Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:18 as permitted sender) client-ip=2620:137:e000::1:18;
Date:   Fri, 6 May 2022 16:34:43 -0700
From:   Peilin Ye <yepeilin.cs@gmail.com>
To:     Stephen Hemminger <stephen@networkplumber.org>
Cc:     "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <eric.dumazet@gmail.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>,
        Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
        David Ahern <dsahern@kernel.org>,
        Jamal Hadi Salim <jhs@mojatatu.com>,
        Cong Wang <xiyou.wangcong@gmail.com>,
        Jiri Pirko <jiri@resnulli.us>,
        Peilin Ye <peilin.ye@bytedance.com>, netdev@vger.kernel.org,
        linux-kernel@vger.kernel.org, Cong Wang <cong.wang@bytedance.com>
Subject: Re: [PATCH RFC v1 net-next 1/4] net: Introduce Qdisc backpressure
 infrastructure
Message-ID: <20220506233443.GA3336@bytedance>
References: <cover.1651800598.git.peilin.ye@bytedance.com>
 <f4090d129b685df72070f708294550fbc513f888.1651800598.git.peilin.ye@bytedance.com>
 <20220506133111.1d4bebf3@hermes.local>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20220506133111.1d4bebf3@hermes.local>
Precedence: bulk

Hi Stephen,

On Fri, May 06, 2022 at 01:31:11PM -0700, Stephen Hemminger wrote:
> On Fri,  6 May 2022 12:44:22 -0700, Peilin Ye <yepeilin.cs@gmail.com> wrote:
> > +static inline void qdisc_backpressure_overlimit(struct Qdisc *sch, struct sk_buff *skb)
> > +{
> > +	struct sock *sk = skb->sk;
> > +
> > +	if (!sk || !sk_fullsock(sk))
> > +		return;
> > +
> > +	if (cmpxchg(&sk->sk_backpressure_status, SK_UNTHROTTLED, SK_OVERLIMIT) == SK_UNTHROTTLED) {
> > +		sock_hold(sk);
> > +		list_add_tail(&sk->sk_backpressure_node, &sch->backpressure_list);
> > +	}
> > +}
> 
> What if socket is closed? You are holding reference but application maybe gone.

Thanks for pointing this out!  I just understood how sk_refcnt works
together with sk_wmem_alloc.

By the time we process this in-flight skb, sk_refcnt may have already
reached 0, which means sk_free() may have already decreased that "extra" 1
sk_wmem_alloc, so skb->destructor() may call __sk_free() while I "hold"
the sock here.  Seems like a UAF.

> Or if output is stalled indefinitely?

It would be better to do a cleanup in sock destroying code, but I am
trying to avoid acquiring Qdisc root_lock there.  I will try to come up
with a better solution.

Thanks,
Peilin Ye