Re: [PATCH 3/3] clk: meson-g12a: Add EE Clock controller driver

[Jian Hu]

Hi Jerome

On 2018/7/11 21:45, Jerome Brunet wrote:
> On Wed, 2018-07-11 at 21:41 +0800, Jian Hu wrote:
> > > > > > +static struct clk_regmap g12a_mpll0 = {
> > > > > > +       .data = &(struct clk_regmap_gate_data){
> > > > > > +               .offset = HHI_MPLL_CNTL1,
> > > > > > +               .bit_idx = 31,
> > > > > > +       },
> > > > > > +       .hw.init = &(struct clk_init_data){
> > > > > > +               .name = "mpll0",
> > > > > > +               .ops = &clk_regmap_gate_ops,
> > > > > > +               .parent_names = (const char *[]){ "mpll0_div" },
> > > > > > +               .num_parents = 1,
> > > > > > +               .flags = CLK_SET_RATE_PARENT,
> > > > > > +       },
> > > > > > +};
> > >
> > > The previous had a predivider (1 or 2) in front of these mpll. Even if the
> > > predivider is usually set to be a passthrough, it is better to model the tree
> > > correctly.
> > >
> > > Is this SoC any different ?
> >
> > I am not sure the difference, I will confirm with IC design guys.
>
> I suggest that you have a look at the (upstream) axg and gxbb clock driver for
> this
>
> Same goes for the fdiv gates.
>
> Last, please trim your replies a bit. It will make easier to see what you are
> replying to.
>
> .

I have confirmed with IC design guys, axg and gxbb clock driver have a 
prediver in frount  of these mpll, according to the Spec, we can get the 
prediver function in HHI_MPLL_CNTL5 bit12, the G12A SoC have not the 
prediver function.