Install ReDroid for playing Android games in Docker containers on Linux PC, with GPU acceleration

中文版

ReDroid (Remote anDroid) is an open source and self-hosting “cloud phone” solution. By running Android in containers on Linux PC, we could access Android desktop by using Scrcpy. And the container could be accessed remotely.

Here is my demonstration of what “self-hosting cloud Android phone” looks like. In this video, I install Scrcpy client on iPhone and connect to Linux PC using Zerotier to play Android games remotely. (sorry my video is in Chinese)

ReDroid is also one of the few solutions to run Android APP on Linux with open source software. Many Android simulators, not to mention cloud phones, are closed source. In contrast, ReDroid is an open source solution except for the ARM translator. Even better, ReDroid supports GPU acceleration plus ARM translations, so you can play more mobile 3D games!

ReDroid is much light-weight and easy to deploy than Android-x86 VM.

If you ask me, there is already a similar, container-based solution, which is known as Waydroid for playing Android games. Why do I choose ReDroid over Waydroid? Well freak Nvidia. Waydroid does not work on Nvidia GPU.

In this article, I will discuss how to use ReDroid to play mobile games on Linux PC. We will setup a ReDroid container, and add ARM translator + Google Service Framework to the ReDroid image for the best experience.

1. Requirements

#

Any Linux distribution should be fine.

It is recommended to use a Linux PC with x86 architecture to run Redroid. The ARM architecture support is not very good.

If you have a x86 PC, it can only run x86 architecture Android APPs. However, many mobile games only have ARM architecture versions, so you needs to install the ARM translator (libndk or libhoudini) in the ReDroid container.

To play mobile games, it is recommended that the computer has at least 8GB of RAM, because sometimes ARM takes up a lot of RAM when translating it into x86 instructions.

For GPU acceleration, it is recommended to use Intel or AMD GPUs. 3D acceleration is available out of the box. Nvidia is not recommended.

Regarding the Android version, the redroid:11.0.0-latest and redroid:12.0.0-latest image files released by the Redroid developer have built-in ARM translator libndk developed by Google. I have tried only Android 11 which is relatively stable. GApps It can also be used, so in this article I will uses the Android 11 image .

2. Install dependencies

#

ReDroid’s Github has installation instructions for other Linux distributions. I used the Ubuntu.

1. First, prepare the binder module. On Ubuntu 24.04 execute the following commands to install the kernel modules:

sudo apt install linux-modules-extra-`uname -r`

sudo modprobe binder_linux devices="binder,hwbinder,vndbinder"

2. Make the modules load automatically at boot

sudo cat <<EOT >> /etc/modules-load.d/redroid.conf
binder_linux
options binder_linux devices="binder,hwbinder,vndbinder"
EOT

3. Install Docker

4. Install the ADB tool

5. Finally install Scrcpy.

If you want to integrate the graphical Scrcpy interface with key mapping, you can modify QtScrcpy. You can also try the web version of ws-scrcpy.

3. Pre-install GApps on the RedDroid image

#

The image released by the developer of ReDroid is native Android and does not have GApps pre-installed. We can use ayasa520’s Remote-Android Script script to automatically pull the Redroid image and install GApps on it.

This script can also be used to install libndk, libhoudini, Magisk, Widevine DRM and other components.

1. Clone the Remote-Android Script repository and create a Python virtual environment

sudo apt install lzip python3 python3-venv python3-pip

git clone https://github.com/ayasa520/redroid-script.git

cd redroid-script

python3 -m venv venv

venv/bin/pip install -r requirements.txt

2. Pull the Android 11 image file install GApps

venv/bin/python3 redroid.py -a 11.0.0 -g

3. This will build the image redroid/redroid:11.0.0_gapps containing GApps.

4. Start the Redroid container with docker-compose

#

1. Create a virtual machine running Redroid (this is only required for Nvidia GPU). Since the Nvidia closed-source driver cannot allow Redroid to use GPU acceleration, you need to run a QEMU virtual machine and install Redroid in it to achieve hardware acceleration through virtio-gpu. Otherwise, Redroid will use software rendering.

qemu-img create -f qcow2 ubuntu.qcow2 64GB

qemu-system-x86_64 -boot d -cdrom "ubuntu-22.04.1-desktop-amd64.iso" -enable-kvm -smp 4 -device intel-hda -device hda-duplex -device virtio-vga-gl -net nic -net user,hostfwd=tcp::5555-:5555 -cpu host -m 4096 -display sdl,gl=on -hda ubuntu.qcow2

qemu-system-x86_64 -enable-kvm -smp 4 -device intel-hda -device hda-duplex -device virtio-vga-gl -net nic -net user,hostfwd=tcp::5555-:5555 -cpu host -m 4096 -display sdl,gl=on -hda ubuntu.qcow2

2. Create a directory to store Android data and create a docker-compose

mkdir ~/redroid
cd redroid
vim docker-compose.yml

3. Fill in the following content. We will use the GApps image that created before.

version: "3"
services:
 redroid:
 image: redroid/redroid:11.0.0_gapps
 stdin_open: true
 tty: true
 privileged: true
 ports:
 - "5555:5555"
 volumes:
 # The data is stored in the current directory
 - ./redroid-11-data:/data
 command:
 # Android resolution
 - androidboot.redroid_width=720
 - androidboot.redroid_height=1280
 - androidboot.redroid_dpi=320
 # Enable host GPU hardware acceleration
 - androidboot.redroid_gpu_mode=host
 # Set libndk parameters
 - ro.product.cpu.abilist0=x86_64,arm64-v8a,x86,armeabi-v7a,armeabi
 - ro.product.cpu.abilist64=x86_64,arm64-v8a
 - ro.product.cpu.abilist32=x86,armeabi-v7a,armeabi
 -ro.dalvik.vm.isa.arm=x86
 - ro.dalvik.vm.isa.arm64=x86_64
 - ro.enable.native.bridge.exec=1
 - ro.dalvik.vm.native.bridge=libndk_translation.so
 - ro.ndk_translation.version=0.2.2

4. Start the container

sudo docker compose up -d

5. Use ADB to connect to the local ReDroid:

adb connect localhost:5555

# If the connection cannot be made, use the following command to see what is going on inside the container.
sudo docker exec <container ID> logcat
sudo docker logs <container ID>

6. Execute Scrcpy and connect to Android:

scrcpy -s localhost:5555

7. You shall see the Android desktop.

   

8. Google Play may display “Device not verified”.

9. Execute the following commands to obtain the Android device ID, go to Google website to register the device, wait 30 minutes and then restart the Redroid container. Then you can log in to Google Play.

adb -s localhost:5555 root

adb -s localhost:5555 shell 'sqlite3 /data/data/com.google.android.gsf/databases/gservices.db \
 "select * from main where name = \"android_id\";"'

5. How to install APK on ReDroid

#

Currently, even if libndk is installed, the Android 11 Play Store still does not allow downloading of ARM-based APPs. Please use other app stores such as APKPure to install the APP.

In addition to downloading the APK using the browser inside the container, you can also use ADB to install the APK into the Redroid container. For example, download Line’s APK from ApkMirror, and then install it with ADB:

adb -s localhost:5555 install "jp.naver.line.android.apk"

You can also use ADB’s pull and push commands to transfer files.

For screen resolution adjustment, please refer to the instructions of ReDroid and Scrcpy.

6. How to “turn on and off” ReDroid

#

If you want to shut down Redroid, close the Scrcpy window and stop the container:

cd ~/redroid

sudo docker compose down

You can use this command to start Redroid later. The /data data of the ReDroid container is located in the ~/redroid/redroid-11-data directory, which can be used to back up files of multiple ReDroid containers.

cd ~/redroid

sudo docker compose up -d

adb connect localhost:5555

scrcpy -s localhost:5555

References

#

• redroid (Remote-Android) is a multi-arch, GPU enabled, Android in Cloud solution - Github
• App 自動化測試（三）ReDroid 安裝與基本使用 - Scott Hsiao
• 在使用 NVIDIA 显卡时为 redroid 开启3d加速
• redroid nVidia GPU support #282
• Install GApps Manually - Google Groups