[Gta04-owner] GTA04-Kernel 3.17.0

Dr. H. Nikolaus Schaller hns at goldelico.com
Mon Oct 27 13:03:48 CET 2014

Am 27.10.2014 um 12:28 schrieb Radek Polak <psonek2 at seznam.cz>:

> On Monday, October 27, 2014 12:04:38 Dr. H. Nikolaus Schaller wrote:
> > Hi Radek,
> > 
> > Am 27.10.2014 um 11:30 schrieb Radek Polak <psonek2 at seznam.cz>:
> > > On Monday, October 27, 2014 10:15:09 Dr. H. Nikolaus Schaller wrote:
> > > > Who is testing what repo.
> > > 
> > > It was always pleasure to test Neil’s repo for me..
> > 
> > is this meant to imply (sometimes it might be important what has not been
> > said), that it is not a pleasure to test the gta04-kernel repo?
> No implying was meant. 

Ok. Would have wondered me.

> But while we are at it: Neil's kernel were always nice stable, he was very responsive on ML and fixing stuff at amazing speed, introducing no regressions. The kernel was better every day…

Isn’t that true for the gta04-kernel as well? It is stable, has a known set of bugs and missing features (see issue tracker) and otherwise works. Aren’t we responsive as well?

The only big regression was introduced by the move to Device Tree. And a smaller one because not all patches from 3.7 could have been ported to 3.12 and had to be rebuilt.

> > If yes, what are the hurdles?
> For me anything than 3.7 was unusable power management wise as far as i am interested in phone not toy. Adding stuff to kernel which is not necessary for phone and going for worse PM is not interesting for me as QtMoko devel.

Yes, I understand. Power Management is the biggest missing feature in my eyes as well. But it is also the most problematic thing to improve - if other parts are not working perfectly.

In most cases our patches for 3.7 were not acceptable upstream (because something was in the work, with device tree it will become better, this is the wrong way to solve etc.).

And, I think we must have a feature complete and bug free kernel first, before we can improve power management.

For example, the move to DT has broken WiFi between 3.12 and 3.14. And HDQ. So we can’t optimize power use at all unless it works again.

So what should we do? Stick with plain old 3.7 and improve there without any chance to get it upstream? Or follow upstream improving the chances?

We had such a discussion more than a year ago and the gta04-kernel team decided to start this rolling kernel work.

And IMHO it has many benefits. The most prominent result is that we have a 3.12 based Replicant 4.2 that works quite well. Would have been impossible with the 3.7 kernel.

The next step of that plan would have been to upstream and improve power management, but the device tree did stop that and we are still suffering from some minor issued with that.

Of course power management does not improve by just following linus/master instead of having a stable base. We have to work on it. And if currently nobody works on it it is not to blame the development model, but that nobody has the time (which is quite usual for volunteers).

> > Since we are constantly upstreaming from the gta04-kernel repo (using this
> > model since approx. 12 months) without significant issues [1], you must be
> > talking about something different.
> > 
> > Therefore I ask for testing and submitting patches. You don’t even have to
> > take care of upstreaming (because we do)
> As far as i understand it, for Neil it’s more comfortable to upstream his changes himself.

Well, if it were about “more comfortable", I could simply lean back and do nothing for the gta04 any more. And just buy an iPhone or Android crap.

But does that help all the gta04-owners and users out there? Does it fit into the spirit of our community? Having a device with long-term support by mainline?

That is the reason why I continue this discussion and am trying to make it more a team effort. Where everyone contributes to the existing stuff as much as possible from his/her area of experience. Even if it is less comfortable for the individual because it needs to follow some rules and disussing who could do what.

Of course it is easier to circumvent the gta04-kernel work and directly submit upstream. But IMHO it ignores the chance to submit something even better.

>  But now i am silent. I am not doing any kernel work which would justify my to continue in the discussion and i dont want to talk for others.

I can also only talk for myself and the gtaß4-kernel team as well as “sell” the “gta04-kernel idea" and ask questions to those who aren’t convinced or contribute.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20141027/efa4bcf2/attachment-0001.html>

More information about the Gta04-owner mailing list