[PATCH net-next v4 0/5] ipv6: tie the dst cookie to the lookup that returned the dst

From
Ralf Lici <ralf@mandelbit.com>
Date
2026-11-20 09:36:44
Also in
lkml
Hi,

This series implements a first part of the model proposed in [1] to
address the late cookie sampling race: a dst can be stored together with
a cookie read after a FIB change that its lookup did not see, and then
passes dst_check until the next change.

The case that made me post it as a series is xfrm. Adding a policy
restamps every fib6 node, so a socket whose lookup ran just before the
insertion keeps its plain dst and leaks plaintext although the policy
matches. I reproduced it for TCP connect and connected UDP, and the same
window lets a more specific blackhole route be bypassed for routes using
nexthop objects. The ovpn patch [2] hit the same pattern in
dst_cache_set_ip6.

Patches 1-3 are the prerequisites: writers that publish a routing
change before advancing the IPv6 generation, and an ip6_dst_check that
compares the cookie of nexthop-object dsts. Patch 4 adds the
lookup-cookie API and patch 5 converts tcp_v6_connect,
ip6_datagram_dst_update and ip6_sk_dst_lookup_flow to it.

Compared to [1], the fallback cookie is 0 instead of the pre-lookup
generation, so fib6_flush_trees is left alone, and nexthop-object dsts
are checked in ip6_dst_check instead of passing the generation down to
ip6_rt_pcpu_alloc.

I am targeting net-next to get feedback on the lookup-cookie API and its
incremental adoption before extending the conversion across the
remaining cache users. I am not requesting stable backports at this
stage.

Known limitations and pre-existing issues, not addressed here:

- The conversion is partial. The remaining ip6_dst_store users
  (inet6_csk_route_socket, inet6_sk_rebuild_header and
  tcp_v6_syn_recv_sock) and the other cache users (dst_cache, sctp,
  ipvs, flowtable, sk_rx_dst) still sample the cookie late.
- xfrm bundles ignore the returned cookie. Their route and path cookies
  are still sampled late in xfrm_bundle_create. I plan to apply the same
  generation bracketing inside xfrm as a follow-up.
- xfrm_policy_insert unlinks a replaced policy after rt_genid_bump_ipv6,
  so a lookup after the bump can still match the old policy. I will send
  a separate fix.
- Route deletion does not advance the generation, and ip6_dst_check does
  not look at rt->from for nexthop-object dsts, so these are not
  invalidated by deletion. This predates the series and the kdoc of the
  new helper does not claim otherwise.
- A newly linked fib6 node holds fn_sernum 0 until fib6_add stamps it,
  so a dst stored with the fallback cookie 0 can pass a check in that
  window. It fails as soon as the node is stamped.

I tested it in a VM with lockdep, PROVE_RCU and DEBUG_ATOMIC_SLEEP,
using a debug-only delay after the lookup to hit the window: no
warnings, and the plaintext and blackhole bypass are gone in these test
cases.

[1] https://lore.kernel.org/netdev/20260901122501.482920-1-ralf@mandelbit.com/ (local)
[2] https://lore.kernel.org/netdev/cover.1789639346.git.ralf@mandelbit.com/ (local)

Thanks,

Ralf Lici
Mandelbit Srl

---
Changes since v3 https://lore.kernel.org/netdev/cover.1794906743.git.ralf@mandelbit.com/ (local)
- Rebase onto current net-next.
- Split v3 2/3 into the writer ordering (2/5), the ip6_dst_check change
  (3/5) and the new API (4/5), each with its own explanation. (David)
- Reword the kdoc of ip6_dst_lookup_flow_cookie to say that it performs
  the xfrm policy lookup (David) and narrow it to changes that advance
  the IPv6 generation. (Sashiko)
- Document the ordering of fib6_new_sernum that the helper relies on and
  name it in the barrier comments. (Xia)
- Note the per-call generation advance in 1/5 and widen the
  rt6_exception_net comment. (Sashiko)
- Drop the Fixes tag from the conversion patch. (Sashiko)
- List the known limitations and pre-existing issues above.

Changes since v2 https://lore.kernel.org/all/cover.1794384146.git.ralf@mandelbit.com/ (local)
- Rebase onto current net-next, preserving the connected-only cache
  lookup introduced by f778e80b3daf ("ipv6: fix cached route reuse for
  control messages"). (Jakub)
- Move the serial update after rt6_multipath_rebalance in fib6_ifup and
  fib6_ifdown. (Sashiko)
- Add Fixes tags and improve wording. (Sashiko)
- Clarify that this is the groundwork for fixing the race and a partial
  conversion to the new API: other users will be covered in a follow-up
  series.

Changes since v1 https://lore.kernel.org/netdev/cover.1793806941.git.ralf@mandelbit.com/ (local)
- Fix kdoc return statements in patch 2/3. (NIPA)

Ralf Lici (5):
  ipv6: stamp the default route node with a fresh serial on link failure
  ipv6: advance the serial after rebalancing in fib6_ifup and
    fib6_ifdown
  ipv6: validate the cookie of nexthop-object dsts in ip6_dst_check
  ipv6: add ip6_dst_lookup_flow_cookie
  ipv6: use lookup cookies in selected socket dst cache paths

 include/net/ip6_route.h | 18 ++++++++----
 include/net/ipv6.h      | 13 +++++++++
 net/ipv6/datagram.c     |  6 ++--
 net/ipv6/ip6_fib.c      |  4 +++
 net/ipv6/ip6_output.c   | 63 ++++++++++++++++++++++++++++++++++++++---
 net/ipv6/route.c        | 38 +++++++++++++------------
 net/ipv6/tcp_ipv6.c     |  5 ++--
 7 files changed, 116 insertions(+), 31 deletions(-)


base-commit: 2208ab00f4f213ade91ab775e7f8c4dacdcc856e
-- 
2.56.0
Thread (6 messages) 6 messages, 1 author, last reply 1d ago
Keyboard shortcuts
hback out one level
jnext message in thread / next series-diff position
kprevious message in thread / previous series-diff position
ldrill in
Escclose help / fold thread tree
?toggle this help