Getting Started

Setting up a Build Environment

This guide highlights the installation / setup process for the luna gateware library. It focuses on getting the Python module (and prerequisites) up and running.

Prerequisites

  • Python 3.7, or later.
  • A working FPGA toolchain. We only officially support a toolchain composed of the Project Trellis ECP5 tools, the yosys synthesis suite, and the NextPNR place-and-route tool. All of these tools must be built from master.
  • A working installation of nMigen. Note that only the official toolchain from @nmigen <https://github.com/nmigen> is supported; the @m-labs <https://github.com/m-labs> derivative is not.

Installation

Currently, the LUNA library is considered a “work-in-progress”; and thus it’s assumed you’ll want to use a local copy of LUNA for development.

The easiest way to set this up is to install the distribution in a virtual environment. From the root of the repository:

# Pull down poetry, our build system.
pip3 install poetry --user

# Install a copy of our local tools into our virtualenv.
poetry install

If you want to install LUNA to your machine globally (not recommended), you can do so using the following single command:

# Create a LUNA package, and install it.
pip3 install . --user

Testing

The easiest way to test your installation is to build one of the test applets. These applets are just Python scripts that construct and program gateware using nMigen; so they can be run like any other script:

# With GSG or self-built LUNA hardware connected; we can run the full test,
# and test both our installation and the attached hardware.
poetry run applets/interactive-test.py

# Without LUNA hardware connected, we'll only build the applet, to exercise
# our toolchain.
poetry run applets/interactive-test.py --dry-run

The apollo utility.

The luna distribution depends on apollo, which includes a utility that can be used to perform various simple functions useful in development; including simple JTAG operations, SVF playback, manipulating the board’s flash, and debug comms.

$ apollo
usage: apollo [-h] command: [[argument]] [[value]]

Utility for LUNA development via an onboard Debug Controller.

positional arguments:
  command:    info       -- Prints information about any connected LUNA-compatible boards
              configure  -- Uploads a bitstream to the device's FPGA over JTAG.
              erase      -- Clears the attached board's configuration flash.
              program    -- Programs the target bitstream onto the attached FPGA.
              jtag-scan  -- Prints information about devices on the onboard JTAG chain.
              flash-scan -- Attempts to detect any attached configuration flashes.
              svf        -- Plays a given SVF file over JTAG.
              spi        -- Sends the given list of bytes over debug-SPI, and returns the response.
              spi-inv    -- Sends the given list of bytes over SPI with inverted CS.
              spi-reg    -- Reads or writes to a provided register over the debug-SPI.
  [argument]  the argument to the given command; often a filename
  [value]     the value to a register write command

To have easy access to the apollo command, you’ll need to ensure that your python binary directory is in your PATH. For macOS/Linux, this often means adding ~/.local/bin to your PATH.