[PATCH net-next] sctp: bring inet(6)_skb_parm back to sctp_input_cb

[PATCH net-next] sctp: bring inet(6)_skb_parm back to sctp_input_cb

[Xin Long]

inet(6)_skb_parm was removed from sctp_input_cb by Commit a1dd2cf2f1ae
("sctp: allow changing transport encap_port by peer packets"), as it
thought sctp_input_cb->header is not used any more in SCTP.

syzbot reported a crash:

  [ ] BUG: KASAN: use-after-free in decode_session6+0xe7c/0x1580
  [ ]
  [ ] Call Trace:
  [ ]  <IRQ>
  [ ]  dump_stack+0x107/0x163
  [ ]  kasan_report.cold+0x1f/0x37
  [ ]  decode_session6+0xe7c/0x1580
  [ ]  __xfrm_policy_check+0x2fa/0x2850
  [ ]  sctp_rcv+0x12b0/0x2e30
  [ ]  sctp6_rcv+0x22/0x40
  [ ]  ip6_protocol_deliver_rcu+0x2e8/0x1680
  [ ]  ip6_input_finish+0x7f/0x160
  [ ]  ip6_input+0x9c/0xd0
  [ ]  ipv6_rcv+0x28e/0x3c0

It was caused by sctp_input_cb->header/IP6CB(skb) still used in sctp rx
path decode_session6() but some members overwritten by sctp6_rcv().

This patch is to fix it by bring inet(6)_skb_parm back to sctp_input_cb
and not overwriting it in sctp4/6_rcv() and sctp_udp_rcv().

Reported-by: syzbot+5be8aebb1b7dfa90ef31@syzkaller.appspotmail.com
Fixes: a1dd2cf2f1ae ("sctp: allow changing transport encap_port by peer packets")
Signed-off-by: Xin Long <lucien.xin@gmail.com>
---
 include/net/sctp/structs.h | 6 ++++++
 net/sctp/ipv6.c            | 2 +-
 net/sctp/protocol.c        | 3 +--
 3 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/include/net/sctp/structs.h b/include/net/sctp/structs.h
index 80f7149..1aa5852 100644
--- a/include/net/sctp/structs.h
+++ b/include/net/sctp/structs.h
@@ -1121,6 +1121,12 @@ static inline void sctp_outq_cork(struct sctp_outq *q)
  * sctp_input_cb is currently used on rx and sock rx queue
  */
 struct sctp_input_cb {
+	union {
+		struct inet_skb_parm    h4;
+#if IS_ENABLED(CONFIG_IPV6)
+		struct inet6_skb_parm   h6;
+#endif
+	} header;
 	struct sctp_chunk *chunk;
 	struct sctp_af *af;
 	__be16 encap_port;
diff --git a/net/sctp/ipv6.c b/net/sctp/ipv6.c
index 814754d..c3e89c7 100644
--- a/net/sctp/ipv6.c
+++ b/net/sctp/ipv6.c
@@ -1074,7 +1074,7 @@ static struct inet_protosw sctpv6_stream_protosw = {
 
 static int sctp6_rcv(struct sk_buff *skb)
 {
-	memset(skb->cb, 0, sizeof(skb->cb));
+	SCTP_INPUT_CB(skb)->encap_port = 0;
 	return sctp_rcv(skb) ? -1 : 0;
 }
 
diff --git a/net/sctp/protocol.c b/net/sctp/protocol.c
index 41f287a..6f2bbfe 100644
--- a/net/sctp/protocol.c
+++ b/net/sctp/protocol.c
@@ -843,7 +843,6 @@ static int sctp_ctl_sock_init(struct net *net)
 
 static int sctp_udp_rcv(struct sock *sk, struct sk_buff *skb)
 {
-	memset(skb->cb, 0, sizeof(skb->cb));
 	SCTP_INPUT_CB(skb)->encap_port = udp_hdr(skb)->source;
 
 	skb_set_transport_header(skb, sizeof(struct udphdr));
@@ -1163,7 +1162,7 @@ static struct inet_protosw sctp_stream_protosw = {
 
 static int sctp4_rcv(struct sk_buff *skb)
 {
-	memset(skb->cb, 0, sizeof(skb->cb));
+	SCTP_INPUT_CB(skb)->encap_port = 0;
 	return sctp_rcv(skb);
 }
 
-- 
2.1.0

Re: [PATCH net-next] sctp: bring inet(6)_skb_parm back to sctp_input_cb

[Marcelo Ricardo Leitner]

On Wed, Nov 04, 2020 at 02:55:32PM +0800, Xin Long wrote:
> inet(6)_skb_parm was removed from sctp_input_cb by Commit a1dd2cf2f1ae
> ("sctp: allow changing transport encap_port by peer packets"), as it
> thought sctp_input_cb->header is not used any more in SCTP.
> 
> syzbot reported a crash:
> 
>   [ ] BUG: KASAN: use-after-free in decode_session6+0xe7c/0x1580
>   [ ]
>   [ ] Call Trace:
>   [ ]  <IRQ>
>   [ ]  dump_stack+0x107/0x163
>   [ ]  kasan_report.cold+0x1f/0x37
>   [ ]  decode_session6+0xe7c/0x1580
>   [ ]  __xfrm_policy_check+0x2fa/0x2850
>   [ ]  sctp_rcv+0x12b0/0x2e30
>   [ ]  sctp6_rcv+0x22/0x40
>   [ ]  ip6_protocol_deliver_rcu+0x2e8/0x1680
>   [ ]  ip6_input_finish+0x7f/0x160
>   [ ]  ip6_input+0x9c/0xd0
>   [ ]  ipv6_rcv+0x28e/0x3c0
> 
> It was caused by sctp_input_cb->header/IP6CB(skb) still used in sctp rx
> path decode_session6() but some members overwritten by sctp6_rcv().
> 
> This patch is to fix it by bring inet(6)_skb_parm back to sctp_input_cb
> and not overwriting it in sctp4/6_rcv() and sctp_udp_rcv().
> 
> Reported-by: syzbot+5be8aebb1b7dfa90ef31@syzkaller.appspotmail.com
> Fixes: a1dd2cf2f1ae ("sctp: allow changing transport encap_port by peer packets")
> Signed-off-by: Xin Long <lucien.xin@gmail.com>

Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>

Thanks Xin.

Re: [PATCH net-next] sctp: bring inet(6)_skb_parm back to sctp_input_cb

[Jakub Kicinski]

On Thu, 5 Nov 2020 00:49:09 -0300 Marcelo Ricardo Leitner wrote:
> On Wed, Nov 04, 2020 at 02:55:32PM +0800, Xin Long wrote:
> > inet(6)_skb_parm was removed from sctp_input_cb by Commit a1dd2cf2f1ae
> > ("sctp: allow changing transport encap_port by peer packets"), as it
> > thought sctp_input_cb->header is not used any more in SCTP.
> > 
> > syzbot reported a crash:
> > 
> >   [ ] BUG: KASAN: use-after-free in decode_session6+0xe7c/0x1580
> >   [ ]
> >   [ ] Call Trace:
> >   [ ]  <IRQ>
> >   [ ]  dump_stack+0x107/0x163
> >   [ ]  kasan_report.cold+0x1f/0x37
> >   [ ]  decode_session6+0xe7c/0x1580
> >   [ ]  __xfrm_policy_check+0x2fa/0x2850
> >   [ ]  sctp_rcv+0x12b0/0x2e30
> >   [ ]  sctp6_rcv+0x22/0x40
> >   [ ]  ip6_protocol_deliver_rcu+0x2e8/0x1680
> >   [ ]  ip6_input_finish+0x7f/0x160
> >   [ ]  ip6_input+0x9c/0xd0
> >   [ ]  ipv6_rcv+0x28e/0x3c0
> > 
> > It was caused by sctp_input_cb->header/IP6CB(skb) still used in sctp rx
> > path decode_session6() but some members overwritten by sctp6_rcv().
> > 
> > This patch is to fix it by bring inet(6)_skb_parm back to sctp_input_cb
> > and not overwriting it in sctp4/6_rcv() and sctp_udp_rcv().
> > 
> > Reported-by: syzbot+5be8aebb1b7dfa90ef31@syzkaller.appspotmail.com
> > Fixes: a1dd2cf2f1ae ("sctp: allow changing transport encap_port by peer packets")
> > Signed-off-by: Xin Long <lucien.xin@gmail.com>  
> 
> Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>

Applied, thanks!