[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20241024191356.38734-1-kuniyu@amazon.com>
Date: Thu, 24 Oct 2024 12:13:56 -0700
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <laoar.shao@...il.com>
CC: <davem@...emloft.net>, <dsahern@...nel.org>, <edumazet@...gle.com>,
<kuba@...nel.org>, <menglong8.dong@...il.com>, <netdev@...r.kernel.org>,
<pabeni@...hat.com>, <kuniyu@...zon.com>, <dxu@...uu.xyz>
Subject: Re: [PATCH 2/2] net: tcp: Add noinline_for_tracing annotation for tcp_drop_reason()
From: Yafang Shao <laoar.shao@...il.com>
Date: Thu, 24 Oct 2024 17:37:42 +0800
> We previously hooked the tcp_drop_reason() function using BPF to monitor
> TCP drop reasons. However, after upgrading our compiler from GCC 9 to GCC
> 11, tcp_drop_reason() is now inlined, preventing us from hooking into it.
> To address this, it would be beneficial to make noinline explicitly for
> tracing.
>
> Link: https://lore.kernel.org/netdev/CANn89iJuShCmidCi_ZkYABtmscwbVjhuDta1MS5LxV_4H9tKOA@mail.gmail.com/
> Suggested-by: Eric Dumazet <edumazet@...gle.com>
> Signed-off-by: Yafang Shao <laoar.shao@...il.com>
> Cc: Menglong Dong <menglong8.dong@...il.com>
I saw a somewhat related post yesterday.
https://x.com/__dxu/status/1849271647989068107
Daniel, could we apply your approach to this issue in the near future ?
Thread link: https://lore.kernel.org/netdev/20241024093742.87681-1-laoar.shao@gmail.com/
> ---
> net/ipv4/tcp_input.c | 4 ++--
> 1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
> index cc05ec1faac8..cf1e2e3a7407 100644
> --- a/net/ipv4/tcp_input.c
> +++ b/net/ipv4/tcp_input.c
> @@ -4894,8 +4894,8 @@ static bool tcp_ooo_try_coalesce(struct sock *sk,
> return res;
> }
>
> -static void tcp_drop_reason(struct sock *sk, struct sk_buff *skb,
> - enum skb_drop_reason reason)
> +noinline_for_tracing static void
> +tcp_drop_reason(struct sock *sk, struct sk_buff *skb, enum skb_drop_reason reason)
> {
> sk_drops_add(sk, skb);
> sk_skb_reason_drop(sk, skb, reason);
> --
> 2.43.5
>
Powered by blists - more mailing lists