[PATCH net v5 2/2] xfrm: fix stale skb->prev after async crypto steals a GSO segment

[Petr Wozniak <petr.wozniak@gmail.com>]

skb_gso_segment() leaves the segment list head with ->prev pointing at
the last segment, an invariant validate_xmit_skb_list() relies on when
it sets its tail pointer (tail = skb->prev).

When validate_xmit_xfrm() walks a GSO list and some segments are stolen
by async crypto (->xmit() returns -EINPROGRESS), those segments are
unlinked from the list but the head ->prev is never updated.  If the
last segment is the one stolen, the returned head still has ->prev
pointing at it, even though it is now owned by the crypto engine and may
be freed.  validate_xmit_skb_list() later does tail->next = skb, writing
through that stale pointer -- a use-after-free.

Repoint skb->prev at the last retained segment before returning.

Fixes: f53c723902d1 ("net: Add asynchronous callbacks for xfrm on layer 2.")
Signed-off-by: Petr Wozniak <petr.wozniak@gmail.com>
---
 net/xfrm/xfrm_device.c | 8 ++++++++
 1 file changed, 8 insertions(+)

diff --git a/net/xfrm/xfrm_device.c b/net/xfrm/xfrm_device.c
index 19c77f09acc9..aec1e1184a71 100644
--- a/net/xfrm/xfrm_device.c
+++ b/net/xfrm/xfrm_device.c
@@ -224,6 +224,14 @@ struct sk_buff *validate_xmit_xfrm(struct sk_buff *skb, netdev_features_t featur
 		pskb = skb2;
 	}
 
+	/* skb_gso_segment() set skb->prev to the last segment, but async
+	 * crypto may have stolen it above without updating ->prev.  Repoint
+	 * it at the last retained segment so validate_xmit_skb_list() does
+	 * not chain onto a segment now owned by the crypto engine.
+	 */
+	if (skb)
+		skb->prev = pskb;
+
 	return skb ? skb : ERR_PTR(-EINPROGRESS);
 }
 EXPORT_SYMBOL_GPL(validate_xmit_xfrm);
-- 
2.51.0