Re: [PATCH v4 17/52] clk: qcom: gcc-qcs615: Tie the CX power domain to controller

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] Enabling runtime PM (`use_rpm = true`) on a clock controller with `CLK_IS_CRITICAL` clocks causes a permanent PM runtime usage count leak, preventing the controller (and its attached CX power domain) from ever suspending.
--

commit 660d870b134f3691d16847f677af3d1122921193
Author: Abel Vesa <abel.vesa@oss.qualcomm.com>

clk: qcom: gcc-qcs615: Tie the CX power domain to controller

This commit enables runtime PM for the QCS615 GCC controller. It aims to allow
the CX power domain to power down once all GCC consumers release their clocks
and GDSCs.

> diff --git a/drivers/clk/qcom/gcc-qcs615.c b/drivers/clk/qcom/gcc-qcs615.c
> index 57f8c80c6f324..914a49573811f 100644
> --- a/drivers/clk/qcom/gcc-qcs615.c
> +++ b/drivers/clk/qcom/gcc-qcs615.c
> @@ -2968,6 +2968,7 @@ static const struct qcom_cc_desc gcc_qcs615_desc = {
>  	.num_resets = ARRAY_SIZE(gcc_qcs615_resets),
>  	.gdscs = gcc_qcs615_gdscs,
>  	.num_gdscs = ARRAY_SIZE(gcc_qcs615_gdscs),
> +	.use_rpm = true,

[Severity: High]
Will this actually allow the CX domain to power down, given the presence of
critical clocks in this controller?

Enabling runtime PM here instructs the clock framework to take PM runtime
references on the GCC device. However, this controller defines several clocks
with the CLK_IS_CRITICAL flag, such as gcc_cpuss_ahb_clk and
gcc_sys_noc_cpuss_ahb_clk:

static struct clk_branch gcc_cpuss_ahb_clk = {
        ...
        .clkr.hw.init = &(const struct clk_init_data) {
                ...
                .flags = CLK_IS_CRITICAL | CLK_SET_RATE_PARENT,
        },
};

When qcom_cc_really_probe() registers these clocks, the clock core sees
CLK_IS_CRITICAL and automatically prepares them via clk_core_prepare(). This
internally calls clk_pm_runtime_get(), which increments the GCC device's PM
runtime usage counter via pm_runtime_resume_and_get().

Because critical clocks are never unprepared, won't this result in a permanent
PM runtime usage count leak that prevents the controller and its attached CX
power domain from ever suspending?

>  };
>  
>  static const struct of_device_id gcc_qcs615_match_table[] = {

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260914-qcom-tie-cx-to-gcc-v4-0-9d5a606ada33@oss.qualcomm.com?part=17