[LEDE-DEV] Policies for changes to LEDE stable branches, or LEDE patches in general?

[Álvaro Fernández Rojas]

Hi Bjørn,

El 25/04/2017 a las 13:43, Bjørn Mork escribió:
> Hello,
> 
> I note that a recent commit to the lede-17.01 branch has added a few
> driver patches which have barely hit the mainline kernel:
> 
> https://github.com/lede-project/source/commit/1ab41265c39354332630bcba0ec704abd2e790f0
> 
> This surprised me quite a bit.  I would expect any such fixes to go the
> normal route from mainline to stable to LEDE, like they do for most
> other current distros. This usually does not take more than a couple of
> weeks anyway, and less if a fix is critical.
You're right, but we're dealing with something that has been accepted upstream and has been tested and discussed by several developers, and I don't think that waiting for a fix to come from linux-stable is mandatory (@lede-dev correct me if I'm wrong).

> 
> It was surprising enough that this hit the master branch.  But going
> into a stable LEDE branch before davem has considered it for stable
> kernels?  Why?
Because it fixes a long standing bug for the Raspberry Pi (and any other devices bridging interfaces that rely on drivers that alter sk_buff data without duplicating its content).

As far as I know, only fixes can be pushed to stable branches.
You're right, davem hasn't yet considered it for stable branches, but in this case the patch has been tested by quite a lot of users (Raspberry Pi community), and I thought this was sufficient to consider it stable (BTW, I've also tested it myself).

> 
> And then there is the issue of making LEDE patches for fixes which can
> be, or already are, upstream: Is that wise use of resources? Reading
> https://lede-project.org/docs/guide-developer/the-source-code I see that
> I am wrong thinking that there are policies for this.
> 
> But maybe there should be?
> 
> 
> 
> Bjørn
> 

~Álvaro.