From: Antonio Quartulli <antonio@openvpn.net>
To: netdev@vger.kernel.org
Cc: Antonio Quartulli <antonio@openvpn.net>,
Sabrina Dubroca <sd@queasysnail.net>,
Ralf Lici <ralf@mandelbit.com>, Jakub Kicinski <kuba@kernel.org>,
Paolo Abeni <pabeni@redhat.com>,
Andrew Lunn <andrew+netdev@lunn.ch>,
"David S. Miller" <davem@davemloft.net>,
Eric Dumazet <edumazet@google.com>
Subject: [PATCH net 0/7] pull request: fixes for ovpn 2026-10-22
Date: Thu, 22 Oct 2026 14:15:43 +0200 [thread overview]
Message-ID: <20261022121606.732029-1-antonio@openvpn.net> (raw)
Hi all!
First of all, this PR includes fixes for older releases.
Given Linus' statements on our recent set of fixes, I have
not understood if this kind of fixes are supposed to still go through
net or rather via net-next (so they don't land in an -rc).
As you see I am targeting net, as I always did, but feel free to let me
know if I should do otherwise.
And this is the PR cover:
Here is a batch of seven ovpn fixes for net. Six are about hardening
how ovpn deals with the userspace-provided transport socket, plus a
selftest fix for a NULL deref crash with musl.
The most important ones close two potential crashes, both reachable by
a CAP_NET_ADMIN user configuring a peer via netlink:
* NULL pointer dereference on TX: an IPv6 remote endpoint was accepted
on an AF_INET UDP socket. The first packet sent to such a peer goes
through ovpn_udp6_output() and ip6_dst_lookup_flow(), which
dereferences inet6_sk(sk), but the latter is NULL for IPv4 sockets.
Such remotes are now rejected via netlink and the socket family is
rechecked on TX.
* Out-of-bounds access: the transport socket type was selected by
sk_protocol only, therefore a raw socket with IPPROTO_UDP could be
attached and then accessed as a struct udp_sock. Only real UDP
datagram or TCP stream sockets are now accepted.
The remaining fixes prevent sending UDP packets with source port 0
(e.g. after connect(AF_UNSPEC) on the socket), and correct the TX
accounting for both transports: UDP byte counters did not include the
OpenVPN header and auth tag, while packets dropped locally were
accounted as sent and refreshed the keepalive timer.
The last patch fixes ovpn-cli crashing on musl, because
freeaddrinfo() was being called on NULL.
Regarding Linus' email commenting our last net PR, I think
he complained that we did not CC stable on any patch.
My (long-lasting) understanding is that for patches to net
we're not supposed to do that.
Let me know if I should do otherwise.
Please pull or let me know of any issue!
Thanks a lot,
Antonio
The following changes since commit f4377bf85887f57b59abb5479f7660a81ffe4449:
tipc: fix broadcast sender hang on last acker departure (2026-10-22 11:05:31 +0200)
are available in the Git repository at:
https://github.com/OpenVPN/ovpn-net-next.git tags/ovpn-net-20261022
for you to fetch changes up to a402a88c1d53ec649c57da6d947a72154f9a6de7:
selftests: ovpn: avoid calling freeaddrinfo() on NULL (2026-10-22 14:07:59 +0200)
----------------------------------------------------------------
Included fixes:
* fix selftest tool segfault on musl caused by calling freeaddrinfo()
on NULL
* fix TCP TX accounting, which counted packets dropped locally as sent
and refreshed the keepalive timer for them
* fix UDP TX byte accounting, which left out the OpenVPN data channel
header and authentication tag
* recheck the UDP socket family on TX, since userspace can still change
it via IPV6_ADDRFORM or IPV6_V6ONLY, and tear the peer down on
mismatch, ensuring the IPv6 TX path never sees a socket without IPv6
state
* reject UDP remotes incompatible with the socket family, fixing a NULL
pointer dereference on TX when an IPv6 remote is configured on an
IPv4 UDP socket
* reject sockets that are not real UDP or TCP sockets, preventing e.g.
a raw socket with IPPROTO_UDP from being treated as a udp_sock and
accessed out of bounds
* avoid sending UDP packets with source port 0, and do not account
locally dropped UDP packets as sent
----------------------------------------------------------------
Antonio Quartulli (1):
selftests: ovpn: avoid calling freeaddrinfo() on NULL
Ralf Lici (6):
ovpn: avoid sending UDP packets with source port 0
ovpn: validate sockets before attaching peer transports
ovpn: reject UDP remotes incompatible with socket family
ovpn: recheck UDP socket family before transmit
ovpn: fix UDP TX byte accounting
ovpn: fix TCP TX accounting for locally dropped packets
drivers/net/ovpn/io.c | 8 +-
drivers/net/ovpn/netlink.c | 129 +++++++++++++++++++++-------
drivers/net/ovpn/peer.c | 19 ++++
drivers/net/ovpn/peer.h | 5 +-
drivers/net/ovpn/socket.c | 16 ++--
drivers/net/ovpn/tcp.c | 78 +++++++++--------
drivers/net/ovpn/tcp.h | 10 +--
drivers/net/ovpn/udp.c | 51 ++++++++---
drivers/net/ovpn/udp.h | 4 +-
tools/testing/selftests/net/ovpn/ovpn-cli.c | 3 +-
10 files changed, 224 insertions(+), 99 deletions(-)
next reply other threads:[~2026-10-22 12:16 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-22 12:15 Antonio Quartulli [this message]
2026-10-22 12:15 ` [PATCH net 1/7] ovpn: avoid sending UDP packets with source port 0 Antonio Quartulli
2026-10-22 12:23 ` netdev-bot+sinfo
2026-10-22 12:31 ` Antonio Quartulli
2026-10-22 12:15 ` [PATCH net 2/7] ovpn: validate sockets before attaching peer transports Antonio Quartulli
2026-10-22 12:15 ` [PATCH net 3/7] ovpn: reject UDP remotes incompatible with socket family Antonio Quartulli
2026-10-23 12:17 ` netdev-bot+sashiko
2026-10-22 12:15 ` [PATCH net 4/7] ovpn: recheck UDP socket family before transmit Antonio Quartulli
2026-10-22 12:15 ` [PATCH net 5/7] ovpn: fix UDP TX byte accounting Antonio Quartulli
2026-10-23 12:17 ` netdev-bot+sashiko
2026-10-23 14:10 ` Antonio Quartulli
2026-10-22 12:15 ` [PATCH net 6/7] ovpn: fix TCP TX accounting for locally dropped packets Antonio Quartulli
2026-10-23 12:17 ` netdev-bot+sashiko
2026-10-22 12:15 ` [PATCH net 7/7] selftests: ovpn: avoid calling freeaddrinfo() on NULL Antonio Quartulli
2026-10-23 12:17 ` netdev-bot+sashiko
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261022121606.732029-1-antonio@openvpn.net \
--to=antonio@openvpn.net \
--cc=andrew+netdev@lunn.ch \
--cc=davem@davemloft.net \
--cc=edumazet@google.com \
--cc=kuba@kernel.org \
--cc=netdev@vger.kernel.org \
--cc=pabeni@redhat.com \
--cc=ralf@mandelbit.com \
--cc=sd@queasysnail.net \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox