[ Upstream commit d5256083f6 ]
While implementing ipvlan l3 and l3s mode for kubernetes CNI plugin,
I ran into the issue that while l3 mode is working fine, l3s mode
does not have any connectivity to kube-apiserver and hence all pods
end up in Error state as well. The ipvlan master device sits on
top of a bond device and hostns traffic to kube-apiserver (also running
in hostns) is DNATed from 10.152.183.1:443 to 139.178.29.207:37573
where the latter is the address of the bond0. While in l3 mode, a
curl to https://10.152.183.1:443 or to https://139.178.29.207:37573
works fine from hostns, neither of them do in case of l3s. In the
latter only a curl to https://127.0.0.1:37573 appeared to work where
for local addresses of bond0 I saw kernel suddenly starting to emit
ARP requests to query HW address of bond0 which remained unanswered
and neighbor entries in INCOMPLETE state. These ARP requests only
happen while in l3s.
Debugging this further, I found the issue is that l3s mode is piggy-
backing on l3 master device, and in this case local routes are using
l3mdev_master_dev_rcu(dev) instead of net->loopback_dev as per commit
f5a0aab84b ("net: ipv4: dst for local input routes should use l3mdev
if relevant") and 5f02ce24c2 ("net: l3mdev: Allow the l3mdev to be
a loopback"). I found that reverting them back into using the
net->loopback_dev fixed ipvlan l3s connectivity and got everything
working for the CNI.
Now judging from 4fbae7d83c ("ipvlan: Introduce l3s mode") and the
l3mdev paper in [0] the only sole reason why ipvlan l3s is relying
on l3 master device is to get the l3mdev_ip_rcv() receive hook for
setting the dst entry of the input route without adding its own
ipvlan specific hacks into the receive path, however, any l3 domain
semantics beyond just that are breaking l3s operation. Note that
ipvlan also has the ability to dynamically switch its internal
operation from l3 to l3s for all ports via ipvlan_set_port_mode()
at runtime. In any case, l3 vs l3s soley distinguishes itself by
'de-confusing' netfilter through switching skb->dev to ipvlan slave
device late in NF_INET_LOCAL_IN before handing the skb to L4.
Minimal fix taken here is to add a IFF_L3MDEV_RX_HANDLER flag which,
if set from ipvlan setup, gets us only the wanted l3mdev_l3_rcv() hook
without any additional l3mdev semantics on top. This should also have
minimal impact since dev->priv_flags is already hot in cache. With
this set, l3s mode is working fine and I also get things like
masquerading pod traffic on the ipvlan master properly working.
[0] https://netdevconf.org/1.2/papers/ahern-what-is-l3mdev-paper.pdf
Fixes: f5a0aab84b ("net: ipv4: dst for local input routes should use l3mdev if relevant")
Fixes: 5f02ce24c2 ("net: l3mdev: Allow the l3mdev to be a loopback")
Fixes: 4fbae7d83c ("ipvlan: Introduce l3s mode")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Mahesh Bandewar <maheshb@google.com>
Cc: David Ahern <dsa@cumulusnetworks.com>
Cc: Florian Westphal <fw@strlen.de>
Cc: Martynas Pumputis <m@lambda.lt>
Acked-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 5dc2d3996a ]
After we change the ipvlan mode from l3 to l2, or vice versa, we only
reset IFF_NOARP flag, but don't flush the ARP table cache, which will
cause eth->h_dest to be equal to eth->h_source in ipvlan_xmit_mode_l2().
Then the message will not come out of host.
Here is the reproducer on local host:
ip link set eth1 up
ip addr add 192.168.1.1/24 dev eth1
ip link add link eth1 ipvlan1 type ipvlan mode l3
ip netns add net1
ip link set ipvlan1 netns net1
ip netns exec net1 ip link set ipvlan1 up
ip netns exec net1 ip addr add 192.168.2.1/24 dev ipvlan1
ip route add 192.168.2.0/24 via 192.168.1.2
ping 192.168.2.2 -c 2
ip netns exec net1 ip link set ipvlan1 type ipvlan mode l2
ping 192.168.2.2 -c 2
Add the same configuration on remote host. After we set the mode to l2,
we could find that the src/dst MAC addresses are the same on eth1:
21:26:06.648565 00:b7:13:ad:d3:05 > 00:b7:13:ad:d3:05, ethertype IPv4 (0x0800), length 98: (tos 0x0, ttl 64, id 58356, offset 0, flags [DF], proto ICMP (1), length 84)
192.168.2.1 > 192.168.2.2: ICMP echo request, id 22686, seq 1, length 64
Fix this by calling dev_change_flags(), which will call netdevice notifier
with flag change info.
v2:
a) As pointed out by Wang Cong, check return value for dev_change_flags() when
change dev flags.
b) As suggested by Stefano and Sabrina, move flags setting before l3mdev_ops.
So we don't need to redo ipvlan_{, un}register_nf_hook() again in err path.
Reported-by: Jianlin Shi <jishi@redhat.com>
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Fixes: 2ad7bf3638 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 30877961b1 ]
Commit 296d485680 ("ipvlan: inherit MTU from master device") adjusted
the mtu from the master device when creating a ipvlan device, but it
would also override the mtu value set in rtnl_create_link. It causes
IFLA_MTU param not to take effect.
So this patch is to not adjust the mtu if IFLA_MTU param is set when
creating a ipvlan device.
Fixes: 296d485680 ("ipvlan: inherit MTU from master device")
Reported-by: Jianlin Shi <jishi@redhat.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[ Upstream commit 92ff426450 ]
Packets that don't have dest mac as the mac of the master device should
not be entertained by the IPvlan rx-handler. This is mostly true as the
packet path mostly takes care of that, except when the master device is
a virtual device. As demonstrated in the following case -
ip netns add ns1
ip link add ve1 type veth peer name ve2
ip link add link ve2 name iv1 type ipvlan mode l2
ip link set dev iv1 netns ns1
ip link set ve1 up
ip link set ve2 up
ip -n ns1 link set iv1 up
ip addr add 192.168.10.1/24 dev ve1
ip -n ns1 addr 192.168.10.2/24 dev iv1
ping -c2 192.168.10.2
<Works!>
ip neigh show dev ve1
ip neigh show 192.168.10.2 lladdr <random> dev ve1
ping -c2 192.168.10.2
<Still works! Wrong!!>
This patch adds that missing check in the IPvlan rx-handler.
Reported-by: Amit Sikka <amit.sikka@ericsson.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <alexander.levin@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When netdev_upper_dev_unlink failed in ipvlan_link_new, need to
unlink the ipvlan dev with upper dev.
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When ipvlan_link_new fails and creates one ipvlan port, it does not
destroy the ipvlan port created. It causes mem leak and the physical
device contains invalid ipvlan data.
Signed-off-by: Gao Feng <fgao@ikuai8.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In a typical IPvlan L3 setup where master is in default-ns and
each slave is into different (slave) ns. In this setup egress
packet processing for traffic originating from slave-ns will
hit all NF_HOOKs in slave-ns as well as default-ns. However same
is not true for ingress processing. All these NF_HOOKs are
hit only in the slave-ns skipping them in the default-ns.
IPvlan in L3 mode is restrictive and if admins want to deploy
iptables rules in default-ns, this asymmetric data path makes it
impossible to do so.
This patch makes use of the l3_rcv() (added as part of l3mdev
enhancements) to perform input route lookup on RX packets without
changing the skb->dev and then uses nf_hook at NF_INET_LOCAL_IN
to change the skb->dev just before handing over skb to L4.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
CC: David Ahern <dsa@cumulusnetworks.com>
Reviewed-by: David Ahern <dsa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The earlier patch c3aaa06d5a (ipvlan: scrub skb before routing
in L3 mode.) did this but only for TX path in L3 mode. This
patch extends it for both the modes for TX/RX path.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In case a qdisc is used on a ipvlan device, we need to use different
lockdep classes to avoid false positives.
Use the new netdev_lockdep_set_classes() generic helper.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When newlink creation fails at device-registration, the port->count
is decremented twice. Francesco Ruggeri (fruggeri@arista.com) found
this issue in Macvlan and the same exists in IPvlan driver too.
While fixing this issue I noticed another issue of missing unregister
in case of failure, so adding it to the fix which is similar to the
macvlan fix by Francesco in commit 3083796075 ("macvlan: fix failure
during registration v3")
Reported-by: Francesco Ruggeri <fruggeri@arista.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
CC: Eric Dumazet <edumazet@google.com>
CC: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
vlan drivers lack proper propagation of gso_max_segs from
lower device.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
1. scope correction for few functions that are used in single file.
2. Adjust variables that are used in fast-path to fit into single cacheline
3. Update rcv_frame() to skip shared check for frames coming over wire
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mode argument was erronusly defined as u32 but it has always
been u16. Also use ipvlan_set_mode() helper to set the mode instead
of assigning directly. This should avoid future erronus assignments /
updates.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Scrub skb before hitting the iptable hooks to ensure packets hit
these hooks. Set the xnet param only when the packet is crossing the
ns boundry so if the IPvlan slave and master belong to the same ns,
the param will be set to false.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
CC: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When we create IPvlan slave; we use ether_setup() and that
sets up default MTU to 1500 while the master device may have
lower / different MTU. Any subsequent changes to the masters'
MTU are reflected into the slaves' MTU setting. However if those
don't happen (most likely scenario), the slaves' MTU stays at
1500 which could be bad.
This change adds code to inherit MTU from the master device
instead of using the default value during the link initialization
phase.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
CC: Tim Hockins <thockins@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David 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: Tom Herbert <tom@herbertland.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ipvlan_handle_frame is a rx_handler, and when it returns a value other
than RX_HANDLER_CONSUMED (here, NET_RX_DROP aka RX_HANDLER_ANOTHER),
__netif_receive_skb_core expects that the skb still exists and will
process it further, but we just freed it.
Fixes: 2ad7bf3638 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pass a **skb to ipvlan_rcv_frame so that if skb_share_check returns a
new skb, we actually use it during further processing.
It's safe to ignore the new skb in the ipvlan_xmit_* functions, because
they call ipvlan_rcv_frame with local == true, so that dev_forward_skb
is called and always takes ownership of the skb.
Fixes: 2ad7bf3638 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the ipv4 outbound path of an ipvlan device in l3 mode, the ifindex is
being grabbed from dev_get_iflink. This works for the physical device
case, since as the documentation of that function notes: "Physical
interfaces have the same 'ifindex' and 'iflink' values.". However, if
the master device is a veth, and the pairs are in separate net
namespaces, the route lookup will fail with -ENODEV due to outer veth
pair being in a separate namespace from the ipvlan master/routing
namespace.
ns0 | ns1 | ns2
veth0a--|--veth0b--|--ipvl0
In ipvlan_process_v4_outbound(), a packet sent from ipvl0 in the above
configuration will pass fl.flowi4_oif == veth0a to
ip_route_output_flow(), but *net == ns1.
Notice also that ipv6 processing is not using iflink. Since there is a
discrepancy in usage, fixup both v4 and v6 case to use local dev
variable.
Tested this with l3 ipvlan on top of veth, as well as with single
physical interface in the top namespace.
Signed-off-by: Brenden Blanco <bblanco@plumgrid.com>
Reviewed-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Compute net once in ipvlan_process_v4_outbound and
ipvlan_process_v6_outbound and store it in a variable so that net does
not need to be recomputed next time it is used.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stop hidding the sk parameter with an inline helper function and make
all of the callers pass it, so that it is clear what the function is
doing.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is confusing and silly hiding a parameter so modify all of
the callers to pass in the appropriate socket or skb->sk if
no socket is known.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
All structures used in traffic forwarding are rcu-protected:
ipvl_addr, ipvl_dev and ipvl_port. Thus we can unhash addresses
without synchronization. We'll anyway hash it back into the same
bucket: in worst case lockless lookup will scan hash once again.
Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
Earlier tricks of setting broadcast bit only when IPv4 address is added
onto interface are not good enough especially when autoconf comes in play.
Setting them on always is performance drag but now that multicast /
broadcast is not processed in fast-path; enabling broadcast will let
autoconf work correctly without affecting performance characteristics of
the device.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Processing multicast / broadcast in fast path is performance draining
and having more links means more cloning and bringing performance
down further.
Broadcast; in particular, need to be given to all the virtual links.
Earlier tricks of enabling broadcast bit for IPv4 only interfaces are not
really working since it fails autoconf. Which means enabling broadcast
for all the links if protocol specific hacks do not have to be added into
the driver.
This patch defers all (incoming as well as outgoing) multicast traffic to
a work-queue leaving only the unicast traffic in the fast-path. Now if we
need to apply any additional tricks to further reduce the impact of this
(multicast / broadcast) type of traffic, it can be implemented while
processing this work without affecting the fast-path.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/usb/asix_common.c
drivers/net/usb/sr9800.c
drivers/net/usb/usbnet.c
include/linux/usb/usbnet.h
net/ipv4/tcp_ipv4.c
net/ipv6/tcp_ipv6.c
The TCP conflicts were overlapping changes. In 'net' we added a
READ_ONCE() to the socket cached RX route read, whilst in 'net-next'
Eric Dumazet touched the surrounding code dealing with how mini
sockets are handled.
With USB, it's a case of the same bug fix first going into net-next
and then I cherry picked it back into net.
Signed-off-by: David S. Miller <davem@davemloft.net>
The goal of this patch is to prepare the removal of the iflink field. It
introduces a new ndo function, which will be implemented by virtual interfaces.
There is no functional change into this patch. All readers of iflink field
now call dev_get_iflink().
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When an ipvlan interface is down, its addresses are not on the hash list.
Fix checks for existence of addresses not to depend on the hash list, walk
through all interface addresses instead.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adding and removing to the 'ipvlans' list is already done using _rcu list
operations.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When ipvlan interface with IP addresses attached is brought down and then
deleted, the assigned addresses are deleted twice from the address hash
list, first on the interface down and second on the link deletion.
Similarly, when an address is added while the interface is down, it is added
second time once the interface is brought up.
When the interface is down, the addresses should be kept off the hash list
for performance reasons. Ensure this is true, which also fixes the double add
problem. To fix the double free, check whether the address is hashed before
removing it.
Reported-by: Dan Williams <dcbw@redhat.com>
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that there are no more users kill dev_rebuild_header and all of it's
implementations.
This is long overdue.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Cosmetic patch to add __percpu qualifier to pcpu_stats
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
They are all either written once or extremly rarely (e.g. from init
code), so we can move them to the .data..read_mostly section.
Signed-off-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The ip6_route_output() always returns a valid dst pointer unlike in IPv4
case. So the validation has to be different from the IPv4 path. Correcting
that error in this patch.
This was picked up by a static checker with a following warning -
drivers/net/ipvlan/ipvlan_core.c:380 ipvlan_process_v6_outbound()
warn: 'dst' isn't an ERR_PTR
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move the port check [ipvlan_dev_master()] and device check
[ipvlan_dev_slave()] functions to netdevice.h and rename them
netif_is_ipvlan_port() and netif_is_ipvlan() resp. to be
consistent with macvlan api naming.
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If a device is already a macvlan port then refuse to use it as
an ipvlan port in the early stage of port creation.
thost1:~# ip link add link eth0 mvl0 type macvlan
thost1:~# echo $?
0
thost1:~# ip link add link eth0 ipvl0 type ipvlan
RTNETLINK answers: Device or resource busy
thost1:~# echo $?
2
thost1:~#
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The free_percpu() function tests whether its argument is NULL and then
returns immediately. Thus the test around the call is not needed.
This issue was detected by using the Coccinelle software.
Signed-off-by: Markus Elfring <elfring@users.sourceforge.net>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This driver uses ip_out_local() and ip6_route_output() which are
defined only if CONFIG_INET and CONFIG_IPV6 are enabled respectively.
Reported-by: Jim Davis <jim.epost@gmail.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix sparse warnings reported by kbuild robot
drivers/net/ipvlan/ipvlan_main.c:172:13: warning: symbol 'ipvlan_start_xmit' was not declared. Should it be static?
drivers/net/ipvlan/ipvlan_main.c:256:33: warning: incorrect type in initializer (different address spaces)
drivers/net/ipvlan/ipvlan_main.c:256:33: expected void const [noderef] <asn:3>*__vpp_verify
drivers/net/ipvlan/ipvlan_main.c:256:33: got struct ipvl_pcpu_stats *<noident>
drivers/net/ipvlan/ipvlan_main.c:544:5: warning: symbol 'ipvlan_link_register' was not declared. Should it be static
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
