Hello,
I get *Failed to start Load Kernel Modules* during the boot.
Everytime Kernel is updated VirtualBox modules fail I have to reboot into the new Kernel, run the following commands and then it starts working.
$ sudo akmods --force $ sudo dracut -v -f $ sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
Then everything starts working again and the boot error goes away. You can found the background of the issue in the link below.
https://bugzilla.redhat.com/show_bug.cgi?id=1254340
My question is how can I automate this task.
On 06/18/18 19:19, Sudhir Khanger wrote:
I get *Failed to start Load Kernel Modules* during the boot.
Everytime Kernel is updated VirtualBox modules fail I have to reboot into the new Kernel, run the following commands and then it starts working.
$ sudo akmods --force $ sudo dracut -v -f $ sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
Then everything starts working again and the boot error goes away. You can found the background of the issue in the link below.
You are using the VirtualBox packages from rpmfusion, yes?
I don't, but I seem to recall they operate differently than the packages from Oracle. The Oracle packages don't rely on akmod and don't put the modules in the initramfs. The vbox modules are checked for existence when vboxdrv.service is run and if they don't exist for the running kernel they are built at that time.
Assuming akmods is being used to build the modules when the kernel is updated there should be some sort of log file in /var/cache/akmods just like there is for nvidia. Have you checked?
Also, have you considered asking about this on the rpmfusion user's list since it is their package?
On Mon, Jun 18, 2018 at 5:58 PM Ed Greshko ed.greshko@greshko.com wrote:
On 06/18/18 19:19, Sudhir Khanger wrote:
I get *Failed to start Load Kernel Modules* during the boot.
Everytime Kernel is updated VirtualBox modules fail I have to reboot
into the new
Kernel, run the following commands and then it starts working.
$ sudo akmods --force $ sudo dracut -v -f $ sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
Then everything starts working again and the boot error goes away. You
can found
the background of the issue in the link below.
You are using the VirtualBox packages from rpmfusion, yes?
I don't, but I seem to recall they operate differently than the packages from Oracle. The Oracle packages don't rely on akmod and don't put the modules in the initramfs. The vbox modules are checked for existence when vboxdrv.service is run and if they don't exist for the running kernel they are built at that time.
Assuming akmods is being used to build the modules when the kernel is updated there should be some sort of log file in /var/cache/akmods just like there is for nvidia. Have you checked?
Also, have you considered asking about this on the rpmfusion user's list since it is their package?
-- Conjecture is just a conclusion based on incomplete information. It isn't a fact.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
Thanks. I will try Oracle's package for now.
I get the same message now that is "Failed to start Load Kernel Modules" but there isn't anything in the logs.
https://gist.github.com/sudhirkhanger/bcd766104e2268cb73a8d80c0152a932
The only service which fails is dbxtool.service
journalctl -b | grep -ie "Failed to start Load Kernel Modules"
The above command yields no results.
On Mon, Jun 18, 2018 at 5:58 PM Ed Greshko ed.greshko@greshko.com wrote:
On 06/18/18 19:19, Sudhir Khanger wrote:
I get *Failed to start Load Kernel Modules* during the boot.
Everytime Kernel is updated VirtualBox modules fail I have to reboot
into the new
Kernel, run the following commands and then it starts working.
$ sudo akmods --force $ sudo dracut -v -f $ sudo grub2-mkconfig -o /boot/efi/EFI/fedora/grub.cfg
Then everything starts working again and the boot error goes away. You
can found
the background of the issue in the link below.
You are using the VirtualBox packages from rpmfusion, yes?
I don't, but I seem to recall they operate differently than the packages from Oracle. The Oracle packages don't rely on akmod and don't put the modules in the initramfs. The vbox modules are checked for existence when vboxdrv.service is run and if they don't exist for the running kernel they are built at that time.
Assuming akmods is being used to build the modules when the kernel is updated there should be some sort of log file in /var/cache/akmods just like there is for nvidia. Have you checked?
Also, have you considered asking about this on the rpmfusion user's list since it is their package?
-- Conjecture is just a conclusion based on incomplete information. It isn't a fact.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
Okay update time. Irrespective of where I install VirtualBox from if a new kernel is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
sudo dracut -v -f
On 07/23/2018 02:04 AM, Sudhir Khanger wrote:
Okay update time. Irrespective of where I install VirtualBox from if a new kernel is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
Yes, the self-compiled modules aren't created at the time the initramfs is created, so you have to add them yourself. I guess virtualbox adds config to force its modules to be loaded and dracut expects those to be available in the initramfs.
On 07/23/2018 10:08 AM, Samuel Sieb wrote:
On 07/23/2018 02:04 AM, Sudhir Khanger wrote:
Okay update time. Irrespective of where I install VirtualBox from if a new kernel is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
Yes, the self-compiled modules aren't created at the time the initramfs is created, so you have to add them yourself. I guess virtualbox adds config to force its modules to be loaded and dracut expects those to be available in the initramfs.
Question, do you have the VirtualBox-kmodsrc RPM installed? I believe that's necessary so kmod can rebuild the modules when the kernel is updated. I don't use VirtualBox myself, so I can't speak to it. ---------------------------------------------------------------------- - Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com - - AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 - - - - Do you know how to save five drowning lawyers? No? GOOD! - ----------------------------------------------------------------------
On Mon, Jul 23, 2018 at 10:58 PM Rick Stevens ricks@alldigital.com wrote:
On 07/23/2018 10:08 AM, Samuel Sieb wrote:
On 07/23/2018 02:04 AM, Sudhir Khanger wrote:
Okay update time. Irrespective of where I install VirtualBox from if a new kernel is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
Yes, the self-compiled modules aren't created at the time the initramfs is created, so you have to add them yourself. I guess virtualbox adds config to force its modules to be loaded and dracut expects those to be available in the initramfs.
Question, do you have the VirtualBox-kmodsrc RPM installed? I believe that's necessary so kmod can rebuild the modules when the kernel is updated. I don't use VirtualBox myself, so I can't speak to it.
- Rick Stevens, Systems Engineer, AllDigital ricks@alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
-Do you know how to save five drowning lawyers? No? GOOD! -
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
That is not the case when you use Oracle-provided VirtualBox builds. Akmod and kmod kernel modules are installed as a part of RPMFusion's version of VirtualBox.
On 07/23/18 17:04, Sudhir Khanger wrote:
Okay update time. Irrespective of where I install VirtualBox from if a new kernel is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
When it comes to VirtualBox from Oracle I find that statement "odd" to say the least.
When I install the VirtualBox from Oracle it installs /lib/systemd/system/vboxdrv.service. When the system starts if there is no driver matching the running kernel it will be built and then loaded.
This results in....
[egreshko@meimei ~]$ systemd-analyze blame 52.215s vboxdrv.service
And in the journal....
Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[728]: vboxdrv.sh: Starting VirtualBox services. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[887]: Starting VirtualBox services. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[728]: vboxdrv.sh: Building VirtualBox kernel modules. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[892]: Building VirtualBox kernel modules. Jul 22 17:01:45 meimei.greshko.com vboxdrv.sh[6011]: VirtualBox kernel modules built. Jul 22 17:01:45 meimei.greshko.com kernel: vboxdrv: Found 4 processor cores Jul 22 17:01:46 meimei.greshko.com kernel: vboxdrv: TSC mode is Invariant, tentative frequency 2660206031 Hz Jul 22 17:01:46 meimei.greshko.com kernel: vboxdrv: Successfully loaded version 5.2.16 (interface 0x00290001) Jul 22 17:01:46 meimei.greshko.com vboxdrv.sh[6052]: VirtualBox services started. Jul 22 17:01:46 meimei.greshko.com audit[1]: SERVICE_START pid=1 uid=0 auid=4294967 295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=vboxdrv comm="systemd " exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
[egreshko@meimei ~]$ lsmod | grep vbox vboxpci 28672 0 vboxnetadp 28672 0 vboxnetflt 32768 1 vboxdrv 491520 4 vboxpci,vboxnetadp,vboxnetflt
[egreshko@meimei ~]$ uname -r 4.17.7-200.fc28.x86_64
And, as you can see, the built driver was created at boot time....
[egreshko@meimei ~]$ ll /usr/lib/modules/4.17.7-200.fc28.x86_64/misc/vboxdrv.ko -rw-r--r--. 1 root root 14121784 Jul 22 17:01 /usr/lib/modules/4.17.7-200.fc28.x86_64/misc/vboxdrv.ko
On Tue, Jul 24, 2018 at 3:49 AM Ed Greshko ed.greshko@greshko.com wrote:
On 07/23/18 17:04, Sudhir Khanger wrote:
Okay update time. Irrespective of where I install VirtualBox from if a
new kernel
is installed then I have to manually create a new initramfs otherwise systemd-modules-load.service fails at the boot.
When it comes to VirtualBox from Oracle I find that statement "odd" to say the least.
When I install the VirtualBox from Oracle it installs /lib/systemd/system/vboxdrv.service. When the system starts if there is no driver matching the running kernel it will be built and then loaded.
This results in....
[egreshko@meimei ~]$ systemd-analyze blame 52.215s vboxdrv.service
And in the journal....
Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[728]: vboxdrv.sh: Starting VirtualBox services. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[887]: Starting VirtualBox services. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[728]: vboxdrv.sh: Building VirtualBox kernel modules. Jul 22 17:00:54 meimei.greshko.com vboxdrv.sh[892]: Building VirtualBox kernel modules. Jul 22 17:01:45 meimei.greshko.com vboxdrv.sh[6011]: VirtualBox kernel modules built. Jul 22 17:01:45 meimei.greshko.com kernel: vboxdrv: Found 4 processor cores Jul 22 17:01:46 meimei.greshko.com kernel: vboxdrv: TSC mode is Invariant, tentative frequency 2660206031 Hz Jul 22 17:01:46 meimei.greshko.com kernel: vboxdrv: Successfully loaded version 5.2.16 (interface 0x00290001) Jul 22 17:01:46 meimei.greshko.com vboxdrv.sh[6052]: VirtualBox services started. Jul 22 17:01:46 meimei.greshko.com audit[1]: SERVICE_START pid=1 uid=0 auid=4294967 295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg='unit=vboxdrv comm="systemd " exe="/usr/lib/systemd/systemd" hostname=? addr=? terminal=? res=success'
[egreshko@meimei ~]$ lsmod | grep vbox vboxpci 28672 0 vboxnetadp 28672 0 vboxnetflt 32768 1 vboxdrv 491520 4 vboxpci,vboxnetadp,vboxnetflt
[egreshko@meimei ~]$ uname -r 4.17.7-200.fc28.x86_64
And, as you can see, the built driver was created at boot time....
[egreshko@meimei ~]$ ll /usr/lib/modules/4.17.7-200.fc28.x86_64/misc/vboxdrv.ko -rw-r--r--. 1 root root 14121784 Jul 22 17:01 /usr/lib/modules/4.17.7-200.fc28.x86_64/misc/vboxdrv.ko
-- Conjecture is just a conclusion based on incomplete information. It isn't a fact.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
systemd-modules-load.service fails at boot and then it probably builds the modules. After the system boots, systemd-modules-load.service has active status and not failed. But the only way to get rid of Failed to start Load Kernel Modules is by rebuilding initramfs via command dracut -v -f. This error crops up everytime you install a new kernel irrespective of VirtualBox's installation source.
On 07/24/18 12:08, Sudhir Khanger wrote:
systemd-modules-load.service fails at boot and then it probably builds the modules. After the system boots, systemd-modules-load.service has active status and not failed. But the only way to get rid of Failed to start Load Kernel Modules**is by rebuilding initramfs via command dracut -v -f. This error crops up everytime you install a new kernel irrespective of VirtualBox's installation source.
Nope....
As a matter of fact, systemd-modules-load.service never runs on my system. Why?
None of these conditions are met....
ConditionDirectoryNotEmpty=|/lib/modules-load.d ConditionDirectoryNotEmpty=|/usr/lib/modules-load.d ConditionDirectoryNotEmpty=|/usr/local/lib/modules-load.d ConditionDirectoryNotEmpty=|/etc/modules-load.d ConditionDirectoryNotEmpty=|/run/modules-load.d ConditionKernelCommandLine=|modules-load ConditionKernelCommandLine=|rd.modules-load
[egreshko@meimei ~]$ ll /lib/modules-load.d total 0 [egreshko@meimei ~]$ ll /usr/lib/modules-load.d total 0 [egreshko@meimei ~]$ ll /usr/local/lib/modules-load.d ls: cannot access '/usr/local/lib/modules-load.d': No such file or directory [egreshko@meimei ~]$ ll /etc/modules-load.d total 0 [egreshko@meimei ~]$ ll /run/modules-load.d ls: cannot access '/run/modules-load.d': No such file or directory
And the kernel command line does not contain either of those parameters.
So.
[egreshko@meimei proc]$ journalctl -u systemd-modules-load.service -- Logs begin at Thu 2017-12-21 09:55:27 CST, end at Tue 2018-07-24 12:39:54 CST. > -- No entries --
I never have to run dracut to get rid of anything.
As I've already shown in my previous message it is vboxdrv.sh that checks/loads the vbox drivers.
It does it here in the script
if ! $MODPROBE vboxdrv > /dev/null 2>&1; then setup
And then goes here if the modprobe fails.
# setup_script setup() { begin_msg "Building VirtualBox kernel modules" console log "Building the main VirtualBox module."
And it creates the /var/log/vbox-setup.log
[egreshko@meimei log]$ cat vbox-setup.log Building the main VirtualBox module. Building the net filter module. Building the net adaptor module. Building the PCI pass-through module.
On Tue, Jul 24, 2018 at 10:25 AM Ed Greshko ed.greshko@greshko.com wrote:
On 07/24/18 12:08, Sudhir Khanger wrote:
systemd-modules-load.service fails at boot and then it probably builds
the modules.
After the system boots, systemd-modules-load.service has active status
and not
failed. But the only way to get rid of Failed to start Load Kernel
Modules**is by
rebuilding initramfs via command dracut -v -f. This error crops up
everytime you
install a new kernel irrespective of VirtualBox's installation source.
Nope....
As a matter of fact, systemd-modules-load.service never runs on my system. Why?
None of these conditions are met....
ConditionDirectoryNotEmpty=|/lib/modules-load.d ConditionDirectoryNotEmpty=|/usr/lib/modules-load.d ConditionDirectoryNotEmpty=|/usr/local/lib/modules-load.d ConditionDirectoryNotEmpty=|/etc/modules-load.d ConditionDirectoryNotEmpty=|/run/modules-load.d ConditionKernelCommandLine=|modules-load ConditionKernelCommandLine=|rd.modules-load
[egreshko@meimei ~]$ ll /lib/modules-load.d total 0 [egreshko@meimei ~]$ ll /usr/lib/modules-load.d total 0 [egreshko@meimei ~]$ ll /usr/local/lib/modules-load.d ls: cannot access '/usr/local/lib/modules-load.d': No such file or directory [egreshko@meimei ~]$ ll /etc/modules-load.d total 0 [egreshko@meimei ~]$ ll /run/modules-load.d ls: cannot access '/run/modules-load.d': No such file or directory
And the kernel command line does not contain either of those parameters.
So.
[egreshko@meimei proc]$ journalctl -u systemd-modules-load.service -- Logs begin at Thu 2017-12-21 09:55:27 CST, end at Tue 2018-07-24 12:39:54 CST. > -- No entries --
I never have to run dracut to get rid of anything.
As I've already shown in my previous message it is vboxdrv.sh that checks/loads the vbox drivers.
It does it here in the script
if ! $MODPROBE vboxdrv > /dev/null 2>&1; then setupAnd then goes here if the modprobe fails.
# setup_script setup() { begin_msg "Building VirtualBox kernel modules" console log "Building the main VirtualBox module."
And it creates the /var/log/vbox-setup.log
[egreshko@meimei log]$ cat vbox-setup.log Building the main VirtualBox module. Building the net filter module. Building the net adaptor module. Building the PCI pass-through module.
-- Conjecture is just a conclusion based on incomplete information. It isn't a fact.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
Like I mentioned earlier that systemd-modules-load.service service fails and starts working after I create a new initramfs. You were unwilling to accept what I experienced on my system irrespective of the cause I shot a video of it which you can find below.
https://www.youtube.com/watch?v=YZHyGlwtPQo&feature=youtu.be
On Sat, Aug 4, 2018 at 11:09 PM Samuel Sieb samuel@sieb.net wrote:
Did you check if there was anything in /etc/modules-load.d/? _______________________________________________ users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
$ cat /etc/modules-load.d/bbswitch.conf # Load bbswitch.ko at boot bbswitch
There's only bbswitch.conf file.
On 08/05/18 01:12, Sudhir Khanger wrote:
Like I mentioned earlier that systemd-modules-load.service service fails and starts working after I create a new initramfs. You were unwilling to accept what I experienced on my system irrespective of the cause I shot a video of it which you can find below.
I'm not unwilling, just telling you that it isn't related to your initial assumption that it has to do with VirtualBox Modules as those aren't needed in the initramfs are aren't loaded by systemd-modules-load.service.
As I pointed out, systemd-modules-load.service is triggered by certain conditions. One of them is...
ConditionDirectoryNotEmpty=|/etc/modules-load.d
And you've answered that in your follow up to another query....
$ cat /etc/modules-load.d/bbswitch.conf # Load bbswitch.ko at boot bbswitch
So, the systemd-modules-load.service failure is due to "bbswitch" not being available.
I don't know anything about how you've installed bumblebee or how it manages the building of a kernel matching module. But that is the area you should be investigating and not anything to do with Virtualbox.
On Sun, Aug 5, 2018 at 3:46 AM Ed Greshko ed.greshko@greshko.com wrote:
On 08/05/18 01:12, Sudhir Khanger wrote:
Like I mentioned earlier that systemd-modules-load.service service fails
and starts
working after I create a new initramfs. You were unwilling to accept
what I
experienced on my system irrespective of the cause I shot a video of it
which you
can find below.
I'm not unwilling, just telling you that it isn't related to your initial assumption that it has to do with VirtualBox Modules as those aren't needed in the initramfs are aren't loaded by systemd-modules-load.service.
As I pointed out, systemd-modules-load.service is triggered by certain conditions. One of them is...
ConditionDirectoryNotEmpty=|/etc/modules-load.d
And you've answered that in your follow up to another query....
$ cat /etc/modules-load.d/bbswitch.conf # Load bbswitch.ko at boot bbswitch
So, the systemd-modules-load.service failure is due to "bbswitch" not being available.
I don't know anything about how you've installed bumblebee or how it manages the building of a kernel matching module. But that is the area you should be investigating and not anything to do with Virtualbox.
-- Conjecture is just a conclusion based on incomplete information. It isn't a fact.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://getfedora.org/code-of-conduct.html List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org/...
Is that the case with both Oracle and RPMFusion's VirtualBox or of only former?
FYI I have dropped a message for the Bumblebee packager. Let's see if he has some thoughts.
https://github.com/Bumblebee-Project/Bumblebee/issues/153#issuecomment-41049...
On 08/05/18 11:22, Sudhir Khanger wrote:
Is that the case with both Oracle and RPMFusion's VirtualBox or of only former?
I don't normally use the RPMFusion packages. I don't recall why that is. Must have had a bad experience at some point. So, today I did install it on a laptop and found....
When it comes to the drivers they do things differently.
The Oracle method is as I've outlined previously.
As for RPMFusion it utilizes the "akmod" facility to build and install the modules when the kernel is updated. This is assuming you've installed akmod-VirtualBox from RPMfusion.
The akmods facility creates the file /etc/kernel/postinst.d/akmodsposttrans which checks for the existence of the driver for the newly installed kernel. It then fires off the process to create the new drivers and this results in the following in /var/cache/akmods/akmods.log.
2018/08/05 12:56:46 akmods: Building and installing VirtualBox-kmod 2018/08/05 12:56:46 akmods: Building RPM using the command '/sbin/akmodsbuild --target x86_64 --kernels 4.17.11-200.fc28.x86_64 /usr/src/akmods/VirtualBox-kmod.latest' 2018/08/05 12:57:50 akmods: Installing newly built rpms 2018/08/05 12:57:50 akmods: DNF detected 2018/08/05 12:58:06 akmods: Successful.
This results in the creation and installation of kmod-VirtualBox-4.17.11-200.fc28.x86_64-5.2.16-1.fc28.x86_64.rpm which contains (abbreviated)
[root@acer VirtualBox]# rpm2cpio kmod-VirtualBox-4.17.11-200.fc28.x86_64-5.2.16-1.fc28.x86_64.rpm | cpio -it ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox/vboxdrv.ko ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox/vboxnetadp.ko ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox/vboxnetflt.ko ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox/vboxpci.ko ./usr/lib/modules/4.17.11-200.fc28.x86_64/extra/VirtualBox/vboxsf.ko
It also does create the file /lib/modules-load.d/VirtualBox.conf which is one of the triggers for systemd-modules-load so you will get this on reboot.
[egreshko@acer ~]$ systemctl status systemd-modules-load ● systemd-modules-load.service - Load Kernel Modules Loaded: loaded (/usr/lib/systemd/system/systemd-modules-load.service; static; v> Active: active (exited) since Sun 2018-08-05 13:08:31 CST; 1min 13s ago Docs: man:systemd-modules-load.service(8) man:modules-load.d(5) Process: 541 ExecStart=/usr/lib/systemd/systemd-modules-load (code=exited, statu> Main PID: 541 (code=exited, status=0/SUCCESS)
Aug 05 13:08:31 acer.greshko.com systemd-modules-load[541]: Inserted module 'vboxd> Aug 05 13:08:31 acer.greshko.com systemd-modules-load[541]: Inserted module 'vboxn> Aug 05 13:08:31 acer.greshko.com systemd-modules-load[541]: Inserted module 'vboxn> Aug 05 13:08:31 acer.greshko.com systemd-modules-load[541]: Inserted module 'vboxp> Aug 05 13:08:31 acer.greshko.com systemd[1]: Started Load Kernel Modules.
But, you've said you get a systemd-modules-load failure when the Oracle packages are installed. I've shown that can't be the case with Oracle.
When comes to RPMFusion the following are now installed on my system.
VirtualBox-server-5.2.16-1.fc28.x86_64 VirtualBox-kmodsrc-5.2.16-1.fc28.noarch kmod-VirtualBox-4.17.9-200.fc28.x86_64-5.2.16-1.fc28.x86_64 kmod-VirtualBox-4.17.11-200.fc28.x86_64-5.2.16-1.fc28.x86_64 akmod-VirtualBox-5.2.16-1.fc28.x86_64
Assuming you've the same installed you can check the /var/cache/akmods/akmods.log for any errors during the creation of the rpm.
Also make sure that you have akmods.service enabled.
FYI I have dropped a message for the Bumblebee packager. Let's see if he has some thoughts.
https://github.com/Bumblebee-Project/Bumblebee/issues/153#issuecomment-41049...
FWIW, on the next kernel update you can temporarily remove the bbswitch.conf file before reboot and see if the error still occurs.
On 08/04/2018 10:37 PM, Ed Greshko wrote:
FWIW, on the next kernel update you can temporarily remove the bbswitch.conf file before reboot and see if the error still occurs.
I expect that you would have to remove it before you do the kernel update. I think dracut copies that file into the initramfs.
On 08/05/18 14:08, Samuel Sieb wrote:
On 08/04/2018 10:37 PM, Ed Greshko wrote:
FWIW, on the next kernel update you can temporarily remove the bbswitch.conf file before reboot and see if the error still occurs.
I expect that you would have to remove it before you do the kernel update. I think dracut copies that file into the initramfs.
Ah, good point. I just got back from being baked in 36C while on a walk. Must have fried a synapse or two.
I have this exact same problem on FC29. All you've said is true regarding the build and install of the kernel modules for the rpmfusion-free-updates repo. The modules get rebuilt by akmod just fine.
I have
dnf provides /usr/lib/modules-load.d/VirtualBox.conf
VirtualBox-server-5.2.22-1.fc29.x86_64 : core part (host server) for VirtualBox Repo : rpmfusion-free-updates Matched from: Filename : /usr/lib/modules-load.d/VirtualBox.conf
The only problem is that dracut doesn't seem to be run after the modules are built. I always get the error during boot until I do a "dracut -f".
My best guess on this is that for some reason dracut is only done for the first akmod build. I have
ls -1 /usr/src/akmods/*.latest
/usr/src/akmods/nvidia-kmod.latest /usr/src/akmods/VirtualBox-kmod.latest
and the error I get is always related to vboxdrv and friends not being there.
On 08/05/18 11:22, Sudhir Khanger wrote:
FYI I have dropped a message for the Bumblebee packager. Let's see if he has some thoughts.
https://github.com/Bumblebee-Project/Bumblebee/issues/153#issuecomment-41049...
While I don't use their packages I did poke around https://linux.itecs.ncsu.edu/redhat/public/ and did find
https://linux.itecs.ncsu.edu/redhat/public/bumblebee/28/x86_64/bbswitch-dkms...
This suggests to me that the bbswitch module is being rebuilt via dkms with a new kernel. You may want to investigate the dkms logs.
FWIW, although I don't currently have any dkms compiled packages, it too should be fired off by the file dkms located in /etc/kernel/postinst.d
-
Ed Greshko -
Rick Stevens -
Samuel Sieb -
Sean McNeil -
Sudhir Khanger