Arch Linux User Repository

@air-g4p that's not what I mean by the dkms.conf being old but anyway - fill your boots :)

@artafinde - glad to learn things work correctly on your end.

However, since you earlier assumed: I think maybe the DKMS modules are old and not updated for latest kernels?

I provided the evidence that your assumption is not true on my fully updated, and re-booted, laptops.

What would make you now believe that building in a clean chroot would correct anything?

I build linux-ck with time makepkg -sric as I have been doing for many years!

Cheers

I am unfamiliar with dkms... can you test your idea about enabling the debug stuff using linux-ck's config and report back?

@air-g4p I'm able to install linux-ck and zfs-dkms without issues, modules generating without errors. Have you tried building those packages in clean chroot?

@graysky - I have tried my best to make it easy for you to track down the root cause of why I am seeing 100% 'dkms install' failures on two significantly different Optimus laptops. Both laptops were fully updated, and rebooted, prior to compiling linux-ck 5.19.12-1 - and both built without error.

Installation failure of the required dkms modules is the issue.

My old laptop is an i5 and is 12+ years old - and still runs perfectly. The relevant 'dkms install' make.logs and installation run output are here: http://ix.io/4bN9

My more modern laptop is an i7 and is 2+ years old. The relevant 'dkms install' make.logs and installation run output for it are here: http://ix.io/4bMJ

As you will see that equates to 7 out of 7 'dkms install' module failures.

You should/will also note that resolve_btfids is involved in every failure.

I've been discussing these multiple 'dkms install' module failures with several people on #archlinux.

One of these fine gentlemen has speculated that these failures may be caused by CONFIG_DEBUG_INFO_BTF=y (which I know for a fact is enabled in the default linux-ck kernel configs), but that one of its dependencies is incorrectly set, possibly causing the 'dkms install' failures.

If you believe that may be a potential cause of these multiple failures, please review:

https://cateee.net/lkddb/web-lkddb/DEBUG_INFO_BTF.html

and

https://github.com/torvalds/linux/blob/master/lib/Kconfig.debug - see Line 339

which show the required, correct dependencies.

Also, as I stated earlier, these exact same dkms modules ALL install without issue against: linux, linux-zen and linux-hardened.

Please advise - it's extremely frustrating, particularly after running linux-ck for many years.

Cheers

@artafinde - Thank you for your follow-up.

The dkms.conf content for each of the 3 failing modules is here: http://ix.io/4bt5

air-g4p this is not installed, see [here]. I think maybe the DKMS modules are old and not updated for latest kernels? Can you share a dkms.conf from /usr/src/<pkg-name>-<pkgver>/dkms.conf for one of them which fail?

The difference with the linux, linux-zen and linux-hardening is that these are build with Debug and stripped after which requires a stupendous amound of RAM :D

here: https://aur.archlinux.org/cgit/aur.git/tree/PKGBUILD?h=linux-ck#n276

@graysky - I built linux-ck 5.19.11-1 without error yesterday. However, all three of my dkms modules {bbswitch, nvidia-390xx and vboxhost} are failing to dkms install.

The relevant make.log files are here: http://ix.io/4bsR

Is there some set of kernel configs I need to be manually configuring prior to compilation?

Also note those same three modules dkms install without error against: linux, linux-zen and linux-hardened.

Any insights appreciated - Cheers

mmm we tried ... but it seems a bit "boring" as a system to use for each of the dozens of aurs installed. anyway thanks. I will follow the advice.

@simona - no yay doesn't build within a clean chroot. I'd suggest install devtools from official repos and use extra-x86_64-build to build this package as it has no other dependency from AUR. See more on the wiki 1