[0/3] clk: meson: add support for PCIE PLL

Message

Neil Armstrong March 7, 2019, 2:14 p.m. UTC

The Amlogic G12A SoCs embeds a dedicated PLL to feed the USB3+PCIE
Combo PHY. This PLL needs a very specific and strict register sequence
in order to correcly enable it and deliver the 100MHz reference clock
to the Analog PHY.

After lot of trials and errors, and since this PLL will ever feed 100MHz
with a static configuration, it is simpler to setup a dedicated ops
structure with a custom _enable() op applying the init register sequence.

The rate calculation ops are kept in order to keep the nominal read
ops as-in, but set_rate is removed.

With this setup, the PLL can be enabled and disable safely and always have
the recommended PLL setup to feed the USB3+PCIE Combo PHY.

Neil Armstrong (3):
  clk: meson-pll: add reduced specific clk_ops for G12A PCIe PLL
  dt-bindings: clk: g12a-clkc: add PCIE PLL clock ID
  clk: meson-g12a: add PCIE PLL clocks

 drivers/clk/meson/clk-pll.c           |  26 ++++++
 drivers/clk/meson/clk-pll.h           |   1 +
 drivers/clk/meson/g12a.c              | 118 ++++++++++++++++++++++++++
 drivers/clk/meson/g12a.h              |   5 +-
 include/dt-bindings/clock/g12a-clkc.h |   1 +
 5 files changed, 150 insertions(+), 1 deletion(-)

Comments

Jerome Brunet March 19, 2019, 10:01 a.m. UTC | #1

On Thu, 2019-03-07 at 15:14 +0100, Neil Armstrong wrote:
> The Amlogic G12A SoCs embeds a dedicated PLL to feed the USB3+PCIE
> Combo PHY. This PLL needs a very specific and strict register sequence
> in order to correcly enable it and deliver the 100MHz reference clock
> to the Analog PHY.
> 
> After lot of trials and errors, and since this PLL will ever feed 100MHz
> with a static configuration, it is simpler to setup a dedicated ops
> structure with a custom _enable() op applying the init register sequence.
> 
> The rate calculation ops are kept in order to keep the nominal read
> ops as-in, but set_rate is removed.
> 
> With this setup, the PLL can be enabled and disable safely and always have
> the recommended PLL setup to feed the USB3+PCIE Combo PHY.
> 
> Neil Armstrong (3):
>   clk: meson-pll: add reduced specific clk_ops for G12A PCIe PLL
>   dt-bindings: clk: g12a-clkc: add PCIE PLL clock ID
>   clk: meson-g12a: add PCIE PLL clocks
> 
>  drivers/clk/meson/clk-pll.c           |  26 ++++++
>  drivers/clk/meson/clk-pll.h           |   1 +
>  drivers/clk/meson/g12a.c              | 118 ++++++++++++++++++++++++++
>  drivers/clk/meson/g12a.h              |   5 +-
>  include/dt-bindings/clock/g12a-clkc.h |   1 +
>  5 files changed, 150 insertions(+), 1 deletion(-)

Well this PLL is indeed a nasty one ;)

Acked-by: Jerome Brunet <[e-mail]>

Neil Armstrong March 19, 2019, 8:23 p.m. UTC | #2

On 19/03/2019 11:01, Jerome Brunet wrote:
> On Thu, 2019-03-07 at 15:14 +0100, Neil Armstrong wrote:
>> The Amlogic G12A SoCs embeds a dedicated PLL to feed the USB3+PCIE
>> Combo PHY. This PLL needs a very specific and strict register sequence
>> in order to correcly enable it and deliver the 100MHz reference clock
>> to the Analog PHY.
>>
>> After lot of trials and errors, and since this PLL will ever feed 100MHz
>> with a static configuration, it is simpler to setup a dedicated ops
>> structure with a custom _enable() op applying the init register sequence.
>>
>> The rate calculation ops are kept in order to keep the nominal read
>> ops as-in, but set_rate is removed.
>>
>> With this setup, the PLL can be enabled and disable safely and always have
>> the recommended PLL setup to feed the USB3+PCIE Combo PHY.
>>
>> Neil Armstrong (3):
>>   clk: meson-pll: add reduced specific clk_ops for G12A PCIe PLL
>>   dt-bindings: clk: g12a-clkc: add PCIE PLL clock ID
>>   clk: meson-g12a: add PCIE PLL clocks
>>
>>  drivers/clk/meson/clk-pll.c           |  26 ++++++
>>  drivers/clk/meson/clk-pll.h           |   1 +
>>  drivers/clk/meson/g12a.c              | 118 ++++++++++++++++++++++++++
>>  drivers/clk/meson/g12a.h              |   5 +-
>>  include/dt-bindings/clock/g12a-clkc.h |   1 +
>>  5 files changed, 150 insertions(+), 1 deletion(-)
> 
> Well this PLL is indeed a nasty one ;)
> 
> Acked-by: Jerome Brunet <[e-mail]>
> 

Applied patch 2 on next/headers, patches 1 & 3 on next/drivers

Neil