Re: [Patch net-next 02/13] ipv4: introduce tracepoint trace_ip_queue_xmit()

[Martin KaFai Lau <kafai@fb.com>]

On Thu, Aug 05, 2021 at 11:57:39AM -0700, Cong Wang wrote:
> From: Qitao Xu <qitao.xu@bytedance.com>
> 
> Tracepoint trace_ip_queue_xmit() is introduced to trace skb
> at the entrance of IP layer on TX side.
> 
> Reviewed-by: Cong Wang <cong.wang@bytedance.com>
> Signed-off-by: Qitao Xu <qitao.xu@bytedance.com>
> ---
>  include/trace/events/ip.h | 42 +++++++++++++++++++++++++++++++++++++++
>  net/ipv4/ip_output.c      | 10 +++++++++-
>  2 files changed, 51 insertions(+), 1 deletion(-)
> 
> diff --git a/include/trace/events/ip.h b/include/trace/events/ip.h
> index 008f821ebc50..553ae7276732 100644
> --- a/include/trace/events/ip.h
> +++ b/include/trace/events/ip.h
> @@ -41,6 +41,48 @@
>  	TP_STORE_V4MAPPED(__entry, saddr, daddr)
>  #endif
>  
> +TRACE_EVENT(ip_queue_xmit,
> +
> +	TP_PROTO(const struct sock *sk, const struct sk_buff *skb),
> +
> +	TP_ARGS(sk, skb),
> +
> +	TP_STRUCT__entry(
> +		__field(const void *, skbaddr)
> +		__field(const void *, skaddr)
> +		__field(__u16, sport)
> +		__field(__u16, dport)
> +		__array(__u8, saddr, 4)
> +		__array(__u8, daddr, 4)
> +		__array(__u8, saddr_v6, 16)
> +		__array(__u8, daddr_v6, 16)
> +	),
> +
> +	TP_fast_assign(
> +		struct inet_sock *inet = inet_sk(sk);
> +		__be32 *p32;
> +
> +		__entry->skbaddr = skb;
> +		__entry->skaddr = sk;
> +
> +		__entry->sport = ntohs(inet->inet_sport);
> +		__entry->dport = ntohs(inet->inet_dport);
> +
> +		p32 = (__be32 *) __entry->saddr;
> +		*p32 = inet->inet_saddr;
> +
> +		p32 = (__be32 *) __entry->daddr;
> +		*p32 =  inet->inet_daddr;
> +
> +		TP_STORE_ADDRS(__entry, inet->inet_saddr, inet->inet_daddr,
> +			      sk->sk_v6_rcv_saddr, sk->sk_v6_daddr);
> +	),
> +
> +	TP_printk("sport=%hu dport=%hu saddr=%pI4 daddr=%pI4 saddrv6=%pI6c daddrv6=%pI6c skbaddr=%px",
> +		  __entry->sport, __entry->dport, __entry->saddr, __entry->daddr,
> +		  __entry->saddr_v6, __entry->daddr_v6, __entry->skbaddr)
> +);
> +
>  #endif /* _TRACE_IP_H */
>  
>  /* This part must be outside protection */
> diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c
> index 6b04a88466b2..dcf94059112e 100644
> --- a/net/ipv4/ip_output.c
> +++ b/net/ipv4/ip_output.c
> @@ -82,6 +82,7 @@
>  #include <linux/netfilter_bridge.h>
>  #include <linux/netlink.h>
>  #include <linux/tcp.h>
> +#include <trace/events/ip.h>
>  
>  static int
>  ip_fragment(struct net *net, struct sock *sk, struct sk_buff *skb,
> @@ -536,7 +537,14 @@ EXPORT_SYMBOL(__ip_queue_xmit);
>  
>  int ip_queue_xmit(struct sock *sk, struct sk_buff *skb, struct flowi *fl)
>  {
> -	return __ip_queue_xmit(sk, skb, fl, inet_sk(sk)->tos);
> +	int ret;
> +
> +	ret = __ip_queue_xmit(sk, skb, fl, inet_sk(sk)->tos);
> +	if (!ret)
> +		trace_ip_queue_xmit(sk, skb);
Instead of adding tracepoints, the bpf fexit prog can be used here and
the bpf prog will have the sk, skb, and ret available (example in fexit_test.c).
Some tracepoints in this set can also be done with bpf fentry/fexit.
Does bpf fentry/fexit work for your use case?