At Tue, 28 Oct 2014 23:24:14 +0100, Michal Kubecek wrote:
On Tue, Oct 28, 2014 at 03:26:10PM +0100, Takashi Iwai wrote:
At Tue, 28 Oct 2014 10:15:07 -0400, Jeff Mahoney wrote:
This is a topic that comes up for every release, but it's always brought up *after* the release. If we want to update the kernel version mid-release, that's a topic for discussion but it needs to happen well before release so it's communicated clearly. It's not a topic I've been interested in driving since there are costs and benefits to both sides and the status quo means less work for me. :)
Well, following the stable kernel would be even less work for you :)
From one point of view, yes. From another... For example, moving from 3.16 to 3.17 brings API changes which break build of VMware and VirtualBox modules (and maybe also other third-party ones). For illustration, this is summary of module audit I've done for the 3.12 (SLE12) based kernel I plan to use for evergreen 13.1:
- 3 modules dropped as "not used any more" - 2 modules dropped as obsoleted by other module/feature - 8 modules merged into one new - 2 modules renamed - 1 module now unconditionally linked into the image
These are only changes from mainline between 3.11 and 3.12 (affecting modules from 13.1 packages), there were also some SuSE specific ones. Add API changes, kABI changes etc.
I'm not trying to say that these issues can't be handled and I'm not trying to say upgrading to newer kernel version would be wrong. I just want to point out that it is wrong to think that "just following stable" is easy. Sure, I used newer kernel on three of my 13.1 systems for some time, 3.16 worked fine without any issue and 3.17 only required patching VMware host modules. But providing newer kernel to everyone in an update requires more careful approach.
Right, kABI change and module changes are the biggest downside of "following stable" method. OTOH, we don't need to move forward immediately when a new stable kernel appears. The only requirement is rather when we won't get any new stable updates. So, we can wait for a while until kABI issues on most drivers get fixed. 3.16.x will be discontinued after 3.18 release. That is, in less than two months after openSUSE 13.2 release, we shall face the problem, and have to decide whether to leave it as wild west like openSUSE 13.1 kernel or not. Takashi -- To unsubscribe, e-mail: opensuse-kernel+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-kernel+owner@opensuse.org