* [PATCH net] net: check untrusted gso_size at kernel entry
@ 2020-05-25 19:07 Willem de Bruijn
2020-05-27 3:25 ` David Miller
0 siblings, 1 reply; 2+ messages in thread
From: Willem de Bruijn @ 2020-05-25 19:07 UTC (permalink / raw)
To: netdev; +Cc: davem, Willem de Bruijn, syzbot
From: Willem de Bruijn <willemb@google.com>
Syzkaller again found a path to a kernel crash through bad gso input:
a packet with gso size exceeding len.
These packets are dropped in tcp_gso_segment and udp[46]_ufo_fragment.
But they may affect gso size calculations earlier in the path.
Now that we have thlen as of commit 9274124f023b ("net: stricter
validation of untrusted gso packets"), check gso_size at entry too.
Fixes: bfd5f4a3d605 ("packet: Add GSO/csum offload support.")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Willem de Bruijn <willemb@google.com>
---
include/linux/virtio_net.h | 14 ++++++++++++--
1 file changed, 12 insertions(+), 2 deletions(-)
diff --git a/include/linux/virtio_net.h b/include/linux/virtio_net.h
index 6f6ade63b04c..88997022a4b5 100644
--- a/include/linux/virtio_net.h
+++ b/include/linux/virtio_net.h
@@ -31,6 +31,7 @@ static inline int virtio_net_hdr_to_skb(struct sk_buff *skb,
{
unsigned int gso_type = 0;
unsigned int thlen = 0;
+ unsigned int p_off = 0;
unsigned int ip_proto;
if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
@@ -68,7 +69,8 @@ static inline int virtio_net_hdr_to_skb(struct sk_buff *skb,
if (!skb_partial_csum_set(skb, start, off))
return -EINVAL;
- if (skb_transport_offset(skb) + thlen > skb_headlen(skb))
+ p_off = skb_transport_offset(skb) + thlen;
+ if (p_off > skb_headlen(skb))
return -EINVAL;
} else {
/* gso packets without NEEDS_CSUM do not set transport_offset.
@@ -92,17 +94,25 @@ static inline int virtio_net_hdr_to_skb(struct sk_buff *skb,
return -EINVAL;
}
- if (keys.control.thoff + thlen > skb_headlen(skb) ||
+ p_off = keys.control.thoff + thlen;
+ if (p_off > skb_headlen(skb) ||
keys.basic.ip_proto != ip_proto)
return -EINVAL;
skb_set_transport_header(skb, keys.control.thoff);
+ } else if (gso_type) {
+ p_off = thlen;
+ if (p_off > skb_headlen(skb))
+ return -EINVAL;
}
}
if (hdr->gso_type != VIRTIO_NET_HDR_GSO_NONE) {
u16 gso_size = __virtio16_to_cpu(little_endian, hdr->gso_size);
+ if (skb->len - p_off <= gso_size)
+ return -EINVAL;
+
skb_shinfo(skb)->gso_size = gso_size;
skb_shinfo(skb)->gso_type = gso_type;
--
2.27.0.rc0.183.gde8f92d652-goog
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH net] net: check untrusted gso_size at kernel entry
2020-05-25 19:07 [PATCH net] net: check untrusted gso_size at kernel entry Willem de Bruijn
@ 2020-05-27 3:25 ` David Miller
0 siblings, 0 replies; 2+ messages in thread
From: David Miller @ 2020-05-27 3:25 UTC (permalink / raw)
To: willemdebruijn.kernel; +Cc: netdev, willemb, syzkaller
From: Willem de Bruijn <willemdebruijn.kernel@gmail.com>
Date: Mon, 25 May 2020 15:07:40 -0400
> From: Willem de Bruijn <willemb@google.com>
>
> Syzkaller again found a path to a kernel crash through bad gso input:
> a packet with gso size exceeding len.
>
> These packets are dropped in tcp_gso_segment and udp[46]_ufo_fragment.
> But they may affect gso size calculations earlier in the path.
>
> Now that we have thlen as of commit 9274124f023b ("net: stricter
> validation of untrusted gso packets"), check gso_size at entry too.
>
> Fixes: bfd5f4a3d605 ("packet: Add GSO/csum offload support.")
> Reported-by: syzbot <syzkaller@googlegroups.com>
> Signed-off-by: Willem de Bruijn <willemb@google.com>
Applied and queued up for -stable, thanks Willem.
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2020-05-27 3:25 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-05-25 19:07 [PATCH net] net: check untrusted gso_size at kernel entry Willem de Bruijn
2020-05-27 3:25 ` David Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).