Netdev List
 help / color / mirror / Atom feed
From: Eric Dumazet <edumazet@google.com>
To: davem@davemloft.net, kuba@kernel.org, pabeni@redhat.com
Cc: horms@kernel.org, dsahern@kernel.org, idosch@nvidia.com,
	 netdev@vger.kernel.org, eric.dumazet@gmail.com,
	 Eric Dumazet <edumazet@google.com>
Subject: [PATCH net v2 5/5] gre: fix out-of-bounds read of erspan metadata in collect_md mode
Date: Wed, 16 Sep 2026 10:01:55 +0000	[thread overview]
Message-ID: <20260916100155.1398403-6-edumazet@google.com> (raw)
In-Reply-To: <20260916100155.1398403-1-edumazet@google.com>

erspan_rcv() and ip6erspan_rcv() copy the ERSPAN metadata out of the
packet for collect_md tunnels:

	pkt_md = (struct erspan_metadata *)(gh + gre_hdr_len +
					    sizeof(*ershdr));
	...
	memcpy(md2, pkt_md, ver == 1 ? ERSPAN_V1_MDSIZE :
				       ERSPAN_V2_MDSIZE);

Both read 8 bytes at 12 bytes from the start of the GRE header, but only
ask pskb_may_pull() for erspan_hdr_len(ver) bytes beyond it, which type I
support made 0 for version 0. Two ways to get there:

- An ERSPAN type I packet has a 4 byte GRE header and no ERSPAN header
  at all, so erspan_rcv() sets ver = 0 and only pulls the GRE header. It
  still falls back to a collect_md tunnel through itn->collect_md_tun,
  and there the ternary above picks ERSPAN_V2_MDSIZE.

- A packet with an 8 byte GRE header and ershdr->ver == 0 is malformed,
  yet neither erspan_rcv() nor ip6erspan_rcv() validates the version
  before using it, so erspan_hdr_len(0) pulls nothing either. A version
  above 2 is not an out-of-bounds read, but it does store a version that
  the transmit side (erspan_fb_xmit(), ip6erspan_tunnel_xmit()) rejects.

Reject a base header whose version is neither 1 nor 2, and skip the
metadata extraction for type I, which has none: ip_tun_rx_dst() hands
out a zeroed option area, so md->version = 0 alone describes it.

Fixes: f989d546a2d5 ("erspan: Add type I version 0 support.")
Cc: stable@vger.kernel.org
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 net/ipv4/ip_gre.c  | 34 ++++++++++++++++++++++------------
 net/ipv6/ip6_gre.c |  2 ++
 2 files changed, 24 insertions(+), 12 deletions(-)

diff --git a/net/ipv4/ip_gre.c b/net/ipv4/ip_gre.c
index 696884f53cdcc65fe87cf04f357f1cc45a7e0736..92f3a52d20d38186c3ce87824a8e15c9e00a925f 100644
--- a/net/ipv4/ip_gre.c
+++ b/net/ipv4/ip_gre.c
@@ -274,7 +274,6 @@ static int erspan_rcv(struct sk_buff *skb, struct tnl_ptk_info *tpi,
 	struct ip_tunnel_net *itn;
 	struct ip_tunnel *tunnel;
 	const struct iphdr *iph;
-	struct erspan_md2 *md2;
 	int ver;
 	int len;
 
@@ -294,6 +293,9 @@ static int erspan_rcv(struct sk_buff *skb, struct tnl_ptk_info *tpi,
 
 		ershdr = (struct erspan_base_hdr *)(skb->data + gre_hdr_len);
 		ver = ershdr->ver;
+		if (unlikely(ver != 1 && ver != 2))
+			return PACKET_REJECT;
+
 		iph = ip_hdr(skb);
 		__set_bit(IP_TUNNEL_KEY_BIT, flags);
 		tunnel = ip_tunnel_lookup(itn, skb->dev->ifindex, flags,
@@ -318,6 +320,7 @@ static int erspan_rcv(struct sk_buff *skb, struct tnl_ptk_info *tpi,
 		if (tunnel->collect_md) {
 			struct erspan_metadata *pkt_md, *md;
 			struct ip_tunnel_info *info;
+			struct erspan_md2 *md2;
 			unsigned char *gh;
 			__be64 tun_id;
 
@@ -334,19 +337,26 @@ static int erspan_rcv(struct sk_buff *skb, struct tnl_ptk_info *tpi,
 			info = &tun_dst->u.tun_info;
 			info->options_len = sizeof(*md);
 
-			/* skb can be uncloned in __iptunnel_pull_header, so
-			 * old pkt_md is no longer valid and we need to reset
-			 * it
-			 */
-			gh = skb_network_header(skb) +
-			     skb_network_header_len(skb);
-			pkt_md = (struct erspan_metadata *)(gh + gre_hdr_len +
-							    sizeof(*ershdr));
 			md = ip_tunnel_info_opts(&tun_dst->u.tun_info);
 			md->version = ver;
-			md2 = &md->u.md2;
-			memcpy(md2, pkt_md, ver == 1 ? ERSPAN_V1_MDSIZE :
-						       ERSPAN_V2_MDSIZE);
+
+			/* Type I has no ERSPAN header, thus no metadata to
+			 * extract: reading it would go past the @len bytes
+			 * pulled above. ip_tun_rx_dst() zeroed @md for us.
+			 */
+			if (!is_erspan_type1(gre_hdr_len)) {
+				/* skb can be uncloned in __iptunnel_pull_header, so
+				 * old pkt_md is no longer valid and we need to reset
+				 * it
+				 */
+				gh = skb_network_header(skb) +
+				     skb_network_header_len(skb);
+				pkt_md = (struct erspan_metadata *)(gh + gre_hdr_len +
+								    sizeof(*ershdr));
+				md2 = &md->u.md2;
+				memcpy(md2, pkt_md, ver == 1 ? ERSPAN_V1_MDSIZE :
+							       ERSPAN_V2_MDSIZE);
+			}
 
 			__set_bit(IP_TUNNEL_ERSPAN_OPT_BIT,
 				  info->key.tun_flags);
diff --git a/net/ipv6/ip6_gre.c b/net/ipv6/ip6_gre.c
index 8ebda0b6a78b2236b439f5499d84f34f652fcbe2..a59fb82c74dad7f0c1d1128338e7bed1e3c7116f 100644
--- a/net/ipv6/ip6_gre.c
+++ b/net/ipv6/ip6_gre.c
@@ -503,6 +503,8 @@ static int ip6erspan_rcv(struct sk_buff *skb,
 	ipv6h = ipv6_hdr(skb);
 	ershdr = (struct erspan_base_hdr *)skb->data;
 	ver = ershdr->ver;
+	if (unlikely(ver != 1 && ver != 2))
+		return PACKET_REJECT;
 
 	tunnel = ip6gre_tunnel_lookup(skb->dev,
 				      &ipv6h->saddr, &ipv6h->daddr, tpi->key,
-- 
2.55.0.1032.g73a4cd73de-goog


  parent reply	other threads:[~2026-09-16 10:02 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-12 15:09 [PATCH net 0/3] ip_gre: fix header lengths and validation on changelink Eric Dumazet
2026-09-12 15:09 ` [PATCH net 1/3] ip_gre: validate netlink attributes before changing the tunnel Eric Dumazet
2026-09-15 12:11   ` netdev-bot+sashiko
2026-09-12 15:09 ` [PATCH net 2/3] ip_gre: compute tunnel lengths absolutely instead of by delta Eric Dumazet
2026-09-15 12:11   ` netdev-bot+sashiko
2026-09-12 15:09 ` [PATCH net 3/3] ip_gre: recompute erspan header lengths after a change Eric Dumazet
2026-09-15 12:11   ` netdev-bot+sashiko
2026-09-15 13:31 ` [PATCH net 0/3] ip_gre: fix header lengths and validation on changelink Eric Dumazet
2026-09-16 10:01 ` [PATCH net v2 0/5] ip_tunnel, ip_gre: fix changelink lengths and ERSPAN receive Eric Dumazet
2026-09-16 10:01   ` [PATCH net v2 1/5] ip_tunnel: do not clear the active encap before validating the new one Eric Dumazet
2026-09-20 10:48     ` netdev-bot+sashiko
2026-09-16 10:01   ` [PATCH net v2 2/5] ip_gre: validate netlink attributes before changing the tunnel Eric Dumazet
2026-09-20 10:48     ` netdev-bot+sashiko
2026-09-16 10:01   ` [PATCH net v2 3/5] ip_gre: compute tunnel lengths absolutely instead of by delta Eric Dumazet
2026-09-20 10:48     ` netdev-bot+sashiko
2026-09-16 10:01   ` [PATCH net v2 4/5] ip_gre: recompute erspan header lengths after a change Eric Dumazet
2026-09-20 10:48     ` netdev-bot+sashiko
2026-09-16 10:01   ` Eric Dumazet [this message]
2026-09-20 10:48     ` [PATCH net v2 5/5] gre: fix out-of-bounds read of erspan metadata in collect_md mode netdev-bot+sashiko
2026-09-16 22:45   ` [PATCH net v2 0/5] ip_tunnel, ip_gre: fix changelink lengths and ERSPAN receive Jakub Kicinski

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260916100155.1398403-6-edumazet@google.com \
    --to=edumazet@google.com \
    --cc=davem@davemloft.net \
    --cc=dsahern@kernel.org \
    --cc=eric.dumazet@gmail.com \
    --cc=horms@kernel.org \
    --cc=idosch@nvidia.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox