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;
From:   Kuniyuki Iwashima <kuniyu@amazon.com>
To:     <martin.lau@linux.dev>
CC:     <andrii@kernel.org>, <ast@kernel.org>, <bpf@vger.kernel.org>,
        <daniel@iogearbox.net>, <davem@davemloft.net>,
        <dsahern@kernel.org>, <edumazet@google.com>, <haoluo@google.com>,
        <joe@cilium.io>, <joe@wand.net.nz>, <john.fastabend@gmail.com>,
        <jolsa@kernel.org>, <kafai@fb.com>, <kpsingh@kernel.org>,
        <kuba@kernel.org>, <linux-kernel@vger.kernel.org>,
        <lmb@isovalent.com>, <netdev@vger.kernel.org>, <pabeni@redhat.com>,
        <sdf@google.com>, <song@kernel.org>,
        <willemdebruijn.kernel@gmail.com>, <yhs@fb.com>,
        <kuniyu@amazon.com>
Subject: Re: [PATCH bpf-next 1/2] bpf, net: Support SO_REUSEPORT sockets with bpf_sk_assign
Date:   Thu, 25 May 2023 18:43:17 -0700
Message-ID: <20230526014317.80715-1-kuniyu@amazon.com>
In-Reply-To: <7188429a-c380-14c8-57bb-9d05d3ba4e5e@linux.dev>
References: <7188429a-c380-14c8-57bb-9d05d3ba4e5e@linux.dev>
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Type: text/plain
Precedence: bulk

From: Martin KaFai Lau <martin.lau@linux.dev>
Date: Thu, 25 May 2023 16:42:46 -0700
> On 5/25/23 1:19 AM, Lorenz Bauer wrote:
> > diff --git a/include/net/inet6_hashtables.h b/include/net/inet6_hashtables.h
> > index 56f1286583d3..3ba4dc2703da 100644
> > --- a/include/net/inet6_hashtables.h
> > +++ b/include/net/inet6_hashtables.h
> > @@ -48,6 +48,13 @@ struct sock *__inet6_lookup_established(struct net *net,
> >   					const u16 hnum, const int dif,
> >   					const int sdif);
> >   
> > +struct sock *inet6_lookup_reuseport(struct net *net, struct sock *sk,
> > +				    struct sk_buff *skb, int doff,
> > +				    const struct in6_addr *saddr,
> > +				    __be16 sport,
> > +				    const struct in6_addr *daddr,
> > +				    unsigned short hnum);
> > +
> >   struct sock *inet6_lookup_listener(struct net *net,
> >   				   struct inet_hashinfo *hashinfo,
> >   				   struct sk_buff *skb, int doff,
> > @@ -85,14 +92,33 @@ static inline struct sock *__inet6_lookup_skb(struct inet_hashinfo *hashinfo,
> >   					      int iif, int sdif,
> >   					      bool *refcounted)
> >   {
> > -	struct sock *sk = skb_steal_sock(skb, refcounted);
> > -
> > +	bool prefetched;
> > +	struct sock *sk = skb_steal_sock(skb, refcounted, &prefetched);
> > +	struct net *net = dev_net(skb_dst(skb)->dev);
> > +	const struct ipv6hdr *ip6h = ipv6_hdr(skb);
> > +
> > +	if (prefetched) {
> > +		struct sock *reuse_sk = inet6_lookup_reuseport(net, sk, skb, doff,
> 
> If sk is TCP_ESTABLISHED, I suspect sk->sk_reuseport is 1 (from sk_clone)?

Exactly, it will cause null-ptr-deref in reuseport_select_sock().
We may want to use rcu_access_pointer(sk->sk_reuseport_cb) in
each lookup_reuseport() instead of adding sk_state check ?


> 
> If it is, it should still work other than an extra inet6_ehashfn. Does it worth 
> an extra sk->sk_state check or it is overkill?
> 
> 
> > +							       &ip6h->saddr, sport,
> > +							       &ip6h->daddr, ntohs(dport));