Re: [PATCH v2] af_key: Add check for the return value of pfkey_sadb2xfrm_user_sec_ctx()

[Paolo Abeni]

On 7/7/25 6:05 PM, Haoxiang Li wrote:
> Add check for the return value of pfkey_sadb2xfrm_user_sec_ctx()
> in pfkey_compile_policy(), and set proper error flag.
> 
> Signed-off-by: Haoxiang Li <haoxiang_li2024@xxxxxxx>
> ---
> Changes in v2:
> - Set error flag '*dir' properly.
> - Hi, Steffen! I know that inside pfkey_sadb2xfrm_user_sec_ctx(), null
> value check has been done. This patch does the null value check after
> pfkey_sadb2xfrm_user_sec_ctx() being called in pfkey_compile_policy().
> Also, set proper error flag if pfkey_sadb2xfrm_user_sec_ctx() returns
> null. This patch code is similar to [1]. Thanks, Steffen!
> 
> [1]https://github.com/torvalds/linux/blob/master/net/key/af_key.c#L2404
> ---
>  net/key/af_key.c | 5 +++++
>  1 file changed, 5 insertions(+)
> 
> diff --git a/net/key/af_key.c b/net/key/af_key.c
> index efc2a91f4c48..9cd14a31a427 100644
> --- a/net/key/af_key.c
> +++ b/net/key/af_key.c
> @@ -3335,6 +3335,11 @@ static struct xfrm_policy *pfkey_compile_policy(struct sock *sk, int opt,
>  		if ((*dir = verify_sec_ctx_len(p)))
>  			goto out;
>  		uctx = pfkey_sadb2xfrm_user_sec_ctx(sec_ctx, GFP_ATOMIC);
> +		if (!uctx) {
> +			*dir = -ENOMEM;
> +			goto out;
> +		}
> +
>  		*dir = security_xfrm_policy_alloc(&xp->security, uctx, GFP_ATOMIC);

AFAICS security_xfrm_policy_alloc() handle safely 'uctx' arguments ...

>  		kfree(uctx);

... and kfree, too.

This patch looks not needed.

/P