current code silently ignores change of port in the request
message. This patch makes sure the port is modified and
notification is sent to userspace.

Fixes: cf6b8e1e ("bridge: add API to notify bridge driver of learned FBD on offloaded device")
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When a new static FDB is added to the bridge a notification is sent to
the driver for offload. In case of successful offload the driver should
notify the bridge back, which in turn should mark the FDB as offloaded.

Currently, externally learned is equivalent for being offloaded which is
not correct due to the fact that FDBs which are added from user-space are
also marked as externally learned. In order to specify if an FDB was
successfully offloaded a new flag is introduced.
Signed-off-by: NArkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently the bridge doesn't notify the underlying devices about new
FDBs learned. The FDB sync is placed on the switchdev notifier chain
because devices may potentially learn FDB that are not directly related
to their ports, for example:

1. Mixed SW/HW bridge - FDBs that point to the ASICs external devices
                        should be offloaded as CPU traps in order to
			perform forwarding in slow path.
2. EVPN - Externally learned FDBs for the vtep device.

Notification is sent only about static FDB add/del. This is done due
to fact that currently this is the only scenario supported by switch
drivers.
Signed-off-by: NArkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: NIvan Vecera <ivecera@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is done as a preparation to moving the switchdev notifier chain
to be atomic. The FDB external learning should be called under rtnl
or rcu.
Signed-off-by: NArkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 7e26bf45 ("net: bridge: allow SW learn to take over HW fdb
entries") added the ability to "take over an entry which was previously
learned via HW when it shows up from a SW port".

However, if an entry was learned via HW and then a control packet
(e.g., ARP request) was trapped to the CPU, the bridge driver will
update the entry and remove the externally learned flag, although the
entry is still present in HW. Instead, only clear the externally learned
flag in case of roaming.

Fixes: 7e26bf45 ("net: bridge: allow SW learn to take over HW fdb entries")
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NArkadi Sharashevsky <arkadis@mellanox.com>
Cc: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Acked-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Recently we added support for SW fdbs to take over HW ones, but that
results in changing a user-visible fdb flag thus we need to send a
notification, also it's consistent with how HW takes over SW entries.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The NTF_EXT_LEARNED flag was added for switchdev and externally learned
entries, but it can also be used for entries learned via a software
in user-space which requires dynamic entries that do not expire.
One such case that we have is with quagga and evpn which need dynamic
entries but also require to age them themselves.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Allow to take over an entry which was previously learned via HW when it
shows up from a SW port. This is analogous to how HW takes over SW learned
entries already.
Suggested-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Andrei reported a false alarm of lockdep at net/bridge/br_fdb.c:109,
this is because in Andrei's case, a spin_bug() was already triggered
before this, therefore the debug_locks is turned off, lockdep_is_held()
is no longer accurate after that. We should use lockdep_assert_held_once()
instead of lockdep_is_held() to respect debug_locks.

Fixes: 410b3d48 ("bridge: fdb: add proper lock checks in searching functions")
Reported-by: NAndrei Vagin <avagin@gmail.com>
Signed-off-by: NCong Wang <xiyou.wangcong@gmail.com>
Acked-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

added_by_external_learn fdb entries are added and expired by
external entities like switchdev driver or external controllers.
ageing is already disabled for such entries. Hence, don't
indicate expiry for such fdb entries.

CC: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Reviewed-by: NIdo Schimmel <idosch@mellanox.com>
Tested-by: NIdo Schimmel <idosch@mellanox.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We can simplify the logic of entries pointing to the bridge by
converging the fdb_delete_by functions, this would allow us to use the
same function for both cases since the fdb's dst is set to NULL if it is
pointing to the bridge thus we can always check for a port match.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to avoid new errors add checks to br_fdb_find and fdb_find_rcu
functions. The first requires hash_lock, the second obviously RCU.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Before this patch we had 3 different fdb searching functions which was
confusing. This patch reduces all of them to one - fdb_find_rcu(), and
two flavors: br_fdb_find() which requires hash_lock and br_fdb_find_rcu
which requires RCU. This makes it clear what needs to be used, we also
remove two abusers of __br_fdb_get which called it under hash_lock and
replace them with br_fdb_find().
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jiffies is volatile so read it once.
Signed-off-by: NStephen Hemminger <sthemmin@microsoft.com>
Acked-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Writing once per jiffy is enough to limit the bridge's false sharing.
After this change the bridge doesn't show up in the local load HitM stats.
Suggested-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Move the fdb garbage collector to a workqueue which fires at least 10
milliseconds apart and cleans chain by chain allowing for other tasks
to run in the meantime. When having thousands of fdbs the system is much
more responsive. Most importantly remove the need to check if the
matched entry has expired in __br_fdb_get that causes false-sharing and
is completely unnecessary if we cleanup entries, at worst we'll get 10ms
of traffic for that entry before it gets deleted.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds vlan and address to warning messages printed
in the bridge fdb code for debuggability.
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

fdb dumps spanning multiple skb's currently restart from the first
interface again for every skb. This results in unnecessary
iterations on the already visited interfaces and their fdb
entries. In large scale setups, we have seen this to slow
down fdb dumps considerably. On a system with 30k macs we
see fdb dumps spanning across more than 300 skbs.

To fix the problem, this patch replaces the existing single fdb
marker with three markers: netdev hash entries, netdevs and fdb
index to continue where we left off instead of restarting from the
first netdev. This is consistent with link dumps.

In the process of fixing the performance issue, this patch also
re-implements fix done by
commit 472681d5 ("net: ndo_fdb_dump should report -EMSGSIZE to rtnl_fdb_dump")
(with an internal fix from Wilson Kok) in the following ways:
- change ndo_fdb_dump handlers to return error code instead
of the last fdb index
- use cb->args strictly for dump frag markers and not error codes.
This is consistent with other dump functions.

Below results were taken on a system with 1000 netdevs
and 35085 fdb entries:
before patch:
$time bridge fdb show | wc -l
15065

real    1m11.791s
user    0m0.070s
sys 1m8.395s

(existing code does not return all macs)

after patch:
$time bridge fdb show | wc -l
35085

real    0m2.017s
user    0m0.113s
sys 0m1.942s
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NWilson Kok <wkok@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Adding fdb entries pointing to the bridge device uses fdb_insert(),
which lacks various checks and does not respect added_by_user flag.

As a result, some inconsistent behavior can happen:
* Adding temporary entries succeeds but results in permanent entries.
* Same goes for "dynamic" and "use".
* Changing mac address of the bridge device causes deletion of
  user-added entries.
* Replacing existing entries looks successful from userspace but actually
  not, regardless of NLM_F_EXCL flag.

Use the same logic as other entries and fix them.

Fixes: 3741873b ("bridge: allow adding of fdb entries pointing to the bridge device")
Signed-off-by: NToshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The missing br_vlan_should_use() test caused creation of an unneeded
local fdb entry on changing mac address of a bridge device when there is
a vlan which is configured on a bridge port but not on the bridge
device.

Fixes: 2594e906 ("bridge: vlan: add per-vlan struct and move to rhashtables")
Signed-off-by: NToshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Acked-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the send skbuff reaches the end, nlmsg_put and friends returns
-EMSGSIZE but it is silently thrown away in ndo_fdb_dump. It is called
within a for_each_netdev loop and the first fdb entry of a following
netdev could fit in the remaining skbuff.  This breaks the mechanism
of cb->args[0] and idx to keep track of the entries that are already
dumped, which results missing entries in bridge fdb show command.
Signed-off-by: NMinoura Makoto <minoura@valinux.co.jp>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

switchdev drivers need to know the netdev on which the switchdev op was
invoked. For example, the STP state of a VLAN interface configured on top
of a port can change while being member in a bridge. In this case, the
underlying driver should only change the STP state of that particular
VLAN and not of all the VLANs configured on the port.

However, current switchdev infrastructure only passes the port netdev down
to the driver. Solve that by passing the original device down to the
driver as part of the required switchdev object / attribute.

This doesn't entail any change in current switchdev drivers. It simply
enables those supporting stacked devices to know the originating device
and act accordingly.
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Problem Description:
We can add fdbs pointing to the bridge with NULL ->dst but that has a
few race conditions because br_fdb_insert() is used which first creates
the fdb and then, after the fdb has been published/linked, sets
"is_local" to 1 and in that time frame if a packet arrives for that fdb
it may see it as non-local and either do a NULL ptr dereference in
br_forward() or attach the fdb to the port where it arrived, and later
br_fdb_insert() will make it local thus getting a wrong fdb entry.
Call chain br_handle_frame_finish() -> br_forward():
But in br_handle_frame_finish() in order to call br_forward() the dst
should not be local i.e. skb != NULL, whenever the dst is
found to be local skb is set to NULL so we can't forward it,
and here comes the problem since it's running only
with RCU when forwarding packets it can see the entry before "is_local"
is set to 1 and actually try to dereference NULL.
The main issue is that if someone sends a packet to the switch while
it's adding the entry which points to the bridge device, it may
dereference NULL ptr. This is needed now after we can add fdbs
pointing to the bridge.  This poses a problem for
br_fdb_update() as well, while someone's adding a bridge fdb, but
before it has is_local == 1, it might get moved to a port if it comes
as a source mac and then it may get its "is_local" set to 1

This patch changes fdb_create to take is_local and is_static as
arguments to set these values in the fdb entry before it is added to the
hash. Also adds null check for port in br_forward.

Fixes: 3741873b ("bridge: allow adding of fdb entries pointing to the bridge device")
Reported-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Acked-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since spinlock is held here, defer the switchdev operation. Also, ensure
that defered switchdev ops are processed before port master device
is unlinked.
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When object is used in deferred work, we cannot use pointers in
switchdev object structures because the memory they point at may be already
used by someone else. So rather do local copy of the value.
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Acked-by: NScott Feldman <sfeldma@gmail.com>
Reviewed-by: NJohn Fastabend <john.r.fastabend@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch enables adding of fdb entries pointing to the bridge device.
This can be used to propagate mac address of vlan interfaces
configured on top of the vlan filtering bridge.

Before:
$bridge fdb add 44:38:39:00:27:9f dev bridge
RTNETLINK answers: Invalid argument

After:
$bridge fdb add 44:38:39:00:27:9f dev bridge
Signed-off-by: NRoopa Prabhu <roopa@cumulusnetworks.com>
Reviewed-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Suggested-by: NScott Feldman <sfeldma@gmail.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Acked-by: NScott Feldman <sfeldma@gmail.com>
Reviewed-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace "void *obj" with a generic structure. Introduce couple of
helpers along that.
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Acked-by: NScott Feldman <sfeldma@gmail.com>
Reviewed-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make the struct name in sync with object id name.
Suggested-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Acked-by: NScott Feldman <sfeldma@gmail.com>
Reviewed-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Suggested-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Acked-by: NScott Feldman <sfeldma@gmail.com>
Reviewed-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Similar to the notifier_call callback of a notifier_block, change the
function signature of switchdev add and del operations to:

    int switchdev_port_obj_add/del(struct net_device *dev,
                                   enum switchdev_obj_id id, void *obj);

This allows the caller to pass a specific switchdev_obj_* structure
instead of the generic switchdev_obj one.

Drivers implementation of these operations and switchdev have been
changed accordingly.
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch changes the bridge vlan implementation to use rhashtables
instead of bitmaps. The main motivation behind this change is that we
need extensible per-vlan structures (both per-port and global) so more
advanced features can be introduced and the vlan support can be
extended. I've tried to break this up but the moment net_port_vlans is
changed and the whole API goes away, thus this is a larger patch.
A few short goals of this patch are:
- Extensible per-vlan structs stored in rhashtables and a sorted list
- Keep user-visible behaviour (compressed vlans etc)
- Keep fastpath ingress/egress logic the same (optimizations to come
  later)

Here's a brief list of some of the new features we'd like to introduce:
- per-vlan counters
- vlan ingress/egress mapping
- per-vlan igmp configuration
- vlan priorities
- avoid fdb entries replication (e.g. local fdb scaling issues)

The structure is kept single for both global and per-port entries so to
avoid code duplication where possible and also because we'll soon introduce
"port0 / aka bridge as port" which should simplify things further
(thanks to Vlad for the suggestion!).

Now we have per-vlan global rhashtable (bridge-wide) and per-vlan port
rhashtable, if an entry is added to a port it'll get a pointer to its
global context so it can be quickly accessed later. There's also a
sorted vlan list which is used for stable walks and some user-visible
behaviour such as the vlan ranges, also for error paths.
VLANs are stored in a "vlan group" which currently contains the
rhashtable, sorted vlan list and the number of "real" vlan entries.
A good side-effect of this change is that it resembles how hw keeps
per-vlan data.
One important note after this change is that if a VLAN is being looked up
in the bridge's rhashtable for filtering purposes (or to check if it's an
existing usable entry, not just a global context) then the new helper
br_vlan_should_use() needs to be used if the vlan is found. In case the
lookup is done only with a port's vlan group, then this check can be
skipped.

Things tested so far:
- basic vlan ingress/egress
- pvids
- untagged vlans
- undef CONFIG_BRIDGE_VLAN_FILTERING
- adding/deleting vlans in different scenarios (with/without global ctx,
  while transmitting traffic, in ranges etc)
- loading/removing the module while having/adding/deleting vlans
- extracting bridge vlan information (user ABI), compressed requests
- adding/deleting fdbs on vlans
- bridge mac change, promisc mode
- default pvid change
- kmemleak ON during the whole time
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NSiva Mannem <siva.mannem.lnx@gmail.com>
Signed-off-by: NScott Feldman <sfeldma@gmail.com>
Acked-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Acked-by: NJiri Pirko <jiri@resnulli.us>
Acked-by: NPremkumar Jonnala <pjonnala@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts commit f1d5ca43, reversing
changes made to 4933d85c.

I applied v2 instead of v3.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The address in the switchdev_obj_fdb structure is currently represented
as a pointer. Replacing it for a 6-byte array allows switchdev to carry
addresses directly read from hardware registers, not stored by the
switch chip driver (as in Rocker).
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a new argument to br_fdb_delete_by_port which allows to specify a
vid to match when flushing entries and use it in nbp_vlan_delete() to
flush the dynamically learned entries of the vlan/port pair when removing
a vlan from a port. Before this patch only the local mac was being
removed and the dynamically learned ones were left to expire.
Note that the do_all argument is still respected and if specified, the
vid will be ignored.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We need to delete from offload the device externally learnded fdbs when any
one of these events happen:

1) Bridge ages out fdb.  (When bridge is doing ageing vs. device doing
ageing.  If device is doing ageing, it would send SWITCHDEV_FDB_DEL
directly).

2) STP state change flushes fdbs on port.

3) User uses sysfs interface to flush fdbs from bridge or bridge port:

	echo 1 >/sys/class/net/BR_DEV/bridge/flush
	echo 1 >/sys/class/net/BR_PORT/brport/flush

4) Offload driver send event SWITCHDEV_FDB_DEL to delete fdb entry.

For rocker, we can now get called to delete fdb entry in wait and nowait
contexts, so set NOWAIT flag when deleting fdb entry.
Signed-off-by: NScott Feldman <sfeldma@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Before this patch the user-specified bridge port was ignored when
deleting an fdb entry and thus one could delete an entry that belonged
to any port.
Example (eth0 and eth1 are br0 ports):
bridge fdb add 00:11:22:33:44:55 dev eth0 master
bridge fdb del 00:11:22:33:44:55 dev eth1 master
(succeeds)

after the patch:
bridge fdb add 00:11:22:33:44:55 dev eth0 master
bridge fdb del 00:11:22:33:44:55 dev eth1 master
RTNETLINK answers: No such file or directory

Based on a patch by Wilson Kok.
Reported-by: NWilson Kok <wkok@cumulusnetworks.com>
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

br_fdb_update() can be called in process context in the following way:
br_fdb_add() -> __br_fdb_add() -> br_fdb_update() (if NTF_USE flag is set)
so we need to disable softirqs because there are softirq users of the
hash_lock. One easy way to reproduce this is to modify the bridge utility
to set NTF_USE, enable stp and then set maxageing to a low value so
br_fdb_cleanup() is called frequently and then just add new entries in
a loop. This happens because br_fdb_cleanup() is called from timer/softirq
context. The spin locks in br_fdb_update were _bh before commit f8ae737d
("[BRIDGE]: forwarding remove unneeded preempt and bh diasables")
and at the time that commit was correct because br_fdb_update() couldn't be
called from process context, but that changed after commit:
292d1398 ("bridge: add NTF_USE support")
Using local_bh_disable/enable around br_fdb_update() allows us to keep
using the spin_lock/unlock in br_fdb_update for the fast-path.
Signed-off-by: NNikolay Aleksandrov <nikolay@cumulusnetworks.com>
Fixes: 292d1398 ("bridge: add NTF_USE support")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This reverts commit 1d7c4903.

Nikolay Aleksandrov has a better version of this fix.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>