Re: [PATCH] bus: qcom-ebi2: unwind clocks on child setup failures

[Konrad Dybcio]

On 6/16/26 5:17 PM, Pengpeng Hou wrote:
> qcom_ebi2_probe() enables both EBI2 clocks before walking child
> chip-select nodes.  A missing child reg property returns directly from
> that loop, bypassing the existing clock-disable labels.
> 
> Route that failure through the existing unwind path.  Also check the
> final child population result, depopulate any partially created children,
> and disable the clocks if population fails.
> 
> Signed-off-by: Pengpeng Hou <pengpeng@xxxxxxxxxxx>
> ---
>  drivers/bus/qcom-ebi2.c | 12 +++++++++---
>  1 file changed, 9 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/bus/qcom-ebi2.c b/drivers/bus/qcom-ebi2.c
> index be8166565e7c..68fe931d1636 100644
> --- a/drivers/bus/qcom-ebi2.c
> +++ b/drivers/bus/qcom-ebi2.c
> @@ -353,7 +353,7 @@ static int qcom_ebi2_probe(struct platform_device *pdev)
>  		/* Figure out the chipselect */
>  		ret = of_property_read_u32(child, "reg", &csindex);
>  		if (ret)
> -			return ret;
> +			goto err_disable_clk;
>  
>  		if (csindex > 5) {
>  			dev_err(dev,
> @@ -372,8 +372,14 @@ static int qcom_ebi2_probe(struct platform_device *pdev)
>  		have_children = true;
>  	}
>  
> -	if (have_children)
> -		return of_platform_default_populate(np, NULL, dev);
> +	if (have_children) {
> +		ret = of_platform_default_populate(np, NULL, dev);
> +		if (ret) {
> +			of_platform_depopulate(dev);

We can use devm_of_platform_populate() instead (which will match any child
node instead of just the simple-bus-y ones, but that's fine) and
devm_clk_get_enabled() to handle what you're addressing here, then we
need no more unwinding

Konrad