After 1b23cdbd2b ("net: protect netdev->napi_list with netdev_lock()")
it makes sense to iterate through dev->napi_list while holding
the device lock.
Also call synchronize_net() at most one time.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250117232113.1612899-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Basically, dev_ifsioc() operates on the passed single netns (except
for netdev notifier chains with lower/upper devices for which we will
need more changes).
Let's hold rtnl_net_lock() for dev_ifsioc().
Now that NETDEV_CHANGENAME is always triggered under rtnl_net_lock()
of the device's netns. (do_setlink() and dev_ifsioc())
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250115095545.52709-4-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
devnet_rename_sem is no longer used since commit
0840556e5a ("net: Protect dev->name by seqlock.").
Also, RTNL serialises dev_change_name().
Let's remove devnet_rename_sem.
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250115095545.52709-3-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The cited commit forgot to add netdev_rename_lock in one of the
error paths in dev_change_name().
Let's hold netdev_rename_lock before restoring the old dev->name.
Fixes: 0840556e5a ("net: Protect dev->name by seqlock.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250115095545.52709-2-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The tool pp_alloc_fail.py tested error recovery by injecting errors
into the function page_pool_alloc_pages(). The page pool allocation
function page_pool_dev_alloc() does not end up calling
page_pool_alloc_pages(). page_pool_alloc_netmems() seems to be the
function that is called by all of the page pool alloc functions in
the API, so move error injection to that function instead.
Signed-off-by: John Daley <johndale@cisco.com>
Link: https://patch.msgid.link/20250115181312.3544-2-johndale@cisco.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Eric Dumazet says:
====================
net: reduce RTNL pressure in unregister_netdevice()
One major source of RTNL contention resides in unregister_netdevice()
Due to RCU protection of various network structures, and
unregister_netdevice() being a synchronous function,
it is calling potentially slow functions while holding RTNL.
I think we can release RTNL in two points, so that three
slow functions are called while RTNL can be used
by other threads.
v1: https://lore.kernel.org/netdev/20250107130906.098fc8d6@kernel.org/T/#m398c95f5778e1ff70938e079d3c4c43c050ad2a6
====================
Link: https://patch.msgid.link/20250114205531.967841-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
One synchronize_net() call is currently done while holding RTNL.
This is source of RTNL contention in workloads adding and deleting
many network namespaces per second, because synchronize_rcu()
and synchronize_rcu_expedited() can use 60+ ms in some cases.
For cleanup_net() use, temporarily release RTNL
while calling the last synchronize_net().
This should be safe, because devices are no longer visible
to other threads after unlist_netdevice() call
and setting dev->reg_state to NETREG_UNREGISTERING.
In any case, the new netdev_lock() / netdev_unlock()
infrastructure that we are adding should allow
to fix potential issues, with a combination
of a per-device mutex and dev->reg_state awareness.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Link: https://patch.msgid.link/20250114205531.967841-6-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Two synchronize_net() calls are currently done while holding RTNL.
This is source of RTNL contention in workloads adding and deleting
many network namespaces per second, because synchronize_rcu()
and synchronize_rcu_expedited() can use 60+ ms in some cases.
For cleanup_net() use, temporarily release RTNL
while calling the last synchronize_net().
This should be safe, because devices are no longer visible
to other threads at this point.
In any case, the new netdev_lock() / netdev_unlock()
infrastructure that we are adding should allow
to fix potential issues, with a combination
of a per-device mutex and dev->reg_state awareness.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Link: https://patch.msgid.link/20250114205531.967841-5-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
flush_all_backlogs() is called from unregister_netdevice_many_notify()
as part of netdevice dismantles.
This is currently called under RTNL, and can last up to 50 ms
on busy hosts.
There is no reason to hold RTNL at this stage, if our caller
is cleanup_net() : netns are no more visible, devices
are in NETREG_UNREGISTERING state and no other thread
could mess our state while RTNL is temporarily released.
In order to provide isolation, this patch provides a separate
'net_todo_list' for cleanup_net().
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Link: https://patch.msgid.link/20250114205531.967841-4-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
flush_all_backlogs() uses per-cpu and static data to hold its
temporary data, on the assumption it is called under RTNL
protection.
Following patch in the series will break this assumption.
Use instead a dynamically allocated piece of memory.
In the unlikely case the allocation fails,
use a boot-time allocated memory.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Link: https://patch.msgid.link/20250114205531.967841-3-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
cleanup_net() is the single thread responsible
for netns dismantles, and a serious bottleneck.
Before we can get per-netns RTNL, make sure
all synchronize_net() called from this thread
are using rcu_synchronize_expedited().
v3: deal with CONFIG_NET_NS=n
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jesse Brandeburg <jbrandeburg@cloudflare.com>
Link: https://patch.msgid.link/20250114205531.967841-2-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
NAPI lifetime, visibility and config are all fully under
netdev_lock protection now.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-12-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Protect the following members of netdev and napi by netdev_lock:
- defer_hard_irqs,
- gro_flush_timeout,
- irq_suspend_timeout.
The first two are written via sysfs (which this patch switches
to new lock), and netdev genl which holds both netdev and rtnl locks.
irq_suspend_timeout is only written by netdev genl.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-11-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Take netdev_lock() in netif_napi_set_irq(). All NAPI "control fields"
are now protected by that lock (most of the other ones are set during
napi add/del). The napi_hash_node is fully protected by the hash
spin lock, but close enough for the kdoc...
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-10-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Now that NAPI instances can't come and go without holding
netdev->lock we can trivially switch from rtnl_lock() to
netdev_lock() for setting netdev->threaded via sysfs.
Note that since we do not lock netdev_lock around sysfs
calls in the core we don't have to "trylock" like we do
with rtnl_lock.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-9-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In prep for dropping rtnl_lock, start locking netdev->lock in netlink
genl ops. We need to be using netdev->up instead of flags & IFF_UP.
We can remove the RCU lock protection for the NAPI since NAPI list
is protected by netdev->lock already.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-8-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Wrap napi_enable() / napi_disable() with netdev_lock().
Provide the "already locked" flavor of the API.
iavf needs the usual adjustment. A number of drivers call
napi_enable() under a spin lock, so they have to be modified
to take netdev_lock() first, then spin lock then call
napi_enable_locked().
Protecting napi_enable() implies that napi->napi_id is protected
by netdev_lock().
Acked-by: Francois Romieu <romieu@fr.zoreil.com> # via-velocity
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-7-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Hold netdev->lock when NAPIs are getting added or removed.
This will allow safe access to NAPI instances of a net_device
without rtnl_lock.
Create a family of helpers which assume the lock is already taken.
Switch iavf to them, as it makes extensive use of netdev->lock,
already.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-6-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Some uAPI (netdev netlink) hide net_device's sub-objects while
the interface is down to ensure uniform behavior across drivers.
To remove the rtnl_lock dependency from those uAPIs we need a way
to safely tell if the device is down or up.
Add an indication of whether device is open or closed, protected
by netdev->lock. The semantics are the same as IFF_UP, but taking
netdev_lock around every write to ->flags would be a lot of code
churn.
We don't want to blanket the entire open / close path by netdev_lock,
because it will prevent us from applying it to specific structures -
core helpers won't be able to take that lock from any function
called by the drivers on open/close paths.
So the state of the flag is "pessimistic", as in it may report false
negatives, but never false positives.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-5-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add helpers for accessing netdevs under netdev_lock().
There's some careful handling needed to find the device and lock it
safely, without it getting unregistered, and without taking rtnl_lock
(the latter being the whole point of the new locking, after all).
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-4-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Protect writes to netdev->reg_state with netdev_lock().
From now on holding netdev_lock() is sufficient to prevent
the net_device from getting unregistered, so code which
wants to hold just a single netdev around no longer needs
to hold rtnl_lock.
We do not protect the NETREG_UNREGISTERED -> NETREG_RELEASED
transition. We'd need to move mutex_destroy(netdev->lock)
to .release, but the real reason is that trying to stop
the unregistration process mid-way would be unsafe / crazy.
Taking references on such devices is not safe, either.
So the intended semantics are to lock REGISTERED devices.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250115035319.559603-3-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When a single buffer XDP is attached, NIC should guarantee only single
page packets will be received.
tcp-data-split feature splits packets into header and payload. single
buffer XDP can't handle it properly.
So attaching single buffer XDP should be disallowed when tcp-data-split
is enabled.
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Link: https://patch.msgid.link/20250114142852.3364986-6-ap420073@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
If driver doesn't support ring parameter or tcp-data-split configuration
is not sufficient, the devmem should not be set up.
Before setup the devmem, tcp-data-split should be ON and hds-thresh
value should be 0.
Tested-by: Stanislav Fomichev <sdf@fomichev.me>
Reviewed-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Link: https://patch.msgid.link/20250114142852.3364986-4-ap420073@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When tcp-data-split is UNKNOWN mode, drivers arbitrarily handle it.
For example, bnxt_en driver automatically enables if at least one of
LRO/GRO/JUMBO is enabled.
If tcp-data-split is UNKNOWN and LRO is enabled, a driver returns
ENABLES of tcp-data-split, not UNKNOWN.
So, `ethtool -g eth0` shows tcp-data-split is enabled.
The problem is in the setting situation.
In the ethnl_set_rings(), it first calls get_ringparam() to get the
current driver's config.
At that moment, if driver's tcp-data-split config is UNKNOWN, it returns
ENABLE if LRO/GRO/JUMBO is enabled.
Then, it sets values from the user and driver's current config to
kernel_ethtool_ringparam.
Last it calls .set_ringparam().
The driver, especially bnxt_en driver receives
ETHTOOL_TCP_DATA_SPLIT_ENABLED.
But it can't distinguish whether it is set by the user or just the
current config.
When user updates ring parameter, the new hds_config value is updated
and current hds_config value is stored to old_hdsconfig.
Driver's .set_ringparam() callback can distinguish a passed
tcp-data-split value is came from user explicitly.
If .set_ringparam() is failed, hds_config is rollbacked immediately.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Link: https://patch.msgid.link/20250114142852.3364986-2-ap420073@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
When __netpoll_setup() is called directly, instead of through
netpoll_setup(), the np->skb_pool list head isn't initialized.
If skb_pool_flush() is later called, then we hit a NULL pointer
in skb_queue_purge_reason(). This can be seen with this repro,
when CONFIG_NETCONSOLE is enabled as a module:
ip tuntap add mode tap tap0
ip link add name br0 type bridge
ip link set dev tap0 master br0
modprobe netconsole netconsole=4444@10.0.0.1/br0,9353@10.0.0.2/
rmmod netconsole
The backtrace is:
BUG: kernel NULL pointer dereference, address: 0000000000000008
#PF: supervisor write access in kernel mode
#PF: error_code(0x0002) - not-present page
... ... ...
Call Trace:
<TASK>
__netpoll_free+0xa5/0xf0
br_netpoll_cleanup+0x43/0x50 [bridge]
do_netpoll_cleanup+0x43/0xc0
netconsole_netdev_event+0x1e3/0x300 [netconsole]
unregister_netdevice_notifier+0xd9/0x150
cleanup_module+0x45/0x920 [netconsole]
__se_sys_delete_module+0x205/0x290
do_syscall_64+0x70/0x150
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Move the skb_pool list setup and initial skb fill into __netpoll_setup().
Fixes: 221a9c1df7 ("net: netpoll: Individualize the skb pool")
Signed-off-by: John Sperbeck <jsperbeck@google.com>
Reviewed-by: Breno Leitao <leitao@debian.org>
Link: https://patch.msgid.link/20250114011354.2096812-1-jsperbeck@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add multicast filtering support for VLAN interfaces in dual EMAC mode
for ICSSG driver.
The driver uses vlan_for_each() API to get the list of available
vlans. The driver then sync mc addr of vlan interface with a locally
mainatined list emac->vlan_mcast_list[vid] using __hw_addr_sync_multiple()
API.
__hw_addr_sync_multiple() is used instead of __hw_addr_sync() to sync
vdev->mc with local list because the sync_cnt for addresses in vdev->mc
will already be set by the vlan_dev_set_rx_mode() [net/8021q/vlan_dev.c]
and __hw_addr_sync() only syncs when the sync_cnt == 0. Whereas
__hw_addr_sync_multiple() can sync addresses even if sync_cnt is not 0.
Export __hw_addr_sync_multiple() so that driver can use it.
Once the local list is synced, driver calls __hw_addr_sync_dev() with
the local list, vdev, sync and unsync callbacks.
__hw_addr_sync_dev() is used with the local maintained list as the list
to synchronize instead of using __dev_mc_sync() on vdev because
__dev_mc_sync() on vdev will call __hw_addr_sync_dev() on vdev->mc and
sync_cnt for addresses in vdev->mc will already be set by the
vlan_dev_set_rx_mode() [net/8021q/vlan_dev.c] and __hw_addr_sync_dev()
only syncs if the sync_cnt of addresses in the list (vdev->mc in this case)
is 0. Whereas __hw_addr_sync_dev() on local list will work fine as the
sync_cnt for addresses in the local list will still be 0.
Based on change in addresses in the local list, sync / unsync callbacks
are invoked. In the sync / unsync API in driver, based on whether the ndev
is vlan or not, driver passes appropriate vid to FDB helper functions.
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
init_dummy_netdev_core() used to cater to net_devices which
did not come from alloc_netdev_mqs(). Since that's no longer
supported remove the init logic which duplicates alloc_netdev_mqs().
While at it rename back to init_dummy_netdev().
Reviewed-by: Joe Damato <jdamato@fastly.com>
Link: https://patch.msgid.link/20250113003456.3904110-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
init_dummy_netdev() can initialize statically declared or embedded
net_devices. Such netdevs did not come from alloc_netdev_mqs().
After recent work by Breno, there are the only two cases where
we have do that.
Switch those cases to alloc_netdev_mqs() and delete init_dummy_netdev().
Dealing with static netdevs is not worth the maintenance burden.
Reviewed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Reviewed-by: Matthieu Baerts (NGI0) <matttbe@kernel.org>
Reviewed-by: Joe Damato <jdamato@fastly.com>
Link: https://patch.msgid.link/20250113003456.3904110-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Passing a sufficient amount of imix entries leads to invalid access to the
pkt_dev->imix_entries array because of the incorrect boundary check.
UBSAN: array-index-out-of-bounds in net/core/pktgen.c:874:24
index 20 is out of range for type 'imix_pkt [20]'
CPU: 2 PID: 1210 Comm: bash Not tainted 6.10.0-rc1 #121
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996)
Call Trace:
<TASK>
dump_stack_lvl lib/dump_stack.c:117
__ubsan_handle_out_of_bounds lib/ubsan.c:429
get_imix_entries net/core/pktgen.c:874
pktgen_if_write net/core/pktgen.c:1063
pde_write fs/proc/inode.c:334
proc_reg_write fs/proc/inode.c:346
vfs_write fs/read_write.c:593
ksys_write fs/read_write.c:644
do_syscall_64 arch/x86/entry/common.c:83
entry_SYSCALL_64_after_hwframe arch/x86/entry/entry_64.S:130
Found by Linux Verification Center (linuxtesting.org) with SVACE.
Fixes: 52a62f8603 ("pktgen: Parse internet mix (imix) input")
Signed-off-by: Artem Chernyshev <artem.chernyshev@red-soft.ru>
[ fp: allow to fill the array completely; minor changelog cleanup ]
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are no module callers of dev_get_by_napi_id(),
and commit d1cacd7477 ("netdev: prevent accessing NAPI instances
from another namespace") proves that getting NAPI by id
needs to be done with care. So hide dev_get_by_napi_id().
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com>
Reviewed-by: Joe Damato <jdamato@fastly.com>
Link: https://patch.msgid.link/20250110004924.3212260-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Dump continuation depends on the NAPI list being sorted.
Broken netlink dump continuation may be rare and hard to debug
so add a warning if we notice the potential problem while walking
the list.
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Reviewed-by: Joe Damato <jdamato@fastly.com>
Link: https://patch.msgid.link/20250110004505.3210140-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Commit 86e25f40aa ("net: napi: Add napi_config") moved napi->napi_id
assignment to a later point in time (napi_hash_add_with_id). This breaks
__xdp_rxq_info_reg which copies napi_id at an earlier time and now
stores 0 napi_id. It also makes sk_mark_napi_id_once_xdp and
__sk_mark_napi_id_once useless because they now work against 0 napi_id.
Since sk_busy_loop requires valid napi_id to busy-poll on, there is no way
to busy-poll AF_XDP sockets anymore.
Bring back the ability to busy-poll on XSK by resolving socket's napi_id
at bind time. This relies on relatively recent netif_queue_set_napi,
but (assume) at this point most popular drivers should have been converted.
This also removes per-tx/rx cycles which used to check and/or set
the napi_id value.
Confirmed by running a busy-polling AF_XDP socket
(github.com/fomichev/xskrtt) on mlx5 and looking at BusyPollRxPackets
from /proc/net/netstat.
Fixes: 86e25f40aa ("net: napi: Add napi_config")
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://patch.msgid.link/20250109003436.2829560-1-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
As pointed out in the original comment, lookup in sockmap can return a TCP
ESTABLISHED socket. Such TCP socket may have had SO_ATTACH_REUSEPORT_EBPF
set before it was ESTABLISHED. In other words, a non-NULL sk_reuseport_cb
does not imply a non-refcounted socket.
Drop sk's reference in both error paths.
unreferenced object 0xffff888101911800 (size 2048):
comm "test_progs", pid 44109, jiffies 4297131437
hex dump (first 32 bytes):
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
80 00 01 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
backtrace (crc 9336483b):
__kmalloc_noprof+0x3bf/0x560
__reuseport_alloc+0x1d/0x40
reuseport_alloc+0xca/0x150
reuseport_attach_prog+0x87/0x140
sk_reuseport_attach_bpf+0xc8/0x100
sk_setsockopt+0x1181/0x1990
do_sock_setsockopt+0x12b/0x160
__sys_setsockopt+0x7b/0xc0
__x64_sys_setsockopt+0x1b/0x30
do_syscall_64+0x93/0x180
entry_SYSCALL_64_after_hwframe+0x76/0x7e
Fixes: 64d85290d7 ("bpf: Allow bpf_map_lookup_elem for SOCKMAP and SOCKHASH")
Signed-off-by: Michal Luczaj <mhal@rbox.co>
Reviewed-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://patch.msgid.link/20250110-reuseport-memleak-v1-1-fa1ddab0adfe@rbox.co
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Linus suggested during one of past maintainer summits (in context of
a DMA_BUF discussion) that symbol namespaces can be used to prevent
unwelcome but in-tree code from using all exported functions.
Create a namespace for netdev.
Export netdev_rx_queue_restart(), drivers may want to use it since
it gives them a simple and safe way to restart a queue to apply
config changes. But it's both too low level and too actively developed
to be used outside netdev.
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Netlink code depends on NAPI instances being sorted by ID on
the netdev list for dump continuation. We need to be able to
find the position on the list where we left off if dump does
not fit in a single skb, and in the meantime NAPI instances
can come and go.
This was trivially true when we were assigning a new ID to every
new NAPI instance. Since we added the NAPI config API, we try
to retain the ID previously used for the same queue, but still
add the new NAPI instance at the start of the list.
This is fine if we reset the entire netdev and all NAPIs get
removed and added back. If driver replaces a NAPI instance
during an operation like DEVMEM queue reset, or recreates
a subset of NAPI instances in other ways we may end up with
broken ordering, and therefore Netlink dumps with either
missing or duplicated entries.
At this stage the problem is theoretical. Only two drivers
support queue API, bnxt and gve. gve recreates NAPIs during
queue reset, but it doesn't support NAPI config.
bnxt supports NAPI config but doesn't recreate instances
during reset.
We need to save the ID in the config as soon as it is assigned
because otherwise the new NAPI will not know what ID it will
get at enable time, at the time it is being added.
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
In commit 66e4c8d950 ("net: warn if transport header was not set")
I added a debug check in skb_transport_header() to detect
if a caller expects the transport_header to be set to a meaningful
value by a prior code path.
Unfortunately, __netif_receive_skb_core() resets the transport header
to the same value than the network header, defeating this check
in receive paths.
Pretending the transport and network headers are the same
is usually wrong.
This patch removes this reset for CONFIG_DEBUG_NET=y builds
to let fuzzers and CI find bugs.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250107144342.499759-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The NAPI IDs were not fully exposed to user space prior to the netlink
API, so they were never namespaced. The netlink API must ensure that
at the very least NAPI instance belongs to the same netns as the owner
of the genl sock.
napi_by_id() can become static now, but it needs to move because of
dev_get_by_napi_id().
Cc: stable@vger.kernel.org
Fixes: 1287c1ae0f ("netdev-genl: Support setting per-NAPI config values")
Fixes: 27f91aaf49 ("netdev-genl: Add netlink framework functions for napi")
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Reviewed-by: Joe Damato <jdamato@fastly.com>
Link: https://patch.msgid.link/20250106180137.1861472-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
(un)?register_netdevice_notifier_dev_net() hold RTNL before triggering
the notifier for all netdev in the netns.
Let's convert the RTNL to rtnl_net_lock().
Note that move_netdevice_notifiers_dev_net() is assumed to be (but not
yet) protected by per-netns RTNL of both src and dst netns; we need to
convert wireless and hyperv drivers that call dev_change_net_namespace().
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250106070751.63146-4-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
(un)?register_netdevice_notifier_net() hold RTNL before triggering the
notifier for all netdev in the netns.
Let's convert the RTNL to rtnl_net_lock().
Note that the per-netns netdev notifier is protected by per-netns RTNL.
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250106070751.63146-3-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In commit d7811e623d ("[NET]: Drop tx lock in dev_watchdog_up")
dev_watchdog_up() became a simple wrapper for __netdev_watchdog_up()
Herbert also said : "In 2.6.19 we can eliminate the unnecessary
__dev_watchdog_up and replace it with dev_watchdog_up."
This patch consolidates things to have only two functions, with
a common prefix.
- netdev_watchdog_up(), exported for the sake of one freescale driver.
This replaces __netdev_watchdog_up() and dev_watchdog_up().
- netdev_watchdog_down(), static to net/sched/sch_generic.c
This replaces dev_watchdog_down().
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Reviewed-by: Jason Xing <kerneljasonxing@gmail.com>
Link: https://patch.msgid.link/20250105090924.1661822-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Let's hold per-netns RTNL of dev_net(dev) in register_netdev()
and unregister_netdev().
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
rtnl_lock_killable() is used only in register_netdev()
and will be converted to per-netns RTNL.
Let's unexport it and add the corresponding helper.
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
We use NAPI ID as the key for continuing dumps. We also depend
on the NAPIs being sorted by ID within the driver list. Tx NAPIs
(which don't have an ID assigned) break this expectation, it's
not currently possible to dump them reliably. Since Tx NAPIs
are relatively rare, and can't be used in doit (GET or SET)
hide them from the dump API as well.
Fixes: 27f91aaf49 ("netdev-genl: Add netlink framework functions for napi")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20250103183207.1216004-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
