Platform: STM32F446xE / NUCLEO-F446RE
This repository contains:
- Applications for running on each control board [apps]
- Custom and external libaries [libs]
- Miscellaneous items [misc]
- Python scripts [scripts]
- Arm MBED OS 6 SDK source [mbed-os]
- Configuration files for each target hardware [targets]
- Test applications for testing code components [test-apps]
- Makefile [makefile]
- Github Actions configurations for automatic build testing [.github]
- Create a branch in the format
yourName/#<issue-number>/featureName
for every feature you are working on - Rebase onto master and test on hardware before merging into master
- Add a Github Actions build target for your application if it is not a test application
- Create a pull request to merge any branch into master and select everyone else working on the feature as reviewers
- Name the pull request
Closes #<issue-number>: FeatureTitle
- Name the pull request
- When merging a pull request that fixes an issue title the commit
Fixes #issueNumber: FeatureTitle
- Clean binaries between making changes to the makefile
- There seems to be an annoying mix of CamelCase and snake_case in MBED but just try to be consistent with whatever code is nearby
- Squash when merging pull requests
-
Download the development toolchains and serial interface software
For Ubuntu (20.04 preferred)
- Update package lists:
sudo apt update
- Install Ninja:
sudo apt install ninja-build
- Install ccache:
sudo apt install ccache
- Install mbed-tools and other python requirements:
pip3 install -r scripts/requirements.txt
sudo apt install screen can-utils
for serial and CAN interfacing- Install/update ARM GCC toolchain:
Note: If you are not using Ubuntu 20.04 and/or bash you may need to modify this script's paths/files.
sudo apt autoremove gcc-arm-none-eabi wget https://developer.arm.com/-/media/Files/downloads/gnu-rm/10-2020q4/gcc-arm-none-eabi-10-2020-q4-major-x86_64-linux.tar.bz2 sudo tar -xvf gcc-arm-none-eabi-10-2020-q4-major-x86_64-linux.tar.bz2 -C /opt/ echo "PATH=/opt/gcc-arm-none-eabi-10-2020-q4-major/bin:\$PATH" >> ~/.bashrc export PATH=/opt/gcc-arm-none-eabi-10-2020-q4-major/bin:$PATH
- Install CMake:
- Follow kitware instructions to add Latest CMake apt repository
sudo apt install cmake
For Windows
- Install Windows Subsystem for Linux (WSL) with Ubuntu 20.04
- Follow Ubuntu setup instructions (optionally instead of
screen
you can use PuTTy, a GUI Windows app)
For Mac (Not Recommended)
- Open Command Line
- Install Homebrew if not installed
/usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"
- Download auto-run script, which will auto install with latest version:
brew tap ARMmbed/homebrew-formulae
- Install ARM GCC toolchain via HomeBrew:
brew install arm-none-eabi-gcc
- Install ZOC for serial interfacing
- Update package lists:
-
Download source code
git clone --recurse-submodules https://github.com/uwrobotics/MarsRover2020-firmware.git cd MarsRover2020-firmware
Note: The repository has a submodule. To update the submodule, use
git submodule update
-
Install mbed-tools and other python requirements:
pip3 install -r scripts/requirements.txt
-
Verify the the toolchains were installed correctly
Open a new Command Prompt / Terminal window and run the following commands:
make --version # Should be v3.8.x or newer mbed-tools --version # Should be v7.3.3 or newer cmake --version # Should be v3.20.0 or newer ninja --version # Should be v1.10.0 or newer arm-none-eabi-gcc --version # Should be v10.2.1 or newer
-
Run make with the target application and board
Ex. Compile the science application for the science board:
make APP=science_2021 TARGET=SCIENCE_REV2_2021
ORmake app=science_2021 target=SCIENCE_REV2_2021
Ex. Compile the arm application for the arm board:
make APP=arm_2021 TARGET=ARM_REV2_2021
ORmake app=arm_2021 target=ARM_REV2_2021
Note: The APP and TARGET variables can be defined in any order.
After compiling an application you should see a message similar to the following:
[100%] Linking CXX executable arm.arm-board.elf Generating bin from elf make[4]: Leaving directory '/home/wmmc88/MarsRover2020-firmware/build-arm-board' [100%] Built target arm.arm-board.elf make[3]: Leaving directory '/home/wmmc88/MarsRover2020-firmware/build-arm-board' make[2]: Leaving directory '/home/wmmc88/MarsRover2020-firmware/build-arm-board' make[1]: Leaving directory '/home/wmmc88/MarsRover2020-firmware/build-arm-board'
Note: Ninja automatically optimizes the build process to most efficiently use all your system's resources to significantly speed up compile time. In some cases, this may slow down other processes during the compilation. You can choose to use fewer threads during build with the following command:
CMAKE_BUILD_PARALLEL_LEVEL=<max number of threads> make APP=<app-name> TARGET=<target-name>
You can add the
CMAKE_BUILD_PARALLEL_LEVEL
to your.bashrc
to have this max thread limit be persistent.echo "export CMAKE_BUILD_PARALLEL_LEVEL=<max number of threads>" >> ~/.bashrc
Tip: You can choose to build all the supported app/target configs at once using
make all
-
Deploy onto board (see below for how to connect to a rover control board)
Note: The following instructions only apply to Nucleo and Rev 1 boards. For Rev 2 boards and beyond, see Using the ST-Link to Program Rover Boards (Rev 2 +)
Find the application .bin file, located in the build--board/apps/ directory.
For Ubuntu
- Install libusb
sudo apt install libusb-1.0-0-dev
- Drag and Drop .bin file into NODE_F446RC device folder
For Windows
- Drag and Drop .bin file into NODE_F446RC device folder OR if this does not work or debugging is required:
- Download st-link utility. Scroll down to Get Software
- Connect USB to nucleo board and open st-link utility
- Load code by going to Target->Program and browse for .bin file
For Mac
- Drag and Drop .bin file into NODE_F446RC disk
After deploying, the Nucleo will begin to flash red and green. Once the LED stays green, power-cycle the board by unplugging and replugging the 5V connector on the Nucleo.
- Install libusb
- To clean the project workspace of app and library build files, run
make clean
- To clean compiled MBED SDK files, run
make clean-mbed
Rev 2 PCBs come with ARM 10-pin SWD headers and can be programmed via ST-Link. A 20-pin to 10-pin adapter is needed to hook the ST-Link 20-pin header to the rover board 10-pin header.
- sudo apt install stlink-tools
- Tested on 1.6.0
- sudo apt install stlink-gui
- Tested on 1.6.0
- sudo apt install openocd
- Tested on 0.10.0
- Download st-link utility
Important: Ensure your PCB is powered prior to connecting the st-link to your computer (this is an ST bug)
Option 1: Use GUI
- Launch the GUI with
stlink-gui
- Click connect
- Click Open and select your .bin file
- Click Flash (leave memory address as 0x8000000)
Option 2: Use CLI
- Ensure that the st-link is connected with
st-info --probe
- Flash with
st-flash write <path to .bin> 0x8000000
- If the flashing was successful, you should see the following message:
Flash written and verified! jolly good!
Pro tips:
To erase the loaded program from flash: st-flash erase
To reset/rerun the program: st-flash reset
- Connect the ST-Link to the rover board and to your computer
- Open ST-Link Utility
- Click "Connect to the target"
- Click "Open file" and select the
.bin
file of the program to flash - Click "Program & verify:
- Click "Start"
In order to use the Nucleo development board as a programmer, the two jumpers (black caps) labelled NUCLEO - ST-LINK will need to be removed. This will sever the ST-LINK debugger portion of the Nucleo from the MCU side, allowing it to be used as a general debugger.
The ST-LINK debugger can then be connected via header CN4 (pins 1-5 with 1 nearest to the SWD label) to a rover board debug header (pins should be labelled) to program it according to the following table:
+-----------------------+-----------------------------------+
| Nucleo CN4 Pin Number | Rover Board Debug Header Pin Name |
+-----------------------+-----------------------------------+
| 1 (VREF) | VCC |
| 2 (SWCLK) | CLK |
| 3 (GND) | GND |
| 4 (SWDIO) | IO |
| 5 (NRST) | RST |
| 6 (SWO) | Not Connected |
+-----------------------+-----------------------------------+
After deploying the binary to the board, the Nucleo's LD1
LED will flash red and green. Programming is complete when the LED stays green, so don't powercycle the board before this.
The 10-pin Serial Wire Debug (SWD) programming interface does not come with UART lines for standard printf usage. Instead, we will use Serial Wire Output (SWO), a single wire interface to transmit trace messages to an external debugger.
See apps/test-logger
for an example of using the SWO-supported logger utility. Ensure that the target you are building for is configured to support SWO logging (see targets/<target name>/include/mbed_config_target.h).
- In one terminal, launch the openocd server and specify runtime parameters with
openocd -f /usr/share/openocd/scripts/interface/stlink-v2.cfg -f /usr/share/openocd/scripts/target/stm32f4x.cfg -c "tpiu config internal - uart off 180000000" -c "itm ports on"
- In a different terminal, run the swo_parser.py under the scripts folder with
python3 <path to swo_parser.py>
- Ensure that the rover board is running and the ST-Link is connected
- In the ST-Link Utility software, click "Print via SWO viewer"
- Set the system clock rate to 180000000Hz and stimulus port to 0
- Click "Start". The SWO print statements should appear in the Serial Wire Viewer console.
Detailed steps for this have not been created yet. Refer to this (link)[https://github.com/stlink-org/stlink/blob/develop/doc/tutorial.md] for guidance
The boards can be communicated with through the serial interface exposed through the debug pins. You can use the USB-serial interface built into the Nucleo dev boards to communicate with the control boards by connecting the TX pin to the board's RX pin and the RX pin to the board's TX pin (transmit to recieve and vice versa). Ensure the program running on the nucleo is not printing too.
On Ubuntu
- Run
screen /dev/serial/by-id/usb-STM* 115200
from the terminal. You may need to prepend this withsudo
.
On Windows
- Device manager, go to Ports (COM & LPT) and find the name of the Nucleo port (ie COM4)
- Open PuTTy, select the Serial radio button, enter the COM port name and the baud rate (default 115200) and click open
The boards can also be communicated with over the CAN bus interfaces. You can use a CANable serial USB-CAN dongle to communicate with them from your development computer. Connect the CAN_H, CAN_L, and GND pins of the CANable to the corresponding pins on the board, and the dongle to your computer.
On Ubuntu
- Run
sudo slcand -o -c -s6 /dev/serial/by-id/*CAN*-if00 can0
to set up the CAN interface- The flag
-s6
sets the bus speed to 500 kbps - The flag
-s8
sets the bus speed to 1 Mbps
- The flag
- Run
sudo ip link set can0 up
to enable the interface - Run
cansend can0 999#DEADBEEF
to send a frame to ID 0x999 with payload 0xDEADBEEF - Run
candump can0
to show all traffic received by can0
See the CANable Getting Started guide for more information including Windows support.
This repository follows the formatting rules outlines in the .clang-format file. You can make your code conformant by using the CLI (documentation here) or by installing a clang-format plugin/extension in your IDE of choice. For example, CLion has built-in Clang-Format support and VS Code has a decent extension.
You may have to edit the settings of the plugin to match the version of clang-format that we are using. The current version can be seen here under clangFormatVersion
.
To download the matching version of LLVM(contains clang-format):
wget https://apt.llvm.org/llvm.sh
chmod +x llvm.sh
sudo ./llvm.sh <version number>
sudo apt install clang-format-<version number>
For VS Code you will need to specify the path to the clang-format executable. To do this:
- In shell script copy the output of the following command:
which clang-format-11
- Open the User Settings in JSON format
- Add the two following settings to the file "editor.formatOnSave": true, "clang-format.executable": "{path copied from step 1}"
For every feature that gets added, a test app that tests the feature in isolation should be written to verify that the feature actually functions properly in our hardware. The author of the feature should ensure that the test app works on all the board targets that make sense. Typically this means that the test app should work on at least the nucleo board and the board that the feature was designed for. For example, the test-can
app should be able to be compiled and run successfully for all of our boards, including a nucleo dev board.
You can take a look at the apps, libraries, and board targets that are already added in existing CMakeLists to see how you declare new CMake targets. If more clarification is needed, you can refer to the official CMake documentation.
Tips:
- mbed_set_post_build(target_name) is a CMake function provided by mbed that generates the bin and hex files from the elf.
Important Note: Whenever an app or board target is added, make sure to also add the relevant configurations to the supported_build_configurations.yaml
. This list is used to keep track of what app-target tuples are "supported" and should always build properly. For example, if you add a new board target, add the board target to each app that should be able to compile for your new board. As another example, if you add a new app, you should add the app to the supported_build_configurations.yaml
and list out all the boards it should be able to be built on.