[v3,0/4] clk: qcom: Add support for SM8150

Add support for rpm clock controller found in SM8150 and while at it update
the driver to support parent data clock scheme as suggested by Stephen.

Changes since v3:
 - Add reviewed-by from Bjorn
 - Update the parent name as xo_board
 - Fix style issue

Changes since v2:
 - Describe parent clocks for rpmhcc
 - Add support for parent data scheme for rpmhcc

Vinod Koul (4):
  dt-bindings: clock: Document the parent clocks
  clk: qcom: clk-rpmh: Convert to parent data scheme
  dt-bindings: clock: Document SM8150 rpmh-clock compatible
  clk: qcom: clk-rpmh: Add support for SM8150

 .../bindings/clock/qcom,rpmh-clk.txt          |  7 +++-
 drivers/clk/qcom/clk-rpmh.c                   | 37 ++++++++++++++++++-
 2 files changed, 41 insertions(+), 3 deletions(-)

On Thu 22 Aug 10:01 PDT 2019, Vinod Koul wrote:

> Convert the rpmh clock driver to use the new parent data scheme by
> specifying the parent data for board clock.
> 
> Signed-off-by: Vinod Koul <vkoul@kernel.org>
> ---
>  drivers/clk/qcom/clk-rpmh.c | 10 ++++++++--
>  1 file changed, 8 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/clk/qcom/clk-rpmh.c b/drivers/clk/qcom/clk-rpmh.c
> index c3fd632af119..0bced7326a20 100644
> --- a/drivers/clk/qcom/clk-rpmh.c
> +++ b/drivers/clk/qcom/clk-rpmh.c
> @@ -95,7 +95,10 @@ static DEFINE_MUTEX(rpmh_clk_lock);
>  		.hw.init = &(struct clk_init_data){			\
>  			.ops = &clk_rpmh_ops,				\
>  			.name = #_name,					\
> -			.parent_names = (const char *[]){ "xo_board" },	\
> +			.parent_data =  &(const struct clk_parent_data){ \
> +					.fw_name = "xo_board",		\
> +					.name = "xo_board",		\

Iiuc .name here refers to the global clock namespace and .fw_name refers
to the device_node local name space. As such I really prefer this to be:

  .fw_name = "xo",
  .name = "xo_board",

This ensures the backwards compatibility (when using global lookup),
without complicating the node-local naming.

Regards,
Bjorn

> +			},						\
>  			.num_parents = 1,				\
>  		},							\
>  	};								\
> @@ -110,7 +113,10 @@ static DEFINE_MUTEX(rpmh_clk_lock);
>  		.hw.init = &(struct clk_init_data){			\
>  			.ops = &clk_rpmh_ops,				\
>  			.name = #_name_active,				\
> -			.parent_names = (const char *[]){ "xo_board" },	\
> +			.parent_data =  &(const struct clk_parent_data){ \
> +					.fw_name = "xo_board",		\
> +					.name = "xo_board",		\
> +			},						\
>  			.num_parents = 1,				\
>  		},							\
>  	}
> -- 
> 2.20.1
>

On 23-08-19, 23:31, Bjorn Andersson wrote:
> On Thu 22 Aug 10:01 PDT 2019, Vinod Koul wrote:
> 
> > Convert the rpmh clock driver to use the new parent data scheme by
> > specifying the parent data for board clock.
> > 
> > Signed-off-by: Vinod Koul <vkoul@kernel.org>
> > ---
> >  drivers/clk/qcom/clk-rpmh.c | 10 ++++++++--
> >  1 file changed, 8 insertions(+), 2 deletions(-)
> > 
> > diff --git a/drivers/clk/qcom/clk-rpmh.c b/drivers/clk/qcom/clk-rpmh.c
> > index c3fd632af119..0bced7326a20 100644
> > --- a/drivers/clk/qcom/clk-rpmh.c
> > +++ b/drivers/clk/qcom/clk-rpmh.c
> > @@ -95,7 +95,10 @@ static DEFINE_MUTEX(rpmh_clk_lock);
> >  		.hw.init = &(struct clk_init_data){			\
> >  			.ops = &clk_rpmh_ops,				\
> >  			.name = #_name,					\
> > -			.parent_names = (const char *[]){ "xo_board" },	\
> > +			.parent_data =  &(const struct clk_parent_data){ \
> > +					.fw_name = "xo_board",		\
> > +					.name = "xo_board",		\
> 
> Iiuc .name here refers to the global clock namespace and .fw_name refers
> to the device_node local name space. As such I really prefer this to be:
> 
>   .fw_name = "xo",
>   .name = "xo_board",
> 
> This ensures the backwards compatibility (when using global lookup),
> without complicating the node-local naming.

Sure, while thinking more on this, should we finalize the name as xo or
cxo, I see latter being also used at few places. It would be great to
get a name and stick to it for longer time :)

Quoting Vinod Koul (2019-08-25 23:21:27)
> On 23-08-19, 23:31, Bjorn Andersson wrote:
> > On Thu 22 Aug 10:01 PDT 2019, Vinod Koul wrote:
> > 
> > > Convert the rpmh clock driver to use the new parent data scheme by
> > > specifying the parent data for board clock.
> > > 
> > > Signed-off-by: Vinod Koul <vkoul@kernel.org>
> > > ---
> > >  drivers/clk/qcom/clk-rpmh.c | 10 ++++++++--
> > >  1 file changed, 8 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/drivers/clk/qcom/clk-rpmh.c b/drivers/clk/qcom/clk-rpmh.c
> > > index c3fd632af119..0bced7326a20 100644
> > > --- a/drivers/clk/qcom/clk-rpmh.c
> > > +++ b/drivers/clk/qcom/clk-rpmh.c
> > > @@ -95,7 +95,10 @@ static DEFINE_MUTEX(rpmh_clk_lock);
> > >             .hw.init = &(struct clk_init_data){                     \
> > >                     .ops = &clk_rpmh_ops,                           \
> > >                     .name = #_name,                                 \
> > > -                   .parent_names = (const char *[]){ "xo_board" }, \
> > > +                   .parent_data =  &(const struct clk_parent_data){ \
> > > +                                   .fw_name = "xo_board",          \
> > > +                                   .name = "xo_board",             \
> > 
> > Iiuc .name here refers to the global clock namespace and .fw_name refers
> > to the device_node local name space. As such I really prefer this to be:
> > 
> >   .fw_name = "xo",
> >   .name = "xo_board",
> > 
> > This ensures the backwards compatibility (when using global lookup),
> > without complicating the node-local naming.
> 
> Sure, while thinking more on this, should we finalize the name as xo or
> cxo, I see latter being also used at few places. It would be great to
> get a name and stick to it for longer time :)
> -- 

I would name it 'cxo' because that's the pin name on this platform.