Jump to content

BananaPi M1 stuck in boot loop after apt-get upgrade


Tomas Rollo

Recommended Posts

Hello,

 

I run Armbian on Banapi M1 (use it as samba file server with OpenMediaVault) and recently after apt-get upgrade and reboot the Bananapi is stuck in boot loop.

 

I have tried the following to eliminate possibility that it's a HW failure:

  1. booted into fresh Armbian image with different SD card: success -> the board is OK
  2. dumped whole Armbian installation as image from the original SD card, overwrote it with fresh Armbian image and booted with it: success -> the SD card is OK
  3. wrote the dumped Armbian installation image onto a different SD card and booted from it: stuck in boot loop
  4. wrote the dumped Armbian installation image onto a second different SD card and booted from it: stuck in boot loop

thus I believe the problem must be in the Armbian installation itself, not in HW

 

I don't know how to get hold of the boot messages since I never get to the login prompt but I've at least filmed the whole boot on the monitor.

 

The system always reboots right after the following lines:

[    4.681580] systemd-udevd[144]: starting version 215
[    5.847373] device-mapper: uevent: version 1.0.3
[    5.849275] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com

Since I have dump of the SD card I mounted the image on a different machine to have access to the filesystem. I reconciled the filmed boot loop with the armhwinfo.log content of previous boots and it seems the following lines are missing compared to successful boots:

...
[    4.237054] mmc0: host does not support reading read-only switch, assuming write-enable
[    4.239588] mmc0: new high speed SDHC card at address b368
[    4.240515] mmcblk0: mmc0:b368 SD    7.52 GiB 
[    4.242469]  mmcblk0: p1
...

and the system never reaches the following point:

...
[    4.619783] Freeing unused kernel memory: 1024K (c0e00000 - c0f00000)
[    4.681580] systemd-udevd[144]: starting version 215
[    5.847373] device-mapper: uevent: version 1.0.3
[    5.849275] device-mapper: ioctl: 4.35.0-ioctl (2016-06-23) initialised: dm-devel@redhat.com
<--- this is the last line, the system reboots right after here after a while
[    6.346419] EXT4-fs (mmcblk0p1): mounted filesystem with writeback data mode. Opts: (null)
<--- assuming this means the root filesystem is mounted, the system never reaches here
[    6.864213] systemd[1]: systemd 215 running in system mode. (+PAM +AUDIT +SELINUX +IMA +SYSVINIT +LIBCRYPTSETUP +GCRYPT +ACL +XZ -SECCOMP -APPARMOR)
[    6.864741] systemd[1]: Detected architecture 'arm'.
[    6.912254] systemd[1]: Set hostname to <bananapi>.
...

What I conclude out of this is that the system has trouble remounting the root filesystem on the SD card (though the SD card is physically OK and the system at least can read the /boot part out of the SD).

 

Unfortunately I have no clue where to look to see why the root filesystem can't be mounted :(

 

Could anyone point me in the right direction? I tried looking into things like apt-get history log and there are several core looking packages that got upgraded recently (like linux-dtb-next-sunxi, linux-u-boot-bananapi-next, others...) but I don't know which ones could cause such problems.

 

Thanks for any hints!

T.

Link to comment
Share on other sites

Hm, I investigated the boot logs further and found two more lines that were in the boot messages in the past but are not contained in the boot-loop-stuck-image boot process I filmed:

[    4.133385] sunxi-mmc 1c0f000.mmc: Got CD GPIO
[    4.189763] sunxi-mmc 1c0f000.mmc: base:0xf08fa000 irq:28

This leads me to belief that there's something wrong with the mmc driver in the kernel (kernel/drivers/mmc/host/sunxi-mmc.ko?) -> the system can't find the mmc0 device -> causes kernel panic later on during the udev part of booting (see above).

 

The question is what could cause such sudden problem when during previous boots the system was working without a problem?

Link to comment
Share on other sites

Anyway, seems to me it would be best just to reinstall Armbian with the latest version - is there a safe way to take a fresh Armbian install and move data, files & config from the old image? Or should I just take the /boot directory from the new fresh install and copy it over to the old image? Are there any dependencies between the content of the /boot dir and the rest of the filesystem that could have changed between Armbian 4.8.4/v5.23 (the old image) and 4.9.7/v5.25 (the fresh verrsion)?

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...

Important Information

Terms of Use - Privacy Policy - Guidelines