The destination ethernet address is netconsole configuration: no other
netpoll user (bonding, team, vlan, bridge, macvlan, dsa) references
np->remote_mac, only netconsole's ethernet framing and its
configfs/cmdline interface do.
Move it into struct netconsole_target and convert push_eth() to take the
netconsole_target; netconsole_print_banner() and
netconsole_parser_cmdline() already take it. The configfs show/store
handlers and alloc_and_init() reach the field directly.
No functional change; the remote_mac sysfs attribute is unchanged.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-9-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The source and destination UDP ports live in struct netpoll but are
netconsole configuration. No other netpoll user (bonding, team, vlan,
bridge, macvlan, dsa) touches np->local_port or np->remote_port; they
only use the netpoll TX/forwarding path. Only netconsole's UDP framing
and its configfs/cmdline interface read these fields.
Move both into struct netconsole_target and convert the three helpers
that read them - push_udp(), netconsole_print_banner() and
netconsole_parser_cmdline() - to take the netconsole_target. The
configfs show/store handlers already have the target in hand.
No functional change; the local_port / remote_port sysfs attributes are
unchanged.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-8-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
These two fields back the fallback skb pool that find_skb() uses.
Every helper that touches them lives in netconsole now (refill_skbs,
refill_skbs_work_handler, netconsole_skb_pool_init,
netconsole_skb_pool_flush, find_skb, netcons_skb_pop), so the data
can move alongside its only consumer.
Add skb_pool and refill_wq to struct netconsole_target, drop them
from struct netpoll.
This will save 48-bytes for every netpoll user instance (except
netconsole that will have it in netconsole target struct).
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-7-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
skb_pool_flush() has no callers left in net/core/netpoll.c after
netconsole took over the pool lifecycle. Inline its body into
netconsole_skb_pool_flush() (the only caller) and drop the function
and its export from netpoll. The prototype goes from
<linux/netpoll.h>.
Pure code motion: cancel_work_sync() + skb_queue_purge_reason()
semantics are unchanged.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-6-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
refill_skbs() is now only called from netconsole (directly via
netconsole_skb_pool_init() and indirectly via the just-moved
refill_skbs_work_handler()), and the MAX_UDP_CHUNK / MAX_SKBS /
MAX_SKB_SIZE macros are private to it. Move them all into
drivers/net/netconsole.c.
MAX_UDP_CHUNK and MAX_SKB_SIZE were promoted to <linux/netpoll.h>
by commit 6c537b845c ("netconsole: do not dequeue pooled skbs that
cannot satisfy len") so find_skb() could detect oversized requests
against the same value refill_skbs() used. With both functions now
local to netconsole, the shared definition no longer needs to live
in the header.
Pure code motion: bodies and pool sizing semantics are unchanged.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-5-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The work handler is wired via INIT_WORK() in
netconsole_skb_pool_init() and has no other callers since the
previous patch took the skb pool lifecycle out of __netpoll_setup().
Move the function body into drivers/net/netconsole.c as a
file-static helper, drop EXPORT_SYMBOL_GPL() and remove the
prototype from <linux/netpoll.h>.
Pure code motion: the body is unchanged and still calls the
exported refill_skbs() in net/core/netpoll.c.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-4-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The fallback skb pool fronted by find_skb() is netconsole's only client:
every other netpoll goes through __netpoll_setup() / netpoll_send_skb()
without ever touching np->skb_pool.
Today __netpoll_setup() and __netpoll_cleanup() create and destroy the
pool for everyone, paying ~48 KB of pre-allocated skbs per netpoll
instance that only netconsole uses, what a waste!
Move the responsibility to netconsole. __netpoll_setup() did this under
the RTNL, but netconsole enables targets from enabled_store() /
alloc_param_target() without it, while the teardown path flushes the pool
(cancel_work_sync() + skb_queue_purge()) under the RTNL from
netconsole_process_cleanups_core(). Initialising the queue head and the
refill work on every enable would therefore race that flush.
They only need initialising once: after a flush the queue head is left
valid and empty and cancel_work_sync() leaves the work re-armable. Set
them up in alloc_and_init(), while the target is not yet reachable, and
let the enable paths only refill the pool via refill_skbs(), which
serialises with the flush through skb_pool.lock. See discussions in [1]
Link: https://lore.kernel.org/all/alDMvD5S7TZnoD_V@gmail.com/ [1]
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-3-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
These three helpers manage the per-netpoll fallback skb pool. They
are file-static today because all of their callers live in
net/core/netpoll.c. Subsequent patches relocate the pool's owner
from struct netpoll to the only consumer that actually uses it
(netconsole), and that work needs netconsole to drive the helpers
directly while the function bodies still live here.
Drop static, add prototypes in <linux/netpoll.h>, and
EXPORT_SYMBOL_GPL() each. No behaviour change.
The exports are transitional. Each helper is moved into
drivers/net/netconsole.c later in this series, and at that point
its EXPORT_SYMBOL_GPL() and prototype are dropped. By the end of
the series no symbol introduced here remains exported.
The goal of this patch is to make the subsequente patches easy to
review.
Signed-off-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20260710-netconsole_move_more-v3-2-6f63f76b28bc@debian.org
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Kuniyuki Iwashima says:
====================
net: Support per-netns device unregistration
The biggest blocker to per-netns RTNL is netdev unregistration.
It starts within a single netns, but it can eventually involve
multiple namespaces.
There are three types of such cross-netns devices:
1. Paired devices (e.g., netkit, veth, vxcan)
-> Unregistering one device also deletes its peer, which
may reside in another netns.
2. Tunnel devices (e.g., bareudp, geneve, etc)
-> Destroying a netns removes devices in another netns if
their backend sockets reside in the dying netns
3. Stacked devices (e.g., ipvlan, macvlan, etc)
-> Removing the lower device also removes multiple upper
devices, each of which may reside in different namespaces.
While the first two device types require at most two rtnl_net_lock()s,
the stacked type has no upper limit. This makes it impossible to
freeze all necessary namespaces in advance.
This series introduces per-netns work, initially suggested at
NetConf 2024, to delegate the unregistration of such cross-netns
devices.
https://netdev.bots.linux.dev/netconf/2024/kuniyu.pdf#page=62
The first half of the series wraps NETDEV_UNREGISTER (in core) with
per-netns RTNL, adds a helper for per-netns device unregistration,
and forces per-netns device unregistration in the core code when
CONFIG_DEBUG_NET_SMALL_RTNL=y.
The latter half picks out one from each type (veth, bareudp, ipvlan)
and converts them to support per-netns device unregistration,
although the operations are **still serialised under RTNL** for now.
Please note that this series focuses only on the device unregistration
paths. For example, there are ASSERT_RTNL() left in other paths, and
Sashiko may point it out, but they are out of scope.
This is just the first step, and we need more incremental changes to
completely remove RTNL anyway.
Now, we can see that unregistering a lower device (veth0 below)
removes upper devices (ipvl2, ipvl3) in different namespaces using
per-netns work with a different PID. The lower device (veth0) is
freed only after all upper ipvlan devices have called netdev_put()
in ipvlan_uninit().
# ip netns add ns1
# ip netns add ns2
# ip netns add ns3
# ip -n ns1 link add veth0 type veth peer veth1
# ip -n ns2 link add ipvl2 link veth0 link-netns ns1 type ipvlan mode l2
# ip -n ns3 link add ipvl3 link veth0 link-netns ns1 type ipvlan mode l2
# ip -n ns1 link del veth0
# bpftrace -e '#include <linux/netdevice.h>
kprobe:ipvlan_uninit,
kprobe:veth_dellink,
kprobe:free_netdev {
$dev = (struct net_device *)arg0;
printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
}'
PID: 2010 | DEV: veth0
veth_dellink+5
rtnl_dellink+1213
rtnetlink_rcv_msg+1791
...
PID: 440 | DEV: ipvl2
ipvlan_uninit+5
unregister_netdevice_many_notify+7129
unregister_netdevice_many_net+1050
rtnl_net_work_func+136
...
PID: 440 | DEV: ipvl2
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
PID: 440 | DEV: ipvl3
ipvlan_uninit+5
unregister_netdevice_many_notify+7129
unregister_netdevice_many_net+1050
rtnl_net_work_func+136
process_scheduled_works+2538
...
PID: 2010 | DEV: veth0
free_netdev+5
netdev_run_todo+4798
rtnl_dellink+1507
rtnetlink_rcv_msg+1791
...
PID: 440 | DEV: ipvl3
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
v1: https://lore.kernel.org/netdev/20260701214334.266991-1-kuniyu@google.com/
====================
Link: https://patch.msgid.link/20260703001009.1572444-1-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When a lower device is unregistered, its upper ipvlan devices
must also be unregistered. However, these upper devices may
reside in different netns than the lower device.
Let's use unregister_netdevice_queue_net() to support per-netns
device unregistration for ipvlan.
The new dying flag in struct ipvl_dev is used to avoid a race
that ipvlan_link_delete() is called while its lower device is
being removed in ipvlan_device_event().
If dying is true in ipvlan_link_delete(), the ipvlan device is
already destructed but not yet unregistered. In this case,
unregistration will be done in __rtnl_net_unlock() of the
->dellink() caller.
Tested:
1. Create veth in ns1 and two ipvlan devices in ns2 and ns3.
# ip netns add ns1
# ip netns add ns2
# ip netns add ns3
# ip -n ns1 link add veth0 type veth peer veth1
# ip -n ns2 link add ipvl2 link veth0 link-netns ns1 type ipvlan mode l2
# ip -n ns3 link add ipvl3 link veth0 link-netns ns1 type ipvlan mode l2
2. Run bpftrace to check that veth is unregistered first but
wait ipvlan to be unregistered
# bpftrace -e '#include <linux/netdevice.h>
kprobe:ipvlan_uninit,
kprobe:veth_dellink,
kprobe:free_netdev {
$dev = (struct net_device *)arg0;
printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
}'
3. Remove the lower veth0 in ns1.
# ip -n ns1 link del veth0
We can see that veth0 is freed after unregistering ipvl2 and ipvl3
in per-netns work because ipvl_port holds refcount of veth0.
PID: 2010 | DEV: veth0
veth_dellink+5
rtnl_dellink+1213
rtnetlink_rcv_msg+1791
...
PID: 440 | DEV: ipvl2
ipvlan_uninit+5
unregister_netdevice_many_notify+7129
unregister_netdevice_many_net+1050
rtnl_net_work_func+136
process_scheduled_works+2538
...
PID: 440 | DEV: ipvl2
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
PID: 440 | DEV: ipvl3
ipvlan_uninit+5
unregister_netdevice_many_notify+7129
unregister_netdevice_many_net+1050
rtnl_net_work_func+136
process_scheduled_works+2538
...
PID: 2010 | DEV: veth0
free_netdev+5
netdev_run_todo+4798
rtnl_dellink+1507
rtnetlink_rcv_msg+1791
...
PID: 440 | DEV: ipvl3
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-15-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
struct ipvl_port is shared between a lower device and its upper
ipvlan devices.
All upper devices are linked to ipvl_port.ipvlans.
Once RTNL is removed, the list can be modified concurrently from
different netns due to device removal.
Let's protect it with a per-port mutex.
NETDEV_PRECHANGEUPPER and NETDEV_CHANGEUPPER are explicitly
skipped to avoid deadlock for netdev_upper_dev_unlink() called
from NETDEV_UNREGISTER.
Note that __ipvtap_dellink_ptr is added for CONFIG_IPVLAN=y
but CONFIG_TAP=m and CONFIG_IPVTAP=m.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-14-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
ipvlan_uninit() for the last ipvlan device resets the lower device's
rx_handler_data to NULL.
Once RTNL is removed, ipvlan_init() would race with ipvlan_uninit(),
which could leak a newly allocated ipvl_port.
ipvlan_init() ipvlan_uninit()
| |- if (refcount_dec_and_test(old_port))
... |- ipvlan_port_destroy(old_port)
| '
|- refcount_inc_not_zero(old_port) <-- fails
|- ipvlan_port_create(phy_dev) .
|- new_port = kzalloc() |
|- phy_dev->rx_handler_data = new_port
|- phy_dev->rx_handler_data = NULL
...
`- kfree(old_port);
Let's synchronise the two by holding the lower device's netdev_lock().
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-13-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
struct ipvl_port is shared between a lower device and its upper
ipvlan devices.
While each upper device can always access ipvl_port safely via
ipvlan_dev.port, the lower device relies on RTNL to access it
via net_device.rx_handler_data.
Once RTNL is removed, the lower device cannot read ipvl_port safely
in ipvlan_device_event() because the port could be freed concurrently
and net_device.rx_handler_data is set to NULL if the last ipvlan
device in another namespace is unregistered.
Let's convert ipvl_port.count to refcount_t and use RCU along with
refcount_inc_not_zero() in ipvlan_device_event().
netdev_put() in ipvlan_port_destroy() is also moved down after
cancel_work_sync(), which is the last user of port->dev.
Note that ipvlan->port is now set in ipvlan_init() so that it can
be used in ipvlan_uninit(), instead of ipvlan_port_get_rtnl()
(rtnl_dereference()).
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-12-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
bareudp_exit_rtnl_net() iterates bareudp devices whose sockets
are in the dying netns and queues them for destruction.
So the devices may reside in different netns.
Let's use unregister_netdevice_queue_net() to support per-netns
device unregistration.
list_del() is changed to list_del_init() to avoid queueing the
same device twice.
Even after bareudp_exit_rtnl_net() queues a cross-netns bareudp
device, bareudp_dellink() could be called concurrently for it
(once RTNL is removed). In such a case, __rtnl_net_unlock() will
perform the unregistration.
Note that bareudp uses register_pernet_subsys() instead of _device(),
so default_device_exit_batch() guarantees that the async per-netns
works are flushed before ->exit().
Tested:
1. Create bareudp device across two netns.
# ip netns add ns1
# ip netns add ns2
# ip -n ns1 link add bareudp0 link-netns ns2 type bareudp \
dstport 9292 ethertype ipv4
2. Run bpftrace to check that bareudp_uninit() is called between
->exit_rtnl() and ->exit().
# bpftrace -e '#include <linux/netdevice.h>
kprobe:bareudp_uninit {
$dev = (struct net_device *)arg0;
printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
}
kprobe:bareudp_exit_rtnl_net,
kprobe:bareudp_exit_net {
printf("PID: %d%s\n", pid, kstack());
}'
3. Remove the netns where the bareudp socket resides
# ip netns del ns2
Now, we can see bareudp0 is unregistered by per-netns work
instead of cleanup_net() and it finishes before ->exit() to
avoid WARN_ON_ONCE(!list_empty(&bn->bareudp_list)) there.
PID: 576
bareudp_exit_rtnl_net+5
ops_undo_list+702
cleanup_net+1122
process_scheduled_works+2538
...
PID: 470 | DEV: bareudp0
bareudp_uninit+5
unregister_netdevice_many_notify+7129
unregister_netdevice_many_net+1050
rtnl_net_work_func+136
process_scheduled_works+2538
...
PID: 576
bareudp_exit_net+5
ops_undo_list+1064
cleanup_net+1122
process_scheduled_works+2538
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-11-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
struct bareudp_dev.net is the netns where the backend bareudp
socket resides.
struct bareudp_dev is linked to the bareudp_net.bareudp_list of
the socket's netns.
During netns dismantle or module unload, bareudp_exit_rtnl_net()
iterates the list and queues devices for destruction regardless
of the devices' netns.
Thus, once RTNL is removed, the list can be modified concurrently
from different netns due to device removal.
Let's protect it with per-netns mutex.
bareudp_newlink() is still protected by rtnl_net_lock()s, so
acquiring bn->lock twice in bareudp_find_dev() and
bareudp_configure() is not a problem.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-10-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Currently, veth_dellink() unregisters both local and peer devices
synchronously under RTNL.
Once RTNL is removed, it can be called concurrently from different
netns.
Let's use xchg() and unregister_netdevice_queue_net() to support
per-netns device unregistration.
This way, each device is queued for destruction only once by
the winner of the race.
Note that the extra netdev_hold() ensures that @peer obtained by
the first xchg() is not freed during the subsequent access to
netdev_priv(peer). The 2nd xchg() overwrites @dev to balance
the refcount.
Tested:
1. Create two veth pairs (veth1-2, veth3-4) between two netns
(ns1 & ns2).
# ip netns add ns1
# ip netns add ns2
# ip -n ns1 link add veth1 type veth peer veth2 netns ns2
# ip -n ns1 link add veth3 type veth peer veth4 netns ns2
2. Run bpftrace to check if the same process does NOT
unregister the paired veth devices
# bpftrace -e '#include <linux/netdevice.h>
kprobe:free_netdev {
$dev = (struct net_device *)arg0;
printf("PID: %d | DEV: %s%s\n", pid, $dev->name, kstack());
}'
3. Remove veth2 in ns2 and check bpftrace output
# ip -n ns2 link del veth2
PID: 2194 | DEV: veth2
free_netdev+5
netdev_run_todo+4798
rtnl_dellink+1507
rtnetlink_rcv_msg+1791
...
PID: 448 | DEV: veth1
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
4. Remove ns2 (thus veth4) and check bpftrace output
# ip netns del ns2
PID: 571 | DEV: veth4
free_netdev+5
netdev_run_todo+4798
default_device_exit_batch+2271
ops_undo_list+993
cleanup_net+1122
process_scheduled_works+2538
...
PID: 441 | DEV: veth3
free_netdev+5
netdev_run_todo+4798
process_scheduled_works+2538
...
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-9-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
For per-netns device unregistration, the list passed to
unregister_netdevice_many() must contain devices from a single
netns only (once all callers are converted).
Let's move collected devices in the following functions to
net->dev_unreg_head and let __rtnl_net_unlock() pass them to
unregister_netdevice_many().
* default_device_exit_batch()
* ops_exit_rtnl_list()
* __rtnl_kill_links()
This allows incremental conversion of each driver to support
per-netns device unregistration without affecting the normal
kernel where CONFIG_DEBUG_NET_SMALL_RTNL is disabled.
Note that this change unbatches synchronize_rcu() etc in
unregister_netdevice_many(), but we can later split it into
multiple stages to batch them again.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-8-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When we need to unregister a netdev in a different netns, we will
delegate its unregistration to per-netns work.
There are three types of such cross-netns devices:
1. Paired devices (e.g., netkit, veth, vxcan)
-> Unregistering one device also deletes its peer, which
may reside in another netns.
2. Tunnel devices (e.g., bareudp, geneve, etc)
-> Destroying a netns removes devices in another netns if
their backend sockets reside in the dying netns
3. Stacked devices (e.g., ipvlan, macvlan, etc)
-> Removing the lower device also removes multiple upper
devices, each of which may reside in different namespaces.
In these cases, we will use unregister_netdevice_queue_net() to
queue such potential cross-netns devices for destruction.
Each driver must not call both unregister_netdevice_queue_net()
and unregister_netdevice_queue() for the same device. See the
subsequent veth/bareudp/ipvlan patches for how they avoid double
queueing.
unregister_netdevice_queue_net() takes net and dev. If dev resides
in the net, it simply calls unregister_netdevice_queue().
If dev_net(dev) is different from the net, it enqueues the device
to dev_net(dev)->dev_unreg_head and schedules the per-netns work.
When __rtnl_net_unlock() is called from the per-netns work (or another
thread already holding the lock), unregister_netdevice_many_net()
collects the queued devices and calls unregister_netdevice_many()
to perform the actual unregistration.
During netns dismantle, rtnl_net_flush_workqueue() is called at the
end of default_device_exit_batch() to ensure that cross-netns
devices in the other alive netns are unregistered.
Once RTNL is removed, a device could be moved to another netns while
being queued to net->dev_unreg_head.
__dev_change_net_namespace() handles this race by acquiring
net->dev_unreg_lock of both the old and new netns after dev_set_net()
and moving the device between their dev_unreg_head lists.
Since dev_set_net() and unregister_netdevice_queue_net() are
synchronised by netdev_lock(), the device is either queued to the
old netns's dev_unreg_head and then moved, or queued directly to
the new netns.
Note that unregister_netdevice_move_net() does not need to call
rtnl_net_queue_work() because __dev_change_net_namespace() is
(supposed to be) called with rtnl_net_lock(). (Not all callers
hold it yet, but the race does not happen until all callers
are converted and RTNL is removed.)
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-7-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Currently, netdev_run_todo() processes pending devices from multiple
namespaces in a batch.
To expand the per-netns RTNL coverage for NETDEV_UNREGISTER, let's
acquire __rtnl_net_lock() in netdev_wait_allrefs_any().
Note that netdev_run_todo() itself will need to be namespacified
before RTNL is removed.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-6-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The biggest blocker to per-netns RTNL is netdev unregistration.
It starts within a single netns (e.g., during a device lookup or
netns dismantle), but it can eventually involve multiple namespaces,
such as when upper ipvlan devices reside in different netns.
This prevents us from acquiring multiple rtnl_net_lock()s beforehand.
When we encounter such a cross-netns device, we must delegate the
unregistration to the work of the netns where the device actually
resides.
Let's add per-netns rtnl_work to support the deferred netdev
unregistration.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-4-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When rtnl_link_unregister() is called during module unload, it
calls __rtnl_kill_links() for every netns.
__rtnl_kill_links() collects all devices of the unloaded module
and passes them to unregister_netdevice_many().
Let's move unregister_netdevice_many() to rtnl_link_unregister()
to unregister all devices across netns in a single batch.
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-3-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
There are a few cases where rtnl_net_lock() is not properly
held in rtnl_newlink().
When either of IFLA_NET_NS_PID / IFLA_NET_NS_FD / IFLA_TARGET_NETNSID
is specified but IFLA_LINK_NETNSID is not, sock_net(skb->sk) is used
as link_net in rtnl_newlink_link_net().
In addition, the do_setlink() path uses sock_net(skb->sk) and one
from the three netns attributes while rtnl_link_get_net_capable()
returns only one of four.
Let's add sock_net(skb->sk) to rtnl_nets in rtnl_newlink().
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
Link: https://patch.msgid.link/20260703001009.1572444-2-kuniyu@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Update tun_fill_info() to read device configuration fields (flags, owner,
group, numqueues, numdisabled) locklessly using READ_ONCE().
Annotate all writes to these fields in the control paths with WRITE_ONCE()
to prevent data races, as these fields can be modified concurrently via
ioctls (TUNSETPERSIST, TUNSETOWNER, TUNSETGROUP, TUNSETIFF) or queue
attaching/detaching.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Link: https://patch.msgid.link/20260706163517.2415530-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Pull networking fixes from Paolo Abeni:
"Including fixes from netfilter, Bluetooth and batman-adv.
Current release - regressions:
- bluetooth: fix using chan->conn as indication to no remote netdev
Current release - new code bugs:
- netfilter: cap to maximum number of expectation per master on
updates
Previous releases - regressions:
- bluetooth:
- fix UAF of hci_conn_params in add_device_complete
- fix null ptr deref in hci_abort_conn()
- igmp: remove multicast group from hash table on device destruction
- batman-adv: prevent TVLV OOB check overflow
- eth: mlx5/mlx5e:
- fix off-by-one in single-FDB error rollback
- skip peer flow cleanup when LAG seq is unavailable
- fix crashes in dynamic per-channel stats and HV VHCA agent
- eth: mana: Sync page pool RX frags for CPU
Previous releases - always broken:
- netfilter:
- mark malformed IPv6 extension headers for hotdrop
- terminate table name before find_table_lock()
- ipvs: use parsed transport offset in TCP state lookup
- sched: act_pedit: fix TOCTOU heap OOB write in tc offload
- ethtool: rss: fix hfunc and input_xfrm parsing on big endian
- ipv4/ipv6: fix UAF and memory leak in IGMP/MLD
- tls: consume empty data records in tls_sw_read_sock()
- eth:
- octeontx2-af: fix VF bringup affecting PF promiscuous state
- gue: validate REMCSUM private option length"
* tag 'net-7.2-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (94 commits)
macsec: don't read an unset MAC header in macsec_encrypt()
dibs: loopback: validate offset and size in move_data()
octeontx2-af: fix VF bringup affecting PF promiscuous state
ethtool: rss: Fix hfunc and input_xfrm parsing on big endian
net/mlx5: Fix L3 tunnel entropy refcount leak
net: macb: drop in-flight Tx SKBs on close
net: mana: Sync page pool RX frags for CPU
net: mana: Validate the packet length reported by the NIC
selftests/net: fix EVP_MD_CTX leak in tcp_mmap
ipvs: ensure inner headers in ICMP errors are in headroom
ipvs: use parsed transport offset in SCTP state lookup
ipvs: use parsed transport offset in TCP state lookup
ipvs: pass parsed transport offset to state handlers
netfilter: handle unreadable frags
netfilter: flowtable: support IPIP tunnel with direct xmit
netfilter: flowtable: IPIP tunnel hardware offload is not yet support
netfilter: flowtable: use dst in this direction when pushing IPIP header
netfilter: ipset: allocate the proper memory for the generic hash structure
netfilter: ipset: cleanup the add/del backlog when resize failed
netfilter: ipset: exclude gc when resize is in progress
...
macsec_encrypt() reads the Ethernet header via eth_hdr(skb)
(skb->head + skb->mac_header) to memmove() the 12 source/destination MAC
bytes forward and make room for the SecTAG.
On the AF_PACKET SOCK_RAW + PACKET_QDISC_BYPASS transmit path the skb
reaches the macsec ndo_start_xmit() with the MAC header unset, so
eth_hdr(skb) resolves to skb->head + (u16)~0 and the read is out of
bounds: a 12-byte heap over-read that is also emitted on the wire as the
frame's outer source/destination MAC. KASAN reports a slab-out-of-bounds
read in macsec_start_xmit() on 6.0; on current mainline a CONFIG_DEBUG_NET
build flags it as an unset mac header in skb_mac_header().
On the TX path the L2 header is at skb->data, so use skb_eth_hdr(), added
by commit 96cc4b6958 ("macvlan: do not assume mac_header is set in
macvlan_broadcast()") for exactly this purpose.
Fixes: c09440f7dc ("macsec: introduce IEEE 802.1AE driver")
Cc: stable@vger.kernel.org
Signed-off-by: Daehyeon Ko <4ncienth@gmail.com>
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://patch.msgid.link/20260703083634.2035145-1-4ncienth@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The loopback move_data() performs a memcpy into the registered DMB
without checking whether offset + size exceeds the DMB length. Unlike
real ISM hardware, which enforces memory region bounds natively, the
software loopback has no such protection.
A peer-supplied out-of-bounds offset or oversized write would result in
an OOB write past the allocated kernel buffer. Add an explicit bounds
check before the memcpy to reject such requests with -EINVAL.
Fixes: f7a22071db ("net/smc: implement DMB-related operations of loopback-ism")
Cc: stable@vger.kernel.org
Reported-by: Federico Kirschbaum <federico.kirschbaum@xbow.com>
Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
Reported-by: Baul Lee <baul.lee@xbow.com>
Link: https://patch.msgid.link/20260707074318.1448662-1-dust.li@linux.alibaba.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Similar to commit 690043b95c ("selftests: drv-net: rss: Add
retries to test_rss_key_indir to reduce flakes"), implement the
retry mechanism for test_rss_context_overlap. This gives the test
more attempts to distribute the flow evenly, as the chance of
flow skewing to one queue is high.
Example failures:
# Check failed 5288 < 7000 traffic on main context (1/2): [2727, 2561, 8961, 6648]
not ok 1 rss_ctx.test_rss_context_overlap
# Check failed 6710 < 7000 traffic on main context (2/2): [9280, 5217, 5358, 1352]
not ok 1 rss_ctx.test_rss_context_overlap
Ran test_rss_context_overlap and test_rss_context_overlap2
over 1,000 consecutive runs with no failures.
Signed-off-by: Zinc Lim <limzhineng2@gmail.com>
Link: https://patch.msgid.link/20260702164932.2832916-1-limzhineng2@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Mbox handling of nix_set_rx_mode for a VF with promiscuous and
all_multi flags set to false causes deletion of the PF's promiscuous
and allmulti MCAM rules. This occurs because the APIs that
enable/disable these rules operate only on the PF, even when the
mbox request is made via a VF interface.
Guard both rvu_npc_enable_allmulti_entry() and
rvu_npc_enable_promisc_entry() disable paths with an is_vf() check so
that a VF bringing up or tearing down its interface cannot inadvertently
clear the PF's MCAM rules.
Fixes: 967db3529e ("octeontx2-af: add support for multicast/promisc packet replication feature")
Signed-off-by: Harman Kalra <hkalra@marvell.com>
Signed-off-by: Nitin Shetty J <nshettyj@marvell.com>
Link: https://patch.msgid.link/20260702045616.3002773-2-nshettyj@marvell.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Florian Westphal says:
====================
netfilter: updates for net
The following patchset contains Netfilter fixes for *net*.
Most of these are LLM fixes for old issues flagged by sashiko/LLMs.
Many of these trigger drive-by-findings in sashiko. In particular:
- many load/store tearing and missing memory barriers, races
etc. in ipset, esp. with GC and resizing.
Keeping the proposed patches spinning for yet-another-iteration
keeps legit fixes back, so I prefer to add these now and follow
up with other reports later.
- flowtable work queue still has possible races with teardown,
but same rationale as with ipset: drive-by findings, not
problems coming with the flowtable IPIP changeset in this PR.
- ever since unreadable frag skb support was added in 6.12, we can no
longer do: BUG_ON(skb_copy_bits( ...): it will fire with such skbs.
Mina Almasry is looking at similar patterns elsewhere in the stack.
1) Guard skb->mac_header adjustment after IPv6 defragmentation in
nf_conntrack_reasm. From Xiang Mei.
2) NUL-terminate ebtables table names before calling find_table_lock() to
prevent stack-out-of-bounds reads. Also from Xiang Mei.
3) Zero the ebtables chainstack array, else error unwind may free bogus
pointer when CPU mask is sparse. All three issues date from 2.6 days.
4) Ensure ebtables module names are c-strings, same bug pattern as 2).
Bug added in 4.6.
5) Fix catchall element handling for inverted lookups in nft_lookup. Fold the
catchall lookup into ext before computing the match status. Was like
this ever since catchall elements got introduced in 5.13.
From Tamaki Yanagawa.
6-9) ipset updates from Jozsef Kadlecsik:
- mark rcu protected areas correctly
- address gc and resize clash in the comment extension
- add/del backlog cleanup in the error path
- allocate right size for the generic hash structure
10-12): IPIP flowtable updates from Pablo Neira Ayuso:
- Use the current direction's route when pushing IPIP headers
Fix incorrect headroom and fragmentation offset calculations.
- Avoid hardware offload for IPIP tunnels due to lack of driver support.
- Support IPIP tunnels with direct xmit in netfilter flowtable.
dst_cache and dst_cookie are moved outside the union to share route
state across flows. This is a followup to work done in 6.19 cycle.
13) Don't BUG() on skb_copy_bits error. Handle unreadable fragments by
either returning an error or restricting the copy operations to linear area,
This became an issue when unreable frag support was merged in 6.12.
14-16): IPVS updates from Yizhou Zhao:
- Pass parsed transport offset to IPVS state handlers.
update callback signatures.
- use correct transport header offset on state lookp in TCP.
As-is it was possible for ipv6 extension header data to be
treated as L4 header.
- same for SCTP. This was also broken since 2.6 days.
17) Ensure inner IP headers in ICMP errors are in the skb headroom after
stripping outer headers. Add more checks for the length of inner headers.
This was broken since 3.7 days.
From Julian Anastasov.
netfilter pull request nf-26-07-08
* tag 'nf-26-07-08' of https://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
ipvs: ensure inner headers in ICMP errors are in headroom
ipvs: use parsed transport offset in SCTP state lookup
ipvs: use parsed transport offset in TCP state lookup
ipvs: pass parsed transport offset to state handlers
netfilter: handle unreadable frags
netfilter: flowtable: support IPIP tunnel with direct xmit
netfilter: flowtable: IPIP tunnel hardware offload is not yet support
netfilter: flowtable: use dst in this direction when pushing IPIP header
netfilter: ipset: allocate the proper memory for the generic hash structure
netfilter: ipset: cleanup the add/del backlog when resize failed
netfilter: ipset: exclude gc when resize is in progress
netfilter: ipset: mark the rcu locked areas properly
netfilter: nft_lookup: fix catchall element handling with inverted lookups
netfilter: ebtables: module names must be null-terminated
netfilter: ebtables: zero chainstack array
netfilter: ebtables: terminate table name before find_table_lock()
netfilter: nf_conntrack_reasm: guard mac_header adjustment after IPv6 defrag
====================
Link: https://patch.msgid.link/20260708140309.19633-1-fw@strlen.de
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
ETHTOOL_A_RSS_HFUNC and ETHTOOL_A_RSS_INPUT_XFRM are NLA_U32 attributes,
but ethnl_rss_set() and ethnl_rss_create_doit() parse them with
ethnl_update_u8(), which reads a single byte.
On little endian this happens to read the least significant byte and
works as long as the value fits in a byte. On big endian it reads the
most significant byte, so the requested value is parsed incorrectly.
The destination fields in struct ethtool_rxfh_param are u8, so the
attribute can't be read directly with ethnl_update_u32().
Cap the hfunc policy at U8_MAX so an out of range value is rejected
instead of being silently truncated into the u8 field, and add
ethnl_update_u8_u32() to read the full u32 and narrow it into the u8
destination.
Fixes: 82ae67cbc4 ("ethtool: rss: support setting hfunc via Netlink")
Fixes: d3e2c7bab1 ("ethtool: rss: support setting input-xfrm via Netlink")
Fixes: a166ab7816 ("ethtool: rss: support creating contexts via Netlink")
Reviewed-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Nimrod Oren <noren@nvidia.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20260706055017.3355806-1-gal@nvidia.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
mlx5_tun_entropy_refcount_inc() counts both VXLAN and L2-to-L3
tunnel reformat entries as entropy-enabling users. The matching
decrement path only handled VXLAN, leaving L2-to-L3 tunnel entries
counted after release.
Handle MLX5_REFORMAT_TYPE_L2_TO_L3_TUNNEL in
mlx5_tun_entropy_refcount_dec() as well so the enabling entry
refcount remains balanced.
Fixes: f828ca6a2f ("net/mlx5e: Add support for hw encapsulation of MPLS over UDP")
Signed-off-by: Li RongQing <lirongqing@baidu.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/20260703141423.1723-1-lirongqing@baidu.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
<linux/mdio.h> itself doesn't use any of the device id structures. The
files #including <linux/mdio.h> use a variety of them:
$ git grep -l '<linux/mdio.h>' | xargs grep --color -E '\<(acpi_device_id|amba_id|ap_device_id|apr_device_id|auxiliary_device_id|bcma_device_id|ccw_device_id|cdx_device_id|coreboot_device_id|css_device_id|dfl_device_id|dmi_(device|system)_id|eisa_device_id|fsl_mc_device_id|hda_device_id|hid_device_id|hv_vmbus_device_id|i2c_device_id|i3c_device_id|ieee1394_device_id|input_device_id|ipack_device_id|isapnp_device_id|ishtp_device_id|mcb_device_id|mdio_device_id|mei_cl_device_id|mhi_device_id|mips_cdmm_device_id|of_device_id|parisc_device_id|pci_device_id|pci_epf_device_id|pcmcia_device_id|platform_device_id|pnp_(card_)?device_id|rio_device_id|rpmsg_device_id|sdio_device_id|sdw_device_id|serio_device_id|slim_device_id|spi_device_id|spmi_device_id|ssam_device_id|ssb_device_id|tb_service_id|tee_client_device_id|typec_device_id|ulpi_device_id|usb_device_id|vchiq_device_id|virtio_device_id|wmi_device_id|x86_(cpu|device)_id|zorro_device_id|cpu_feature)\>'
...
but none of them relies on <linux/mdio.h>'s #include.
<linux/mod_devicetable.h> is a bad header mixing many different device
id structures and thus is an unfortunate dependency because if e.g.
struct wmi_device_id is modified all users of <linux/mdio.h> need to be
recompiled despite none of them using struct wmi_device_id.
So drop the unused header.
Signed-off-by: Uwe Kleine-König (The Capable Hub) <u.kleine-koenig@baylibre.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/ca270a534d0f230a939a3fb4a661808b35d6436d.1783329817.git.u.kleine-koenig@baylibre.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add Static and Dynamic Interrupt Moderation (DIM) support for
Rx and Tx.
Update queue creation procedure with new data struct with the related
settings.
Add functions to collect stat for DIM, and workers to update DIM data
and settings.
Update ethtool handler to get/set the moderation settings from a user.
To avoid detach/re-attach ops, ring DIM doorbell to change settings
at run time.
By default, adaptive-rx/tx (DIM) are enabled if supported by HW.
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20260702220123.815018-1-haiyangz@linux.microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The MACB driver has since forever leaked the outgoing SKBs that
have not yet been marked as completed. They live in queue->tx_skb
which gets freed without remorse nor checking.
macb_free_consistent() gets called in a few codepaths, but only close will
trigger the added expressions. In macb_open() and macb_alloc_consistent()
failure cases, queues' tx_skb just got allocated and are empty.
Fixes: 89e5785fc8 ("[PATCH] Atmel MACB ethernet driver")
Cc: stable@vger.kernel.org
Reviewed-by: Nicolai Buchwitz <nb@tipi-net.de>
Signed-off-by: Théo Lebrun <theo.lebrun@bootlin.com>
Link: https://patch.msgid.link/20260702-macb-drop-tx-v4-1-1c833eebdbc8@bootlin.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
MANA allocates RX buffers from page pool fragments when frag_count is
greater than 1. In that case the buffers remain DMA mapped by page pool
and the RX completion path does not call dma_unmap_single(). As a result,
the implicit sync-for-CPU normally performed by dma_unmap_single() is
missing before the packet data is passed to the networking stack.
This breaks RX on configurations which require explicit DMA syncing, for
example when booted with swiotlb=force.
Fix this by recording the page pool page and DMA sync offset when the RX
buffer is allocated, and syncing the received packet range for CPU access
before handing the RX buffer to the stack.
Fixes: 730ff06d3f ("net: mana: Use page pool fragments for RX buffers instead of full pages to improve memory efficiency.")
Cc: stable@vger.kernel.org
Reviewed-by: Haiyang Zhang <haiyangz@microsoft.com>
Signed-off-by: Dexuan Cui <decui@microsoft.com>
Link: https://patch.msgid.link/20260702041237.617719-3-decui@microsoft.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
In tcp_mmap.c, both child_thread() and main() allocate an EVP_MD_CTX
via EVP_MD_CTX_new() when integrity checking is enabled, but neither
function releases the context. child_thread() misses the free in its
common cleanup block, and main() returns without freeing the context.
This results in a SHA256 context leak on every run that uses the
‑i (integrity) option. Add the missing EVP_MD_CTX_free() calls to
the appropriate cleanup paths to fix the leak.
Fixes: 5c5945dc69 ("selftests/net: Add SHA256 computation over data sent in tcp_mmap")
Signed-off-by: Wang Yan <wangyan01@kylinos.cn>
Link: https://patch.msgid.link/20260702025949.442523-1-wangyan01@kylinos.cn
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Pull HID fixes from Jiri Kosina:
- OOB, UAF, NULL-deref fixes in core and picolcd, logitech, letsketch,
appleir and multitouch drivers (Georgiy Osokin, HyeongJun An, Lee
Jones, Manish Khadka, Maoyi Xie and Trung Nguyen)
- fix for integer wraparound (and corresponding regression selftest) in
hid-bpf (Yiyang Chen)
* tag 'hid-for-linus-2026070801' of git://git.kernel.org/pub/scm/linux/kernel/git/hid/hid:
selftests/hid: multitouch: test a large ContactCountMaximum
HID: multitouch: fix out-of-bounds bit access on mt_io_flags
selftests/hid: Cover hid_bpf_get_data() size overflow
selftests/hid: Load only requested struct_ops maps
HID: bpf: Fix hid_bpf_get_data() range check
HID: lg-g15: cancel pending work on remove to fix a use-after-free
HID: logitech-dj: Fix maxfield check in DJ short report validation
HID: core: Fix OOB read in hid_get_report for numbered reports
HID: picolcd: prevent NULL pointer dereference in picolcd_send_and_wait()
HID: appleir: fix UAF on pending key_up_timer in remove()
HID: letsketch: fix UAF on inrange_timer at driver unbind
Sashiko points out that after stripping the outer headers
with pskb_pull() we should ensure the inner IP headers
in ICMP errors from tunnels are present in the skb headroom
for functions like ipv4_update_pmtu(), icmp_send() and
IP_VS_DBG().
Also, add more checks for the length of the inner headers.
Fixes: f2edb9f770 ("ipvs: implement passive PMTUD for IPIP packets")
Link: https://sashiko.dev/#/patchset/20260702073430.67680-1-zhaoyz24%40mails.tsinghua.edu.cn
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Florian Westphal <fw@strlen.de>
set_sctp_state() reads the SCTP chunk header again in order to drive the
IPVS SCTP state table. For IPv6 it computes the offset with
sizeof(struct ipv6hdr), while the surrounding IPVS code uses iph.len from
ip_vs_fill_iph_skb(), where ipv6_find_hdr() has already skipped
extension headers and found the real transport header.
This makes the state machine read from the wrong offset for IPv6 SCTP
packets that carry extension headers. For example, an INIT packet with an
8-byte destination options header can be scheduled correctly by
sctp_conn_schedule(), but set_sctp_state() reads the first byte of the
SCTP verification tag as a DATA chunk type. The connection then moves
from NONE to ESTABLISHED instead of INIT1, gets the longer established
timeout, and updates the active/inactive destination counters
incorrectly. This happens even though the SCTP handshake has not
completed.
Use the parsed transport offset passed down from ip_vs_set_state() for
the SCTP chunk-header lookup. For IPv4 and IPv6 packets without
extension headers this preserves the existing offset.
Fixes: 2906f66a56 ("ipvs: SCTP Trasport Loadbalancing Support")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/netdev/20260705123040.35755-1-zhaoyz24@mails.tsinghua.edu.cn/
Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn>
Reported-by: Ao Wang <wangao@seu.edu.cn>
Reported-by: Xuewei Feng <fengxw06@126.com>
Reported-by: Qi Li <qli01@tsinghua.edu.cn>
Reported-by: Ke Xu <xuke@tsinghua.edu.cn>
Assisted-by: Claude Code:GLM-5.2
Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Florian Westphal <fw@strlen.de>
TCP state handling reparses the skb to find the TCP header. For IPv6 it
uses sizeof(struct ipv6hdr), while the surrounding IPVS code already
parsed the packet with ip_vs_fill_iph_skb() and has the real
transport-header offset in iph.len.
This makes TCP state handling look at the wrong bytes when an IPv6
packet carries extension headers. Use the parsed transport offset passed
down from ip_vs_set_state() when reading the TCP header.
For IPv4 and for IPv6 packets without extension headers, the passed
offset matches the previous value.
Fixes: 0bbdd42b7e ("IPVS: Extend protocol DNAT/SNAT and state handlers")
Link: https://lore.kernel.org/netdev/20260705125659.37744-1-zhaoyz24@mails.tsinghua.edu.cn/
Reported-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Reported-by: Yuxiang Yang <yangyx22@mails.tsinghua.edu.cn>
Reported-by: Ao Wang <wangao@seu.edu.cn>
Reported-by: Xuewei Feng <fengxw06@126.com>
Reported-by: Qi Li <qli01@tsinghua.edu.cn>
Reported-by: Ke Xu <xuke@tsinghua.edu.cn>
Assisted-by: Claude Code:GLM-5.2
Signed-off-by: Yizhou Zhao <zhaoyz24@mails.tsinghua.edu.cn>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Florian Westphal <fw@strlen.de>
