On Tue 01 Oct 2019 at 20:53, Martin Blumenstingl <martin.blumenstingl@xxxxxxxxxxxxxx> wrote:
>
> [...]
>> > +static struct clk_hw_onecell_data meson8_ddr_clk_hw_onecell_data = {
>> > + .hws = {
>> > + [DDR_CLKID_DDR_PLL_DCO] = &meson8_ddr_pll_dco.hw,
>> > + [DDR_CLKID_DDR_PLL] = &meson8_ddr_pll.hw,
>>
>> I wonder if onecell is not overkill for this driver. We won't expose the
>> DCO, so only the post divider remains
>>
>> Do you expect this provider to have more than one leaf clock ?
>> If not, maybe you could use of_clk_hw_simple_get() instead ?
> there's some more clock bits in DDR_CLK_CNTL - the public A311D
> datasheet has a description for these bits but I'm not sure they do
> the same on Meson8/Meson8b/Meson8m2
> all I know is that some magic bytes are written to DDR_CLK_CNTL in the
> old u-boot code
>
> that's why I don't want to make any assumptions and play safe here (by
> using a onecell clock provider)
Understood. Let's keep onecell then.
