Here is a quick overview of the process to create a Ubuntu system image that is bootable with Android chroot. It uses the rootstock utility to setup the initial image, including release version, username/password, image size, as well as to seed the desired packages to be pre-installed with the distro. For complete options with rootstock, consult the man pages.

Quickstart:

mkdir ubuntu
cd ubuntu
sudo rootstock \
--fqdn ubuntu \
--login ubuntu \
--password ubuntu \
--imagesize 4G \
--dist maverick \
--seed linux-image-omap,build-essential,mysql-server,tightvncserver,lxde,\
mysql-server-core-5.5,mysql-server-5.5,libmysqlclient16,mysql-common,\
mysql-client-core-5.5
dd if=/dev/zero of=ubuntu.img bs=1MB count=0 seek=4096
mke2fs -F ubuntu.img
mkdir ubuntumnt
sudo mount -o loop ubuntu.img ubuntumnt
sudo tar -C ubuntumnt -zxf armel-rootfs-XXXXXXXXXXXX.tgz
sudo umount ubuntumnt
sudo rm armel-rootfs-XXXXXXXXXXXX.tgz
sudo rm -rf ubuntumnt
cd ..
adb push ubuntu/ /sdcard/ubuntu

NOTE: XXXXXXXXXXXX is the timestamp for the file creation, for example: armel-rootfs-201203201350.tgz

Steps explained:

First, make a directory to store the ubuntu image and scripts

mkdir ubuntu
cd ubuntu

Next, execute rootstock as root. The options can be configured to create the image of your choosing. For example, if you wont be using a GUI in the chroot, you may want to omit the ‘lxde’ and ‘tightvncserver’ packages. You can also modify the image size if you desire as well, but remember that the maximum filesize on a FAT32 filesystem is 4GB (and your sdcard is very likely formatted as FAT32).

sudo rootstock \
--fqdn ubuntu \
--login ubuntu \
--password ubuntu \
--imagesize 4G \
--dist oneiric \
--seed linux-image-omap,build-essential,mysql-server,tightvncserver,lxde,\
mysql-server-core-5.5,mysql-server-5.5,libmysqlclient16,mysql-common,\
mysql-client-core-5.5

Now we need to create a blank filesystem image to extract the rootstock onto. For this task, we use the dd command. Remember to set the seek paramater to match the imagesize you created in the previous step.

dd if=/dev/zero of=ubuntu.img bs=1MB count=0 seek=4096

Use mke2fs to create a new filesystem on the created image file.

mke2fs -F ubuntu.img

Make a directory to serve as a mountpoint for the ubuntu image.

mkdir ubuntumnt

Next we need access to the filesystem we just created. This is accomplished by mounting the image file as a disk using the loopback device.

sudo mount -o loop ubuntu.img ubuntumnt

Now we use tar to extract the contents of the created ARM root filesystem to the image mounted on the loopback. Replace XXXXXXXXXXXX with the timestamp that was created with rootstock. E.g. armel-rootfs-201104151837.tgz

sudo tar -C ubuntumnt -zxf armel-rootfs-XXXXXXXXXXXX.tgz

Now that were finished with the extraction, we can unmount the system image.

sudo umount ubuntumnt

Were now finished with the desktop portion of the install, so can safely remove the tar of the image, and its mountpoint.

sudo rm armel-rootfs-XXXXXXXXXXXX.tgz
sudo rm -rf ubuntumnt

Switch to the parent directory, then use ADB to push the contents of the chroot install to the sdcard.

cd ..
adb push ubuntu/ /sdcard/ubuntu

Clone the scripts from my github repo:

git clone git://github.com/mitchtech/chroot_android.git -b ubuntu
cd chroot_android
tar -cvf ubuntu.tar *
./adb push ubuntu.tar /sdcard/ubuntu/

ADB shell into the device

./adb shell

Get root and change into the Ubuntu directory

su
cd /sdcard/ubuntu

Uncompress the image and scripts:

tar -xvf ubuntu.tar

Next run the installer script.

sh ./installubuntu.sh

Now, to start Ubuntu type ‘startubuntu’. Once Ubuntu started, to gain shell type ‘ubuntu’. To shutdown type ‘stopubuntu’.

startubuntu
ubuntu

If all goes well, you’ll be in the Ubuntu chroot. If you get

root@localhost:

then you know it’s working!


comments powered by Disqus