tap: add missing verification for short frame

commit ed7f2af upstream.

The cited commit missed to check against the validity of the frame length
in the tap_get_user_xdp() path, which could cause a corrupted skb to be
sent downstack. Even before the skb is transmitted, the
tap_get_user_xdp()-->skb_set_network_header() may assume the size is more
than ETH_HLEN. Once transmitted, this could either cause out-of-bound
access beyond the actual length, or confuse the underlayer with incorrect
or inconsistent header length in the skb metadata.

In the alternative path, tap_get_user() already prohibits short frame which
has the length less than Ethernet header size from being transmitted.

This is to drop any frame shorter than the Ethernet header size just like
how tap_get_user() does.

CVE: CVE-2024-41090
Link: https://lore.kernel.org/netdev/1717026141-25716-1-git-send-email-si-wei.liu@oracle.com/
Fixes: 0efac27 ("tap: accept an array of XDP buffs through sendmsg()")
Cc: stable@vger.kernel.org
Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Jason Wang <jasowang@redhat.com>
Link: https://patch.msgid.link/20240724170452.16837-2-dongli.zhang@oracle.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers/net/tap.c

@@ -1153,6 +1153,11 @@ static int tap_get_user_xdp(struct tap_queue *q, struct xdp_buff *xdp)
  struct sk_buff *skb;
  int err, depth;
 
+ if (unlikely(xdp->data_end - xdp->data < ETH_HLEN)) {
+ err = -EINVAL;
+ goto err;
+ }
+
  if (q->flags & IFF_VNET_HDR)
  vnet_hdr_len = READ_ONCE(q->vnet_hdr_sz);