[yocto] ***SPAM*** Re: [meta-raspberrypi] Current master broken

[Herve Jourdain]

Sorry, a bit tired...
I meant we need to track the changes happening to the versions we decide we
want to support, and update the recipes accordingly.
My understanding is we're interested in 4.4, 4.8 and 4.9. And I have a
personal interest in 4.7 for the moment, so I can try to keep the 4.7 recipe
up to date with the changes.
If 4.4 is "stable", then we need some people to focus on tracking the
changes happening on 4.8 and 4.9.

Cheers,
Herve

-----Original Message-----
From: Andrei Gherzan [mailto:andrei at gherzan.ro] 
Sent: mardi 6 décembre 2016 19:11
To: Herve Jourdain <herve.jourdain at neuf.fr>
Cc: 'Khem Raj' <raj.khem at gmail.com>; yocto at yoctoproject.org; 'Gary Thomas'
<gary at mlbassoc.com>
Subject: Re: ***SPAM*** Re: [yocto] [meta-raspberrypi] Current master broken

On Tue, Dec 06, 2016 at 07:04:31PM +0100, Herve Jourdain wrote:
> Hi Andrei,
>
> But I believe that the core of the problem is not necessarily the 
> patch in oe-core to detect the kernel version itself, but like Khem 
> mentioned the fact that a "well known" SRCREV is only valid for a 
> brief period of time under the RPI tree "constant" rebasing we've been
experiencing lately.
> So basically, the SRCREV that is used is not corresponding to what it 
> used to be, if it's still there in the tree, and the oe-core detects 
> that - the SRCREV thought it was one version of the kernel, and it 
> just happened that it corresponds to a different codebase with a different
linux version.
> Killing the error reporting by a patch will probably not prevent the 
> mismatch happening behind the scene.
> So like Khem mentioned, either we find a way to "prevent" the rebasing 
> at the source, or we need to track down every version we want to 
> support, and update SRCREV and corresponding kernel version for those.

Got it now. What do you mean by "track down every version we want"?
--
Andrei Gherzan