[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20221006171038.68453-1-kuniyu@amazon.com>
Date: Thu, 6 Oct 2022 10:10:38 -0700
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <pabeni@...hat.com>
CC: <davem@...emloft.net>, <dsahern@...nel.org>, <edumazet@...gle.com>,
<kuba@...nel.org>, <kuni1840@...il.com>, <kuniyu@...zon.com>,
<linux-kernel@...r.kernel.org>, <netdev@...r.kernel.org>,
<syzkaller-bugs@...glegroups.com>, <vyasevic@...hat.com>,
<yoshfuji@...ux-ipv6.org>
Subject: Re: [PATCH v4 net 3/5] tcp/udp: Call inet6_destroy_sock() in IPv6 sk->sk_destruct().
From: Paolo Abeni <pabeni@...hat.com>
Date: Thu, 06 Oct 2022 11:19:53 +0200
> On Tue, 2022-10-04 at 10:18 -0700, Kuniyuki Iwashima wrote:
> > Originally, inet6_sk(sk)->XXX were changed under lock_sock(), so we were
> > able to clean them up by calling inet6_destroy_sock() during the IPv6 ->
> > IPv4 conversion by IPV6_ADDRFORM. However, commit 03485f2adcde ("udpv6:
> > Add lockless sendmsg() support") added a lockless memory allocation path,
> > which could cause a memory leak:
> >
> > setsockopt(IPV6_ADDRFORM) sendmsg()
> > +-----------------------+ +-------+
> > - do_ipv6_setsockopt(sk, ...) - udpv6_sendmsg(sk, ...)
> > - lock_sock(sk) ^._ called via udpv6_prot
> > - WRITE_ONCE(sk->sk_prot, &tcp_prot) before WRITE_ONCE()
> > - inet6_destroy_sock()
> > - release_sock(sk) - ip6_make_skb(sk, ...)
> > ^._ lockless fast path for
> > the non-corking case
> >
> > - __ip6_append_data(sk, ...)
> > - ipv6_local_rxpmtu(sk, ...)
> > - xchg(&np->rxpmtu, skb)
> > ^._ rxpmtu is never freed.
> >
> > - lock_sock(sk)
> >
> > For now, rxpmtu is only the case, but not to miss the future change
> > and a similar bug fixed in commit e27326009a3d ("net: ping6: Fix
> > memleak in ipv6_renew_options()."), let's set a new function to IPv6
> > sk->sk_destruct() and call inet6_cleanup_sock() there. Since the
> > conversion does not change sk->sk_destruct(), we can guarantee that
> > we can clean up IPv6 resources finally.
> >
> > We can now remove all inet6_destroy_sock() calls from IPv6 protocol
> > specific ->destroy() functions, but such changes are invasive to
> > backport. So they can be posted as a follow-up later for net-next.
> >
> > Fixes: 03485f2adcde ("udpv6: Add lockless sendmsg() support")
> > Signed-off-by: Kuniyuki Iwashima <kuniyu@...zon.com>
> > ---
> > Cc: Vladislav Yasevich <vyasevic@...hat.com>
> > ---
> > include/net/ipv6.h | 1 +
> > include/net/udp.h | 2 +-
> > include/net/udplite.h | 8 --------
> > net/ipv4/udp.c | 9 ++++++---
> > net/ipv4/udplite.c | 8 ++++++++
> > net/ipv6/af_inet6.c | 9 ++++++++-
> > net/ipv6/udp.c | 15 ++++++++++++++-
> > net/ipv6/udp_impl.h | 1 +
> > net/ipv6/udplite.c | 9 ++++++++-
> > 9 files changed, 47 insertions(+), 15 deletions(-)
> >
> > diff --git a/include/net/ipv6.h b/include/net/ipv6.h
> > index dfa70789b771..e7ec3e8cd52e 100644
> > --- a/include/net/ipv6.h
> > +++ b/include/net/ipv6.h
> > @@ -1179,6 +1179,7 @@ void ipv6_local_error(struct sock *sk, int err, struct flowi6 *fl6, u32 info);
> > void ipv6_local_rxpmtu(struct sock *sk, struct flowi6 *fl6, u32 mtu);
> >
> > void inet6_cleanup_sock(struct sock *sk);
> > +void inet6_sock_destruct(struct sock *sk);
> > int inet6_release(struct socket *sock);
> > int inet6_bind(struct socket *sock, struct sockaddr *uaddr, int addr_len);
> > int inet6_getname(struct socket *sock, struct sockaddr *uaddr,
> > diff --git a/include/net/udp.h b/include/net/udp.h
> > index 5ee88ddf79c3..fee053bcd17c 100644
> > --- a/include/net/udp.h
> > +++ b/include/net/udp.h
> > @@ -247,7 +247,7 @@ static inline bool udp_sk_bound_dev_eq(struct net *net, int bound_dev_if,
> > }
> >
> > /* net/ipv4/udp.c */
> > -void udp_destruct_sock(struct sock *sk);
> > +void udp_destruct_common(struct sock *sk);
> > void skb_consume_udp(struct sock *sk, struct sk_buff *skb, int len);
> > int __udp_enqueue_schedule_skb(struct sock *sk, struct sk_buff *skb);
> > void udp_skb_destructor(struct sock *sk, struct sk_buff *skb);
> > diff --git a/include/net/udplite.h b/include/net/udplite.h
> > index 0143b373602e..299c14ce2bb9 100644
> > --- a/include/net/udplite.h
> > +++ b/include/net/udplite.h
> > @@ -25,14 +25,6 @@ static __inline__ int udplite_getfrag(void *from, char *to, int offset,
> > return copy_from_iter_full(to, len, &msg->msg_iter) ? 0 : -EFAULT;
> > }
> >
> > -/* Designate sk as UDP-Lite socket */
> > -static inline int udplite_sk_init(struct sock *sk)
> > -{
> > - udp_init_sock(sk);
> > - udp_sk(sk)->pcflag = UDPLITE_BIT;
> > - return 0;
> > -}
> > -
> > /*
> > * Checksumming routines
> > */
> > diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
> > index 560d9eadeaa5..48adb418e404 100644
> > --- a/net/ipv4/udp.c
> > +++ b/net/ipv4/udp.c
> > @@ -1598,7 +1598,7 @@ int __udp_enqueue_schedule_skb(struct sock *sk, struct sk_buff *skb)
> > }
> > EXPORT_SYMBOL_GPL(__udp_enqueue_schedule_skb);
> >
> > -void udp_destruct_sock(struct sock *sk)
> > +void udp_destruct_common(struct sock *sk)
> > {
> > /* reclaim completely the forward allocated memory */
> > struct udp_sock *up = udp_sk(sk);
> > @@ -1611,10 +1611,14 @@ void udp_destruct_sock(struct sock *sk)
> > kfree_skb(skb);
> > }
> > udp_rmem_release(sk, total, 0, true);
> > +}
> > +EXPORT_SYMBOL_GPL(udp_destruct_common);
> >
> > +static void udp_destruct_sock(struct sock *sk)
> > +{
> > + udp_destruct_common(sk);
> > inet_sock_destruct(sk);
> > }
> > -EXPORT_SYMBOL_GPL(udp_destruct_sock);
> >
> > int udp_init_sock(struct sock *sk)
> > {
> > @@ -1622,7 +1626,6 @@ int udp_init_sock(struct sock *sk)
> > sk->sk_destruct = udp_destruct_sock;
> > return 0;
> > }
> > -EXPORT_SYMBOL_GPL(udp_init_sock);
> >
> > void skb_consume_udp(struct sock *sk, struct sk_buff *skb, int len)
> > {
> > diff --git a/net/ipv4/udplite.c b/net/ipv4/udplite.c
> > index 6e08a76ae1e7..4785ac4a8719 100644
> > --- a/net/ipv4/udplite.c
> > +++ b/net/ipv4/udplite.c
> > @@ -17,6 +17,14 @@
> > struct udp_table udplite_table __read_mostly;
> > EXPORT_SYMBOL(udplite_table);
> >
> > +/* Designate sk as UDP-Lite socket */
> > +static inline int udplite_sk_init(struct sock *sk)
>
> You should avoid the 'inline' specifier in c files.
I'll fix it.
> > +{
> > + udp_init_sock(sk);
> > + udp_sk(sk)->pcflag = UDPLITE_BIT;
> > + return 0;
> > +}
> > +
> > static int udplite_rcv(struct sk_buff *skb)
> > {
> > return __udp4_lib_rcv(skb, &udplite_table, IPPROTO_UDPLITE);
> > diff --git a/net/ipv6/af_inet6.c b/net/ipv6/af_inet6.c
> > index 83b9e432f3df..ce5378b78ec9 100644
> > --- a/net/ipv6/af_inet6.c
> > +++ b/net/ipv6/af_inet6.c
> > @@ -109,6 +109,13 @@ static __inline__ struct ipv6_pinfo *inet6_sk_generic(struct sock *sk)
> > return (struct ipv6_pinfo *)(((u8 *)sk) + offset);
> > }
> >
> > +void inet6_sock_destruct(struct sock *sk)
> > +{
> > + inet6_cleanup_sock(sk);
> > + inet_sock_destruct(sk);
> > +}
> > +EXPORT_SYMBOL_GPL(inet6_sock_destruct);
>
> I'm sorry for not noticing this before, but it looks like the above
> export is not needed? only used by udp, which is in the same binary
> (either kernel of ipv6 module) as af_inet6
Ah, please don't be sorry, I appreciate your review!
Exactly, it compiled without exporting the symbol, I'll remove it in v5.
Powered by blists - more mailing lists