On Thu, Jun 20, 2013 at 10:27:44AM +0200, Michal Simek wrote:
> On 06/20/2013 10:02 AM, Arnd Bergmann wrote:
> > On Thursday 20 June 2013, Michal Simek wrote:
> >> On 06/19/2013 08:46 PM, Arnd Bergmann wrote:
> >>> On Wednesday 19 June 2013, Soren Brinkmann wrote:
> >>>> I don't know how much a defconfig is supposed to provide, hence as RFC.
> >>>> This patches are needed for booting Zynq into a minimum ramfs based
> >>>> system with a serial console.
> >>>
> >>> In my opinion we should provide enable all the platform specific drivers
> >>> in the defconfigs, as well as everything needed to boot the system,
> >>> to get proper compile coverage as well as the ability to test changes
> >>> easily. Your patches look good. Michal, would you apply them and
> >>> send another pull request or should I just take them directly?
> >>
> >> Soren asked me 2 days ago if make sense to create zynq defconfig or not.
> >> I just suggested him to better extend this multi_v7_defconfig.
> >> But still question is if we can/should create zynq specific defconfig?
> >> Or are you going to remove all of these platform specific defconfig?
> >
> > We don't have a consistent policy across platforms at the moment.
> > Traditionally we had multiple defconfigs per platform, in some cases
> > one per board, but moving towards one defconfig per platform at
> > the moment.
>
> That's what I though but on the other hand in this process
> all these defconfigs should be removed.
>
> > I guess whether or not to have a separate defconfig for one platform
> > or to use only multi_*_defconfig is a question of how many people
> > would use a zynq_defconfig in practice.
>
> The point is if you look at zynq users than they will just use this
> zynq_defconfig
> because they know that it is for zynq and also they don't want to
> compile drivers for other platforms which zynq can't use.
> From distribution point of view they want to use only one image because it is
> just
> easier.
>
> Based on this if there is an option to also add just zynq defconfig, I would
> prefer
> to also add it.
>
> >> Definitely agree that multi_v7 defconfig should enable everything needed
> >> to boot the system.
> >> Does it also mean that we should also enable all zynq drivers
> >> to get better compile coverage?
> >
> > I would say yes.
> >
> > My feeling is that multi_v7_defconfig should enable all hardware
> > support for the platforms in it, and that users would take it
> > as a starting point if they want to have a configuration for
> > an embedded system, disabling everything they don't need.
>
> I just wanted to be sure because you wrote just drivers for booting
> it means any "minimal" configuration to get it boot not all drivers.
>
> If you are ok, Soren will prepare also specific zynq defconfig file
> and check if there are any missing drivers which are not enabled for zynq
> for multi_v7. I will collect them in one branch and will send pull request.
I can check. But I don't think it makes too much sense currently. Even
though multi_v7_defconfig targets several SOCs its pretty minimal. I
think there are just a few SOC BSPs and serial drivers selected. Due to
lacking driver support in mainline, a Zynq specific config would not be
that different, IMHO.
But this does hopefully change with growing driver support for Zynq in
mainline.
Sören
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
