Openvswitch allows to drop a packet's Ethernet header, therefore
skb_mpls_push() and skb_mpls_pop() might be called with ethernet=true
and mac_len=0. In that case the pointer passed to skb_mod_eth_type()
doesn't point to an Ethernet header and the new Ethertype is written at
unexpected locations.
Fix this by verifying that mac_len is big enough to contain an Ethernet
header.
Fixes: fa4e0f8855fc ("net/sched: fix corrupted L2 header with MPLS 'push' and
'pop' actions")
Signed-off-by: Guillaume Nault <gna...@redhat.com>
---
Notes:
- Found by code inspection.
- Using commit fa4e0f8855fc for the Fixes tag because mac_len is
needed for the test. The problem probably exists since openvswitch
can pop the Ethernet header though.
net/core/skbuff.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 6faf73d6a0f7..2b48cb0cc684 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -5622,7 +5622,7 @@ int skb_mpls_push(struct sk_buff *skb, __be32 mpls_lse,
__be16 mpls_proto,
lse->label_stack_entry = mpls_lse;
skb_postpush_rcsum(skb, lse, MPLS_HLEN);
- if (ethernet)
+ if (ethernet && mac_len >= ETH_HLEN)
skb_mod_eth_type(skb, eth_hdr(skb), mpls_proto);
skb->protocol = mpls_proto;
@@ -5662,7 +5662,7 @@ int skb_mpls_pop(struct sk_buff *skb, __be16 next_proto,
int mac_len,
skb_reset_mac_header(skb);
skb_set_network_header(skb, mac_len);
- if (ethernet) {
+ if (ethernet && mac_len >= ETH_HLEN) {
struct ethhdr *hdr;
/* use mpls_hdr() to get ethertype to account for VLANs. */
--
2.21.3
