Fix bpf_get/setsockopt failed when TCP over IPv4 via INET6 API

[PATCH bpf-next v3 1/2] bpf: Fix bpf_get/setsockopt to tos not take effect when TCP over IPv4 via INET6 API

Posted by Feng zhou 2 months, 2 weeks ago

From: Feng Zhou <zhoufeng.zf@bytedance.com>

when TCP over IPv4 via INET6 API, bpf_get/setsockopt with ipv4 will
fail, because sk->sk_family is AF_INET6. With ipv6 will success, not
take effect, because inet_csk(sk)->icsk_af_ops is ipv6_mapped and
use ip_queue_xmit, inet_sk(sk)->tos.

Bpf_get/setsockopt use sk_is_inet() helper to fix this case.

Signed-off-by: Feng Zhou <zhoufeng.zf@bytedance.com>
---
 net/core/filter.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/net/core/filter.c b/net/core/filter.c
index e4a4454df5f9..90f4dbb8d2b5 100644
--- a/net/core/filter.c
+++ b/net/core/filter.c
@@ -5399,7 +5399,12 @@ static int sol_ip_sockopt(struct sock *sk, int optname,
 			  char *optval, int *optlen,
 			  bool getopt)
 {
-	if (sk->sk_family != AF_INET)
+
+	/*
+	 * SOL_IP socket options are available on AF_INET and AF_INET6, for
+	 * example, TCP over IPv4 via INET6 API.
+	 */
+	if (!sk_is_inet(sk))
 		return -EINVAL;
 
 	switch (optname) {
-- 
2.30.2

Re: [PATCH bpf-next v3 1/2] bpf: Fix bpf_get/setsockopt to tos not take effect when TCP over IPv4 via INET6 API

Posted by Martin KaFai Lau 1 month, 4 weeks ago

On 9/14/24 3:32 AM, Feng zhou wrote:
> From: Feng Zhou <zhoufeng.zf@bytedance.com>
> 
> when TCP over IPv4 via INET6 API, bpf_get/setsockopt with ipv4 will

I think you meant bpf_get/setsockopt with SOL_IP will fail. so s/ipv4/SOL_IP/?

> fail, because sk->sk_family is AF_INET6. With ipv6 will success, not
> take effect, because inet_csk(sk)->icsk_af_ops is ipv6_mapped and
> use ip_queue_xmit, inet_sk(sk)->tos.

Change lgtm.

Patch 2 has a conflict, so can you please reword this commit message to reflect 
the latest change. e.g. afaik, this is no longer specific to mapped address or not.

> 
> Bpf_get/setsockopt use sk_is_inet() helper to fix this case.
> 
> Signed-off-by: Feng Zhou <zhoufeng.zf@bytedance.com>
> ---
>   net/core/filter.c | 7 ++++++-
>   1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/net/core/filter.c b/net/core/filter.c
> index e4a4454df5f9..90f4dbb8d2b5 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -5399,7 +5399,12 @@ static int sol_ip_sockopt(struct sock *sk, int optname,
>   			  char *optval, int *optlen,
>   			  bool getopt)
>   {
> -	if (sk->sk_family != AF_INET)
> +
> +	/*
> +	 * SOL_IP socket options are available on AF_INET and AF_INET6, for
> +	 * example, TCP over IPv4 via INET6 API.
> +	 */
> +	if (!sk_is_inet(sk))
>   		return -EINVAL;
>   
>   	switch (optname) {

Re: [External] Re: [PATCH bpf-next v3 1/2] bpf: Fix bpf_get/setsockopt to tos not take effect when TCP over IPv4 via INET6 API

Posted by Feng Zhou 1 month, 2 weeks ago

在 2024/10/1 10:27, Martin KaFai Lau 写道:
> On 9/14/24 3:32 AM, Feng zhou wrote:
>> From: Feng Zhou <zhoufeng.zf@bytedance.com>
>>
>> when TCP over IPv4 via INET6 API, bpf_get/setsockopt with ipv4 will
> 
> I think you meant bpf_get/setsockopt with SOL_IP will fail. so 
> s/ipv4/SOL_IP/?
> 
>> fail, because sk->sk_family is AF_INET6. With ipv6 will success, not
>> take effect, because inet_csk(sk)->icsk_af_ops is ipv6_mapped and
>> use ip_queue_xmit, inet_sk(sk)->tos.
> 
> Change lgtm.
> 
> Patch 2 has a conflict, so can you please reword this commit message to 
> reflect the latest change. e.g. afaik, this is no longer specific to 
> mapped address or not.
> 

Sorry for taking so long to reply.

Will do, thanks.

>>
>> Bpf_get/setsockopt use sk_is_inet() helper to fix this case.
>>
>> Signed-off-by: Feng Zhou <zhoufeng.zf@bytedance.com>
>> ---
>>   net/core/filter.c | 7 ++++++-
>>   1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/net/core/filter.c b/net/core/filter.c
>> index e4a4454df5f9..90f4dbb8d2b5 100644
>> --- a/net/core/filter.c
>> +++ b/net/core/filter.c
>> @@ -5399,7 +5399,12 @@ static int sol_ip_sockopt(struct sock *sk, int 
>> optname,
>>                 char *optval, int *optlen,
>>                 bool getopt)
>>   {
>> -    if (sk->sk_family != AF_INET)
>> +
>> +    /*
>> +     * SOL_IP socket options are available on AF_INET and AF_INET6, for
>> +     * example, TCP over IPv4 via INET6 API.
>> +     */
>> +    if (!sk_is_inet(sk))
>>           return -EINVAL;
>>       switch (optname) {
>

Re: [PATCH bpf-next v3 1/2] bpf: Fix bpf_get/setsockopt to tos not take effect when TCP over IPv4 via INET6 API

Posted by Eric Dumazet 2 months, 2 weeks ago

On Sat, Sep 14, 2024 at 12:32 PM Feng zhou <zhoufeng.zf@bytedance.com> wrote:
>
> From: Feng Zhou <zhoufeng.zf@bytedance.com>
>
> when TCP over IPv4 via INET6 API, bpf_get/setsockopt with ipv4 will
> fail, because sk->sk_family is AF_INET6. With ipv6 will success, not
> take effect, because inet_csk(sk)->icsk_af_ops is ipv6_mapped and
> use ip_queue_xmit, inet_sk(sk)->tos.
>
> Bpf_get/setsockopt use sk_is_inet() helper to fix this case.
>
> Signed-off-by: Feng Zhou <zhoufeng.zf@bytedance.com>

Reviewed-by: Eric Dumazet <edumazet@google.com>