Re: [PATCH 0/4] arm64: dts: allwinner: add A733/Cubie A7S DTS support

[Vinicius Pedrosa <vinicius.eduardo.pedrosa@gmail.com>]

Hi Enzo,

Two findings from building this series locally on top of the A733 CCU v2,
plus one note for the DTSI move you already agreed to with Jernej. Both
findings reproduce on your own Cubie A7S DTB. Junhui and Jerome on Cc.

1. Schema evidence for the clock list Jerome already questioned.

Jerome raised the xtal rate and the naming of the node feeding the CCU. A
full CHECK_DTBS=y run - the whole schema set, not a selected binding -
rejects that node outright:

  sun60i-a733-cubie-a7s.dtb: clock-controller@2002000
  (allwinner,sun60i-a733-ccu):
    clock-names: ['hosc', 'losc', 'iosc'] is too short
    clocks: [[3], [4], [5]] is too short
      from schema: allwinner,sun60i-a733-ccu.yaml

The posted binding wants more clock entries than the DTSI supplies. This
may be the same problem Jerome described rather than a second one, and
whatever resolves his point probably resolves this too. I am sending the
diagnostic because it gives you a check that either passes or does not.

2. /soc raises a W=1 warning.

  sun60i-a733.dtsi:116.6-197.4: Warning (unit_address_vs_reg): /soc: node
  has a reg or ranges property, but no unit name

It is the only diagnostic this series adds over an unmodified tree in my
W=1 comparison - 499 unique diagnostics before, 500 after.

3. When the UART0 pins move into the DTSI, they will need
allwinner,pinmux.

allwinner,sun55i-a523-pinctrl.yaml has listed allwinner,pinmux under
required since d626d248caae, and the A733 compatible joins that schema in
the pinctrl RFC. uart0_pb9_pb10_pins currently sets only pins and
function. Nothing catches it today, because the A733 compatible has no
schema yet - but it will fail validation once the pinctrl series lands,
which is likely around when you send v2.

Static evidence only - no A733 hardware here.

Vinicius