Re: [PATCH 2/3] clk: meson: g12a: Add support for G12B CPUB clocks

[Neil Armstrong]

On 21/05/2019 19:49, Martin Blumenstingl wrote:
> Hi Neil,
> 
> On Tue, May 21, 2019 at 5:02 PM Neil Armstrong <narmstrong@xxxxxxxxxxxx> wrote:
>>
>> Update the Meson G12A Clock driver to support the Amlogic G12B SoC.
>>
>> G12B clock driver is very close, the main differences are :
>> - the clock tree is duplicated for the both clusters, and the
>>   SYS_PLL are swapped between the clusters
>> - G12A has additional clocks like for CSI an other components
> should this also be G12B?

Indeed, another leftover from previous patchset...

> 
> [...]
>> +static struct clk_regmap g12b_cpub_clk_apb_div = {
> if you also think that it's worth it then please add a comment stating
> that this is called "PCLK_mux" in the datasheet
> same goes for the ATB and AXI clocks below as the naming in the driver
> and datasheet differs
> 
>> +       .data = &(struct clk_regmap_div_data){
>> +               .offset = HHI_SYS_CPUB_CLK_CNTL1,
>> +               .shift = 3,
>> +               .width = 3,
>> +               .flags = CLK_DIVIDER_POWER_OF_TWO,
>> +       },
>> +       .hw.init = &(struct clk_init_data){
>> +               .name = "cpub_clk_apb_div",
>> +               .ops = &clk_regmap_divider_ro_ops,
>> +               .parent_names = (const char *[]){ "cpub_clk" },
>> +               .num_parents = 1,
>> +       },
>> +};
> I'm assuming you checked that this is really a power of two divider,
> on the Meson8/8b/8m2 SoCs this is a mux between div[2..8]
> (the same goes for the ATB, AXI and trace div clocks below)

Indeed it's not a power of 2 here, it's mux between div[2..8]

I'll also need to update the first cluster aswell...

> 
>> +
>> +static struct clk_regmap g12b_cpub_clk_apb = {
>> +       .data = &(struct clk_regmap_gate_data){
>> +               .offset = HHI_SYS_CPUB_CLK_CNTL1,
>> +               .bit_idx = 16,
> the public S922X datasheet calls this "PCLK_dis", does this mean you
> need a flag here?
>   .flags = CLK_GATE_SET_TO_DISABLE,

The first cluster register has some description, but with the same
fields naming :
APB_CLK_DIS: set to 1 to manually disable the APB clock...

So you are right, I'll also fix the first cluster clocks.

> 
> [...]
>> +static struct clk_regmap g12b_cpub_clk_atb = {
>> +       .data = &(struct clk_regmap_gate_data){
>> +               .offset = HHI_SYS_CPUB_CLK_CNTL1,
>> +               .bit_idx = 17,
> the public S922X datasheet calls this "ATCLK_clk_dis", does this mean
> you need a flag here?
>   .flags = CLK_GATE_SET_TO_DISABLE,

Exact

> 
> [...]
>> +static struct clk_regmap g12b_cpub_clk_axi = {
>> +       .data = &(struct clk_regmap_gate_data){
>> +               .offset = HHI_SYS_CPUB_CLK_CNTL1,
>> +               .bit_idx = 18,
> the public S922X datasheet calls this "ACLKM_clk_dis", does this mean
> you need a flag here?
>   .flags = CLK_GATE_SET_TO_DISABLE,

Exact

> 
> [...]
>> +static struct clk_regmap g12b_cpub_clk_trace = {
>> +       .data = &(struct clk_regmap_gate_data){
>> +               .offset = HHI_SYS_CPUB_CLK_CNTL1,
>> +               .bit_idx = 23,
> the public S922X datasheet calls this "Trace_clk_dis", does this mean
> you need a flag here?
>   .flags = CLK_GATE_SET_TO_DISABLE,

Exact

Thanks for the review !

Neil

> 
> 
> Regards
> Martin
>