Kernel testing on GA Releases.

Josh Boyer jwboyer at redhat.com
Tue Jul 2 17:44:19 UTC 2013


On Tue, Jul 02, 2013 at 04:59:08PM +0000, "Jóhann B. Guðmundsson" wrote:
> On 07/02/2013 02:35 PM, Justin M. Forbes wrote:
> >On Tue, 2013-07-02 at 10:07 +0000, "Jóhann B. Guðmundsson" wrote:
> >>>I'm not sure if you guys are aware of this but a lot of reporters test
> >>>and run development and more recent kernels then those that are
> >>>available via the usual GA update mechanism and recent changes [1] which
> >>>introduce hard dependency on dracut and systemd versions on kernel
> >>>installs are preventing the ability for reporters to either a) enable
> >>>fedora-rawhide-kernel-nodebug.repo or b) manually install specific
> >>>kernel release from koji and test more recent kernels on already
> >>>existing GA releases.
> >We are very supportive of this, and frequently point people towards
> >rawhide-nodebug. If people don't want to run full rawhide, we would love
> >for them to at least run kernel snapshots.
> >
> >>>If we now have hard dependency's on specific dracut/systemd versions we
> >>>need to start looking into sync-ing kernel/dracut/systemd releases
> >>>better in GA ( arguably the entire base/core stack but that's a topic
> >>>for a later time) or back port the necessary bits that the kernel
> >>>install requires into the kernel/systemd/dracut releases in GA.
> >>>
> >This is a rare instance, and it hit several people.  If I had the option
> >of putting the required changes into F17/F18 I would have, but Harald
> >did put them in 19 at least.  This is a situation we really prefer to
> >avoid, but sometimes there is no other way.
> 
> Well this could have in it's most simple form be avoided with F18
> related #ifdef in the kernel.spec.

We took the patch Harald provided.

> I pinged Harald in irc about this and he went through dracut changes
> and backported/rebased to 029 [1]
> So that takes care of the dracut dependency which leaves out systemd
> and I've pinged Michal but he has been to busy to look into it yet
> which leaves out grubby and looking at [2] grubby seems to include
> only fixes and those changes so I dont see why Peter should not be
> able to backport/rebase to 8.26 for F18.

You might want to file an RFE bug.  I don't believe Peter reads this
list.

josh


More information about the kernel mailing list