Date:   Fri, 31 Dec 2021 18:22:23 -0800
From:   Jakub Kicinski <kuba@kernel.org>
To:     Menglong Dong <menglong8.dong@gmail.com>
Cc:     Steven Rostedt <rostedt@goodmis.org>,
        David Ahern <dsahern@kernel.org>, mingo@redhat.com,
        David Miller <davem@davemloft.net>,
        Neil Horman <nhorman@tuxdriver.com>,
        Eric Dumazet <edumazet@google.com>,
        Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
        jonathan.lemon@gmail.com, alobakin@pm.me,
        Kees Cook <keescook@chromium.org>,
        Paolo Abeni <pabeni@redhat.com>, talalahmad@google.com,
        haokexin@gmail.com, Menglong Dong <imagedong@tencent.com>,
        atenart@kernel.org, bigeasy@linutronix.de,
        Wei Wang <weiwan@google.com>, arnd@arndb.de, vvs@virtuozzo.com,
        Cong Wang <cong.wang@bytedance.com>,
        LKML <linux-kernel@vger.kernel.org>,
        netdev <netdev@vger.kernel.org>,
        Mengen Sun <mengensun@tencent.com>, mungerjiang@tencent.com
Subject: Re: [PATCH v2 net-next 1/3] net: skb: introduce
 kfree_skb_with_reason()
Message-ID: <20211231182223.43afa349@kicinski-fedora-PC1C0HJN.hsd1.ca.comcast.net>
In-Reply-To: <CADxym3aonWQoR=XkoLqn_taEhjBYeMf7f2Tgjgnq7fCNT2kHNw@mail.gmail.com>
References: <20211230093240.1125937-1-imagedong@tencent.com>
        <20211230093240.1125937-2-imagedong@tencent.com>
        <20211230172619.40603ff3@kicinski-fedora-PC1C0HJN.hsd1.ca.comcast.net>
        <CADxym3aonWQoR=XkoLqn_taEhjBYeMf7f2Tgjgnq7fCNT2kHNw@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Fri, 31 Dec 2021 14:35:31 +0800 Menglong Dong wrote:
> > >  void skb_release_head_state(struct sk_buff *skb);
> > >  void kfree_skb(struct sk_buff *skb);  
> >
> > Should this be turned into a static inline calling
> > kfree_skb_with_reason() now? BTW you should drop the
> > '_with'.
> >  
> 
> I thought about it before, but I'm a little afraid that some users may trace
> kfree_skb() with kprobe, making it inline may not be friendly to them?

Hm, there is a bpf sample which does that, but that's probably 
not commonly used given there is a tracepoint. If someone is 
using a kprobe they can switch to kprobing kfree_skb*reason().