Cubieboard 2 goes into boot loop after upgrading kernel

[phutidus]

I have installed Armbian on my Cubieboard 2 and Banana Pi, and a few hours ago for some reason I want to upgrade the kernel from 4.1.2 to 4.1.6 on both of them, then the story begins...

 

I.

 

First, below is a list of partitions on Banana Pi (same on Cubieboard 2)

(notice that I attached a SATA hard drive on it, and created 2 primary partitions)

➜  ~  df -h
Filesystem       Size  Used Avail Use% Mounted on
/dev/sda1        ~~~G  ~~~G  ~~~G  16% /          (where system actually resides)
/dev/sda2        ~~~G  ~~~G  ~~~G  81% /data      (data storage)
/media/mmc/boot  7.3G  813M  6.2G  12% /boot      (SD card, used for booting) 

I installed my system on /dev/sda1 using

/root/nand-sata-install

To upgrade kernel, I followed the instruction on website, i.e., run the following command as root:

wget -q -O - http://upgrade.armbian.com | bash

After the upgrade completed, I immediately (didn't reboot) used apt-get to upgrade packages 

apt-get update && apt-get dist-upgrade 

List of installed packages

# installed when I run wget -q -O - http://upgrade.armbian.com | bash
2015-09-15 17:12:24 install libboost-iostreams1.54.0:armhf <none> 1.54.0-4ubuntu3.1
2015-09-15 17:12:25 install libept1.4.12:armhf <none> 1.0.12
2015-09-15 17:12:26 install libsigc++-2.0-0c2a:armhf <none> 2.2.10-0.2ubuntu2
2015-09-15 17:12:26 install aptitude-common:all <none> 0.6.8.2-1ubuntu4
2015-09-15 17:12:28 install libcwidget3:armhf <none> 0.5.16-3.5ubuntu1
2015-09-15 17:12:29 install libxapian22:armhf <none> 1.2.16-2ubuntu1
2015-09-15 17:12:30 install aptitude:armhf <none> 0.6.8.2-1ubuntu4
# installed when I run apt-get update && apt-get dist-upgrade
2015-09-15 17:46:07 install linux-image-next-sunxi:armhf <none> 4.2
2015-09-15 17:48:59 install linux-firmware-image-next-sunxi:armhf <none> 4.2
2015-09-15 17:49:00 install linux-headers-next-sunxi:armhf <none> 4.2
2015-09-15 17:49:21 install linux-u-boot-bananapi:armhf <none> 4.2
2015-09-15 17:49:36 install linux-dtb-next-sunxi:armhf <none> 4.2
2015-09-15 17:49:37 install linux-trusty-root-next-bananapi:armhf <none> 4.2 

II.

 

Then I reboot. Banana Pi goes into a infinite boot loop (Cubieboard 2 also goes into the same loop after upgrading)

first, the screen shows some message:

U-Boot 2015.07-dirty (Aug 25 2015 #$@%$^%)
(...some text I think not related to the issue...) 

and the screen flashed into next screen, again with some message:

Kernel panic - not syncing: No working init found. Try passing init= option to kernel. See Linux Documentation/init.txt for guidance.

(...messages...)

Rebooting in 10 secconds.. 

then the first screen, second, first, second, first, second...

 

 

After some investigation, I found that /boot/boot.cmd seems to be the culprit of the boot loop, the first line in the file gave the wrong root partition:

setenv bootargs console=tty1 root=/dev/sda2 rootwait rootfstype=ext4 ......

It should be /dev/sda1, so I corrected that to /dev/sda1 and run the following command, then reboot, everything goes fine.

mkimage -C none -A arm -T script -d /boot/boot.cmd /boot/boot.scr

I know how to fix it, but has anyone find out the root cause of this issue?

Thanks.

[Igor]

I guess you hit the bug of the upgrade script. I made some updates today but not in this area. I guess I need one more streak to fix those problems. 

[phutidus]

I guess you hit the bug of the upgrade script. I made some updates today but not in this area. I guess I need one more streak to fix those problems. 

 

This function in script downloaded from http://upgrade.armbian.com seems to find the wrong root partition

get_root_device ()
{
rootdevice="/dev/"$(lsblk -idn -o NAME | grep $1)
partitions=$(($(fdisk -l $rootdevice | grep $rootdevice | wc -l)-1))
rootdevice="/dev/"$(lsblk -idn -o NAME | grep $1)$2$partitions
}

If I'm right, the /root/nand-sata-install script always install Armbian to /dev/sda1 if user decides to install to SATA device, while this function have the last partition in SATA drive as the root partition (in my case /dev/sda2).

[Igor]

I prepared script that is functional if we have more than one SD card partition. If you have some 3rd party installation.

 

That's the reason of this functions. But the usage might be false. I'll look into it and fix.

 

Thanks for detailed report. It helps!

[Igor]

Now this part is fixed + I found and fix some other case bug. (booting from NAND)

 

I run an upgrade over one of my production server today & upgrade succeeded.