[meta-intel] Use of overrides in linux-yocto bbappends

[Paul Eggleton]

On Friday 11 April 2014 08:49:17 Darren Hart wrote:
> On 4/11/14, 8:37, "Paul Eggleton" <paul.eggleton at linux.intel.com> wrote:
> >Koen reported that with a recent change to meta-minnow where
> >LINUX_VERSION =
> >"something" changed to LINUX_VERSION_minnow = "something" (which was a
> >good
> >change), the result was that the version of the linux-yocto kernel
> >packages
> >for an unrelated machine in another layer went backwards, because the
> >value
> >that was previously leaking over into that machine's kernel recipe wasn't
> >anymore.
> >
> >This in itself is OK, since the change was to make the append set the
> >variable
> >correctly. However it prompted me to look through meta-intel to see how
> >we're
> >setting LINUX_VERSION. I see there are a number of bbappends for
> >linux-yocto*
> >recipes that set LINUX_VERSION with no machine override; we need to stop
> >doing
> >this. I can throw together a patch to fix it, but we'll also need to make
> >sure
> >new BSPs added in future use overrides as well. Any suggestions on how to
> >handle the latter? Is this a documentation issue?
> 
> I've already asked Nitin to make this change across meta-intel and have
> corrected patches that have been coming in without the MACHINE override.
> So yes, we're on it.

Ah, and in fact when I switch back from dora to master I see that it has 
already been fixed there. Great - sorry for the noise.

> I suppose this should be added to the linux kernel development manual.

Right, that sounds like a good place.

Cheers,
Paul

-- 

Paul Eggleton
Intel Open Source Technology Centre