NAK: [SRU][m:unstable-meta][RFC 1/1] UBUNTU: [Packaging] Drop wireguard DKMS

Andrea Righi andrea.righi at canonical.com
Wed Oct 4 11:19:05 UTC 2023


On Wed, Oct 04, 2023 at 10:55:57AM +0200, Juerg Haefliger wrote:
> On Wed, 4 Oct 2023 07:51:36 +0200
> Andrea Righi <andrea.righi at canonical.com> wrote:
> 
> > On Tue, Oct 03, 2023 at 03:46:25PM -0500, John Cabaj wrote:
> > > On 10/2/23 6:11 PM, Dimitri John Ledkov wrote:  
> > > > On Mon, 2 Oct 2023 at 19:21, John Cabaj <john.cabaj at canonical.com> wrote:  
> > > >>
> > > >> Commit ed294cea82a6 in the main kernel package dropped
> > > >> Wireguard DKMS support. Reflecting that in the meta
> > > >> package to avoid error message during package building
> > > >> ("modinfo: ERROR: Module /lib/modules/<abi>/kernel/drivers/net/wireguard/wireguard.ko not found")  
> > > > 
> > > > 
> > > > This is an error in your packaging, since you are probably disabling
> > > > module compression, this check too should be adjusted to look for
> > > > uncompressed module.  
> > > 
> > > While I did notice this on a package I was creating, this error occurred while attempting a 
> > > normal mantic:linux crank as well (open, close, pull-source, build-source), which had module 
> > > compression enabled.  
> > 
> > IIUC we get this error when creating the source package, because
> > neither wireguard.ko.zst nor wireguard.ko is available, but at build
> > time this should be resolved correctly.
> > 
> > This is ugly and we should definitely get rid of this confusing logic.
> > Since there are still userspace components that may require the
> > wireguard-modules we can't completely get rid of it, but maybe we can
> > simply provide an unversioned wireguard-modules dependency, so
> > user-space should be happy and we can get rid of the logic that extracts
> > the version from wireguard.ko[.zst].
> 
> Why not hard-code it to 1.0.0 since that's what the module advertises and
> it's likely to never change.

This should work as well and it's very easy to do.

-Andrea



More information about the kernel-team mailing list