# Re: \[PATCH RFC 3/5\] rtnetlink: do\_setlink: Use sockaddr\_storage

- From: Eric Dumazet \<hidden\>
- Date: 2024-11-05 10:59:53 
- List: netdev
- Page: https://ratatoskr.run/netdev/2024/11/2813110
- Thread: https://ratatoskr.run/netdev/2024/11/2812451/t.md

```text

On 11/4/24 11:25 PM, Kees Cook wrote:
> Instead of a heap allocation use a stack allocated sockaddr_storage to
> support arbitrary length addr_len value (but bounds check it against the
> maximum address length).
>
> Signed-off-by: Kees Cook <kees@kernel.org>
> ---
>   net/core/rtnetlink.c | 12 ++++--------
>   1 file changed, 4 insertions(+), 8 deletions(-)
>
> diff --git a/net/core/rtnetlink.c b/net/core/rtnetlink.c
> index f0a520987085..eddd10b74f06 100644
> --- a/net/core/rtnetlink.c
> +++ b/net/core/rtnetlink.c
> @@ -2839,21 +2839,17 @@ static int do_setlink(const struct sk_buff *skb,
>   	}
>   
>   	if (tb[IFLA_ADDRESS]) {
> -		struct sockaddr *sa;
> -		int len;
> +		struct sockaddr_storage addr;
> +		struct sockaddr *sa = (struct sockaddr *)&addr;


We already use too much stack space.


Please move addr into struct rtnl_newlink_tbs ?


>   
> -		len = sizeof(sa_family_t) + max_t(size_t, dev->addr_len,
> -						  sizeof(*sa));
> -		sa = kmalloc(len, GFP_KERNEL);
> -		if (!sa) {
> +		if (dev->addr_len > sizeof(addr.__data)) {
>   			err = -ENOMEM;
>   			goto errout;
>   		}
>   		sa->sa_family = dev->type;
> -		memcpy(sa->sa_data, nla_data(tb[IFLA_ADDRESS]),
> +		memcpy(addr.__data, nla_data(tb[IFLA_ADDRESS]),
>   		       dev->addr_len);
>   		err = dev_set_mac_address_user(dev, sa, extack);
> -		kfree(sa);
>   		if (err)
>   			goto errout;
>   		status |= DO_SETLINK_MODIFIED;
```