Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20230105100014.1935-1-cuiyunhui@bytedance.com> <20230105090807.51213cea@gandalf.local.home>
In-Reply-To: <20230105090807.51213cea@gandalf.local.home>
From:   =?UTF-8?B?6L+Q6L6J5bSU?= <cuiyunhui@bytedance.com>
Date:   Fri, 6 Jan 2023 11:24:18 +0800
Message-ID: <CAEEQ3wmnUpuZxLbrMwPCk+pnbC27fo4bi-nN-848o6QGO6Xt+Q@mail.gmail.com>
Subject: Re: [External] Re: [PATCH v2] sock: add tracepoint for send recv length
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     mhiramat@kernel.org, davem@davemloft.net, edumazet@google.com,
        kuba@kernel.org, pabeni@redhat.com, duanxiongchun@bytedance.com,
        kuniyu@amazon.com, linux-kernel@vger.kernel.org,
        linux-trace-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Precedence: bulk

On Thu, Jan 5, 2023 at 10:08 PM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Thu,  5 Jan 2023 18:00:14 +0800
> Yunhui Cui <cuiyunhui@bytedance.com> wrote:
>
> > @@ -715,6 +716,10 @@ static inline int sock_sendmsg_nosec(struct socket=
 *sock, struct msghdr *msg)
> >                                    inet_sendmsg, sock, msg,
> >                                    msg_data_left(msg));
> >       BUG_ON(ret =3D=3D -EIOCBQUEUED);
> > +
> > +     trace_sock_sendmsg_length(sock->sk, sock->sk->sk_family,
> > +                               sock->sk->sk_protocol,
> > +                               ret > 0 ? ret : 0, ret > 0 ? 0 : ret, 0=
);
>
> The length and error seem confusing. Basically, length is ret > 0 and err=
or
> is ret < 0, right?
>
> That means we don't need both. We could simply have:
>
> > +     TP_ARGS(sk, family, protocol, length, error, flags),
>
>         TP_ARGS(sk, family, protocol, ret, flags)
>

Hi Steve, thank you for your advice=EF=BC=8Ci'll modify it on v3.

> >
> > @@ -992,9 +997,17 @@ INDIRECT_CALLABLE_DECLARE(int inet6_recvmsg(struct=
 socket *, struct msghdr *,
> >  static inline int sock_recvmsg_nosec(struct socket *sock, struct msghd=
r *msg,
> >                                    int flags)
> >  {
> > -     return INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg,
> > +     int ret =3D INDIRECT_CALL_INET(sock->ops->recvmsg, inet6_recvmsg,
> >                                 inet_recvmsg, sock, msg, msg_data_left(=
msg),
> >                                 flags);
> > +
> > +     trace_sock_recvmsg_length(sock->sk, sock->sk->sk_family,
> > +                               sock->sk->sk_protocol,
> > +                               (ret > 0 && !(flags & MSG_PEEK)) ?
> > +                               ret : 0,
> > +                               (ret > 0 && !(flags & MSG_PEEK)) ? 0 : =
ret,
>
> Since both length and error are zero when flags has MSG_PEEK set:
>
>         trace_sock_recvmsg_length(sock->sk, sock->sk->sk_family,
>                                   sock->sk->sk_protocol,
>                                   !(flags & MSG_PEEK) ? ret : 0,
>
> -- Steve

Regardless of whether the MSG_PEEK flag is set or not, it is possible
to return -errno,
but based on your suggestion, I plan to modify it like this:

trace_sock_recvmsg_length(sock->sk, sock->sk->sk_family,
                          sock->sk->sk_protocol,
                          !(flags & MSG_PEEK) ? ret : (ret < 0 ? ret : 0),

what do you think?

Thanks,
Yunhui