Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 5 Next »

Foreword

You can refrain from downloading anything on the linked websites unless it is in the Installation section, we have set up a template package that already contains most of the libraries and it should be straightforward to get set up on the rest of the development environment. This guide is meant to help you understand the layout of the template and understand the toolchain required for developing embedded software for the solar car. It aims to be a comprehensive guide to getting you to the point where you can start developing as soon as you feel comfortable with the C paradigms for embedded programming. 

Prework

Before we get started if you have not yet done so, please read up on Software 101 in particular, the Coding Standards! These will be necessary to actually do any development on the car. It is also recommended that you spend some time working with the MSP430 launchpads which were part of the architecture from the previous vehicle MSXI and will provide a good basis for getting started. 

Architecture and Libraries

The architecture we will be using on MSXII is the stm32f0xx variant of the ARM Cortex M0 MCU. This architecture is supported by both the CMSIS ARM Core M0 Library and STM Peripherals Library. Specific documentation for these libraries can be found on the Software Resources page of confluence. 

Supported Development Toolchain

  • Compiler: GCC ARM
    • A compiler builds human readable source code into a machine readable target language, usually a machine executable or binary for a specific architecture. GCC ARM compiles C code into machine readable .elf, .hex or .bin files which can be flashed onto the MCU. GCC ARM is the standard open source compiler for ARM architecture. Compilers are powerful and often catch typos, errors, and warnings; they also optimize code for efficiency or size. The compiler flags set these behaviors of the compiler.
  • Build Tools: GNU Make and GNU Linker aka ld
    • Build tools usually instruct the compiler on where the source files it is building are located and which files to look at in order to build the target program. Often a Linker will be used with a Makefile to build a collection of C and header files into a standard library which can be included in your program. GNU Make and Linker are long-standing open source standard build tools used by many developers, although many alternatives do exist.
  • Linter: cpplint 
    • A linter is a program that checks the style of source code for errors. Some linters also perform static analysis, ours does not, which look to catch programming errors before the code is executed. cpplint is Google's open source linter for C/C++ programs, note that our Coding Standards closely follow Google's and as a result, this linter works well for us. It will often be more pedantic than we need but it is a good tool to use.
  • Flasher: st-link/GDB or OpenOCD
    • A flasher will transfer firmware (our compiled program) onto the MCU. Typically, this is done with a special cable or by a chip on the PCB and requires flashing software to transfer the contents of the source code into the memory of the MCU. For our purposes, we will be using st-link with GDB for debug which is the chip manufacturer supported method or OpenOCD the open source version.
  • IDEs: Eclipse or any text editor (Core members like: Vim, Sublime Text and Visual Studio Code)
    • Anyone just starting out and especially those using Windows are strongly encouraged to use the Eclipse IDE which is one of the few ways to get a working Toolchain on a Windows machine. Note that we will not be supporting any issues you may encounter using other methods on Windows although, we will also support Linux if you want a VM or dual boot but you will need to figure out the install on your own. 
  • Source Control: GitHub
    • We use GitHub for our source and versioning control. In order to contribute to the codebase, you will also need to start using GitHub. If you are new to GitHub you may want to check out these guides. We have tight controls on out git repositories and require all commits to come from pull requests and that they are squashed prior to submitting.

Installation

General Instructions:

Step 1: If you don't already have one, get a GitHub account!

Step 2: Join our GitHub organization uw-midsun

Step 3: Follow your system specific instructions below

Windows:

Word of Warning: This install is probably the most painful and likely to break do your best and if you get stuck just ask

Step 1: Download and install Eclipse this will require at minimum a JRE or JDK version 1.7.0 or higher. I had issues with the x86 or 64 bit install so you may want to consider something just installing the x64 or 32 bit version first. Also note: when you get to the Eclipse installer you only need the C/C++ version of Eclipse.

Step 2: Follow this guide to install the correct build tools and the GCC ARM compiler and set them up in your environment. You may want to disregard the comments about not adding the programs to the system environment path as I had trouble without doing that. Also, make sure you prepend the path variables if you do add them as if they are at the end they have a tendency not to work.

Step 3: Install Python version 2.7 and get cpplint using pip from the command line: (You can also clone it off the GitHub repo here)

$ pip install cpplint

You will need to add both Python2.7 and cpplint to your environment path so that the makefile runs the lint correctly.

Step 4: In theory that should be everything. Try it by cloning a copy of the template from  TODO "here" and building it in Eclipse.

Linux:

In theory, this is the easiest install method. For that reason, those interested in running a VM or dual-booting their machine to get Linux are welcome to do so. Although, we don't explicitly support either we can probably provide guidance if you get really stuck. However, if you break anything doing it we are not liable!

This step by step is currently for Ubuntu adjust to your system as needed

Step 1: Install 32 bit C libraries and some of the build tools if you don't have them

$ sudo apt-get -y install lib32z1 lib32ncurses5 lib32bz2-1.0 binutils gdb python

Step 2: Download and unpack the GCC ARM Linux tarball from here. Alternatively install it from your package repository if you have one.

Step 3: Download cpplint using pip or from the package repository 

$ pip install cpplint

Step 4: Install st-link or OpenOCD, they may be in your package repository but you can always build from source. 

Step 5: In theory that is everything assuming your distribution is sane and already has gcc and make installed. Feel free to grab an IDE or text editor if you don't already have one you like. The link in Step 2 of the Windows guide has a tutorial for setting up in Eclipse.

Step 6: Try cloning a copy of the template from TODO "here" and building it by navigating to the directory and typing 

$ make

If it compiles you did everything right (except perhaps the linter) if not you need to fix or add something. You should produce a file called main.elf.

Mac OSX:

I don't own a Mac nor do any of the software core members. All the tools should, in theory, be OS X compatible so you can try building the same tools that those using the Linux guide do. Alternatively, you can follow the Windows instructions although you shouldn't have to fiddle with Path variables as much.


Project Template

The package template contains five directories, a Makefile and a README. Use this for starting new projects only. Existing projects will have their own repositories which will share a copy of libraries, linker, and extra. 

Makefile

As mentioned, a Makefile tells the compiler what to build and where to find those items. This Makefile is no different, it automatically links and builds a file called main.elf and the STM Peripheral library (THIS LIBRAY FEATURES A HAL DO NOT USE IT WE ARE MAKING OUR OWN) which is located in the libraries directory. The libraries can be included without needing a relative path as can any header files you write and place in the inc directory. The linker directory contains the linker files to build the libraries. The extra folder contains configuration options for making an OpenOCD binary. Any program files you create should go in the src directory and any drivers should go into inc. The README will contain more detailed information and the Makefile does have options for selecting source files.




  • No labels