Re: [PATCH 06/14] clk: qcom: clk-branch: Add support for BRANCH_HALT_POLL flag

[Stephen Boyd <sboyd@kernel.org>]

Quoting Dmitry Baryshkov (2024-10-17 09:56:56)
> From: Kalpak Kawadkar <quic_kkawadka@quicinc.com>
> 
> On some platforms branch clock will be enabled before Linux.
> It is expectated from the clock provider is to poll on the clock

Unfortunately 'expectated' is not a word. The sentence is also
grammatically incorrect.

> to ensure it is indeed enabled and not HW gated, thus add
> the BRANCH_HALT_POLL flag.
[...]
> diff --git a/drivers/clk/qcom/clk-branch.c b/drivers/clk/qcom/clk-branch.c
> index 229480c5b075a0e70dc05b1cb15b88d29fd475ce..c4c7bd565cc9a3926e24bb12ed6355ec6ddd19fb 100644
> --- a/drivers/clk/qcom/clk-branch.c
> +++ b/drivers/clk/qcom/clk-branch.c
> @@ -76,6 +76,7 @@ static int clk_branch_wait(const struct clk_branch *br, bool enabling,
>                 udelay(10);
>         } else if (br->halt_check == BRANCH_HALT_ENABLE ||
>                    br->halt_check == BRANCH_HALT ||
> +                  br->halt_check == BRANCH_HALT_POLL ||

The name is confusing. The halt check is already "polling", i.e. this
isn't a different type of halt check. This is really something like
another branch flag that doesn't have to do with the halt checking and
only to do with skipping writing the enable bit. Maybe we should
introduce another clk_ops for these types of clks instead.

>                    (enabling && voted)) {
>                 int count = 200;
>  
> @@ -97,6 +98,10 @@ static int clk_branch_toggle(struct clk_hw *hw, bool en,
>         struct clk_branch *br = to_clk_branch(hw);
>         int ret;
>  
> +       if (br->halt_check == BRANCH_HALT_POLL) {

Remove braces

> +               return  clk_branch_wait(br, en, check_halt);

Remove extra space      ^

> +       }
> +