: Sint32
def __init__(self):
self.value = Sint32()
Details on defining data types can be found here.
Step 2: Definition of the Service
A SOME/IP method is part of a service and so we will define a Service
as the next step using the method ID and the major version of the service. In the third step, this service will be used for creating a ClientServiceInstance
on which we can call the SOME/IP method. The ServiceBuilder
class offers a fluent API, which is used for creation of the Service
object.
SAMPLE_SERVICE_ID = 0x1234
addition_service = (
ServiceBuilder()
.with_service_id(SAMPLE_SERVICE_ID)
.with_major_version(1)
.build()
)
Step 3: Instantiate the Service
The previously defined Service
can be instantiated into one or multiple service instances. Since we want to call (and not offer) a method, we will instantiate a ClientServiceInstance
. The construct_client_service_instance
is a coroutine since it uses asyncio
internally and therefore has to be await
ed.
- You need to pass the instance ID (
SAMPLE_INSTANCE_ID
) of the server service instance that offers the method. - The endpoint that is passed is the endpoint (ip address and port) of the client and not of the server.
- The
ttl
parameter is the lifetime of a potential subscription in seconds. This is only relevant if you want to subscribe to an event. - It is assumed that the
service_discovery
object was instantiated beforehand. For more information on that topic, read Service Discovery Configuration Parameters. - You can choose either UDP or TCP as the transport protocol. Make sure, that it matches to the service offered by the server.
We will also attach the ClientServiceInstance
to the ServiceDiscovery
object. This allows the ClientServiceInstance
to be notified about service discovery messages.
SAMPLE_INSTANCE_ID = 0x5678
client_instance_addition = await construct_client_service_instance(
service=addition_service,
instance_id=SAMPLE_INSTANCE_ID,
endpoint=(ipaddress.IPv4Address(interface_ip), 3002),
ttl=5,
sd_sender=service_discovery,
protocol=TransportLayerProtocol.UDP,
)
service_discovery.attach(client_instance_addition)
Step 4: Calling the Method
Finally, we need to setup the method parameters for the request and call the SOME/IP method offered by the server. In this case, the parameter to the method is an Addends
object. After creating the Addends
object, we will call the method on the ClientServiceInstance
using the call_method
function. call_method
is a coroutine which has to be awaited and will not block until the response from the server is received. This allows other tasks to be scheduled while waiting for a response. The call_method
function expects a method ID identifying the method on the server to be called. A server could offer multiple methods inside the same service. The second parameter is the payload to be sent with the request: The Addends
object is serialized into a bytes
object and passed to the call.
The call_method
function returns a MethodResult
object with following members:
- message_type (
MessageType
): The MessageType is an enum and can be eitherMessageType.RESPONSE
orMessageType.ERROR
. - return_code (
ReturnCode
): TheReturnCode
enum reflects the return codes defined in the SOME/IP protocol specification. For indicating a successful method call to the client,E_OK
is returned by the server. - payload (
bytes
): The payload is abytes
object which can be deserialized into the expected returned message structure. In this case it will be deserialized into aSum
object.
The call_method
function can raise a RuntimeError
or an asyncio.TimeoutError
. A RuntimeError
is raised in case the service instance offered by the server was not found yet. In this case no method request can be sent since the destination IP address and port are not available yet. The asyncio.TimeoutError
is raised if no response is received by the server after sending the request or in case of TCP, the TCP connection cannot be established.
To avoid the RuntimeError
it is possible to test whether the service was already found by using the service_found
method on the ClientServiceInstance
.
method_parameter = Addends(addend1=1, addend2=2)
try:
# You can query if the service offering the method was already found via SOME/IP service discovery
print(f"Service found: {client_instance_addition.service_found()}")
while not client_instance_addition.service_found():
print("Waiting for service..")
await asyncio.sleep(0.5)
# The call_method function can raise an error, e.g. if no TCP connection to the server can be established
# In case there is an application specific error in the server, the server still returns a response and the
# message_type and return_code are evaluated.
method_result = await client_instance_addition.call_method(
SAMPLE_METHOD_ID, method_parameter.serialize()
)
if method_result.message_type == MessageType.RESPONSE:
print(
f"Received result for method: {' '.join(f'0x{b:02x}' for b in method_result.payload)}"
)
if method_result.return_code == ReturnCode.E_OK:
sum = Sum().deserialize(method_result.payload)
print(f"Sum: {sum.value.value}")
else:
print(
f"Method call returned an error: {method_result.return_code}"
)
elif method_result.message_type == MessageType.ERROR:
print("Server returned an error..")
# In case the server includes an error message in the payload, it can be deserialized and printed
except Exception as e:
print(f"Error during method call: {e}")