Re: [PATCH] clk: fix memory leak in hisi_clk_init.

[Jonathan Neuschäfer]

Hello,

sorry for the delay, but I found this unmerged patch while cleaning up
my mailbox.

On Thu, Dec 22, 2022 at 09:12:21AM +0000, Tao Lan wrote:
> From: taolan <taolan@xxxxxxxxxx>
> 
> when clk_data create fail, we also need to release base.
> 
> Signed-off-by: taolan <taolan@xxxxxxxxxx>
> ---
>  drivers/clk/hisilicon/clk.c | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/drivers/clk/hisilicon/clk.c b/drivers/clk/hisilicon/clk.c
> index 54d9fdc93599..9ca4fc05fa57 100644
> --- a/drivers/clk/hisilicon/clk.c
> +++ b/drivers/clk/hisilicon/clk.c
> @@ -82,6 +82,10 @@ struct hisi_clock_data *hisi_clk_init(struct device_node *np,
>  	of_clk_add_provider(np, of_clk_src_onecell_get, &clk_data->clk_data);
>  	return clk_data;
>  err_data:
> +	if (base) {
> +		iounmap(base);
> +		base = NULL;
> +	}

This is inaccurate. Consider the case when kzalloc fails:

	clk_data = kzalloc(sizeof(*clk_data), GFP_KERNEL);
	if (!clk_data)
		goto err;

base has already been mapped, but the code jumps to 'err', which doesn't
have the iounmap.

To address this properly, you'd have to add another label between
err_data and err.


>  	kfree(clk_data);
>  err:
>  	return NULL;


Best regards,
jn

Attachment: signature.asc
Description: PGP signature