All of lore.kernel.org
 help / color / mirror / Atom feed
* [RFC net-next 0/3] net: fix route flush for routes using nexthop objects
@ 2026-11-20 20:01 Stephen Hemminger
  2026-11-20 20:01 ` [RFC next-net 1/3] net: accept legacy nexthop attributes when deleting by nexthop id Stephen Hemminger
                   ` (2 more replies)
  0 siblings, 3 replies; 4+ messages in thread
From: Stephen Hemminger @ 2026-11-20 20:01 UTC (permalink / raw)
  To: netdev; +Cc: dsahern, Stephen Hemminger

"ip route flush" fails on any route that uses a nexthop object:

  # ip nexthop add id 101 via 192.0.2.1 dev dummy0
  # ip route add 203.0.113.0/24 nhid 101
  # ip route flush to 203.0.113.0/24
  Failed to send flush request: Invalid argument

The first attempt at fixing that was to modify iproute but
that is just working around a kernel bug. Flush should be
able to be done as dump-and-echo by flipping to set
RTM_DELROUTE. This always worked until nexthop compat mode
was introduced.

This series fixes thes by changing where fib handling
is done on delete. I don't use next hop id anywhere
so this is just a proposal.

Patch 1 ignores the legacy nexthop attributes on RTM_DELROUTE when
RTA_NH_ID is present, for both families. The id is already
authoritative in the matching code, so nothing downstream looks at the
fields being cleared. RTM_NEWROUTE stays strict.

Patch 2 accepts RTN_BLACKHOLE on delete when the alias uses a blackhole
nexthop object, so a route can be removed using the type the kernel
advertised. Dropping the dump-side rewrite instead would change what
every existing reader sees.

Patch 3 adds the selftest coverage that was missing. fib_nexthops.sh
exercises "ip nexthop flush" thoroughly but never "ip route flush"
against nhid routes, which is how the two paths drifted apart in the
first place.

Testing: the new tests fail 15 assertions against an unpatched kernel
and all 26 pass with the series applied, booted under qemu with
nexthop_compat_mode=1. The existing fib_nexthops cases, including the
legacy delete paths patch 1 must not disturb, still pass: 93
assertions across ipv4_fcnal, ipv6_fcnal, the compat_mode pair and the
fcnal_runtime pair.

Stephen Hemminger (3):
  net: accept legacy nexthop attributes when deleting by nexthop id
  ipv4: accept the reported route type when deleting a blackhole nexthop
    route
  selftests: fib_nexthops: test route flush with nexthop objects

 net/ipv4/fib_frontend.c                     |  17 ++-
 net/ipv4/fib_lookup.h                       |   1 +
 net/ipv4/fib_semantics.c                    |  14 +++
 net/ipv4/fib_trie.c                         |   2 +-
 net/ipv6/route.c                            |  14 ++-
 tools/testing/selftests/net/fib_nexthops.sh | 126 ++++++++++++++++++++
 6 files changed, 169 insertions(+), 5 deletions(-)

-- 
2.53.0


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2026-11-20 20:06 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-20 20:01 [RFC net-next 0/3] net: fix route flush for routes using nexthop objects Stephen Hemminger
2026-11-20 20:01 ` [RFC next-net 1/3] net: accept legacy nexthop attributes when deleting by nexthop id Stephen Hemminger
2026-11-20 20:01 ` [RFC next-net 2/3] ipv4: accept the reported route type when deleting a blackhole nexthop route Stephen Hemminger
2026-11-20 20:01 ` [RFC next-net 3/3] selftests: fib_nexthops: test route flush with nexthop objects Stephen Hemminger

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.