jumpnow / bbb-upgrader Goto Github PK

Some scripts to support upgrading BeagleBone Black systems.

Additional documentation can be found here.

I'm only testing with Yocto built BBB systems

• that are not using an initrd
• that have a 4GB eMMC (Rev. C boards)

The interesting scripts are

• emmc_uEnv.txt
• bbb_upgrade.sh
• manage_boot_flag_partition.sh

Because the eMMC is not accessible externally, the remaining scripts are here to support the initial system install to an SD card and then the initial eMMC install from the booted SD card system.

Just kidding...

You have to build some binaries for the BBB with Yocto first.

The files you need are

• MLO-beaglebone
• u-boot-beaglebone.img
• <some-image-name>-image-beaglebone.tar.xz

If you already have them, then you are ready to get started.

If not, there are some instructions here.

The scripts can handle the binaries in the same location where they are run or you can export an environment variable called OETMP that points to your bitbake TMPDIR.

The files will be searched for under

${OETMP}/deploy/images/beaglebone/

That assumes you haven't changed the bitbake DEPLOY_DIR. If you have, then figure out yourself how to specify OETMP or just modify the copy_ scripts. They are not difficult.

You will also need the name of the dtb you want the kernel to use.

The normal procedure is to include it in your kernel recipe so that the Yocto kernel scripts will have built and installed in onto the rootfs for you.

Update the fdtfile variable in uEnv.txt and emmc_uEnv.txt with the dtb you want to use instead of the default bbb-nohdmi.dtb that I use for testing.

Assuming your SD card shows up at /dev/sdb run the following

sudo ./mk2parts sdb
./copy_boot.sh sdb
./copy_rootfs.sh sdb <some-image-name>	
./copy_emmc_install.sh sdb <some-image-name> [<hostname>]

Put the card in the BBB and boot the system holding down S2.

From a shell session on the BBB that is now running off the SD card

root@beaglebone:~# cd emmc
root@beaglebone:~/emmc# ./emmc_install.sh <some-image-name> [<hostname>]	

When that completes, remove the SD card and boot the eMMC system.

The system is now ready for the bbb_upgrade.sh script.

The eMMC was partitioned as follows by the emmc_mk5parts.sh script run by the emmc_install.sh script

Device         Boot   Start     End Sectors  Size Id Type
/dev/mmcblk0p1 *        128  131199  131072   64M  c W95 FAT32 (LBA)
/dev/mmcblk0p2       133120 2230271 2097152    1G 83 Linux
/dev/mmcblk0p3      2230272 4327423 2097152    1G 83 Linux
/dev/mmcblk0p4      4327424 7667711 3340288  1.6G  5 Extended
/dev/mmcblk0p5      4329472 4460543  131072   64M  c W95 FAT32 (LBA)
/dev/mmcblk0p6      4462592 7667711 3205120  1.5G 83 Linux

The current root partition is /dev/mmcblk0p2.

The bbb_upgrade.sh script alternately uses /dev/mmcblk0p2 and /dev/mmcblk0p3 as the new rootfs partition when it performs an upgrade.

It chooses the new rootfs partition based on whichever partition is not the current rootfs.

The /dev/mmcblk0p5 partition is used by the uEnv.txt u-boot script to read and write some flag files to determine which partition to load as the rootfs.

manage_boot_flag_partition.sh is the script that confirms the new rootfs actually worked. It does this by modifying files on /dev/mmcblk0p5.

If this script isn't run the first time after booting a new rootfs following an upgrade, the system will revert back to the old rootfs on the next boot.

The manage_boot_flag_partition.sh script is safe to run anytime, it does nothing if there is nothing to do. My thinking is it would be added to the scripts run by the system init process, whatever type of init you are running.

The bbb_upgrade.sh is responsible for formatting /dev/mmcblk0p5 and populating the initial flag files.

A location is needed for the new image tarball, so make use of /dev/mmcblk0p6

mkfs.ext4 -q -L DATA /dev/mmcblk0p6
mkdir /data
echo "/dev/mmcblk0p6 /data auto defaults 0 0" > /etc/fstab
mount /dev/mmcblk0p6

Copy the tarball image and the bbb_upgrade.sh and manage_boot_flag_partition.sh script to the new /data directory on the BBB.

For example

scp console-image-beaglebone.tar.xz bbb_upgrade.sh manage_boot_flag_partition.sh root@<bbb-ip>:/data

Then on the BBB run bbb_upgrade.sh like this

root@bbb:~# cd /data
root@bbb:/data# ./bbb_upgrade.sh console-image-beaglebone.tar.xz
Checking for an eMMC : OK
Checking that there is no SD card : OK
Finding the current root partition : /dev/mmcblk0p2
The new root will be : /dev/mmcblk0p3
Checking the new root partition size : OK
Checking for a /dev/mmcblk0p5 partition : OK
Checking the /dev/mmcblk0p5 flag partition size : OK
Check that /dev/mmcblk0p5 is not in use : OK
Checking for a /dev/mmcblk0p6 partition : OK
Checking that /mnt is not in use : OK
Formatting partition /dev/mmcblk0p3 as ext4 : OK
Mounting /dev/mmcblk0p3 on /mnt : OK
Extracting new root filesystem console-image-beaglebone.tar.xz to /mnt : OK
Copying config files from current system : OK
Unmounting /dev/mmcblk0p3 : OK
Formatting the flag partition /dev/mmcblk0p5 :
mkfs.fat 3.0.28 (2015-05-16)
OK
Mounting the flag partition /dev/mmcblk0p5 on /mnt : OK
Creating file '/mnt/three' : OK
Unmounting /dev/mmcblk0p5 from /mnt : OK

A new system was installed onto : /dev/mmcblk0p3

Reboot to use the new system.
root@bbb:/data#

After rebooting note the new rootfs

root@bbb:~# cat /proc/cmdline
console=ttyO0,115200n8 consoleblank=0 root=/dev/mmcblk0p3 ro rootfstype=ext4 rootwait

root@bbb:~# mount | grep mmcblk0
/dev/mmcblk0p3 on / type ext4 (rw,relatime,data=ordered)
/dev/mmcblk0p6 on /data type ext4 (rw,relatime,data=ordered)

Now would be the time to run manage_boot_flag_partition.sh or the next reboot will revert back to /dev/mmcblk0p2.

The upgrade script copied over the old /etc/fstab and a /data mount point for /dev/mmcblk0p6 so /data is still there

root@bbb:~# cd /data
root@bbb:/data# ./manage_boot_flag_partition.sh

Checking for an eMMC : OK
Checking that there is no SD card : OK
Finding the current root partition : /dev/mmcblk0p3
Checking there is a /dev/mmcblk0p5 partition : OK
Checking that /dev/mmcblk0p5 is not in use : OK
Checking that /mnt is not in use : OK
Mounting /dev/mmcblk0p5 on /mnt : OK
Updating flag files on /dev/mmcblk0p5 : OK
Unmounting /dev/mmcblk0p5 : OK
root@bbb:/data#

There is still script cleanup and simplification to be done. And more testing.

But the basic framework is in place and seems to be working.