[CentOS-devel] RFC: Stream Kernel SIG Proposal

Wed Jan 20 20:41:21 UTC 2021
redbaronbrowser <redbaronbrowser at protonmail.com>

On Wednesday, January 20, 2021 1:38 PM, Matthew Miller <mattdm at mattdm.org> wrote:

> On Wed, Jan 20, 2021 at 07:11:25PM +0000, Phil Perry wrote:
>
> > > On this point: the Fedora kernel packages are probably a fine place to start
> > > and it would be nice to see this SIG collaborate with the Fedora kernel
> > > maintainers and possibly the Fedora Server SIG rather than just being a
> > > disconnected downstream.
> >
> > No need, this already exists both in the form of the current
> > mainline kernel and an LTS offering, and has done so for many years.
> > Assuming such mainline and LTS kernels built on RHEL will install
> > seamlessly on Stream, there seems little point reinventing the
> > wheel.
> > https://elrepo.org/linux/kernel/el8/x86_64/RPMS/
>
> I would put that a different way: better to connect these things up than to
> have them in different places.

I'm going through LTS kernel commits already to identify back-ported patches applied to the Stream kernel.  By packaging LTS kernels, I get a better starting point from a group with a long standing respect for real openness.

I also was working on tooling for pulling kpatch'ing for the LTS kernels and so far they are written to confirm against the same GPG key.  If you are willing to eventually review my work and consider publishing/sign your own LTS kpatches, I would be open to doing it that way.  Otherwise, I would have to come up with a secure method of having the LTS kernel and the kpatch provider verified via two different GPG keys.

Overall, I need a fall back for the SIG from a dependable source.  I don't know where Mike McGrath's new found love of openness is coming from.  This seems like something that should have been a priority on the November 11th meeting.  Or a priority before Karsten Wade's December 19th claim to be closing the openness gap.  Or a priority when I brought the issue up back on December 23rd.  The bottom line is the deadline was set on CentOS 8 without any prerequisites for the state of Stream once that date is reached.  There is no commitment from Red Hat to follow through last week or today or in a "few weeks."

I am willing to do my best to work with anyone to avoid reinventing the wheel, but to proceed with Goal #4, I need a solid base of patches to work from.