OpenSUSE tumbleweed Update kernel 4.6.0 caused by failed boot system

Updated Linux kernel to the 4.6.0-1.1 version of the OpenSUSE Tumbleweed routine upgrade yesterday. Kernel panic is encountered with the new kernel during reboot and cannot enter the login interface. Choose to use the old version of 4.5.4-1-default in Grub to start and log into the system successfully. This gave me a bit of a sense of crisis.

Because recently very busy, no time to toss, and the computer is the main productivity tools, can not afford to toss, so can not help but blind to some countermeasures. First the temporary record, back slowly try and modify the supplement.

Btrfs's snapshot¶

Give up first, no time to toss.

OpenSUSE has already started promoting Btrfs. Btrfs's snapshot is a file system thermal mirroring feature that produces a snapshot that restores the file system to the mirror when a problem is encountered-a bit like an RPG game archive. It's like a time machine. If you do a large change before each of these mirrors, with a few days after no problem to delete the mirror, there are problems to recover, it should be very convenient.

Originally Btrfs Snapshot is a good tool to prevent similar problems, unfortunately I did not use Btrfs but chose Ext3 as the root partition file system. Originally Btrfs is the default, you say why I want to change the hand cheap.

EXT3 can be converted to Btrfs, but the partition to be converted needs to be uninstalled before conversion. Not so convenient for the system partition (/) (example). First you need to reboot the system into the recovery mode (for example, start with a Live CD) before you can convert it.

Keep multiple versions of the kernel ¶

OpenSUSE By default, the boot manager (which I use GRUB2) retains the last version and the previous version, as well as the running kernel startup items, to address similar problems when upgrading kernel. This is the multi-version kernel system. Select Advance options for OpenSUSE tumbleweed at startup.

To prevent the next time you accidentally remove the running 4.5.4-1-default, specifically specify that you want to keep the boot entry for this kernel.

Edit/etc/zypp/zypp.conf with root, confirming that multiversion = Provides:multiversion (kernel) is not commented out, and then

Multiversion.kernels = latest,latest-1,running
followed by 4.5.4-1-default into,

Multiversion.kernels = Latest,latest-1,running,4.5.4-1-default
Save!

If you are really accidentally cleared, you may also be able to install the old version of the kernel by itself. Never tried!

Use the old startup item temporarily by default ¶

Don't get it, it doesn't feel necessary, it's usually dormant rather than shut down. The reference link is reserved.

About Virtualbox¶

After such a toss, VirtualBox could not start the virtual machine. Error

Kernel driver not installed (rc=-1908)

The VirtualBox Linux kernel driver (vboxdrv) is either isn't loaded or there is a permission problem with/dev/vboxdrv. Please reinstall the kernel module by executing

'/sbin/rcvboxdrv Setup '

As Root. If It is available in your distribution, you should install dkms the package. This package keeps track of Linux kernel changes and recompiles the Vboxdrv kernel module if necessary.

Where:suplibosinit what:3 verr_vm_driver_not_installed (-1908) –the support DRIVER is not installed. On Linux, Open returned enoent.

All right, follow the prompts and run.

# sudo rcvboxdrv setup
recompiling VirtualBox kernel module, not. It has been packaged. Done
As a result, you cannot start any virtual machines.

And then query the status of the Vboxdrv service,

# sudo systemctl status vboxdrv
Vboxdrv.service-lsb:virtualbox Linux Module
Loaded:loaded (/etc/init.d/vboxdrv; bad; vendor preset:disabled)
Active:active (exited) since Sat 2016-06-04 08:49:44 BST; 3h 49min ago
Docs:man:systemd-sysv-generator (8)
process:1410 execstart=/etc/init.d/vboxdrv Start (code=exited, status=0/success)
tasks:0 (limit:512)

June 08:49:43 Laptop systemd[1]: Starting Lsb:virtualbox Linux module ...
June 08:49:44 Laptop vboxdrv[1410]: Starting VirtualBox kernel modules. Failed
June 08:49:44 Laptop vboxdrv[1410]: (Modprobe vboxdrv failed. Use ' DMESG ' T...hy)
June 08:49:44 Laptop Systemd[1]: Started Lsb:virtualbox Linux module.
Hint:some lines were ellipsized, use-l.
And it turns out that the following modules are missing,

# modprobe Vboxdrv
# modprobe Vboxnetflt
# modprobe VBOXNETADP
The results are prompt,

Modprobe:FATAL:Module <module> not found in Directory/lib/modules/4.5.4-1-default
Feel reinstall VirtualBox may be effective,

# Zypper In-f VirtualBox
This-F is forced to reinstall the installed program (probably equivalent to dpkg-reconfigure in Debian, see here)

But it's not going to work!

It turned out that this time it was the new release of the two components of VirtualBox,

Virtualbox-host-kmp-default
Virtualbox-guest-kmp-default
The version is ' 5.0.18_k4.6.0_1-2.3 ', which is completely targeted at kernel 4.6.0. And you can't go back to the previous version from the Opensuse-tumbleweed-oss source via YaST2. It seems that we have to solve the problem of the new kernel.

Then look further at the contents of the Opensuse-tumbleweed-oss source to find a VirtualBox version for the 4.5.4 kernel:

virtualbox-5.0.18-2.1.x86_64.rpm
virtualbox-qt-5.0.18-2.1.x86_64.rpm
virtualbox-host-kmp-default-5.0.18_k4.5.4_1-2.1.x86_64.rpm
Specify the version number to install try,

# sudo zypper in virtualbox-5.0.18-2.1 virtualbox-qt-5.0.18-2.1 virtualbox-host-kmp-default-5.0.18_k4.5.4_1-2.1

Package ' virtualbox-5.0.18-2.1.x86_64 ' not found.
Package ' virtualbox-host-kmp-default-5.0.18_k4.5.4_1-2.1.x86_64 ' not found.
Package ' virtualbox-qt-5.0.18-2.1.x86_64 ' not found.
Resolving package dependencies ...

Nothing todo.
It seems not so, the RPM installation try,

# sudo rpm-i http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/virtualbox-host-kmp-default-5.0.18_ k4.5.4_1-2.1.x86_64.rpm http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/virtualbox-5.0.18-2.1.x86_ 64.rpm http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/virtualbox-qt-5.0.18-2.1.x86_64.rpm
But there's something. so file is wrong.

The final approach is to remove the new kernel first, and then uninstall the VirtualBox for the new kernel at the same time, and then reinstall the VirtualBox. The feeling should be able (because the latter method was tried) to use the RPM installation directly from the command line, but ignore the libdevmapper.so dependency error,

# sudo rpm-i--nodeps http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/ virtualbox-host-kmp-default-5.0.18_k4.5.4_1-2.1.x86_64.rpm http://download.opensuse.org/tumbleweed/repo/oss/ suse/x86_64/virtualbox-5.0.18-2.1.x86_64.rpm http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/ virtualbox-qt-5.0.18-2.1.x86_64.rpm
And actually changing the VirtualBox and virtualbox-qt to the suffix 2.3 version is fine.

I have tested the installation process is also a bit fastidious:

First use RPM to install virtualbox-host-kmp-default-5.0.18_k4.5.4_1-2.1.x86_64 from the command line,
# sudo rpm-i http://download.opensuse.org/tumbleweed/repo/oss/suse/x86_64/virtualbox-host-kmp-default-5.0.18_ k4.5.4_1-2.1.x86_64.rpm
Then lock the package in YaST2 Software management (not to upgrade it),
Then check the VirtualBox to install (only tick this one, the rest will be automatically checked), but specifically will automatically check the virtualbox-guest-kmp-default-5.0.18_k4.6.0_1-2.3 * uncheck *, if there Kernel-default-base this, also need to cancel the check.
Installation can be.
Yes, that's it. You might want to do the following,
sudo rcvboxdrv setup
sudo systemctl start vboxdrv
Anyway, there's finally a VirtualBox to use. I'm not going to upgrade the kernel any more recently, I've already locked the 4.5.4 Kernel Package:D

Remove new kernel

To remove the new kernel-related packages,

# zypper RM kernel-default-4.6.0-1.1 kernel-default-base-4.6.0-1.1 kernel-default-devel-4.6.0-1.1 kernel-devel-4.6.0-1.1 kernel-syms-4.6.0-1.1

The following 8 packages are going to be removed:
kernel-default-4.6.0-1.1 kernel-default-devel-4.6.0-1.1 kernel-devel-4.6.0-1.1
kernel-syms-4.6.0-1.1 VirtualBox Virtualbox-guest-kmp-default Virtualbox-host-kmp-default
Virtualbox-qt
Kernel-macros Although also 4.6.0-1.1 version, but must keep, this bag is for kernel-default-4.5.4 use. Reboot the computer when it is finished so that the kernel changes take effect. Restarting the default is kernel 4.5.4.

Expect the next kernel update to solve the problem of kernel 4.6.0 's inability to work on my ThinkPad t420s.

Updated 2016.06.11

The subsequent update of yesterday included the kernel 4.6.1. I didn't update the kernel, but I did update the other ones. Then came a puzzling question:

Insert the U disk on the hint is not authorized to mount the device (you are not authorized to mount this device);
After restarting the wireless network cannot connect, click the SSID and enter the password prompt "could not add connection–no sessions found for UID 1000″.
Can not load u disk also forget, but there is no network that is not good work. After half a day can not be resolved after the decision to reinstall, by the way to change the root partition for the Btrfs file system.

It was later found that the problems above were highly likely to be caused by a package called Pam-config. You can look at this discussion list in detail. The solution seems to be simple, I have not tried, because it is in the process of reloading the system to find information,

sudo pam-config--add--systemd
And if this does not solve the problem, you can at least configure the network (in YaST2-> network setting) to be managed by wicked instead of the default networkm Anager), and then edit the properties of the wireless network to configure the connection. It may even be necessary to manually add a domain name resolution server to the/etc/resolv.conf. ©