]> git.ipfire.org Git - thirdparty/kernel/linux.git/commit
ipv6: fix NULL pointer deref in ip6_rt_get_dev_rcu()
authorJakub Kicinski <kuba@kernel.org>
Sun, 1 Mar 2026 19:45:48 +0000 (11:45 -0800)
committerJakub Kicinski <kuba@kernel.org>
Wed, 4 Mar 2026 01:14:48 +0000 (17:14 -0800)
commit2ffb4f5c2ccb2fa1c049dd11899aee7967deef5a
tree15764f85bc4fed99fba6ee2178e48697fc6f3fc8
parent1a86a1f7d88996085934139fa4c063b6299a2dd3
ipv6: fix NULL pointer deref in ip6_rt_get_dev_rcu()

l3mdev_master_dev_rcu() can return NULL when the slave device is being
un-slaved from a VRF. All other callers deal with this, but we lost
the fallback to loopback in ip6_rt_pcpu_alloc() -> ip6_rt_get_dev_rcu()
with commit 4832c30d5458 ("net: ipv6: put host and anycast routes on
device with address").

  KASAN: null-ptr-deref in range [0x0000000000000108-0x000000000000010f]
  RIP: 0010:ip6_rt_pcpu_alloc (net/ipv6/route.c:1418)
  Call Trace:
   ip6_pol_route (net/ipv6/route.c:2318)
   fib6_rule_lookup (net/ipv6/fib6_rules.c:115)
   ip6_route_output_flags (net/ipv6/route.c:2607)
   vrf_process_v6_outbound (drivers/net/vrf.c:437)

I was tempted to rework the un-slaving code to clear the flag first
and insert synchronize_rcu() before we remove the upper. But looks like
the explicit fallback to loopback_dev is an established pattern.
And I guess avoiding the synchronize_rcu() is nice, too.

Fixes: 4832c30d5458 ("net: ipv6: put host and anycast routes on device with address")
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://patch.msgid.link/20260301194548.927324-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net/ipv6/route.c