Code processing skb from neigh->arp_queue can access its
nf_bridge->physindev, which can already be freed, leading to crash.

So, as Florian suggests, we can put physinif on nf_bridge and peek into
the original device with dev_get_by_index_rcu(), so that we can be sure
that device is not freed under us.

This is a second attempt to fix this issue, first attempt:

"neighbour: purge nf_bridged skb from foreign device neigh"
https://lore.kernel.org/netdev/20240108085232.95437-1-ptikhomirov@virtuozzo.com/

v3: resend it to proper lists and recipients

Pavel Tikhomirov (4):
  netfilter: nfnetlink_log: use proper helper for fetching physinif
  netfilter: nf_queue: remove excess nf_bridge variable
  netfilter: propagate net to nf_bridge_get_physindev
  netfilter: bridge: replace physindev with physinif in nf_bridge_info

 include/linux/netfilter_bridge.h           |  6 ++--
 include/linux/skbuff.h                     |  2 +-
 net/bridge/br_netfilter_hooks.c            | 42 +++++++++++++++++-----
 net/bridge/br_netfilter_ipv6.c             | 14 +++++---
 net/ipv4/netfilter/nf_reject_ipv4.c        |  9 +++--
 net/ipv6/netfilter/nf_reject_ipv6.c        | 11 ++++--
 net/netfilter/ipset/ip_set_hash_netiface.c |  8 ++---
 net/netfilter/nf_log_syslog.c              | 13 +++----
 net/netfilter/nf_queue.c                   |  6 ++--
 net/netfilter/nfnetlink_log.c              |  8 ++---
 net/netfilter/xt_physdev.c                 |  2 +-
 11 files changed, 80 insertions(+), 41 deletions(-)

-- 
2.43.0

On Thu, Jan 11, 2024 at 11:06:36PM +0800, Pavel Tikhomirov wrote:
> Code processing skb from neigh->arp_queue can access its
> nf_bridge->physindev, which can already be freed, leading to crash.
> 
> So, as Florian suggests, we can put physinif on nf_bridge and peek into
> the original device with dev_get_by_index_rcu(), so that we can be sure
> that device is not freed under us.
> 
> This is a second attempt to fix this issue, first attempt:
> 
> "neighbour: purge nf_bridged skb from foreign device neigh"
> https://lore.kernel.org/netdev/20240108085232.95437-1-ptikhomirov@virtuozzo.com/

I have applied this series to nf.git

I have added a Fixed: tag sufficiently old to the patch fix so it can
reach -stable at some point.

My understanding is that this problem has been always there for
br_netfilter.

Pablo Neira Ayuso <pablo@netfilter.org> wrote:
> My understanding is that this problem has been always there for
> br_netfilter.

Right.