Start the Fuchsia emulator

This guide provides instructions on how to set up and launch the Fuchsia emulator (FEMU) on your machine.

The steps are:

  1. Prerequisites.
  2. Build Fuchsia for FEMU.
  3. Enable VM acceleration (Optional).
  4. Start FEMU.
  5. Discover FEMU.

1. Prerequisites

Running FEMU requires that you've completed the following guides:

2. Build Fuchsia for FEMU

To run FEMU, you first need to build a Fuchsia system image that supports the emulator environment. This guide uses x64 for the board and workbench_eng for the product as an example.

To build a FEMU Fuchsia image, do the following:

  1. Set the Fuchsia build configuration:

    fx set workbench_eng.x64 --release
    
  2. Build Fuchsia:

    fx build
    

For more information on supported boards and products, see the Fuchsia emulator (FEMU) overview page.

3. Enable VM acceleration (Optional)

(Linux only) Most Linux machines support VM acceleration through KVM, which greatly improves the performance and usability of the emulator.

  1. If KVM is available on your machine, add yourself to the kvm group on your machine:

    sudo usermod -a -G kvm ${USER}
    
  2. Log out of all desktop sessions to your machine and then log in again.

  3. To verify that KVM is configured correctly, run the following command:

    if [[ -r /dev/kvm ]] && grep '^flags' /proc/cpuinfo | grep -qE 'vmx|svm'; then echo 'KVM is working'; else echo 'KVM not working'; fi
    

    Verify that this command prints the following line:

    KVM is working
    

    If you see KVM not working, you may need to reboot your machine for the permission change to take effect.

4. Start FEMU

Start the package server

Prior to starting the emulator, start the package server.

To start the package server, run the following command:

fx serve

Once this command is run, it does not terminate and keeps the Fuchsia package server running on the host machine (unless the terminal is closed or Ctrl+C is pressed on the terminal, which kills the command).

Alternatively you can background the fx serve process.

Start the emulator

Start the Fuchsia emulator on your Linux machine.

If your machine is behind a firewall, you may need to apply some additional configuration to allow the emulator to access the network. This is typically accomplished by running an "upscript", which sets up the interfaces and firewall access rules for the current process. If you're on a corporate network, check with your internal networking team to see if they have an existing upscript for you to use.

However, even if you're not behind a firewall, there's still some configuration needed to enable tun/tap networking. The example upscript at FUCHSIA_ROOT/scripts/start-unsecure-internet.sh should work for the majority of non-corporate users.

To start the emulator, do the following:

  1. Start a new terminal.

  2. To configure the upscript, run the following command:

    ffx config set emu.upscript FUCHSIA_ROOT/scripts/start-unsecure-internet.sh
    
    • start-unsecure-internet.sh is an example upscript.
    • FUCHSIA_ROOT is the path to your Fuchsia directory.
  3. Start the emulator with access to external networks:

    ffx emu start --net tap
    

    --net specifies the networking mode for the emulator. --net tap attaches to a Tun/Tap interface.

    Or, you can start the emulator without access to external networks:

    ffx emu start --net none
    

    Starting the emulator opens a new window with the title Fuchsia Emulator.

5. Discover FEMU

To discover the Fuchsia emulator as a running Fuchsia device, run the following command:

ffx target list

This command prints output similar to the following:

$ ffx target list
NAME                SERIAL       TYPE                 STATE      ADDRS/IP                            RCS
fuchsia-emulator    <unknown>    workbench_eng.x64    Product    [fe80::866a:a5ea:cd9e:69f6%qemu]    N

fuchsia-emulator is the default node name of the Fuchsia emulator.

The output of ffx target list is influenced by the --net option in the following ways:

  • --net none disables networking, which causes the device to be not discoverable by ffx target list.

  • --net tap and --net user allow the device to be discoverable when running ffx target list.

Add target manually

If no target is found after running ffx target list, the target may need to be added manually by running ffx target add:

ffx target add device-ip:device-port

Next steps

To learn more about Fuchsia device commands and Fuchsia workflows, see Explore Fuchsia.

Appendices

This section provides additional FEMU options.

See all available flags

To see a full list of the emulator's supported flags, run the following command:

ffx emu start --help

Run FEMU without GUI support

If you don't need graphics or working under the remote workflow, you can run FEMU in headless mode:

ffx emu start --headless

Reboot FEMU

To reboot FEMU, run the following ffx command:

ffx target reboot

Stop FEMU

To stop FEMU, run the following ffx command:

ffx emu stop

Configure IPv6 network

This section provides instructions on how to configure an IPv6 network for FEMU on Linux machine using TUN/TAP.

To enable networking in FEMU using tap networking, do the following:

  1. Set up tuntap:

    sudo ip tuntap add dev qemu mode tap user $USER
    
  2. Enable the network for qemu:

    sudo ip link set qemu up
    

Specify GPU mode for FEMU (Experimental)

By default, the FEMU launcher uses SwiftShader Vulkan ICD for host graphics rendering. With the --gpu flag, you can set the emulator to use the host GPU hardware for rendering. See the following options:

GPU emulation method Description Flag
SwiftShader Use SwiftShader libraries to simulate GPU processing. This is the default mode. ffx emu start --gpu swiftshader_indirect
Hardware (host GPU) Use the host machine's GPU directly to perform GPU processing. ffx emu start --gpu host
Auto Resolve to host if there is a hardware GPU available or swiftshader_indirect if there isn't a hardware GPU available. ffx emu start --gpu auto

The host and auto GPU emulation modes are for experimental use only and are not officially supported at the moment. You may see graphics artifacts, testing failures or emulator crashes when using these two modes.