Re: [PATCH 08/10 net-next v2] bpf: remove ipv6_bpf_stub completely and use direct function calls

[Daniel Borkmann]

On 3/10/26 4:34 PM, Fernando Fernandez Mancera wrote:
> As IPv6 is built-in only, the ipv6_bpf_stub can be removed completely.
>
> Convert all ipv6_bpf_stub usage to direct function calls instead. The
> fallback functions introduced previously will prevent linkage errors
> when CONFIG_IPV6 is disabled.
>
> Signed-off-by: Fernando Fernandez Mancera <fmancera@xxxxxxx>

[...]
> @@ -6000,11 +5996,9 @@ BPF_CALL_3(bpf_bind, struct bpf_sock_addr_kern *, ctx, struct sockaddr *, addr,
>   			return err;
>   		if (((struct sockaddr_in6 *)addr)->sin6_port == htons(0))
>   			flags |= BIND_FORCE_ADDRESS_NO_PORT;
> -		/* ipv6_bpf_stub cannot be NULL, since it's called from
> -		 * bpf_cgroup_inet6_connect hook and ipv6 is already loaded
> -		 */
> -		return ipv6_bpf_stub->inet6_bind(sk, (struct sockaddr_unsized *)addr,
> -						 addr_len, flags);
> +
> +		return inet6_bind_flags(sk, (struct sockaddr_unsized *)addr,
> +					addr_len, flags);

nit: You're adding the inet6_bind_flags just as an alias for __inet6_bind, might as
well just call the latter directly like we do in IPv4 case further above.

[...]> @@ -6283,12 +6277,11 @@ static int bpf_ipv6_fib_lookup(struct net *net, struct bpf_fib_lookup *params,
>   			params->tbid = 0;
>   		}
>   
> -		tb = ipv6_stub->fib6_get_table(net, tbid);
> +		tb = fib6_get_table(net, tbid);
>   		if (unlikely(!tb))
>   			return BPF_FIB_LKUP_RET_NOT_FWDED;
>   
> -		err = ipv6_stub->fib6_table_lookup(net, tb, oif, &fl6, &res,
> -						   strict);
> +		err = fib6_table_lookup(net, tb, oif, &fl6, &res, strict);
>   	} else {
>   		if (flags & BPF_FIB_LOOKUP_MARK)
>   			fl6.flowi6_mark = params->mark;

Love it, for the bpf_ipv6_fib_lookup we're now able to get rid of 3 subsequent
indirect calls in fast-path (ipv6_stub->fib6_lookup + ipv6_stub->fib6_select_path +
ipv6_bpf_stub->ipv6_dev_get_saddr in case BPF prog requested to fill src IP).

[...]
> diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
> index 448be9704313..75a9d9fe1308 100644
> --- a/net/ipv6/af_inet6.c
> +++ b/net/ipv6/af_inet6.c
> @@ -460,6 +460,12 @@ int inet6_bind_sk(struct sock *sk, struct sockaddr_unsized *uaddr, int addr_len)
>   	return __inet6_bind(sk, uaddr, addr_len, flags);
>   }
>   
> +int inet6_bind_flags(struct sock *sk, struct sockaddr_unsized *uaddr,
> +		     int addr_len, u32 flags)
> +{
> +	return __inet6_bind(sk, uaddr, addr_len, flags);
> +}
> +
(wrt above comment)

Acked-by: Daniel Borkmann <daniel@xxxxxxxxxxxxx>

Thanks,
Daniel