wiki:ubuntu

Version 30 (modified by Tim Harvey, 13 months ago) ( diff )

added rootfs generation sections

  1. Ubuntu
    1. Ubuntu root filesystem
      1. Downloading pre-built rootfs tarball
      2. Creating Ubuntu rootfs using qemu and chroot
      3. Creating Ubuntu rootfs using deboostrap
    2. Ubuntu Webmin
    3. Ubuntu Packages
    4. Gateworks Packages
    5. Modem Support
    6. SSH Server
    7. Wireless
      1. Access Point (AP)
      2. Client using wpa_supplicant
    8. Predictable Interface Names (systemd/udev)
    9. Network Configuration
    10. Network Time Protocol (NTP) Configuration
    11. Real Time Clock (RTC) Configuration
    12. Watchdog
    13. Cross compiling example
    14. Native compiling example

Ubuntu

Gateworks supports the popular Ubuntu OS on Newport and Ventana products using an Ubuntu built root filesystem with a Gateworks kernel. In some cases you may be able to use an Ubuntu built kernel.

Gateworks always tries to use the LTS releases (Long Term Support).

  • Jammy Jellyfish 22.04
  • Focal Fossa 20.04

In addition to the information on this page, you likely want to look at product family specific details on the following pages for your product:

Ubuntu root filesystem

Creating a root filesystem comprised of Canonical built packages can be done in a couple of ways:

  • download a pre-built rootfs tarball
  • create one using deboostrap
  • create one using qemu/chroot

This refers to the rootfs only. To easily build an entire system image, including the boot firmware and Ubuntu, look at the BSP for the various product families:

Downloading pre-built rootfs tarball

Gateworks has pre-built Ubuntu based rootfs tarballs at ​http://dev.gateworks.com/ubuntu/ that are created with the ​ubuntu-rootfs script.

Creating Ubuntu rootfs using qemu and chroot

You can use qemu to create an Ubuntu based rootfs with very little in the way of dependencies. To do this you must boot a virtual machine for your architecture therefore you will need a kernel and a minimal root filesystem which can be easily built using buildroot.

Example:

  • aarch64 rootfs:
  1. Start by defining some parameters 
    IMG=ubuntu.ext4
SIZE=500M
UBUNTU=22.04
UBUNTU_REL=.3
  2. Create a ramdisk based image (similar concept as rescue image but we add a file here via BR2_ROOTFS_OVERLAY) 
    sudo apt install build-essential git
git clone http://github.com/buildroot/buildroot.git
cd buildroot
cat << EOF > configs/buildroot_defconfig
BR2_aarch64=y
BR2_cortex_a72=y
BR2_TOOLCHAIN_BUILDROOT_GLIBC=y
BR2_TOOLCHAIN_BUILDROOT_CXX=y
BR2_TARGET_GENERIC_HOSTNAME="localhost"
BR2_ROOTFS_OVERLAY="overlay/"
BR2_LINUX_KERNEL=y
BR2_LINUX_KERNEL_USE_ARCH_DEFAULT_CONFIG=y
BR2_PACKAGE_BUSYBOX_SHOW_OTHERS=y
BR2_PACKAGE_E2FSPROGS=y
# BR2_PACKAGE_E2FSPROGS_FSCK is not set
BR2_PACKAGE_I2C_TOOLS=y
BR2_PACKAGE_RNG_TOOLS=y
BR2_PACKAGE_BASH=y
BR2_PACKAGE_TAR=y
BR2_PACKAGE_UTIL_LINUX_HWCLOCK=y
BR2_TARGET_ROOTFS_CPIO=y
BR2_TARGET_ROOTFS_CPIO_GZIP=y
EOF
# create init script
mkdir -p overlay/etc/init.d
cat <<EOF > overlay/etc/init.d/S99boostrap-ubuntu.sh
#!/bin/sh

if [ "\$1" != "start" ]; then
    exit
fi

# setup
mkfs.ext4 -F /dev/vda -b 4096
mount /dev/vda /mnt
cd /mnt/
udhcpc -i eth0
# install base rootfs from release tarball
wget -c -P /tmp/ http://cdimage.ubuntu.com/ubuntu-base/releases/${UBUNTU}/release/ubuntu-base-${UBUNTU}${UBUNTU_REL}-base-arm64.tar.gz
tar zxf /tmp/ubuntu-base-${UBUNTU}${UBUNTU_REL}-base-arm64.tar.gz -C /mnt
# setup chroot environment
mount -o bind /proc /mnt/proc/
mount -o bind /sys/ /mnt/sys/
mount -o bind /dev/ /mnt/dev/
mount -o bind /dev/pts /mnt/dev/pts
mount -t tmpfs tmpfs /mnt/var/lib/apt/
mount -t tmpfs tmpfs /mnt/var/cache/apt/
echo "nameserver 8.8.8.8" > /mnt/etc/resolv.conf
echo "localhost" > /mnt/etc/hostname
echo "127.0.0.1 localhost" > /mnt/etc/hosts
export DEBIAN_FRONTEND=noninteractive DEBCONF_NONINTERACTIVE_SEEN=true LC_ALL=C LANGUAGE=C LANG=C
# install additional packages
chroot /mnt apt update
chroot /mnt apt upgrade -y
chroot /mnt apt install --no-install-recommends -y \
     apt-utils locales less wget procps \
     psmisc less sudo kmod rng-tools \
     ifupdown net-tools isc-dhcp-client ntpdate \
     openssh-server iproute2 iptables iputils-ping \
     i2c-tools can-utils usbutils pciutils \
     screen picocom \
     vim nano
# set root password
echo -e "root\nroot" | chroot /mnt passwd
# teardown chroot env
umount /mnt/var/lib/apt/
umount /mnt/var/cache/apt
chroot /mnt apt clean
chroot /mnt apt autoclean
reboot
EOF
chmod +x overlay/etc/init.d/S99boostrap-ubuntu.sh
make buildroot_defconfig
make -j8
    • generic aarch64 kernel Image and rootfs.cpio.gz will be in output/images
  1. Use qemu to kick off a virtual aarch64 machine which will run the init script and create root filesystem on a virtual disk 
    apt install qemu-system-arm
truncate -s $SIZE $IMG
qemu-system-aarch64 -m 1G -M virt -cpu cortex-a57 -nographic -smp 1 -no-reboot \
   -device virtio-net-device,netdev=eth0 -netdev user,id=eth0 \
   -device virtio-blk-device,drive=hd0 -drive file=$IMG,if=none,format=raw,id=hd0 \
   -append "console=ttyAMA0 ip=dhcp" \
   -kernel output/images/Image \
   -initrd output/images/rootfs.cpio.gz
    • ubuntu.ext4 contains your rootfs - add desired changes to the setup script

Creating Ubuntu rootfs using deboostrap

A popular way to create an Ubuntu root filesystem is to use the deboostrap utility on a Debian or Ubuntu host. This tool provides a 2-stage install where the second stage is within a chroot environment using qemu.

Gateworks uses a script to do this which you may find at ​http://github.com/Gateworks/ubuntu-rootfs

Requirements:

  • Linux Ubuntu or Debian System with network connection and sudo permissions

Important notes:

  • We set and use target and distro env variables in step 2 and use those env variables in the remaining steps to make this tutorial more version-agnostic. Please be aware of this and do not deviate from the steps unless or until you completely understand what you are doing.
  • These steps are not always exactly what we do in our script but give you an idea of how you would go about doing it yourself if you wanted to customize something

Steps:

  1. Install pre-requisites: 
    sudo apt-get install qemu-user-static debootstrap binfmt-support
  1. Perform first stage install of minimal filesystem for arm64 architecture: 
    distro=focal
arch=arm64
target=${distro}-${arch}
qemu_arch=aarch64
sudo debootstrap --arch=$arch --foreign $distro $target
# copy qemu binary for the binfmt packages to find it and copy in resolv.conf from host
sudo cp /usr/bin/qemu-${qemu_arch}-static $target/usr/bin
    • See ​http://ports.ubuntu.com/ubuntu-ports/dists/ for a list of current Ubuntu releases: oe 20.04=focal (latest LTS)
    • this minimal rootfs is still missing some core packages and configuration before it can be booted. These steps are taken care of in a 2nd stage install within a chroot shell
    • the chroot shell below will provide network support (inherited from the host)
  1. We now have a minimal Ubuntu rootfs - chroot to it and perform the 2nd stage install: 
    sudo chroot $target
# now we are in the chroot - setup env matching the distro above
distro=focal
export LANG=C
# setup second stage
/debootstrap/debootstrap --second-stage
    • this is the most minimal rootfs we would recommend
  1. (optional) add additional apt package repos: 
    cat <<EOT > /etc/apt/sources.list
deb http://ports.ubuntu.com/ubuntu-ports $distro main restricted universe multiverse
deb http://ports.ubuntu.com/ubuntu-ports $distro-updates main restricted universe multiverse
deb http://ports.ubuntu.com/ubuntu-ports $distro-security main restricted universe multiverse
EOT
    • you may want to customize the above list, depending on your needs. See below for more detail on Ubuntu package feeds
  1. (optional) update package database and setup locales (do not skip this step if you are needing to install any packages for the steps below or otherwise) 
    apt-get update
apt-get -f install # fixup missing package dependencies
apt-get install locales dialog
dpkg-reconfigure locales
  1. Set hostname: 
    echo ${distro}-$(uname -m) > /etc/hostname
  1. set a root passwd so you can login 
    passwd
    • or consider adding a user via adduser: 
      adduser myuser
usermod -a -G tty myuser # add to tty group for tty access
usermod -a -G dialout myuser # add to dialout group for UART access
usermod -a -G sudo myuser # add to sudo group for root access
  1. (optional) configure networking:
    • wired ethernet with DHCP on eth0 
      apt-get install net-tools ifupdown
cat <<EOF >> /etc/network/interfaces
allow-hotplug eth0
auto eth0
iface eth0 inet dhcp

EOF
    • or static IP: 
      apt-get install net-tools ifupdown
cat <<EOF >> /etc/network/interfaces
allow-hotplug eth0
auto eth0
iface eth0 inet static
address 192.168.1.1
netmask 255.255.255.0
gateway 192.168.1.254

EOF
    • or wireless (requires ~3MB of additional packages): 
      apt-get install wpasupplicant iw
cat << EOF >> /etc/network/interfaces
# Wireless interface
auto wlan0
iface wlan0 inet dhcp
        wireless_mode managed
        wireless_essid any
        wpa-driver nl80211
        wpa-conf /etc/wpa_supplicant.conf

EOF
wpa_passphrase <myssid> <mypass> >> /etc/wpa_supplicant.conf
  1. (optional) install some useful packages 
    apt-get install openssh-server # ssh server for remote access
apt-get install can-utils i2c-tools usbutils pciutils # cmdline tools for various hardware support
    • Note that by default root ssh access is disabled for security. See This link for info on enabling it
  1. Exit the chroot shell and remove files we no longer need 
    exit
sudo rm $target/usr/bin/qemu-$qemu_arch-static

At this point you have a directory containing a root filesystem (without kernel) and likely want to install it onto removable storage or the on-board FLASH of a target board. Some intermediate formats that are useful to keep around would be a tarball, perhaps an ext4 filesystem image, or a compressed disk image suitable for flashing in the U-Boot bootloader.

To create a tarball which is the most flexible storage format and can be used for a variety of future installation uses: 

sudo tar --keep-directory-symlink -cvJf focal-venice.tar.xz -C rootfs/ .
  • '--numeric-owner' is required to store user/group as a number instead of a name, your specific use case may require this switch.
  • the '-C rootfs/' is required to eliminate the rootfs directory prefix
  • the sudo is needed to be able to read the root owned files
  • '--keep-directory-symlink' will preserve symbolic links

To create an ext4 filesystem from the directory or tarball requires you choose a size for the filesystem. This size can be increased at runtime using resize2fs as long as the partition table has room for it to grow. The advantage of using an as small as possible size is that the time necessary to flash it onto storage is reduced to a minimum (when flashing you have to write the entire ext4 fs but when formatting or resizing it only has to write periodic markers to FLASH).

For a given size (see SIZEMB variable below) you can create a rootfs with: 

SIZEMB=1536 # 1.5GB - expandable later with resize2fs
OUT=rootfs.ext4
# create a file of specific size
truncate -s ${SIZEMB}M ${OUT}
# format it as an ext4 filesystem
mkfs.ext4 -q -F -L rootfs ${OUT}
# mount it to a temporary mount point
tmp_mnt=$(mktemp -d -p/tmp)
mount ${OUT} ${tmp_mnt}
# copy files to it
cp -rup rootfs/* ${tmp_mnt}
# and/or extract files from a tarball
tar -C ${tmp_mnt} -xf linux-venice.tar.xz --keep-directory-symlink
# unmount temporary mount point
umount ${tmp_mnt}
sync
# compress it
gzip -k -f ${OUT}

Ubuntu Webmin

A web GUI for system management: Ubuntu Webmin

Ubuntu Packages

Ubuntu inherits its package management from Debian Linux using the 'apt' packaging system and 'deb' packages. The list of package feeds is in /etc/apt/sources.list: 

# cat rootfs/etc/apt/sources.list
deb http://ports.ubuntu.com/ubuntu-ports vivid main
deb http://ports.ubuntu.com/ubuntu-ports vivid universe

You can search for Ubuntu packages at ​https://launchpad.net/ubuntu. The search results will show what Ubuntu versions (by name) the package is available in and clicking on the resulting package will show information as to the package feed its contained in.

The standard Ubuntu package feeds are located at ​http://ports.ubuntu.com/ubuntu-ports/ and you will find packages in the dist/<ubuntu-version>/<feed> directories. Ubuntu breaks up feeds into the following:

  • Main - Officially supported software.
  • Restricted - Supported software that is not available under a completely free license.
  • Universe - Community maintained software, i.e. not officially supported software.
  • Multiverse - Software that is not free. (meaning licensing)

If you are trying to find out what package an application belongs to you have a few choices:

  1. Use dpkg -S on your Ubuntu development host. For example to find the package that contains ifconfig: 
    $ dpkg -S $(which ifconfig)
net-tools: /sbin/ifconfig
    • ifconfig is in /sbin/ifconfig and is part of the net-tools package
    • dpkg -L net-tools will show you everything else contained in that package
  2. Use apt-cache search: 
    $ apt-cache search ifconfig
iproute2 - networking and traffic control tools
net-tools - The NET-3 networking toolkit
gnome-nettool - network information tool for GNOME
inetutils-tools - base networking utilities (experimental package)
libnet-ifconfig-wrapper-perl - multiplatform Perl wrapper for ifconfig
moreutils - additional Unix utilities
wmifinfo - Dockapp that shows information for all interfaces
  3. Googling the question 'what package contains <xyz>'

Personal Package Archives (PPAs) are package feeds that are not part of Ubuntu and can be used by people to distribute their own personally built packages. To use a PPA you need to first add it to your repository list and update your package sources.

Some additional useful apt command examples:

  • update package feeds: 
    apt-get update
  • upgrade all installed packages to their latest versions: 
    apt-get upgrade
  • show what version of a package is available (regardless of if its installed), for example bluez: 
    apt-cache show bluez
  • clean out any old packages that have newer versions installed and no longer are dependencies of anything else installed: 
    apt-get autoremove
  • clean out apt caches: 
    apt-get clean
  • remove a package (leaves its configs), for example 'bluez': 
    apt-get remove bluez
  • remove a package including any config files (as they normally don't get removed), for example 'bluez': 
    apt-get purge bluez
  • show all available packages including version info: 
    apt list
  • show all installed packages including version info: 
    apt list --installed
  • show upgradable packages: 
    apt list --upgradable

References:

Gateworks Packages

Gateworks maintains a set of packages that are either custom Gateworks applications or packages that we patch or configure differently than those offered by the official Ubuntu package repositories.

You can find these at ​https://launchpad.net/~gateworks-software/+archive/ubuntu/packages which you can add to your Ubuntu system by:

  • Add Gateworks PPA 
    sudo add-apt-repository ppa:gateworks-software/packages
sudo apt-get update
  • Prioritize Gateworks PPA to have higher priority than others 
    cat <<\EOF > /etc/apt/preferences.d/gateworks
Package: *
pin: release o=LP-PPA-gateworks-software-packages
Pin-Priority: 1010

EOF
    • Note that these steps are already done on the Gateworks prebuilt Ubuntu images

Packages that Gateworks provides in this PPA include:

  • brcm-patchram - a userspace app needed to load firmware onto Broadcom USB chips such as that found on the GW5913
  • gsc-update - a userspace app used to compute checksums of Gateworks System Controller firmware images and update firmware
  • gst-gateworks-apps - Example Gstreamer applications
  • gwsoc - a userspace app used to update firmware and configure GW16113
  • hostapd-conf - a shell script that can be used to interactively create a config file for hostapd for Wireless AP's
  • openocd - an updated build of OpenOCD as they have not had a new release since 0.10 released in Jan 2017
  • openssl - Ubuntu's openssl built with AFALG and Cryptodev support enabled

Modem Support

Aleksander Morgado (​​https://aleksander.es), a key developer behind the ModemManager, libqmi, and libmbim projects that provide modem support on Ubuntu provides up-to-date Ubuntu PPA's for Ubuntu.

See modem#ubuntu for more info on how to use these packages and how to retrieve the latest versions of them from his PPA

SSH Server

The openssh-server package provides an ssh daemon suitable for secure shell (ssh) and secure copy (scp): 

apt-get install openssh-server

During development it may be useful to enable root ssh capability, which is disabled by default. To do this edit /etc/ssh/sshd_config, and:

  1. comment out the following line: 
    PermitRootLogin without-password
  2. Just below it, add the following line: 
    PermitRootLogin yes
  3. Then reload SSH config: 
    service ssh reload

Wireless

It helps to check if your wlan0 device is present before continuing with these steps, this can be done by: 

ls /sys/class/net/

is wlan0 in this directory? If so proceed, if not consult wifi wiki.

Access Point (AP)

These instructions below are for creating a wireless Access Point (AP) that will allow nearby client connections.

  1. Install required packages 
    apt-get update
apt-get install iw hostapd linux-firmware
  2. Download the Gateworks hostapd-conf script from our Yocto BSP, this script will produce a 'basic' configuration file to set up your AP: 
    wget https://raw.githubusercontent.com/Gateworks/meta-gateworks/master/recipes-support/hostapd-conf/hostapd-conf/hostapd-conf
chmod +x ./hostapd-conf
  3. Run the script with your parameters 
    # See usage
./hostapd-conf

# See available channel information for device
./hostapd-conf wlan0 ssid-name # parses information from 'iw list' command

# Enter full AP configuration
./hostapd-conf wlan0 test-ssid 161 VHT80
  4. Run hostapd with your configuration file 
    hostapd -B hostapd-phy0.conf # use -dd flag for more debug output
# "wlan0: interface state UNINITIALIZED->HT_SCAN" shows AP has been set up

Note that the hostapd-phy0.conf file created for you via the hostapd-conf script is meant to be a starting point and does not allow you to create every possible combination of configurations. For more information on the hostapd.conf file and its options, see the ​hostapd.conf documentation.

Client using wpa_supplicant

These instructions below are for creating a wireless client that will connect to a nearby access point.

  1. Install required packages: 
    apt-get update
apt-get install iw wpasupplicant udhcpc
  2. Edit /etc/network/interfaces and edit/add the following to configure wlan0 on boot wpa_supplicant: 
    cat << EOF >> /etc/network/interfaces
# Wireless interface
auto wlan0
iface wlan0 inet dhcp
        wireless_mode managed
        wireless_essid any
        wpa-driver nl80211
        wpa-conf /etc/wpa_supplicant.conf

EOF
  3. Create WPA Supplicant Configuration File at /etc/wpa_supplicant.conf using the wpa_passphrase application (see the man page on wpa_supplicant.conf if you need any special configurations): 
    wpa_passphrase <myssid> <mypass> >> /etc/wpa_supplicant.conf
  4. Sync and power cycle the board 
    sync

Troubleshooting:

  • Make sure you only have one wpa_supplicant running via ps -ef | grep wpa_supplicant
  • To scan for available access points in range you can use the iw tool: 
    iw dev wlan0 scan # scan for AP's
  • To manually connect to wireless if auto connection is not working: While the above will configure wireless Client mode on bootup, if you want to start it manually you can do so by manually starting/re-starting wpa_supplicant (ifup will do this for you on bootup if configured in /etc/network/interfaces): 
    ifconfig wlan0 up
killall wpa_supplicant
wpa_supplicant -i wlan0 -D nl80211 -c /etc/wpa_supplicant.conf -B # manage wireless client
udhcpc -i wlan0 # obtain dhcp lease

Predictable Interface Names (systemd/udev)

The Linux kernel creates Ethernet network interfaces named 'eth<n>' where <n> starts at 0 and increments for each device registered by various drivers. The kernel makes no guarantee that these names remain constant across kernel versions or even power cycles. It is not uncommon for example for USB network interfaces to get registered in an inconsistent fashion between power cycles. It is also not uncommon for a kernel change (ie making a built-in driver be a module instead) to cause interface names move around due to changes in the registration order.

To combat this uncertainty a feature referred to as 'Predictable Interface Names' is implemented in Linux distros which use systemd and udev. Specifically this feature has been in udevd since v197. In this case the network interfaces are re-named based on udev rules according to their bus topology. For example under this scheme a PCI based Ethernet network interface on bus 5, slot 0, function 1 would be names 'enp5s0f1'. The best documentation for the naming scheme appears to be the ​udev code comments

Predictable interface names is enabled by default on systems using systemd/udev however you can disable it in a couple of ways:

  • Pass in a kernel command-line parameter of net.ifnames=0. Note that this is 'not' a kernel feature (thus is independent of what kernel version you are using) but the systemd/udev components that implement this feature look for that on the command-line. (default on Gateworks)
  • Remove /lib/systemd/network/99-default.link from the root filesystem.

Alternatively you can also create your own link files to name interfaces to your liking. A link file defines matching criteria and link criteria for naming. You can use 'udevadm info <device>' to obtain details that udev knows about devices (ie 'udevadm info /sys/class/net/eth0' or /udevadm info /dev/ttyUSB0'). Your link files should be placed in /etc/systemd/network and should sort before 99-default.link from /lib/systemd/network.

An example to force a wlan0 name by MAC address c0:ee:40:99: 

root@gw7301:/lib/systemd/network# cat 90-wlan0.link
[Match]
MACAddress=c0:ee:40:99

[Link]
Name=wlan0

References:

Network Configuration

There are various resources on the Internet dealing with Ubuntu network interface configuration which we don't intend to compete with here.

In general:

  • See /etc/network/interfaces for the default network interface configuration
  • See /etc/dhcp/dhclient.conf for the default DHCP client configuration
  • Typically the Gateworks prebuilt root filesystems will configure networking to bring up eth0 by default using dhcp for network configuration with a 10 second timeout. This is done by setting 'timeout 10;' in /etc/dhcp/dhclient.conf: 
    sed -i 's/^timeout.*/timeout 10;/' /etc/dhcp/dhclient.conf

Note that DHCP only retries every 5 minutes to give an IP address per the man pages: ​https://manpages.ubuntu.com/manpages/trusty/man5/dhclient.conf.5.html

Network Time Protocol (NTP) Configuration

Network Time can set your system time based on network time servers:

  • bionic: 
    apt-get install chrony
  • xenial: 
    apt-get install ntp

Real Time Clock (RTC) Configuration

System Time can be set from a Real Time Clock such as the DS1672 that the Gateworks System Controller emulates but this does not seem to work on modern Ubuntu out of the box. While the util-linux package that provides hwclock installs a service /etc/init.d/hwclock.sh it won't run because systemd is installed and the service is masked.

You can work around this by installing a hwclock.service: 

cat << EOF > /etc/systemd/system/hwclock.service
[Unit]
Description=Hardware clock synchronization
[Service]
Type=oneshot
ExecStart=/sbin/hwclock --hctosys --utc --adjfile=/etc/adjtime
ExecStop=/sbin/hwclock --systohc --utc --adjfile=/etc/adjtime
[Install]
WantedBy=multi-user.target
EOF
systemctl enable hwclock

Watchdog

The standard Linux watchdog daemon can be installed on ubuntu and configured.

Note that these steps are already done for you when using the Gateworks pre-built Ubuntu based images:

  • install package 
    # install
apt-get install watchdog
  • create conf file: 
    cat << EOF > /etc/watchdog.conf
watchdog-device = /dev/watchdog
realtime = yes
priority = 1
interval = 5
watchdog-timeout = 30
EOF

Cross compiling example

If you wish to compile applications on a development host intended to run on a target with a different architecture (ie you have an x86 host, and an arm64 target like Venice or Newport) you can use Ubuntu's cross toolchain which consits of a gcc cross-compiler, libc, and binutils.

Example:

  • cross compile for ARM64 (Venice / Newport) on X86 Ubuntu host: 
    # install Ubuntu cross-build package
sudo apt update
sudo apt install crossbuild-essential-arm64 # installs gcc as well as binutils and libs
aarch64-linux-gnu-gcc -v # see gcc version if interested
# build a simple app, using shared libraries
cat << EOF >> helloworld.c
#include <stdio.h>

int main (int argc, char** argv)
{
   printf("Hello World\n");
   return 0;
}
EOF
aarch64-linux-gnu-gcc helloworld.c -o helloworld
file helloworld # shows ELF 64-bit LSB shared object, ARM aarch64, dynamically linked with interpreter /lib/ld-linux-aarch64.so.1
aarch64-linux-gnu-objdump -x helloworld | grep NEEDED # show libs needed on target
    • this was tested on both an Ubuntu bionic dev host (providing gcc-v7.5) as well as an Ubuntu focal dev host (providing gcc-v9.4.0)
  • cross compile for ARM32 (Ventana) on X86 Ubuntu host: 
    # install Ubuntu cross-build package
sudo apt update
sudo apt install crossbuild-essential-armel # installs gcc as well as binutils and libs
arm-linux-gnueabi-gcc -v # see gcc version if interested
# build a simple app, using shared libraries
cat << EOF >> helloworld.c
#include <stdio.h>

int main (int argc, char** argv)
{
   printf("Hello World\n");
   return 0;
}
EOF
arm-linux-gnueabi-gcc helloworld.c -o helloworld
file helloworld # shows ELF 32-bit LSB ARM EABI5v1 shared object, dynamically linked with interpreter /lib/ld-linux.so.3
arm-linux-gnueabi-objdump -x helloworld | grep NEEDED # show libs needed on target

Native compiling example

While typically not as fast as cross-compiling on a higher end processing Linux development host it can be simple to compile ANSI-C code natively on a Gateworks board running an Ubuntu OS.

To natively compile a .c file on a Newport board using Ubuntu, follow the instructions:

  • First, install build-essential 
    apt-get update
apt-get install build-essential
  • Create your C application with the editor of your choice (vi, vim, nano, sublime etc). Here we just use cat so you can cut-and-paste: 
    #include <stdio.h>

int main(void)
{
     printf("hello, world\n");
}
  • Compile the file: 
    ~# gcc -o hello helloworld.c
  • Run the file 
    ~# ./hello
hello, world

Attachments (2)

Download all attachments as: .zip

Note: See TracWiki for help on using the wiki.