SocketCAN#

The SocketCAN documentation can be found in the Linux kernel docs in the networking directory. Quoting from the SocketCAN Linux documentation:

The socketcan package is an implementation of CAN protocols (Controller Area Network) for Linux. CAN is a networking technology which has widespread use in automation, embedded devices, and automotive fields. While there have been other CAN implementations for Linux based on character devices, SocketCAN uses the Berkeley socket API, the Linux network stack and implements the CAN device drivers as network interfaces. The CAN socket API has been designed as similar as possible to the TCP/IP protocols to allow programmers, familiar with network programming, to easily learn how to use CAN sockets.

Important

python-can versions before 2.2 had two different implementations named socketcan_ctypes and socketcan_native. These were removed in version 4.0.0 after a deprecation period.

Socketcan Quickstart#

The CAN network driver provides a generic interface to setup, configure and monitor CAN devices. To configure bit-timing parameters use the program ip.

The virtual CAN driver (vcan)#

The virtual CAN interfaces allow the transmission and reception of CAN frames without real CAN controller hardware. Virtual CAN network devices are usually named ‘vcanX’, like vcan0 vcan1 vcan2.

To create a virtual can interface using socketcan run the following:

sudo modprobe vcan
# Create a vcan network interface with a specific name
sudo ip link add dev vcan0 type vcan
sudo ip link set vcan0 up

Real Device#

vcan should be substituted for can and vcan0 should be substituted for can0 if you are using real hardware. Setting the bitrate can also be done at the same time, for example to enable an existing can0 interface with a bitrate of 1MB:

sudo ip link set can0 up type can bitrate 1000000

CAN over Serial / SLCAN#

SLCAN adapters can be used directly via CAN over Serial / SLCAN, or via SocketCAN with some help from the slcand utility which can be found in the can-utils package.

To create a socketcan interface for an SLCAN adapter run the following:

slcand -f -o -c -s5 /dev/ttyAMA0
ip link set up slcan0

Names of the interfaces created by slcand match the slcan\d+ regex. If a custom name is required, it can be specified as the last argument. E.g.:

slcand -f -o -c -s5 /dev/ttyAMA0 can0
ip link set up can0

PCAN#

Kernels >= 3.4 supports the PCAN adapters natively via SocketCAN, so there is no need to install any drivers. The CAN interface can be brought like so:

sudo modprobe peak_usb
sudo modprobe peak_pci
sudo ip link set can0 up type can bitrate 500000

Intrepid#

The Intrepid Control Systems, Inc provides several devices (e.g. ValueCAN) as well as Linux module and user-space daemon to make it possible to use them via SocketCAN.

Refer to below repositories for installation instructions:

Send Test Message#

The can-utils library for Linux includes a cansend tool which is useful to send known payloads. For example to send a message on vcan0:

cansend vcan0 123#DEADBEEF

CAN Errors#

A device may enter the “bus-off” state if too many errors occurred on the CAN bus. Then no more messages are received or sent. An automatic bus-off recovery can be enabled by setting the “restart-ms” to a non-zero value, e.g.:

sudo ip link set canX type can restart-ms 100

Alternatively, the application may realize the “bus-off” condition by monitoring CAN error frames and do a restart when appropriate with the command:

ip link set canX type can restart

Note that a restart will also create a CAN error frame.

List network interfaces#

To reveal the newly created can0 or a vcan0 interface:

ifconfig

Display CAN statistics#

ip -details -statistics link show vcan0

Network Interface Removal#

To remove the network interface:

sudo ip link del vcan0

Wireshark#

Wireshark supports socketcan and can be used to debug python-can messages. Fire it up and watch your new interface.

To spam a bus:

import time
import can

interface = 'socketcan'
channel = 'vcan0'

def producer(id):
    """:param id: Spam the bus with messages including the data id."""
    bus = can.Bus(channel=channel, interface=interface)
    for i in range(10):
        msg = can.Message(arbitration_id=0xc0ffee, data=[id, i, 0, 1, 3, 1, 4, 1], is_extended_id=False)
        bus.send(msg)

    time.sleep(1)

producer(10)

With debugging turned right up this looks something like this:

../_images/wireshark.png

The process to follow bus traffic is even easier:

for message in Bus(can_interface):
    print(message)

Reading and Timeouts#

Reading a single CAN message off of the bus is simple with the bus.recv() function:

import can

bus = can.Bus(channel='vcan0', interface='socketcan')
message = bus.recv()

By default, this performs a blocking read, which means bus.recv() won’t return until a CAN message shows up on the socket. You can optionally perform a blocking read with a timeout like this:

message = bus.recv(1.0)  # Timeout in seconds.

if message is None:
    print('Timeout occurred, no message.')

If you set the timeout to 0.0, the read will be executed as non-blocking, which means bus.recv(0.0) will return immediately, either with a Message object or None, depending on whether data was available on the socket.

Filtering#

The implementation features efficient filtering of can_id’s. That filtering occurs in the kernel and is much much more efficient than filtering messages in Python.

Broadcast Manager#

The socketcan interface implements thin wrappers to the linux broadcast manager socket api. This allows the cyclic transmission of CAN messages at given intervals. The overhead for periodic message sending is extremely low as all the heavy lifting occurs within the linux kernel.

The BusABC initialized for socketcan interface transparently handles scheduling of CAN messages to Linux BCM via send_periodic():

with can.interface.Bus(interface="socketcan", channel="can0") as bus:
    task = bus.send_periodic(...)

More examples that uses send_periodic() are included in python-can/examples/cyclic.py.

The task object returned by send_periodic() can be used to halt, alter or cancel the periodic message task:

class can.interfaces.socketcan.CyclicSendTask(bcm_socket, task_id, messages, period, duration=None)[source]#

A SocketCAN cyclic send task supports:

  • setting of a task duration

  • modifying the data

  • stopping then subsequent restarting of the task

Construct and start() a task.

Parameters:
  • bcm_socket (socket) – An open BCM socket on the desired CAN channel.

  • task_id (int) – The identifier used to uniquely reference particular cyclic send task within Linux BCM.

  • messages (Sequence[Message] | Message) – The messages to be sent periodically.

  • period (float) – The rate in seconds at which to send the messages.

  • duration (float | None) – Approximate duration in seconds to send the messages for.

modify_data(messages)[source]#

Update the contents of the periodically sent CAN messages by sending TX_SETUP message to Linux kernel.

The number of new cyclic messages to be sent must be equal to the original number of messages originally specified for this task.

Note

The messages must all have the same arbitration_id like the first message.

Parameters:

messages (Sequence[Message] | Message) – The messages with the new can.Message.data.

Return type:

None

start()[source]#

Restart a periodic task by sending TX_SETUP message to Linux kernel.

It verifies presence of the particular BCM task through sending TX_READ message to Linux kernel prior to scheduling.

Raises:

ValueError – If the task referenced by task_id is already running.

Return type:

None

stop()[source]#

Stop a task by sending TX_DELETE message to Linux kernel.

This will delete the entry for the transmission of the CAN-message with the specified task_id identifier. The message length for the command TX_DELETE is {[bcm_msg_head]} (only the header).

Return type:

None

Buffer Sizes#

Currently, the sending buffer size cannot be adjusted by this library. However, this issue describes how to change it via the command line/shell.

Bus#

The SocketcanBus specializes BusABC to ensure usage of SocketCAN Linux API. The most important differences are:

  • usage of SocketCAN BCM for periodic messages scheduling;

  • filtering of CAN messages on Linux kernel level;

  • usage of nanosecond timings from the kernel.

class can.interfaces.socketcan.SocketcanBus(channel='', receive_own_messages=False, local_loopback=True, fd=False, can_filters=None, ignore_rx_error_frames=False, **kwargs)[source]#

A SocketCAN interface to CAN.

It implements can.BusABC._detect_available_configs() to search for available interfaces.

Creates a new socketcan bus.

If setting some socket options fails, an error will be printed but no exception will be thrown. This includes enabling:

  • that own messages should be received,

  • CAN-FD frames and

  • error frames.

Parameters:
  • channel (str) – The can interface name with which to create this bus. An example channel would be ‘vcan0’ or ‘can0’. An empty string ‘’ will receive messages from all channels. In that case any sent messages must be explicitly addressed to a channel using can.Message.channel.

  • receive_own_messages (bool) – If transmitted messages should also be received by this bus.

  • local_loopback (bool) – If local loopback should be enabled on this bus. Please note that local loopback does not mean that messages sent on a socket will be readable on the same socket, they will only be readable on other open sockets on the same machine. More info can be read on the socketcan documentation: See https://www.kernel.org/doc/html/latest/networking/can.html#socketcan-local-loopback1

  • fd (bool) – If CAN-FD frames should be supported.

  • can_filters (Sequence[CanFilter | CanFilterExtended] | None) – See can.BusABC.set_filters().

  • ignore_rx_error_frames – If incoming error frames should be discarded.

RECV_LOGGING_LEVEL = 9#

Log level for received messages

channel_info = 'unknown'#

a string describing the underlying bus and/or channel

property filters: Sequence[CanFilter | CanFilterExtended] | None#

Modify the filters of this bus. See set_filters() for details.

flush_tx_buffer()#

Discard every message that may be queued in the output buffer(s).

Return type:

None

recv(timeout=None)#

Block waiting for a message from the Bus.

Parameters:

timeout (float | None) – seconds to wait for a message or None to wait indefinitely

Returns:

None on timeout or a Message object.

Raises:

CanOperationError – If an error occurred while reading

Return type:

Message | None

send(msg, timeout=None)[source]#

Transmit a message to the CAN bus.

Parameters:
  • msg (Message) – A message object.

  • timeout (float | None) – Wait up to this many seconds for the transmit queue to be ready. If not given, the call may fail immediately.

Raises:

CanError – if the message could not be written.

Return type:

None

send_periodic(msgs, period, duration=None, store_task=True)#

Start sending messages at a given period on this bus.

The task will be active until one of the following conditions are met:

  • the (optional) duration expires

  • the Bus instance goes out of scope

  • the Bus instance is shutdown

  • stop_all_periodic_tasks() is called

  • the task’s stop() method is called.

Parameters:
  • msgs (Message | Sequence[Message]) – Message(s) to transmit

  • period (float) – Period in seconds between each message

  • duration (float | None) – Approximate duration in seconds to continue sending messages. If no duration is provided, the task will continue indefinitely.

  • store_task (bool) – If True (the default) the task will be attached to this Bus instance. Disable to instead manage tasks manually.

Returns:

A started task instance. Note the task can be stopped (and depending on the backend modified) by calling the task’s stop() method.

Return type:

CyclicSendTaskABC

Note

Note the duration before the messages stop being sent may not be exactly the same as the duration specified by the user. In general the message will be sent at the given rate until at least duration seconds.

Note

For extremely long running Bus instances with many short lived tasks the default api with store_task==True may not be appropriate as the stopped tasks are still taking up memory as they are associated with the Bus instance.

set_filters(filters=None)#

Apply filtering to all messages received by this Bus.

All messages that match at least one filter are returned. If filters is None or a zero length sequence, all messages are matched.

Calling without passing any filters will reset the applied filters to None.

Parameters:

filters (Sequence[CanFilter | CanFilterExtended] | None) –

A iterable of dictionaries each containing a “can_id”, a “can_mask”, and an optional “extended” key:

[{"can_id": 0x11, "can_mask": 0x21, "extended": False}]

A filter matches, when <received_can_id> & can_mask == can_id & can_mask. If extended is set as well, it only matches messages where <received_is_extended> == extended. Else it matches every messages based only on the arbitration ID and mask.

Return type:

None

shutdown()[source]#

Stops all active periodic tasks and closes the socket.

Return type:

None

property state: BusState#

Return the current state of the hardware

stop_all_periodic_tasks(remove_tasks=True)#

Stop sending any messages that were started using send_periodic().

Note

The result is undefined if a single task throws an exception while being stopped.

Parameters:

remove_tasks (bool) – Stop tracking the stopped tasks.

Return type:

None