The values are shared between VXLAN-GPE and NSH. Originally probably by
coincidence but I notified both working groups about this last year and they
seem to keep the values in sync since then.

Hopefully they'll get a single IANA registry for the values, too. (I asked
them for that.)

Factor out the code to be shared by the NSH implementation.

NSH and MPLS values are added in this patch, too. For MPLS, the drafts
incorrectly assign only a single value, while we have two MPLS ethertypes.
I raised the problem with both groups. For now, I assume the value is for
unicast.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

refcount_t type and corresponding API should be
used instead of atomic_t when the variable is used as
a reference counter. This allows to avoid accidental
refcounter overflows that might lead to use-after-free
situations.
Signed-off-by: NElena Reshetova <elena.reshetova@intel.com>
Signed-off-by: NHans Liljestrand <ishkamiel@gmail.com>
Signed-off-by: NKees Cook <keescook@chromium.org>
Signed-off-by: NDavid Windsor <dwindsor@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It's not a good idea to add the same hlist_node to two different hash lists.
This leads to various hard to debug memory corruptions.

Fixes: b1be00a6 ("vxlan: support both IPv4 and IPv6 sockets in a single vxlan device")
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

* Multicast addresses are never valid as local address
* Link-local IPv6 unicast addresses may only be used as remote when the
  local address is link-local as well
* Don't allow link-local IPv6 local/remote addresses without interface

We also store in the flags field if link-local addresses are used for the
follow-up patches that actually make VXLAN over link-local IPv6 work.
Signed-off-by: NMatthias Schiffer <mschiffer@universe-factory.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is no good reason to keep the flags twice in vxlan_dev and
vxlan_config.
Signed-off-by: NMatthias Schiffer <mschiffer@universe-factory.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NPravin B Shelar <pshelar@ovn.org>
Acked-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When vxlan device is closed vxlan socket is freed. This
operation can race with vxlan-xmit function which
dereferences vxlan socket. Following patch uses RCU
mechanism to avoid this situation.
Signed-off-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add utility functions to convert a 32 bits key into a 64 bits tunnel and
vice versa.
These functions will be used instead of cloning code in GRE and VXLAN,
and in tc act_iptunnel which will be introduced in a following patch in
this patchset.
Signed-off-by: NAmir Vadai <amir@vadai.me>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Reviewed-by: NShmulik Ladkani <shmulik.ladkani@gmail.com>
Acked-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Now that we have all the drivers using udp_tunnel_get_rx_ports,
ndo_add_udp_enc_rx_port, and ndo_del_udp_enc_rx_port we can drop the
function calls that were specific to VXLAN and GENEVE.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch merges the notifiers for VXLAN and GENEVE into a single UDP
tunnel notifier.  The idea is that we will want to only have to make one
notifier call to receive the list of ports for VXLAN and GENEVE tunnels
that need to be offloaded.

In addition we add a new set of ndo functions named ndo_udp_tunnel_add and
ndo_udp_tunnel_del that are meant to allow us to track the tunnel meta-data
such as port and address family as tunnels are added and removed.  The
tunnel meta-data is now transported in a structure named udp_tunnel_info
which for now carries the type, address family, and port number.  In the
future this could be updated so that we can include a tuple of values
including things such as the destination IP address and other fields.

I also ended up going with a naming scheme that consisted of using the
prefix udp_tunnel on function names.  I applied this to the notifier and
ndo ops as well so that it hopefully points to the fact that these are
primarily used in the udp_tunnel functions.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch makes it so that we add udp_tunnel.h to vxlan.h and geneve.h
header files.  This is useful as I plan to move the generic handlers for
the port offloads into the udp_tunnel header file and leave the vxlan and
geneve headers to be a bit more protocol specific.

I also went through and cleaned out a number of redundant includes that
where in the .h and .c files for these drivers.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We need to perform an additional check on the inner headers to determine if
we can offload the checksum for them.  Previously this check didn't occur
so we would generate an invalid frame in the case of an IPv6 header
encapsulated inside of an IPv4 tunnel.  To fix this I added a secondary
check to vxlan_features_check so that we can verify that we can offload the
inner checksum.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently all drivers depend and autoload the vxlan module because how
vxlan_get_rx_port is linked into them. Remove this dependency:

By using a new event type in the netdevice notifier call chain we proxy
the request from the drivers to flush and resetup the vxlan ports not
directly via function call but by the already existing netdevice
notifier call chain.

I added a separate new event type, NETDEV_OFFLOAD_PUSH_VXLAN, to do so.
We don't need to save those ids, as the event type field is an unsigned
long and using specialized event types for this purpose seemed to be a
more elegant way. This also comes in beneficial if in future we want to
add offloading knobs for vxlan.

Cc: Jesse Gross <jesse@kernel.org>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Due to the fact that the udp socket is destructed asynchronously in a
work queue, we have some nondeterministic behavior during shutdown of
vxlan tunnels and creating new ones. Fix this by keeping the destruction
process synchronous in regards to the user space process so IFF_UP can
be reliably set.

udp_tunnel_sock_release destroys vs->sock->sk if reference counter
indicates so. We expect to have the same lifetime of vxlan_sock and
vxlan_sock->sock->sk even in fast paths with only rcu locks held. So
only destruct the whole socket after we can be sure it cannot be found
by searching vxlan_net->sock_list.

Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Jiri Benc <jbenc@redhat.com>
Cc: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Adapt vxlan_gro_receive, vxlan_gro_complete to take a socket argument.
Set these functions in tunnel_config.  Don't set udp_offloads any more.
Signed-off-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Implement VXLAN-GPE. Only COLLECT_METADATA is supported for now (it is
possible to support static configuration, too, if there is demand for it).

The GPE header parsing has to be moved before iptunnel_pull_header, as we
need to know the protocol.

v2: Removed what was called "L2 mode" in v1 of the patchset. Only "L3 mode"
    (now called "raw mode") is added by this patch. This mode does not allow
    Ethernet header to be encapsulated in VXLAN-GPE when using ip route to
    specify the encapsulation, IP header is encapsulated instead. The patch
    does support Ethernet to be encapsulated, though, using ETH_P_TEB in
    skb->protocol. This will be utilized by other COLLECT_METADATA users
    (openvswitch in particular).

    If there is ever demand for Ethernet encapsulation with VXLAN-GPE using
    ip route, it's easy to add a new flag switching the interface to
    "Ethernet mode" (called "L2 mode" in v1 of this patchset). For now,
    leave this out, it seems we don't need it.

    Disallowed more flag combinations, especially RCO with GPE.
    Added comment explaining that GBP and GPE cannot be set together.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Sparse reports false positives for the header manipulation inlines. Annotate
them correctly.

Tested by sparse on a little endian and big endian machine.

Fixes: 54bfd872 ("vxlan: keep flags and vni in network byte order")
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This work adds support for setting the IPv6 flow label for vxlan per
device and through collect metadata (ip_tunnel_key) frontends. The
vxlan dst cache does not need any special considerations here, for
the cases where caches can be used, the label is static per cache.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This change makes it so that if UDP CSUM is not specified we will default
to enabling it.  The main motivation behind this is the fact that with the
use of outer checksum we can greatly improve the performance for VXLAN
tunnels on devices that don't know how to parse tunnel headers.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Acked-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The tun_id field in struct ip_tunnel_key is __be64, not __be32. We need to
convert the vni to tun_id correctly.

Fixes: 54bfd872 ("vxlan: keep flags and vni in network byte order")
Reported-by: NPaolo Abeni <pabeni@redhat.com>
Tested-by: NPaolo Abeni <pabeni@redhat.com>
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NThadeu Lima de Souza Cascardo <cascardo@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Prevent repeated conversions from and to network order in the fast path.

To achieve this, define all flag constants in big endian order and store VNI
as __be32. To prevent confusion between the actual VNI value and the VNI
field from the header (which contains additional reserved byte), strictly
distinguish between "vni" and "vni_field".
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, pointer to the vxlan header is kept in a local variable. It has
to be reloaded whenever the pskb pull operations are performed which usually
happens somewhere deep in called functions.

Create a vxlan_hdr function and use it to reference the vxlan header
instead.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In case of UDP traffic with datagram length
below MTU this give about 3% performance increase
when tunneling over ipv4 and about 70% when
tunneling over ipv6.
Signed-off-by: NPaolo Abeni <pabeni@redhat.com>
Suggested-and-acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RCO and GBP are VXLAN extensions, not specified in RFC 7348. Because of
that, they need to be explicitly enabled when creating vxlan interface. By
default, those extensions are not used and plain VXLAN header is sent and
received.

Reflect this in vxlan.h: first, the plain VXLAN header is defined. Following
it, RCO is documented and defined, and likewise for GBP.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

VNI_HASH_BITS and VNI_HASH_SIZE are defined twice. Remove the extra
definitions.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

include/net/vxlan.h is a kernel header, no need to prefix fixed size types
with double underscore.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The name NETIF_F_ALL_CSUM is a misnomer. This does not correspond to the
set of features for offloading all checksums. This is a mask of the
checksum offload related features bits. It is incorrect to set both
NETIF_F_HW_CSUM and NETIF_F_IP_CSUM or NETIF_F_IPV6 at the same time for
features of a device.

This patch:
  - Changes instances of NETIF_F_ALL_CSUM to NETIF_F_CSUM_MASK (where
    NETIF_F_ALL_CSUM is being used as a mask).
  - Changes bonding, sfc/efx, ipvlan, macvlan, vlan, and team drivers to
    use NEITF_F_HW_CSUM in features list instead of NETIF_F_ALL_CSUM.
Signed-off-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 3511494c ("vxlan: Group Policy extension") changed definition of
VXLAN_HF_RCO from 0x00200000 to BIT(24). This is obviously incorrect. It's
also in violation with the RFC draft.

Fixes: 3511494c ("vxlan: Group Policy extension")
Cc: Thomas Graf <tgraf@suug.ch>
Cc: Tom Herbert <therbert@google.com>
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For metadata based vxlan interface, open both IPv4 and IPv6 socket. This is
much more user friendly: it's not necessary to create two vxlan interfaces
and pay attention to using the right one in routing rules.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The vxlan_get_sk_family inline function was added after the last #endif,
making multiple inclusion of net/vxlan.h fail. Move it to the proper place.
Reported-by: NMark Rustad <mark.d.rustad@intel.com>
Fixes: 705cc62f ("vxlan: provide access function for vxlan socket address family")
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add calls to gro_cells infrastructure to do GRO when receiving on a tunnel.

Testing:

Ran 200 netperf TCP_STREAM instance

  - With fix (GRO enabled on VXLAN interface)

    Verify GRO is happening.

    9084 MBps tput
    3.44% CPU utilization

  - Without fix (GRO disabled on VXLAN interface)

    Verified no GRO is happening.

    9084 MBps tput
    5.54% CPU utilization
Signed-off-by: NTom Herbert <tom@herbertland.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IFLA_VXLAN_FLOWBASED is useless without IFLA_VXLAN_COLLECT_METADATA,
so combine them into single IFLA_VXLAN_COLLECT_METADATA flag.
'flowbased' doesn't convey real meaning of the vxlan tunnel mode.
This mode can be used by routing, tc+bpf and ovs.
Only ovs is strictly flow based, so 'collect metadata' is a better
name for this tunnel mode.
Signed-off-by: NAlexei Starovoitov <ast@plumgrid.com>
Acked-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This gets rid of all OVS specific VXLAN code in the receive and
transmit path by using a VXLAN net_device to represent the vport.
Only a small shim layer remains which takes care of handling the
VXLAN specific OVS Netlink configuration.

Unexports vxlan_sock_add(), vxlan_sock_release(), vxlan_xmit_skb()
since they are no longer needed.
Signed-off-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NPravin B Shelar <pshelar@nicira.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This factors out the device configuration out of the RTNL newlink
API which allows for in-kernel creation of VXLAN net_devices.
Signed-off-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Allows putting a VXLAN device into a new flow-based mode in which
skbs with a ip_tunnel_info dst metadata attached will be encapsulated
according to the instructions stored in there with the VXLAN device
defaults taken into consideration.

Similar on the receive side, if the VXLAN_F_COLLECT_METADATA flag is
set, the packet processing will populate a ip_tunnel_info struct for
each packet received and attach it to the skb using the new metadata
dst.  The metadata structure will contain the outer header and tunnel
header fields which have been stripped off. Layers further up in the
stack such as routing, tc or netfitler can later match on these fields
and perform forwarding. It is the responsibility of upper layers to
ensure that the flag is set if the metadata is needed. The flag limits
the additional cost of metadata collecting based on demand.

This prepares the VXLAN device to be steered by the routing and other
subsystems which allows to support encapsulation for a large number
of tunnel endpoints and tunnel ids through a single net_device which
improves the scalability.

It also allows for OVS to leverage this mode which in turn allows for
the removal of the OVS specific VXLAN code.

Because the skb is currently scrubed in vxlan_rcv(), the attachment of
the new dst metadata is postponed until after scrubing which requires
the temporary addition of a new member to vxlan_metadata. This member
is removed again in a later commit after the indirect VXLAN receive API
has been removed.
Signed-off-by: NThomas Graf <tgraf@suug.ch>
Signed-off-by: NPravin B Shelar <pshelar@nicira.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

That was we can make sure the output path of ipv4/ipv6 operate on
the UDP socket rather than whatever random thing happens to be in
skb->sk.

Based upon a patch by Jiri Pirko.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>

commit dfd8645e wrongly assumes that VXLAN_VDI_MASK includes
eight lower order reserved bits of VNI field that are using for remote
checksum offload.

Right now, when VNI number greater then 0xffff, vxlan_udp_encap_recv()
will always return with 'bad_flag' error, reducing the usable vni range
from 0..16777215 to 0..65535. Also, it doesn't really check whether RCO
bits processed or not.

Fix it by adding new VNI mask which has all 32 bits of VNI field:
24 bits for id and 8 bits for other usage.
Signed-off-by: NAlexey Kodanev <alexey.kodanev@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Flags are used in the return path rather than the return patch.

Fixes: af33c1ad ("vxlan: Eliminate dependency on UDP socket in transmit path")
Signed-off-by: NSimon Horman <simon.horman@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Change remote checksum handling to set checksum partial as default
behavior. Added an iflink parameter to configure not using
checksum partial (calling csum_partial to update checksum).
Signed-off-by: NTom Herbert <therbert@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>