On 27/07/2018 10:03:22-0700, Stephen Boyd wrote:

Quoting Alexandre Belloni (2018-07-17 15:27:41)

quoted

This is the promised rework of the at91 PMC clocks driver. It is mainly
necessary to remove the DTC warnings but it also complies with the CCF
rule that there should be one node per controller instead of one node
per clock.

This only handles the PMC, I'm planning to also rework the SCKC bindings
later (without breaking the DT ABI).

The series is based on top of clk-next plus at91-dt so I don't think it
is convenient to have it this cycle. However, I would really like to
ensure we agree on the new bindings this cycle before converting all the
other platforms as this is a bit tedious.

The first two patches are actually fixes and may be considered for this
cycle.

One nice note:
at91-sama5d2_xplained.dtb goes from 29351 bytes to 22082 bytes

Patches look mostly good. Rob is happy with the bindings and so am I.

One general question is why the drivers can't be moved to real platform
drivers instead of using OF_CLK_DECLARE?

I tried, this makes the clocksource drivers fail with -EPROBE_DEFER and
so the kernel just stops there.

-- 
Alexandre Belloni, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com