Re: [RFC PATCH 3/5] linux-kernel-base: add support for Linux 3.x

From: Anders Darander <anders.darander@gmail.com>
To: Patches and discussions about the oe-core layer
	<openembedded-core@lists.openembedded.org>
Subject: Re: [RFC PATCH 3/5] linux-kernel-base: add support for Linux 3.x
Date: Wed, 22 Jun 2011 12:41:47 +0200	[thread overview]
Message-ID: <BANLkTim13H6v6T5eovEOYVXoiW=4bOtLXg@mail.gmail.com> (raw)
In-Reply-To: <1308738156.21613.80.camel@phil-desktop>

On Wed, Jun 22, 2011 at 12:22, Phil Blundell <pb@pbcl.net> wrote:
> On Tue, 2011-06-21 at 21:10 +0200, Anders Darander wrote:
>> If we keep the kernel_majorversion, we'll need to have something
>> similar to this,
>> as the major version up to 2.6 was determined by X.Y. From 3.0, only the first
>> digit represents the major number; while the second digit (0 in 3.0)
>> is equal to
>> x in 2.6.x. Thus, the function to determine the major version has to
>> return either
>> 2.6 or 3.
>
> Well, the thing is that OE's "kernelmajorversion" was always something
> of an artificial construct.  It was introduced purely as a way of coping
> with some of the differences between 2.4 and 2.6: this was primarily the
> different module format, but it also happened to coincide with major
> rewrites of the iPAQ and Zaurus kernels which were the main ones we were
> using in OE at the time.  Returning "3" here isn't going to be helpful
> since none of the classes will know what to do with that value: all they
> ever do is compare it against either 2.4 or 2.6.  The exact form of
> those tests has never been standardised, so there is a risk that you
> might encounter

I assumed that the reason was that; and it's nice to get a better grip of the
older history. (At that time I was more concerned with Buildroot and uCLinux).

> Since there are no material differences between Linux 2.6 and Linux 3.0
> that OE needs to be aware of, I think it would make most sense for
> kernelmajorversion() to continue returning "2.6" for Linux 3.0 and
> future versions if we did want to keep that mechanism.

Yes, and that's why I completely removed the tests in the bbclasses that
I've touch sofar.

> However, since the whole purpose of the mechanism was to ease the
> transition from 2.4 to 2.6, I am fairly confident that it can just be
> deleted altogether if support for 2.4 is being removed.

Sounds good. I'll grep through a few more classes to see if the tests seems
to occur in more places. Unless I find that the major version is used for
something else, I'll remove it.

Regards,
Anders