Skip to content

Latest commit

 

History

History
259 lines (209 loc) · 8.83 KB

get_started.md

File metadata and controls

259 lines (209 loc) · 8.83 KB
Raw

Getting Started

Minimum required version of dependencies are:

Name Version
CMake 3.13
Python 3.8

Install Dependencies

  • Ubuntu

    • install tools
      sudo apt install build-essential cmake ninja-build libc6-i386 libc6-i386-cross libstdc++6-i386-cross
    • install python3 (3.8.5 minimum) and pip
      sudo apt install python3 python3-pip

  • Windows

    • Windows Command Prompt The following commands assume you are using cmd.exe, it might differ if you are using PowerShell.

      • Install Chocolatey (https://chocolatey.org/) It is a package manager for Windows, with which it's not that difficult to install native Windows dependencies.
      1. Install Chocolatey by the following instructions on the Chocolatey Install (https://chocolatey.org/install) page.
      2. Open "cmd.exe" as "Administrator"
      3. Disable global confirmation to avoid having to confirm installation of individual programs:
        choco feature enable -n allowGlobalConfirmation
      1. Install CMake
        choco install cmake --installargs 'ADD_CMAKE_TO_PATH=System'
      1. Install other tools:
        choco install git python ninja
      1. Close the Administrator command prompt window.

  • Prepare Toolchain & Environment Variables

    • Supported toolchains:
      • gnu-gcc <-- default toolchain
      • nds-gcc
      • zcc
    • Toolchain setup:
      • gnu-gcc:
        1. Grab a copy of toolchain zip package and unzip to certain path, take TOOLCHAIN_PATH for example, (riscv32-unknown-elf-gcc is supposed to be found in TOOLCHAIN_PATH/bin)
        2. Declare a system environment variable of "GNURISCV_TOOLCHAIN_PATH" to the path of toolchain:
        • Linux, taking zsh for example (replace TOOLCHAIN_PATH with the path of toolchain on your workstation):
          export GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  export HPM_SDK_TOOLCHAIN_VARIANT=
        • Windows command prompt:
          set GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  set HPM_SDK_TOOLCHAIN_VARIANT=
      • nds-gcc:
        1. Grab a copy of toolchain zip package and unzip to certain path, take TOOLCHAIN_PATH for example, (riscv32-elf-gcc is supposed to be found in TOOLCHAIN_PATH/bin)
        2. Declare two system environment variables: "GNURISCV_TOOLCHAIN_PATH" to the path of toolchain; "HPM_SDK_TOOLCHAIN_VARIANT" to "nds-gcc":
        • Linux, taking zsh for example (replace TOOLCHAIN_PATH with the path of toolchain on your workstation):
          export GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  export HPM_SDK_TOOLCHAIN_VARIANT=nds-gcc
        • Windows command prompt:
          set GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  set HPM_SDK_TOOLCHAIN_VARIANT=nds-gcc

      Note: For windows, Andes compiler needs following libraries: - cygwin1.dll - cygncursesw-10.dll make sure its path is appended to the system environment variable "PATH".

      • zcc:
        1. Grab a copy of toolchain zip package and unzip to certain path, take TOOLCHAIN_PATH for example
        2. Declare a system environment variable of "GNURISCV_TOOLCHAIN_PATH" to the path of toolchain:
        • Linux, taking zsh for example (replace TOOLCHAIN_PATH with the path of toolchain on your workstation):
          export GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  export HPM_SDK_TOOLCHAIN_VARIANT=zcc
        • Windows command prompt:
          set GNURISCV_TOOLCHAIN_PATH=TOOLCHAIN_PATH
  set HPM_SDK_TOOLCHAIN_VARIANT=zcc
    • Environment Variables:
      • Using provided scripts to set the environment variable:
        • Linux:
          $ source env.sh
        • Windows command prompt:
          env.cmd
      • Manually declare a environment variable of "HPM_SDK_BASE" to the path of SDK root:
        • Linux, taking zsh for example (assume SDK is located at $HOME/hpm_sdk):
          export HPM_SDK_BASE=$HOME/hpm_sdk
        • Windows command prompt (assume SDK is located at c:\hpm_sdk):
          set HPM_SDK_BASE=c:\hpm_sdk

  • Install python dependencies

    • Linux:
      pip3 install --user -r "$HPM_SDK_BASE/scripts/requirements.txt"
    • Window (by default, python3/pip3 is not available after Python 3.x installed on Windows, but only python/pip):
      pip install --user -r "%HPM_SDK_BASE%/scripts/requirements.txt"

  • Build An Application with GNU GCC toolchain On finishing the steps mentioned above, SDK projects can be generated and built. The following steps describe how a demo can be built:

    1. Go to application directory, taking hello_world for example:
      cd samples/hello_world
    1. create a directory for build
    • Linux:
      mkdir build
    • Windows:
      md build
    1. Change directory to "build"
      cd build
    1. Generate build files for Ninja:
      cmake -GNinja -DBOARD=hpm6750evk ..

    Note: if it complains about "CMAKE_MAKE_PROGRAM is not set", please append -DCMAKE_MAKE_PROGRAM=YOUR_MAKE_EXECUTABLE_PATH to the previous command (NINJA_PATH is the folder in which ninja can be found):

      cmake -GNinja -DBOARD=hpm6750evk -DCMAKE_MAKE_PROGRAM=NINJA_PATH/ninja ..
    1. Building:
      ninja

    When it's done the elf and other application related files can be found in the directory of "output", like map file, assembly source or binary file

  • Quick Guide to Run/Debug An Application (hello_world):

    1. Wire up the board, including debug probe (by default it supports jlink) and serial port
    2. power up the board
    3. open console connecting to the debug console (target serial port) with baudrate of 115200
    4. get a copy of openocd. it can be installed via package management system or downloaded from sourceforge or github. But please make sure its revision is > 0.11
    5. Go to SDK root directory, run provided environment variable scripts:

    • Linux:

        $ source env.sh

    • Windows command prompt:

        env.cmd

      or setup environment variable OPENOCD_SCRIPTS manually:

    • Linux:

        $ export OPENOCD_SCRIPTS=${HPM_SDK_BASE}/boards/openocd

    • Windows:

        set OPENOCD_SCRIPTS=%HPM_SDK_BASE%\boards\openocd
    1. Start openocd with several configuration files in order of type of probe, type of core, type of board. For example, the following command will setup an openocd gdb server with ft2232 to single core on hpm6750evk
      openocd -f probes/ft2232.cfg -f soc/hpm6750-single-core.cfg -f boards/hpm6750evk.cfg

    Note: If using FTDI debugger and meet Error: libusb_open() failed with LIBUSB_ERROR_NOT_FOUND , please check the FTDI usb driver. If it is not installed correctly, use zadig to update:

    Open zadig, click Options-> List All Devices.

    List All Devices

    Select Dual RS232-HS (Interface 0).

    Select Dual RS232-HS (Interface 0)

    Then click Install Driver or Replace Driver.

    Replace Driver

    1. Go to hello_world directory
      cd samples/hello_world
    1. open up another terminal to start a gdb client
    • gnu-gcc:
      TOOLCHAIN_PATH/bin/riscv32-unknown-elf-gdb
    • nds-gcc:
      TOOLCHAIN_PATH/bin/riscv32-elf-gdb
    1. connect gdb client to the gdbserver started by openocd, (by default, gdbserver port is 3333)
      gdb> file build/output/demo.elf
  gdb> target remote localhost:3333
  gdb> load
  gdb> b main
  gdb> c
    1. on the debug console, "hello_world" is printed.

  • Build An Application with Segger Embedded Studio

    • Segger Embedded Studio for RISC-V can be downloaded from Segger
    • Project file for Segger Embedded Studio will be generated while generating build files for Ninja mentioned in "Build An Application with GNU GCC toolchain"->"4. Generate build files for Ninja"
    • The project file (.emProject) can be found at build/segger_embedded_studio/.

    Note: openocd executable needs to be found in the PATH variable of current console, otherwise debug configuration will not be generated to project file and needs to be configured manually in Segger Embedded Studio later.