Re: [PATCH V2] clk: imx: gate off peripheral clock slice

[Abel Vesa <abel.vesa@nxp.com>]

On 21-10-25 20:29:02, Peng Fan (OSS) wrote:
> From: Peng Fan <peng.fan@nxp.com>
> 
> The Peripheral clocks are default enabled when SoC power on, and
> bootloader not gate off the clocks when booting Linux Kernel.
> 
> So Linux Kernel is not aware the peripheral clocks are enabled and
> still take them as disabled because of enable count is zero.
> 
> Then Peripheral clock's source without clock gated off could be
> changed when have assigned-parents in device tree
> 
> However, per i.MX8M* reference mannual, "Peripheral clock slices must
> be stopped to change the clock source", so need to gate off the
> the peripheral clock when registering the clocks to avoid glitch.
> 
> Tested boot on i.MX8MM/P-EVK board
> 
> Fixes: d3ff9728134e ("clk: imx: Add imx composite clock")
> Signed-off-by: Peng Fan <peng.fan@nxp.com>
> ---
> 
> V2:
>  Add Fixes tag
> 
>  drivers/clk/imx/clk-composite-8m.c | 9 ++++++++-
>  1 file changed, 8 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/clk/imx/clk-composite-8m.c b/drivers/clk/imx/clk-composite-8m.c
> index 2dfd6149e528..ee41fbf90589 100644
> --- a/drivers/clk/imx/clk-composite-8m.c
> +++ b/drivers/clk/imx/clk-composite-8m.c
> @@ -184,6 +184,7 @@ struct clk_hw *__imx8m_clk_hw_composite(const char *name,
>  	struct clk_mux *mux = NULL;
>  	const struct clk_ops *divider_ops;
>  	const struct clk_ops *mux_ops;
> +	u32 val;
>  
>  	mux = kzalloc(sizeof(*mux), GFP_KERNEL);
>  	if (!mux)
> @@ -216,8 +217,14 @@ struct clk_hw *__imx8m_clk_hw_composite(const char *name,
>  		div->width = PCG_PREDIV_WIDTH;
>  		divider_ops = &imx8m_clk_composite_divider_ops;
>  		mux_ops = &clk_mux_ops;
> -		if (!(composite_flags & IMX_COMPOSITE_FW_MANAGED))
> +		if (!(composite_flags & IMX_COMPOSITE_FW_MANAGED)) {
>  			flags |= CLK_SET_PARENT_GATE;
> +			if (!(flags & CLK_IS_CRITICAL)) {
> +				val = readl(reg);
> +				val &= ~BIT(PCG_CGC_SHIFT);
> +				writel(val, reg);
> +			}
> +		}

Though I'm usually against special cases like this one. I think the clock
core needs some generic flag that would read the state from HW on probe
and/or another generic flag for disabling on probe.

But for now, I'm OK with this:

Reviewed-by: Abel Vesa <abel.vesa@nxp.com>

>  	}
>  
>  	div->lock = &imx_ccm_lock;
> -- 
> 2.30.0
>