Re: [PATCHv4 0/9] *** ARM: Update arch-vt8500 to Devicetree ***

On Fri, 2012-08-24 at 09:54 +0400, Alexey Charkov wrote:

2012/8/24 Tony Prisk [off-list ref]:

quoted

On Fri, 2012-08-24 at 05:40 +0000, Arnd Bergmann wrote:

quoted

On Thursday 23 August 2012, Tony Prisk wrote:

quoted

I was about to say that if Mike has any issues with the driver that I
can fix the patch conflict at the same time, but I just realised that
its more work than I originally thought :)

I was going to move the arch-vt8500 part of the clocks patch back into
the clocks patch - but that will just move the issue from arm-soc to
clocks because Mike's branch won't have the arch-vt8500 patch so the
patch will fail.

But that's ok, because without the arch-v8500 part, nothing uses
the vt8500-clock driver, so nothing breaks. You can introduce broken
code in the meantime as long as it's impossible to enable and it will
work afterwards.

      Arnd

You lost me a little bit there :)
If it stays as is now, the arch-vt8500 patch will introduce an
unresolved symbol.
If I move the clock code from the arch-vt8500 patch to the clock patch,
then the clock patch won't apply because it needs the arch-vt8500 patch
first and Mike won't have that patch, but it would be fine if both
patches went in via your tree.

Why don't you add a #define in the first patch that makes your
to-be-unresolved symbol a no-op and then drop it within the clock
patch?

Best,
Alexey

Actually - that doesn't resolve the problem unless it all goes in via
the same tree still because the clock patch can only drop the #define if
the arch-vt8500 patch exists still - same as problem #2 I mentioned.

If the clock patch goes via Mikes tree he will need the arch-vt8500
patch as well - don't think it can be avoided.

Still easier to send the whole lot via arm-soc, then at least we can
resolve the patch conflicts without having to apply the patches twice.

Regards
Tony P