Re: Kernel 6.18 fist tests...

[Adam Gibbons <adam.gibbons@ipfire.org>]

[-- Attachment #1: Type: text/plain, Size: 9540 bytes --]

Hi Arne, Adolf, all,
Just chiming in with my own testing experience.

I updated to the latest nightly (which has the dracut-ng initqueue fix) and then applied the 6.18 kernel. Small things I hit: the readme had "-C " instead of "-C /" for the tar command, and there was a typo in the dracut command ("regenrate" instead of "regenerate"). Easy fixes, though. :)

Since then it's been running solidly for over 24 hours now. No odd behaviour at all – everything feels normal, and Observium (which monitors the box) isn't showing anything unusual either.

I've noticed that Linux 6.18.1 was released recently too, so I'll grab Arne's build of that next and give it a go soon.

For reference, my current test box is an older i5-4200U with four Intel I211 NICs. I also have a newer Intel N305 (with I226-V NICs) ready, so I'll move the test over to that soon – mainly curious to see how p-states and power management are behaving in 6.18.

Thanks Arne for building these, really helpful to get it tested early.

Cheers,
Adam

On 15 December 2025 11:46:18 GMT, Adolf Belka <adolf.belka@ipfire.org> wrote:
>Hi Arne,
>
>With the patch fix for the initqueue issue in place I have been running IPFire with that new kernel and testing out various things.
>
>I am not sure what aspects of IPfire would be most likely to be affected by the change to the kernel, so I have just been testing various elements of IPFire. Everything I have tried has worked without any issues.
>
>OpenVPN, WireGuard and IPSec connections have all worked, both RW and N2N with the latter being between a system with 6.18.0 and all other IPFire VM's with 6.12.58
>
>Regards,
>
>Adolf.
>
>
>On 12/12/2025 14:57, Adolf Belka wrote:
>> Hi all,
>> 
>> On 12/12/2025 12:30, Adolf Belka wrote:
>>> Hi Michael & all,
>>> 
>>> On 11/12/2025 23:35, Adolf Belka wrote:
>>>> Hi Michael,
>>>> 
>>>> On 11/12/2025 22:01, Michael Tremer wrote:
>>>>> Hello Adolf,
>>>>> 
>>>>> The initqueue module is probably quite important. As far as I know it kind of replaces some systemd functionality which we don’t have.
>>>>> 
>>>>> But looking at our root file, it is adding some systemd unit files:
>>>>>>>    https://git.ipfire.org/?p=ipfire-2.x.git;a=blob;f=config/rootfiles/common/dracut-ng;hb=d656b7af6411d0306513b853a552f56d6713c174#l484
>>>>> 
>>> 
>>> Three of those files are related to systemd but one is the same as used to be in the base collection but was moved to this separate set in version 108.
>>> 
>>> When I did the package update, I commented out all new entries in the rootfile that weren't previously in the rootfile. Therefore the initqueue set are all commented out but as it had been moved out from the base it needed to be included.
>>> 
>>> I am running a build with all of the initqueue entries uncommented (although probably, the ones related to systemd could be left commented) and I will then do an install with that build and then try the kernel update followed by the dracut command again and see if it then works for the initqueue module.
>>> 
>>> Well report back when I have the results.
>> 
>> I can report that all required modules are now able to be included, including initqueue and all modules are now correctly installed including btrfs, lvm and mdraid that had a problem before.
>> 
>> I will submit a patch to correct the dracut-ng rootfile.
>> 
>> Regards,
>> 
>> Adolf.
>> 
>>> 
>>> Regards,
>>> 
>>> Adolf.
>>> 
>>> 
>>>>> We might have found a breaking change in dracut-ng if we don’t properly support mdraid any more.
>>>> 
>>>> but that should then be presumably raised as an issue with dracut-ng as if it worked with the old dracut it should work with dracut-ng as version 100 was said to be a drop-in replacement for the old version.
>>>> 
>>>> We are currently on version 108 and I just found in the changelog for that version the following change.
>>>> 
>>>> --------------------------
>>>> feat(initqueue): factor out initqueue into its own module
>>>> This PR allows excluding dracut initqueue module, while
>>>> including the base or the dracut-systemd module.
>>>> 
>>>> Factoring out functionality into its own module
>>>> allows to better control to include or exclude the
>>>> functionality.
>>>> --------------------------
>>>> 
>>>> https://github.com/dracut-ng/dracut-ng/commit/3daf67830d8db3b12bc45eef9f8f29f606192db9
>>>> 
>>>> So it looks like it is a module that has to be explicitly specified now whereas in the past it was part of dracut base.
>>>> 
>>>> Regards,
>>>> 
>>>> Adolf.
>>>> 
>>>>> 
>>>>> Lacking support for btrfs in the initramdisk could be less severe, but that also needs to be tested. @Stefan, will you do this for c199?
>>>>> 
>>>>> -Michael
>>>>> 
>>>>>> On 11 Dec 2025, at 18:04, Adolf Belka <adolf.belka@ipfire.org> wrote:
>>>>>> 
>>>>>> Hi Arne,
>>>>>> 
>>>>>> On 11/12/2025 18:28, Adolf Belka wrote:
>>>>>>> Hi All,
>>>>>>> On 11/12/2025 18:20, Adolf Belka wrote:
>>>>>>>> Hi Arne,
>>>>>>>> 
>>>>>>>> I have tried it out on a vm clone running CU199 Testing.
>>>>>>>> 
>>>>>>>> In the readme.txt file I think you meant to have
>>>>>>>> 
>>>>>>>> tar xvaf KERNEL-ARCHIV.tar.xz -C /
>>>>>>>> 
>>>>>>>> instead of
>>>>>>>> 
>>>>>>>> tar xvaf KERNEL-ARCHIV.tar.xz -C \
>>>>>>>> 
>>>>>>>> At least it only extracted for me with the / and not the \.
>>>>>>>> 
>>>>>>>> I then ran the dracut command and got the following messages.
>>>>>>>> 
>>>>>>>> sudo dracut --regenerate-all --force
>>>>>>>> dracut[I]: Executing: /usr/bin/dracut --kver=6.12.58-ipfire --force
>>>>>>>> dracut[E]: Module 'btrfs' depends on module 'initqueue', which can't be installed
>>>>>>>> dracut[E]: Module 'lvm' depends on module 'initqueue', which can't be installed
>>>>>>>> dracut[E]: Module 'mdraid' depends on module 'initqueue', which can't be installed
>>>>>>>> dracut[I]: *** Including module: modsign ***
>>>>>>>> dracut[I]: *** Including module: i18n ***
>>>>>>>> dracut[I]: *** Including module: btrfs ***
>>>>>>>> dracut[I]: *** Including module: dm ***
>>>>>>>> dracut[I]: *** Including module: fs-lib ***
>>>>>>>> dracut[I]: *** Including module: kernel-modules ***
>>>>>>>> dracut[I]: *** Including module: kernel-modules-extra ***
>>>>>>>> dracut[I]: *** Including module: lvm ***
>>>>>>>> dracut[I]: *** Including module: mdraid ***
>>>>>>>> dracut[I]: *** Including module: qemu ***
>>>>>>>> dracut[I]: *** Including module: rootfs-block ***
>>>>>>>> dracut[I]: *** Including module: terminfo ***
>>>>>>>> dracut[I]: *** Including module: udev-rules ***
>>>>>>>> dracut[I]: *** Including module: base ***
>>>>>>>> dracut[I]: *** Including modules done ***
>>>>>>>> dracut[I]: *** Installing kernel module dependencies ***
>>>>>>>> dracut[I]: *** Installing kernel module dependencies done ***
>>>>>>>> dracut[I]: *** Resolving executable dependencies ***
>>>>>>>> dracut[I]: *** Resolving executable dependencies done ***
>>>>>>>> dracut[I]: *** Generating early-microcode cpio image ***
>>>>>>>> dracut[I]: *** Constructing AuthenticAMD.bin ***
>>>>>>>> dracut[I]: *** Constructing GenuineIntel.bin ***
>>>>>>>> dracut[I]: *** Store current command line parameters ***
>>>>>>>> dracut[I]: *** Creating image file '/boot/initramfs-6.12.58-ipfire.img.tmp' ***
>>>>>>>> dracut[I]: *** Hardlinking files ***
>>>>>>>> dracut[I]: *** Hardlinking files done ***
>>>>>>>> 
>>>>>>>> 
>>>>>>>> Not sure about the "module 'initqueue', which can't be installed" bits.
>>>>>>>> 
>>>>>>>> After the above last line I then got
>>>>>>>> 
>>>>>>>> cp: error writing '/boot/initramfs-6.12.58-ipfire.img.tmp': No space left on device
>>>>>>>> 
>>>>>>>> So I suspect I need to make the IPFire vm disk larger.
>>>>>>> The problem was that I made a clone of a system that I had tested out upgrading from CU150 to CU198 and the boot size was obviously smaller back in CU150 as I only have 110MB but the more current systems have 488MB so I will create a clone of one of my newer systems and repeat the test.
>>>>>> 
>>>>>> Repeated with a newer IPFire vm which had a 488MB boot partition and the dracut section went fine with no problems beside the initqueue module not being able to be installed. However that did not seem to cause a problem with the grub-mkconfig command or the subsequent reboot.
>>>>>> 
>>>>>> I will evaluate the install tomorrow to check if anything has a problem.
>>>>>> 
>>>>>> Regards,
>>>>>> 
>>>>>> Adolf.
>>>>>> 
>>>>>> 
>>>>>>> Regards,
>>>>>>> Adolf.
>>>>>>>> 
>>>>>>>> Will give that a try later but would like to know if there is anything I need to do with regard to the initqueue module that is mentioned in the dracut output.
>>>>>>>> 
>>>>>>>> Regards,
>>>>>>>> 
>>>>>>>> Adolf.
>>>>>>>> 
>>>>>>>> On 11/12/2025 14:57, Adam Gibbons wrote:
>>>>>>>>> Nice!
>>>>>>>>> 
>>>>>>>>> I will test this over the weekend.
>>>>>>>>> 
>>>>>>>>> Thanks,
>>>>>>>>> Adam
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> On 11 December 2025 12:00:23 GMT, Arne Fitzenreiter <arne_f@ipfire.org> wrote:
>>>>>>>>> 
>>>>>>>>>      Hi,
>>>>>>>>> 
>>>>>>>>>      i have compiled a first test kernel 6.18 for x86_64.
>>>>>>>>> 
>>>>>>>>>      https://people.ipfire.org/~arne_f/highly-experimental/kernel/ <https://people.ipfire.org/~arne_f/highly-experimental/kernel/>
>>>>>>>>> 
>>>>>>>>>      please help testing and report your findings...
>>>>>>>>> 
>>>>>>>>>      Arne
>>>>> 
>>>>> 
>>>> 
>>> 
>> 
>
>

[-- Attachment #2: Type: text/html, Size: 12071 bytes --]