CanCat

CanCat is an open source multi-purpose tool for interacting and experimenting with Controller Area Networks (CAN), such as those that are often used in cars, building automation, etc.

Description

CanCat has two main parts:

1. Firmware for compatible CAN-transceivers,
2. Python client to talk to CanCat,

The CAN-transceiver combinations that are currently supported by CanCat are:

• Arduino with SeeedStudio's CANBUS Shield
• Arduino DUE with Togglebit.net's CAN shield
• Macchina M2 (Under-the-Hood)
• Macchina M2 (Under-the-Dash)

The goals of CanCat are to provide:

• a way to capture and transmit messages on an arbitrary CAN bus (whatever the speed, as supported by hardware)
• an architecture for analyzing and identifying messages on a CAN bus
• a manner for data to be shared (Ford sends different messages from GM, Saab, Toyota, Honda, Hyundai, Kia, etc.) (either in the form of lookup tables or in the form of accessor codes.)

Software

Required:

• Python 3.6+

Recommended:

• CANBUS_Shield from SeeedStudio (for Firmware compilation)
  https://github.com/Seeed-Studio/CAN_BUS_Shield

• vstructs From the Vivisect framework in the Python Path:
  (required for J1939 module)
  https://github.com/vivisect/vivisect

• ipython

Installation

1. Install pyserial:

$ pip install --user pyserial

2. (OPTIONAL) Install ipython if you want to use CanCat interactively.

$ pip install --user ipython

3. Install the Arduino IDE.

4. (OPTIONAL) If you are using a Macchina M2 follow the getting started guide for the M2 to install the M2 hardware definitions in the Arduino tool.

5. (OPTIONAL) If you are on a Linux system, you may choose to install the arduino-cli for your platform. The arduino-cli tool can be used to compile and flash your CAN device without opening the Arudino IDE.

6. Clone CanCat and build the desired firmware. If you are not using the arduino-cli tool, use the Arduino IDE as normal to build and flash the sketch onto your target device. If you have installed the arguing-cli tool you can compile and flash the CanCat firmware with the following steps:

$ git clone https://github.com/atlas0fd00m/CanCat
$ arduino-cli config init
$ arduino-cli config add board_manager.additional_urls https://macchina.cc/package_macchina_index.json
$ arduino-cli core update-index
$ arduino-cli core install arduino:sam
$ arduino-cli core install macchina:sam
$ cd CanCat/sketches
$ make m2

7. Ensure that your CAN-transceiver is not in bootloader mode by unplugging its USB connector and then plugging it back in again.

8. Connect to your CAN-transceiver with CanCat

9. Use CanCat.

Connecting to your CAN-transceiver with CanCat

Once you have the required software installed and your device is flashed, you can use CanCat with your CAN-transceiver.

Connect to the CAN-transceiver with CanCat [Linux]:

$ ./CanCat.py -p /dev/ttyACM0  # if CanCat device is /dev/ttyACM0

'CanCat, the greatest thing since J2534!'

Research Mode: enjoy the raw power of CanCat

currently your environment has an object called "c" for CanCat.  this is how
you interact with the CanCat tool:
    >>> c.ping()
    >>> c.placeBookmark('')
    >>> c.snapshotCanMsgs()
    >>> c.printSessionStats()
    >>> c.printCanMsgs()
    >>> c.printCanSessions()
    >>> c.CANxmit('message', )
    >>> c.CANreplay()
    >>> c.saveSessionToFile('file_to_save_session_to')
    >>> help(c)

Connect to the CAN-transceiver with CanCat [Linux and other systems]:

>>> import cancatlib

>>> CANalysis = cancat.CanInterface('/dev/ttyACM0', 115200) # your device may vary

>>> CANalysis.ping()

>>> and In [#]: are used interchangeably in this instruction guide.

>>> is the default interactive python prompt, and commands using this prompt will use the CANalysis object.

In [#]: is the ipython prompt, and commands using this prompt will begin with the c object.

Other than the different in prompt type, the commands for CanCat will look the same.

Getting Started

Pinging your CAN-transceiver

To see if CanCat is communicating correctly with your computer and the connected CAN-transceiver, use the ping() command.

In [1]: c.ping()

Setting the Baud Rate

Once you connect to your CAN-transceiver, you will want to use CanCat to set the CAN bus interface baud rate on the device. (Note: 500kbps is the most likely baud rate for devices that you will interact with.)

>>> CANalysis.setCanBaud(cancat.CAN_125KBPS)

After you have set the baud rate on your CAN-transceiver, CanCat will automatically capture any messages it sees on the CAN bus it is attached to. CanCat will store these messages in the current session for analysis. Note: Unless you save the CanCat capture, the messages you have captured will no longer be stored once you end your CanCat session.

Saving CanCat captures

CanCat will only save what it has captured when you tell it to save, so make sure to save your capture session / analysis periodically.

>>> CANalysis.saveSessionToFile('filename_for_this_session')

Once you save for the first time, the file name will be cached so you can simply save your capture to the same file again by typing:

>>> CANalysis.saveSessionToFile()

CanCat help and tips

To access the help function in CanCat:

>>> help(cancat)

If help doesn't provide you with what you are looking for, you can do a tab-complete search to bring up each of the possible CanCat commands.

In [6]: c.<PRESS_TAB_KEY>
c.CANrecv
c.genCanMsgs
c.ping
c.recv
c.saveSessionToFile
c.CANreplay
c.getArbitrationIds
c.placeCanBookmark
c.recvall
c.setCanBaud
c.CANsniff
c.getBookmarkFromMsgIndex
c.port
c.register_handler
c.setCanBookmarkComment
c.CANxmit
c.getCanMsgCount
c.printAsciiStrings
c.remove_handler
c.setCanBookmarkCommentByMsgIndex
c.bookmark_info
c.getMsgIndexFromBookmark
c.printBookmarks
c.reprBookmark
c.setCanBookmarkName
c.bookmarks
c.getSessionStats
c.printCanMsgs
c.reprBookmarks
c.setCanBookmarkNameByMsgIndex
c.clearCanMsgs
c.getSessionStatsByBookmark
c.printCanMsgsByBookmark
c.reprCanMsgs
c.snapshotCanMessages
c.comments
c.loadFromFile
c.printCanSessions
c.reprCanMsgsByBookmark
c.verbose
c.filterCanMsgs
c.log
c.printSessionStats
c.restoreSession
c.filterCanMsgsByBookmark
c.name
c.printSessionStatsByBookmark
c.saveSession

UDS Module

CanCat has a UDS module for doing UDS diagnostics. Basic usage is as follows:

In [1]: import uds

In [2]: u = uds.UDS(c, 0x760, 0x768) # 11-bit IDs

In [3]: u = uds.UDS(c, 0x18da98f1, 0x18daf198, extflag=1) # 29-bit IDs

When initializing the module, pass in the CanCat object you are using (c, in our case), then the arbitration IDs that you will be transmitting and receiving on. Also set extflag to 1 if using 29-bit identifiers.

Now that our UDS module is initialized, we can use it to perform UDS diagnostics on a vehicle. Let's start with something simple:

In [4]: u.ReadDID(0xF190)

This will read Data Identifier (DID) 0xF190, which should contain the VIN for the vehicle. If the read succeeds then the requested data is returned by the ReadDID function. If the read fails, then a NegativeResponseException will be thrown, which will print out the UDS error message that was received.

Since CanCat can be scripted easily, scanning for UDS servers on CAN is easily accomplished. The following loop will send a ReadDID request to every arbitration ID from 0x700 to 0x7F7 and print out which servers respond, and which time out.

for i in range(0x700, 0x7f8):  
    u = uds.UDS(c, i, i+8)
    print("Trying ", hex(i))
    try:
        u.ReadDID(0xf190)
    except:
        print("Error reading DID 0xF190, server exists at this address")

If a timeout is received then no UDS server responded on the address. If a positive response or negative respons is received, then you have discovered a UDS server. Other functionality can be scripted as well, such as scanning DIDs to see which are implemented. This code will scan all the DIDs in the range from 0xF180 to 0xF19F, which contains useful information such as hardware and software part numbers, VINs, and other identifying information for this UDS server.

for i in range(0xf180, 0xf1a0):
    try:                  
        print(u.ReadDID(i))
    except:          
        print(hex(i), " Returned error")

Other UDS functionality includes:

• xmit_recv - Transmit and receive any data
• SendTesterPresent - Sends the tester present message one time
• StartTesterPresent - Starts sending the tester present message periodically
• StopTesterPresent - Stop sending the tester present message periodically
• DiagnosticSessionControl - Change the diagnostic session
• ReadMemoryByAddress - Read a memory address
• ReadDID - Read a DID
• WriteDID - Write a DID
• RequestDownload - Start a data download
• writeMemoryByAddress - Write a memory address
• RequestUpload - Start a data upload
• EcuReset - Reset an ECU
• ScanDIDs - A built-in function for scannin DIDs
• SecurityAccess - Starts a security access request
• _key_from_seed - This method can be overloaded to provide the algorithm for turning a seed into a key. Unimplemented by default.

An additional function provided is printUDSSession, which takes a CanCat variable, and the RX and TX arbitration IDs and parses UDS traffic from the CAN traffic captured by CanCat.

CCP Module

CanCat has a CCP (CAN Calibration Protocol) module. CCP is a predecessor to XCP and is used for calibration and data acquisition.

CCP allows a single master device (called "leaders" in this implementation) connect to multiple slave devices (called "followers" in this implementation).

Basic usage is as follows:

In [1]: import ccp

In [2]: cl = ccp.CCPLeader(c) # if you are acting as a leader device, communicating with a follower device

In [3]: cf = ccp.CCPFollower(c) # if you are acting as a follower device, communicating with a leader device

CanCat's CCP module implements "parse" and "generate" functions for each of the CCP commands defined in the CCP specification (https://automotivetechis.files.wordpress.com/2012/06/ccp211.pdf) unless otherwise noted:

• Leader-side parsing of received event messages (errors) and data acquisition CRMs have not been implemented at this time.
• Data Acquisition (DAQ) functions have been added to the follower according to the spec but have not been tested.

In addition to the "parse" and "generate" functions, CCP sequences are defined at the end of ccp_leader.py.

CCP’s endianness is implementation-specific (aside from CONNECT, DISCONNECT, and TEST messages), so you may need to make modifications to this code. Other implementation-specific settings are noted in the documentation above each function.

Before calibration or data acquisition ("DAQ") can happen, a logical point-to-point connection needs to be established between the leader and follower devices using a CONNECT message.

In CCP, all messages (and data responses) are packed into "message objects" with up to 8 bytes of data. The available message types are as follows:

CRO (Command Receive Object): message sent from the leader device to the follower device(s). CRM (Command Return Message): one type of message sent from the follower device to the leader device containing command / error code and command counter. DTO (Data Transmission Object): message sent from the follower device to the leader device (Command Return Message or Event Message or Data Acquisition Message).

Other CanCat Uses

Using CanCat to Analyze Previous Captures

If all you want to do is analyze a previous CanCat capture you can skip the hardware set up steps mentioned in the Installation section, clone the CanCat repository, add the file you wish to analyze to the CanCat folder, and run the command:

$ ./CanCat.py -f filename_of_previous_capture  # no CanCat device required

CAN-in-the-Middle

CAN-in-the-Middle is another way to utilize your CanCat. It requires two CAN shields on one arduino. One of the CAN shields needs to be modified so that the CS pin of the MCP2515 CAN controller is on D10, rather than D9. This is accomplished by cutting a trace on the CAN shield PCB and bridging (solder bridge or 0-ohm resistor) the pads for CS and D10. Instructions are also on the seeedstudio Wiki, although their board differed slightly from mine, mostly in that the pads are on the bottom of the board on mine and on the top of the board in their example.

Once you have a properly modified CAN Bus shield, you'll be able to isolate components connected to the CAN bus to see which messages a specific device is sending, without changing the conditions by fully removing it from the CAN Bus. This can be very helpful for certain reverse engineering tasks.

Flash the CAN_in_the_middle firmware to the Arduino. Hook the CAN wires up so that the device you are trying to isolate is connected to the modified CAN shield that uses D10 for CS, and the vehicle CAN bus (with the rest of the devices) is connected to the unmodified CAN shield. These are referred to as the Isolation network (ISO) and the Vehicle network (VEH) respectively.

Start CAN_in_the_middle with the following command:

./CanCat.py -I CanInTheMiddle -p /dev/tty.usbmodem1411 -S 500K

( where the -p option is your port and -S is the CAN Baud rate.)

Most of the commands for Can-in-the-Middle are the same as the normal CanCat interface. Functions that report only what has been received on the Isolation side have Iso appended to the end of the function name. For example:

$ citm.getCanMsgCount() # The number of CAN packets seen in aggregate

$ citm.getCanMsgCountIso() # The number of CAN packets received on the Isolation network

$ citm.printCanMsgs() # Prints all CAN messages

$ citm.printCanMsgsIso() # prints all CAN messages received on the Isolation network

Placing a bookmark places a bookmark simultaneously on both the Isolation information (Iso interface messages) and the aggregate information (standard CAN interface messages).

canmap

Canmap is a tool built on CanCat to scan a CAN bus for various UDS capabilities.
Canmap is built on top of the cancat.uds.UDS class. Canmap has many different options to control what type of scans are performed, and how the scans are performed, but the basic information required is the type of scan to run, the port the CanCat device is present on, and the bus speed:

The most basic scan is an ECU scan to identify what ECUs are on the bus

$ ./canmap -p /dev/ttyACM0 -b 500K -sE

Additional scan modes are:

• DIDs
• Sessions

All items can be scanned with this command:

$ ./canmap -p /dev/ttyACM0 -b 500K -sEDS

Saving canmap scan output

The results of a canmap scan can be saved as a configuration yaml file with the -o (--output-file) option. This yaml file can be used as an input to future scans with the -i (--input-file) option. If an input config file is provided information that is already in the config will not be scanned again unless the -r (--rescan) option is provided. For example an aborted DID scan can later be resumed and any ECUs that DIDs were found for will not be searched for again:

$ ./canmap -p /dev/ttyACM0 -b 500K -sEDS -o scan_results.yml
<output>
^C
$ ./canmap -p /dev/ttyACM0 -b 500K -sEDS -i scan_results.yml -o scan_results.yml

The configuration file saves some additional scan parameters such as the baud rate, and timeout parameters. These parameters are re-used when the config file is provided as an input config.

The config file contain a notes field that indicates the command(s) used to create that config file.

The raw can messages can also be saved as a CanCat session with the -c (--can-sesison-file) option. This can be useful to diagnose strange responses found during the scanning:

$ ./canmap -p /dev/ttyACM0 -b 500K -sEDS -c scan_with_weird_errors.sess

The session can then be opened with the normal CanCat options:

$ ./CanCat.py -f scan_with_weird_errors.sess

ECU Scanning

The Range of ECUs to scan can be specified with the -E option, the default range is 00-FF for both standard (11-bit) and extended (29-bit) CAN addressing. The bus mode can be set with the -m (-bus-mode) option.

To scan only a subset of the ECU range in 11-bit mode the command would be:

$ ./canmap -p /dev/ttyACM0 -b 500K -sE -E 60-A0 -m std

ECU scanning works by sending a read DID request (cancat.uds.UDS.ReadDID) or a session control request (cancat.uds.UDS.DiagnosticSessionControl) and waiting for a timeout, a negative response or a positive response. The default method is to attempt to read the VIN from each UDS address
(cancat.uds.UDS.ReadDID(0xF190)). Different methods are available because different methods have different degrees of success on different vehicles.

The other factor that can affect the success rate is how quickly ECUs respond.
The UDS standard timeout is 3 seconds, scanning both bus modes with 3 second timeouts could take up to 25 minutes. Instead the default timeout for ECU scanning is 0.2 seconds, if fewer ECUs than expected are identified it may be worth re-trying the scan with an increased timeout by setting the -T (--timeout`) option.

DID Scanning

DID scanning can take a while depending on the behavior of the ECUs. By default only the UDS standard identification DIDs (F180-F18E,F190-F1FF) are searched for. Testing has shown that searching a range of F000-FFFF can take around 2 minutes for cooperative ECUs, but much much longer for ECUs which allow requests to timeout rather than sending negative responses. A larger range can be specified with the -D option:

$ ./canmap -p /dev/ttyACM0 -b 500K -sD -D F000-FFFF -i known_ecus.yml

DIDs are only scanned on ECUs that have already been identified. If a DID scan is run and there are no known ECUs then no messages will be sent.

Session Scanning

It is assumed that the default session for each ECU is session 1. DIDs identified through scanning are associated with session 1. By default the full range of diagnostic sessions is searched (02-7F). I have found on some ECUs that sessions can only be entered after already being in another prerequesite session. Searching for these recursive diagnostic sessions is enabled by default but can be disabled with the -n (--no-recursive-session-scanning) option.

Depending on the ECU behavior, session scanning can take a varying amount of time and/or produce strange error conditions.

Unit Tests

Unit tests can be run with:

python -m unittest discover -v

Acknowledgments

This project is made possible through collaboration with researchers at GRIMM (SMFS, Inc.), most notably Matt Carpenter and Tim Brom.

Happy Hacking!