OVS_ACTION_ATTR_TRUNC currently stores a delta from the original skb
length in OVS_CB(skb)->cutlen. When a later userspace action segments a
GSO skb, queue_gso_packets() reuses that delta for each smaller segment.
A segment can then reach queue_userspace_packet() with cutlen greater
than skb->len, underflowing the length passed to skb_zerocopy().
Store the maximum preserved length instead and bound each consumer
against the current skb length. Use U32_MAX as the no-truncation
sentinel so the value remains valid if skb geometry changes before a
consumer handles it.
Fixes: f2a4d086ed ("openvswitch: Add packet truncation support.")
Cc: stable@vger.kernel.org
Assisted-by: Codex:gpt-5.5
Signed-off-by: Kyle Zeng <kylebot@openai.com>
Reviewed-by: Ilya Maximets <i.maximets@ovn.org>
Reviewed-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/20260707221635.27489-1-kylebot@openai.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When a packet enters OVS datapath and there is no flow to handle it,
packet goes to userspace through a MISS upcall. With per-CPU upcall
dispatch mechanism, we're using the current CPU id to select the
Netlink PID on which to send this packet. This allows us to send
packets from the same traffic flow through the same handler.
The handler will process the packet, install required flow into the
kernel and re-inject the original packet via OVS_PACKET_CMD_EXECUTE.
While handling OVS_PACKET_CMD_EXECUTE, however, we may hit a
recirculation action that will pass the (likely modified) packet
through the flow lookup again. And if the flow is not found, the
packet will be sent to userspace again through another MISS upcall.
However, the handler thread in userspace is likely running on a
different CPU core, and the OVS_PACKET_CMD_EXECUTE request is handled
in the syscall context of that thread. So, when the time comes to
send the packet through another upcall, the per-CPU dispatch will
choose a different Netlink PID, and this packet will end up processed
by a different handler thread on a different CPU.
The process continues as long as there are new recirculations, each
time the packet goes to a different handler thread before it is sent
out of the OVS datapath to the destination port. In real setups the
number of recirculations can go up to 4 or 5, sometimes more.
There is always a chance to re-order packets while processing upcalls,
because userspace will first install the flow and then re-inject the
original packet. So, there is a race window when the flow is already
installed and the second packet can match it and be forwarded to the
destination before the first packet is re-injected. But the fact that
packets are going through multiple upcalls handled by different
userspace threads makes the reordering noticeably more likely, because
we not only have a race between the kernel and a userspace handler
(which is hard to avoid), but also between multiple userspace handlers.
For example, let's assume that 10 packets got enqueued through a MISS
upcall for handler-1, it will start processing them, will install the
flow into the kernel and start re-injecting packets back, from where
they will go through another MISS to handler-2. Handler-2 will install
the flow into the kernel and start re-injecting the packets, while
handler-1 continues to re-inject the last of the 10 packets, they will
hit the flow installed by handler-2 and be forwarded without going to
the handler-2, while handler-2 still re-injects the first of these 10
packets. Given multiple recirculations and misses, these 10 packets
may end up completely mixed up on the output from the datapath.
Let's allow userspace to specify on which Netlink PID the packets
should be upcalled while processing OVS_PACKET_CMD_EXECUTE.
This makes it possible to ensure that all the packets are processed
by the same handler thread in the userspace even with them being
upcalled multiple times in the process. Packets will remain in order
since they will be enqueued to the same socket and re-injected in the
same order. This doesn't eliminate re-ordering as stated above, since
we still have a race between kernel and the userspace thread, but it
allows to eliminate races between multiple userspace threads.
Userspace knows the PID of the socket on which the original upcall is
received, so there is no need to send it up from the kernel.
Solution requires storing the value somewhere for the duration of the
packet processing. There are two potential places for this: our skb
extension or the per-CPU storage. It's not clear which is better,
so just following currently used scheme of storing this kind of things
along the skb. We still have a decent amount of space in the cb.
Signed-off-by: Ilya Maximets <i.maximets@ovn.org>
Acked-by: Flavio Leitner <fbl@sysclose.org>
Acked-by: Eelco Chaudron <echaudro@redhat.com>
Acked-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/20250702155043.2331772-1-i.maximets@ovn.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
PERCPU_MODULE_RESERVE defines the maximum size that can by used for the
per-CPU data size used by modules. This is 8KiB.
Commit 035fcdc4d2 ("openvswitch: Merge three per-CPU structures into
one") restructured the per-CPU memory allocation for the module and
moved the separate alloc_percpu() invocations at module init time to a
static per-CPU variable which is allocated by the module loader.
The size of the per-CPU data section for openvswitch is 6488 bytes which
is ~80% of the available per-CPU memory. Together with a few other
modules it is easy to exhaust the available 8KiB of memory.
Allocate ovs_pcpu_storage dynamically at module init time.
Reported-by: Gal Pressman <gal@nvidia.com>
Closes: https://lore.kernel.org/all/c401e017-f8db-4f57-a1cd-89beb979a277@nvidia.com
Fixes: 035fcdc4d2 ("openvswitch: Merge three per-CPU structures into one")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/20250613123629.-XSoQTCu@linutronix.de
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
ovs_pcpu_storage is a per-CPU variable and relies on disabled BH for its
locking. Without per-CPU locking in local_bh_disable() on PREEMPT_RT
this data structure requires explicit locking.
The data structure can be referenced recursive and there is a recursion
counter to avoid too many recursions.
Add a local_lock_t to the data structure and use
local_lock_nested_bh() for locking. Add an owner of the struct which is
the current task and acquire the lock only if the structure is not owned
by the current task.
Cc: Aaron Conole <aconole@redhat.com>
Cc: Eelco Chaudron <echaudro@redhat.com>
Cc: Ilya Maximets <i.maximets@ovn.org>
Cc: dev@openvswitch.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reviewed-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/20250512092736.229935-9-bigeasy@linutronix.de
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Some field descriptions were missing, some were not very accurate.
Not touching the uAPI header or .c files for now.
Formatting of those comments isn't great in general, but at least
they are not missing anything now.
Before:
$ ./scripts/kernel-doc -none -Wall net/openvswitch/*.h 2>&1 | wc -l
16
After:
$ ./scripts/kernel-doc -none -Wall net/openvswitch/*.h 2>&1 | wc -l
0
Signed-off-by: Ilya Maximets <i.maximets@ovn.org>
Acked-by: Eelco Chaudron <echaudro@redhat.com>
Reviewed-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/20250320224431.252489-1-i.maximets@ovn.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently, ovs_ct_set_labels() is only called for confirmed conntrack
entries (ct) within ovs_ct_commit(). However, if the conntrack entry
does not have the labels_ext extension, attempting to allocate it in
ovs_ct_get_conn_labels() for a confirmed entry triggers a warning in
nf_ct_ext_add():
WARN_ON(nf_ct_is_confirmed(ct));
This happens when the conntrack entry is created externally before OVS
increments net->ct.labels_used. The issue has become more likely since
commit fcb1aa5163 ("openvswitch: switch to per-action label counting
in conntrack"), which changed to use per-action label counting and
increment net->ct.labels_used when a flow with ct action is added.
Since there’s no straightforward way to fully resolve this issue at the
moment, this reverts the commit to avoid breaking existing use cases.
Fixes: fcb1aa5163 ("openvswitch: switch to per-action label counting in conntrack")
Reported-by: Jianbo Liu <jianbol@nvidia.com>
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Aaron Conole <aconole@redhat.com>
Link: https://patch.msgid.link/1bdeb2f3a812bca016a225d3de714427b2cd4772.1741457143.git.lucien.xin@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Similar to commit 70f06c115b ("sched: act_ct: switch to per-action
label counting"), we should also switch to per-action label counting
in openvswitch conntrack, as Florian suggested.
The difference is that nf_connlabels_get() is called unconditionally
when creating an ct action in ovs_ct_copy_action(). As with these
flows:
table=0,ip,actions=ct(commit,table=1)
table=1,ip,actions=ct(commit,exec(set_field:0xac->ct_label),table=2)
it needs to make sure the label ext is created in the 1st flow before
the ct is committed in ovs_ct_commit(). Otherwise, the warning in
nf_ct_ext_add() when creating the label ext in the 2nd flow will
be triggered:
WARN_ON(nf_ct_is_confirmed(ct));
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Reviewed-by: Aaron Conole <aconole@redhat.com>
Acked-by: Florian Westphal <fw@strlen.de>
Link: https://patch.msgid.link/6b9347d5c1a0b364e88d900b29a616c3f8e5b1ca.1723483073.git.lucien.xin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently tc skb extension is used to send miss info from
tc to ovs datapath module, and driver to tc. For the tc to ovs
miss it is currently always allocated even if it will not
be used by ovs datapath (as it depends on a requested feature).
Export the static key which is used by openvswitch module to
guard this code path as well, so it will be skipped if ovs
datapath doesn't need it. Enable this code path once
ovs datapath needs it.
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The Open vSwitch kernel module uses the upcall mechanism to send
packets from kernel space to user space when it misses in the kernel
space flow table. The upcall sends packets via a Netlink socket.
Currently, a Netlink socket is created for every vport. In this way,
there is a 1:1 mapping between a vport and a Netlink socket.
When a packet is received by a vport, if it needs to be sent to
user space, it is sent via the corresponding Netlink socket.
This mechanism, with various iterations of the corresponding user
space code, has seen some limitations and issues:
* On systems with a large number of vports, there is a correspondingly
large number of Netlink sockets which can limit scaling.
(https://bugzilla.redhat.com/show_bug.cgi?id=1526306)
* Packet reordering on upcalls.
(https://bugzilla.redhat.com/show_bug.cgi?id=1844576)
* A thundering herd issue.
(https://bugzilla.redhat.com/show_bug.cgi?id=1834444)
This patch introduces an alternative, feature-negotiated, upcall
mode using a per-cpu dispatch rather than a per-vport dispatch.
In this mode, the Netlink socket to be used for the upcall is
selected based on the CPU of the thread that is executing the upcall.
In this way, it resolves the issues above as:
a) The number of Netlink sockets scales with the number of CPUs
rather than the number of vports.
b) Ordering per-flow is maintained as packets are distributed to
CPUs based on mechanisms such as RSS and flows are distributed
to a single user space thread.
c) Packets from a flow can only wake up one user space thread.
The corresponding user space code can be found at:
https://mail.openvswitch.org/pipermail/ovs-dev/2021-July/385139.html
Bugzilla: https://bugzilla.redhat.com/1844576
Signed-off-by: Mark Gray <mark.d.gray@redhat.com>
Acked-by: Flavio Leitner <fbl@sysclose.org>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a counter that counts the number of masks cache hits, and
export it through the megaflow netlink statistics.
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch reorders the masks array every 4 seconds based on their
usage count. This greatly reduces the masks per packet hit, and
hence the overall performance. Especially in the OVS/OVN case for
OpenShift.
Here are some results from the OVS/OVN OpenShift test, which use
8 pods, each pod having 512 uperf connections, each connection
sends a 64-byte request and gets a 1024-byte response (TCP).
All uperf clients are on 1 worker node while all uperf servers are
on the other worker node.
Kernel without this patch : 7.71 Gbps
Kernel with this patch applied: 14.52 Gbps
We also run some tests to verify the rebalance activity does not
lower the flow insertion rate, which does not.
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Tested-by: Andrew Theurer <atheurer@redhat.com>
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In kernel datapath of Open vSwitch, there are only 1024
buckets of meter in one datapath. If installing more than
1024 (e.g. 8192) meters, it may lead to the performance drop.
But in some case, for example, Open vSwitch used as edge
gateway, there should be 20K at least, where meters used for
IP address bandwidth limitation.
[Open vSwitch userspace datapath has this issue too.]
For more scalable meter, this patch use meter array instead of
hash tables, and expand/shrink the array when necessary. So we
can install more meters than before in the datapath.
Introducing the struct *dp_meter_instance, it's easy to
expand meter though changing the *ti point in the struct
*dp_meter_table.
Cc: Pravin B Shelar <pshelar@ovn.org>
Cc: Andy Zhou <azhou@ovn.org>
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When using the kernel datapath, the upcall don't
include skb hash info relatived. That will introduce
some problem, because the hash of skb is important
in kernel stack. For example, VXLAN module uses
it to select UDP src port. The tx queue selection
may also use the hash in stack.
Hash is computed in different ways. Hash is random
for a TCP socket, and hash may be computed in hardware,
or software stack. Recalculation hash is not easy.
Hash of TCP socket is computed:
tcp_v4_connect
-> sk_set_txhash (is random)
__tcp_transmit_skb
-> skb_set_hash_from_sk
There will be one upcall, without information of skb
hash, to ovs-vswitchd, for the first packet of a TCP
session. The rest packets will be processed in Open vSwitch
modules, hash kept. If this tcp session is forward to
VXLAN module, then the UDP src port of first tcp packet
is different from rest packets.
TCP packets may come from the host or dockers, to Open vSwitch.
To fix it, we store the hash info to upcall, and restore hash
when packets sent back.
+---------------+ +-------------------------+
| Docker/VMs | | ovs-vswitchd |
+----+----------+ +-+--------------------+--+
| ^ |
| | |
| | upcall v restore packet hash (not recalculate)
| +-+--------------------+--+
| tap netdev | | vxlan module
+---------------> +--> Open vSwitch ko +-->
or internal type | |
+-------------------------+
Reported-at: https://mail.openvswitch.org/pipermail/ovs-dev/2019-October/364062.html
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Offloaded OvS datapath rules are translated one to one to tc rules,
for example the following simplified OvS rule:
recirc_id(0),in_port(dev1),eth_type(0x0800),ct_state(-trk) actions:ct(),recirc(2)
Will be translated to the following tc rule:
$ tc filter add dev dev1 ingress \
prio 1 chain 0 proto ip \
flower tcp ct_state -trk \
action ct pipe \
action goto chain 2
Received packets will first travel though tc, and if they aren't stolen
by it, like in the above rule, they will continue to OvS datapath.
Since we already did some actions (action ct in this case) which might
modify the packets, and updated action stats, we would like to continue
the proccessing with the correct recirc_id in OvS (here recirc_id(2))
where we left off.
To support this, introduce a new skb extension for tc, which
will be used for translating tc chain to ovs recirc_id to
handle these miss cases. Last tc chain index will be set
by tc goto chain action and read by OvS datapath.
Signed-off-by: Paul Blakey <paulb@mellanox.com>
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Based on 1 normalized pattern(s):
this program is free software you can redistribute it and or modify
it under the terms of version 2 of the gnu general public license as
published by the free software foundation this program is
distributed in the hope that it will be useful but without any
warranty without even the implied warranty of merchantability or
fitness for a particular purpose see the gnu general public license
for more details you should have received a copy of the gnu general
public license along with this program if not write to the free
software foundation inc 51 franklin street fifth floor boston ma
02110 1301 usa
extracted by the scancode license scanner the SPDX license identifier
GPL-2.0-only
has been chosen to replace the boilerplate/reference in 21 file(s).
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Alexios Zavras <alexios.zavras@intel.com>
Reviewed-by: Allison Randal <allison@lohutok.net>
Reviewed-by: Richard Fontana <rfontana@redhat.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190529141334.228102212@linutronix.de
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Currently, nf_conntrack_max is used to limit the maximum number of
conntrack entries in the conntrack table for every network namespace.
For the VMs and containers that reside in the same namespace,
they share the same conntrack table, and the total # of conntrack entries
for all the VMs and containers are limited by nf_conntrack_max. In this
case, if one of the VM/container abuses the usage the conntrack entries,
it blocks the others from committing valid conntrack entries into the
conntrack table. Even if we can possibly put the VM in different network
namespace, the current nf_conntrack_max configuration is kind of rigid
that we cannot limit different VM/container to have different # conntrack
entries.
To address the aforementioned issue, this patch proposes to have a
fine-grained mechanism that could further limit the # of conntrack entries
per-zone. For example, we can designate different zone to different VM,
and set conntrack limit to each zone. By providing this isolation, a
mis-behaved VM only consumes the conntrack entries in its own zone, and
it will not influence other well-behaved VMs. Moreover, the users can
set various conntrack limit to different zone based on their preference.
The proposed implementation utilizes Netfilter's nf_conncount backend
to count the number of connections in a particular zone. If the number of
connection is above a configured limitation, ovs will return ENOMEM to the
userspace. If userspace does not configure the zone limit, the limit
defaults to zero that is no limitation, which is backward compatible to
the behavior without this patch.
The following high leve APIs are provided to the userspace:
- OVS_CT_LIMIT_CMD_SET:
* set default connection limit for all zones
* set the connection limit for a particular zone
- OVS_CT_LIMIT_CMD_DEL:
* remove the connection limit for a particular zone
- OVS_CT_LIMIT_CMD_GET:
* get the default connection limit for all zones
* get the connection limit for a particular zone
Signed-off-by: Yi-Hung Wei <yihung.wei@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
OVS kernel datapath so far does not support Openflow meter action.
This is the first stab at adding kernel datapath meter support.
This implementation supports only drop band type.
Signed-off-by: Andy Zhou <azhou@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Later patches will invoke get_dp() outside of datapath.c. Export it.
Signed-off-by: Andy Zhou <azhou@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows reliable identification of netdevice interfaces connected
to openvswitch bridges. In particular, user space queries the netdev
interfaces belonging to the ports for statistics, up/down state, etc.
Datapath dump needs to provide enough information for the user space to be
able to do that.
Currently, only interface names are returned. This is not sufficient, as
openvswitch allows its ports to be in different name spaces and the
interface name is valid only in its name space. What is needed and generally
used in other netlink APIs, is the pair ifindex+netnsid.
The solution is addition of the ifindex+netnsid pair (or only ifindex if in
the same name space) to vport get/dump operation.
On request side, ideally the ifindex+netnsid pair could be used to
get/set/del the corresponding vport. This is not implemented by this patch
and can be added later if needed.
Signed-off-by: Jiri Benc <jbenc@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For sw_flow_actions, the actions_len only represents the kernel part's
size, and when we dump the actions to the userspace, we will do the
convertions, so it's true size may become bigger than the actions_len.
But unfortunately, for OVS_PACKET_ATTR_ACTIONS, we use the actions_len
to alloc the skbuff, so the user_skb's size may become insufficient and
oops will happen like this:
skbuff: skb_over_panic: text:ffffffff8148fabf len:1749 put:157 head:
ffff881300f39000 data:ffff881300f39000 tail:0x6d5 end:0x6c0 dev:<NULL>
------------[ cut here ]------------
kernel BUG at net/core/skbuff.c:129!
[...]
Call Trace:
<IRQ>
[<ffffffff8148be82>] skb_put+0x43/0x44
[<ffffffff8148fabf>] skb_zerocopy+0x6c/0x1f4
[<ffffffffa0290d36>] queue_userspace_packet+0x3a3/0x448 [openvswitch]
[<ffffffffa0292023>] ovs_dp_upcall+0x30/0x5c [openvswitch]
[<ffffffffa028d435>] output_userspace+0x132/0x158 [openvswitch]
[<ffffffffa01e6890>] ? ip6_rcv_finish+0x74/0x77 [ipv6]
[<ffffffffa028e277>] do_execute_actions+0xcc1/0xdc8 [openvswitch]
[<ffffffffa028e3f2>] ovs_execute_actions+0x74/0x106 [openvswitch]
[<ffffffffa0292130>] ovs_dp_process_packet+0xe1/0xfd [openvswitch]
[<ffffffffa0292b77>] ? key_extract+0x63c/0x8d5 [openvswitch]
[<ffffffffa029848b>] ovs_vport_receive+0xa1/0xc3 [openvswitch]
[...]
Also we can find that the actions_len is much little than the orig_len:
crash> struct sw_flow_actions 0xffff8812f539d000
struct sw_flow_actions {
rcu = {
next = 0xffff8812f5398800,
func = 0xffffe3b00035db32
},
orig_len = 1384,
actions_len = 592,
actions = 0xffff8812f539d01c
}
So as a quick fix, use the orig_len instead of the actions_len to alloc
the user_skb.
Last, this oops happened on our system running a relative old kernel, but
the same risk still exists on the mainline, since we use the wrong
actions_len from the beginning.
Fixes: ccea74457b ("openvswitch: include datapath actions with sampled-packet upcall to userspace")
Cc: Neil McKee <neil.mckee@inmon.com>
Signed-off-by: Liping Zhang <zlpnobody@gmail.com>
Acked-by: Pravin B Shelar <pshelar@ovn.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
