* Request: 44378d02c5aa and 73e08d50eb13 (virtio UDP tunnel GSO validation) for 6.18.y and 7.3.y
@ 2026-11-18 20:24 James Sawyer
2026-11-19 16:08 ` Sasha Levin
0 siblings, 1 reply; 2+ messages in thread
From: James Sawyer @ 2026-11-18 20:24 UTC (permalink / raw)
To: stable@vger.kernel.org
Cc: Eric Dumazet, Michael S. Tsirkin, Willem de Bruijn, Paolo Abeni,
netdev@vger.kernel.org
Hi,
Please consider these two mainline commits for 6.18.y and 7.3.y, in
this order:
44378d02c5aa ("net: always dissect GSO packets in __virtio_net_hdr_to_skb()")
73e08d50eb13 ("net: virtio: validate outer protocol of UDP tunnel GSO packets")
73e08d50eb13 is in v7.4-rc3. It fixes a2fb4bc4e2a6 ("net: implement
virtio helpers to handle UDP GSO tunneling."), first released in v6.17.
It has a Fixes: tag but no Cc: stable. 44378d02c5aa is CVE-2026-98954.
Neither is in 6.18.64 or 7.3.5, and I did not find them in stable-queue.
Why: tun accepts a UDP-tunnel GSO virtio-net header without checking
the frame's outer headers, so a VM guest (through vhost) can make the
host build its segments at the wrong offset. On an unpatched v7.0-based
distro kernel I reproduced a full host panic (skb_under_panic, "Fatal
exception in interrupt") when the frame is bridged out through a GRE or
VXLAN port. Without a header push I saw an out-of-bounds read that
hardened usercopy stops in the receiving guest's vhost worker. Details
went to the patch authors privately.
Dry-run apply against the current trees:
7.3.y (7.3.5): 73e08d50eb13 applies cleanly
6.18.y (6.18.64): 73e08d50eb13 does not apply on its own; with
44378d02c5aa applied first, both apply cleanly
7.2.y: already has 44378d02c5aa; 73e08d50eb13 applies
cleanly, if 7.2.y is still maintained
I have only checked that they apply, not built or boot-tested the
stable trees.
Thanks,
James Sawyer
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: Request: 44378d02c5aa and 73e08d50eb13 (virtio UDP tunnel GSO validation) for 6.18.y and 7.3.y
2026-11-18 20:24 Request: 44378d02c5aa and 73e08d50eb13 (virtio UDP tunnel GSO validation) for 6.18.y and 7.3.y James Sawyer
@ 2026-11-19 16:08 ` Sasha Levin
0 siblings, 0 replies; 2+ messages in thread
From: Sasha Levin @ 2026-11-19 16:08 UTC (permalink / raw)
To: stable@vger.kernel.org
Cc: Sasha Levin, Eric Dumazet, Michael S. Tsirkin, Willem de Bruijn,
Paolo Abeni, netdev@vger.kernel.org, James Sawyer
> Please consider these two mainline commits for 6.18.y and 7.3.y, in
> this order:
>
> 44378d02c5aa ("net: always dissect GSO packets in __virtio_net_hdr_to_skb()")
> 73e08d50eb13 ("net: virtio: validate outer protocol of UDP tunnel GSO packets")
Queued both for 6.18 and 7.3, thanks (7.3 already had the first one). I also
queued the tun skb->protocol follow-up on both.
--
Thanks,
Sasha
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2026-11-19 16:08 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-18 20:24 Request: 44378d02c5aa and 73e08d50eb13 (virtio UDP tunnel GSO validation) for 6.18.y and 7.3.y James Sawyer
2026-11-19 16:08 ` Sasha Levin
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox