Re: [PATCH 2/7] clk: qcom: Add Global Clock Controller driver for IPQ9574

[Devi Priya]

On 1/24/2023 3:23 PM, Konrad Dybcio wrote:
> On 24.01.2023 08:27, Devi Priya wrote:
> > On 1/13/2023 7:39 PM, Konrad Dybcio wrote:
> > > On 13.01.2023 14:21, Devi Priya wrote:
> > > > On 1/10/2023 6:07 PM, Konrad Dybcio wrote:
> > > > > On 10.01.2023 13:13, devi priya wrote:
> > > > > > Add Global Clock Controller (GCC) driver for ipq9574 based devices
> > > > > >
> > > > > > Co-developed-by: Anusha Rao <quic_anusha@xxxxxxxxxxx>
> > > > > > Signed-off-by: Anusha Rao <quic_anusha@xxxxxxxxxxx>
> > > > > > Signed-off-by: devi priya <quic_devipriy@xxxxxxxxxxx>
> > > > > > ---
> > > [...]
> > >
> > > > > > +static struct clk_branch gcc_blsp1_qup6_i2c_apps_clk = {
> > > > > > +    .halt_reg = 0x07024,
> > > > > > +    .clkr = {
> > > > > > +        .enable_reg = 0x07024,
> > > > > > +        .enable_mask = BIT(0),
> > > > > > +        .hw.init = &(struct clk_init_data) {
> > > > > > +            .name = "gcc_blsp1_qup6_i2c_apps_clk",
> > > > > > +            .parent_hws = (const struct clk_hw *[]) {
> > > > > > +                    &blsp1_qup6_i2c_apps_clk_src.clkr.hw },
> > > > > > +            .num_parents = 1,
> > > > > > +            .flags = CLK_SET_RATE_PARENT | CLK_IGNORE_UNUSED,
> > > > > Sounds very much like a hack..
> > > > Got it, will remove the clock entry as it is not being used in linux
> > > I'm not sure removing it is the best option, somebody might have a
> > > funky board where they use this particular QUP for I2C for whatever
> > > reason and then the clock would have to be re-added..
> > Sure, Understood
> > This clock is used by the RPM component to communicate with PMIC and we
> > would add the critical flag here
> Okay, so this SoC is intended to ship with some RPM PMICs and
> *always* with an I2C companion that's required for some basic
> functionality, correct?
>
> Otherwise, if it's just for wifi/multimedia/etc (like PM8008ij
> on some newer devices), you should not make it critical and
> simply rely on Linux keeping it alive like so:
>
> consumer takes a regulator
> the regulator does not go to sleep because it's consumed
> the PMIC is active because a regulator on it is being used
> the I2C bus is active because its child PMIC is used
> the I2C clocks are alive because there's an active user
Yes correct, the SoC would always have an I2C companion
> Konrad
> > > Thanks for addressing all of the review comments so thoroughly!
> > >
> > > Konrad
> >
> > Best Regards,
> > Devi Priya