[PATCH resend 2/4] ARM: dts: sunxi: unify APB1 clock

[Kevin Hilman]

Maxime Ripard <maxime.ripard at free-electrons.com> writes:

> Arnd,
>
> On Fri, Nov 21, 2014 at 05:08:08PM +0100, Arnd Bergmann wrote:
>> On Friday 21 November 2014 15:35:57 Arnd Bergmann wrote:
>> > On Friday 21 November 2014 15:29:03 Maxime Ripard wrote:
>> > > On Fri, Nov 21, 2014 at 12:57:02PM +0100, Arnd Bergmann wrote:
>> > > > On Thursday 20 November 2014 14:04:01 Kevin Hilman wrote:
>> > > > > Chen-Yu Tsai <wens at csie.org> writes:
>> > > > > 
>> > > > > > From: Emilio López <emilio at elopez.com.ar>
>> > > > > >
>> > > > > > With the new factors infrastructure in place, we can unify apb1 and
>> > > > > > apb1_mux as a single clock now.
>> > > > > >
>> > > > > > Signed-off-by: Emilio López <emilio at elopez.com.ar>
>> > > > > > [wens at csie.org: Change apb1 node label to "apb1"; reword commit title]
>> > > > > > Signed-off-by: Chen-Yu Tsai <wens at csie.org>
>> > > > > 
>> > > > > Boot breakage in arm-soc/for-next on sun4i-a10-cubieboard and
>> > > > > sun7i-a20-cubieboard2[1] was bisected down to this patch[1].
>> > > > > 
>> > > > > Reverting $SUBJECT on top of arm-soc gets things booting again.
>> > > > > 
>> > > > 
>> > > > As this looks like it was intended as a cleanup without functional
>> > > > changes, I would go ahead and revert it in next/dt.
>> > > > 
>> > > > Any objections?
>> > > 
>> > > Yeah, you'd break linux-next as well doing so, as the clock driver now
>> > > requires this from the DT.
>> > > 
>> > > I can merge it through the clock tree though if you prefer it that
>> > > way.
>> > 
>> > Do you know why this commit breaks booting then?
>> > 
>> 
>> I have now reverted the entire branch, to get things working again.
>> 
>> Please send a new pull request once you have a version that you 
>> have actually tested.
>
> This was tested and working. And again, the linux-next proves it.
>
> I know very well why it doesn't work, and it's actually expected: some
> clock was refactored, the DT needed to be changed, only half of it was
> merged through arm-soc.
>
> It really is just because one part got through arm-soc, the other
> through the clock tree, nothing more.

Which branch (already in linux-next) would be needed in arm-soc/for-next
to resolve the dependency?

Kevin