Cover the paths that come via bpf system call and XSK bind.
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-10-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Most of them are already covered by the converted dev_xxx APIs.
Add the locking wrappers for the remaining ones.
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-9-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Convert all ndo_eth_ioctl invocations to dev_eth_ioctl which does the
locking. Reflow some of the dev_siocxxx to drop else clause.
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-8-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To preserve the atomicity, hold the lock while applying multiple
attributes. The major issue with a full conversion to the instance
lock are software nesting devices (bonding/team/vrf/etc). Those
devices call into the core stack for their lower (potentially
real hw) devices. To avoid explicitly wrapping all those places
into instance lock/unlock, introduce new API boundaries:
- (some) existing dev_xxx calls are now considered "external"
(to drivers) APIs and they transparently grab the instance
lock if needed (dev_api.c)
- new netif_xxx calls are internal core stack API (naming is
sketchy, I've tried netdev_xxx_locked per Jakub's suggestion,
but it feels a bit verbose; but happy to get back to this
naming scheme if this is the preference)
This avoids touching most of the existing ioctl/sysfs/drivers paths.
Note the special handling of ndo_xxx_slave operations: I exploit
the fact that none of the drivers that call these functions
need/use instance lock. At the same time, they use dev_xxx
APIs, so the lower device has to be unlocked.
Changes in unregister_netdevice_many_notify (to protect dev->state
with instance lock) trigger lockdep - the loop over close_list
(mostly from cleanup_net) introduces spurious ordering issues.
netdev_lock_cmp_fn has a justification on why it's ok to suppress
for now.
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-7-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
For the drivers that use queue management API, switch to the mode where
core stack holds the netdev instance lock. This affects the following
drivers:
- bnxt
- gve
- netdevsim
Originally I locked only start/stop, but switched to holding the
lock over all iterations to make them look atomic to the device
(feels like it should be easier to reason about).
Reviewed-by: Eric Dumazet <edumazet@google.com>
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-6-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Introduce new dev_setup_tc for nft ndo_setup_tc paths.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-3-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
For the drivers that use shaper API, switch to the mode where
core stack holds the netdev lock. This affects two drivers:
* iavf - already grabs netdev lock in ndo_open/ndo_stop, so mostly
remove these
* netdevsim - switch to _locked APIs to avoid deadlock
iavf_close diff is a bit confusing, the existing call looks like this:
iavf_close() {
netdev_lock()
..
netdev_unlock()
wait_event_timeout(down_waitqueue)
}
I change it to the following:
netdev_lock()
iavf_close() {
..
netdev_unlock()
wait_event_timeout(down_waitqueue)
netdev_lock() // reusing this lock call
}
netdev_unlock()
Since I'm reusing existing netdev_lock call, so it looks like I only
add netdev_unlock.
Cc: Saeed Mahameed <saeed@kernel.org>
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/20250305163732.2766420-2-sdf@fomichev.me
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It could be hard to understand why the netlink command fails. For example,
if dev->netns_immutable is set, the error is "Invalid argument".
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The name 'netns_local' is confusing. A following commit will export it via
netlink, so let's use a more explicit name.
Reported-by: Eric Dumazet <edumazet@google.com>
Suggested-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
In fact, these two are not tied closely to each other. The only
requirements to GRO are to use it in the BH context and have some
sane limits on the packet batches, e.g. NAPI has a limit of its
budget (64/8/etc.).
Move purely GRO fields into a new structure, &gro_node. Embed it
into &napi_struct and adjust all the references.
gro_node::cached_napi_id is effectively the same as
napi_struct::napi_id, but to be used on GRO hotpath to mark skbs.
napi_struct::napi_id is now a fully control path field.
Three Ethernet drivers use napi_gro_flush() not really meant to be
exported, so move it to <net/gro.h> and add that include there.
napi_gro_receive() is used in more than 100 drivers, keep it
in <linux/netdevice.h>.
This does not make GRO ready to use outside of the NAPI context
yet.
Tested-by: Daniel Xu <dxu@dxuuu.xyz>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
A common task for most drivers is to remember the user-set CPU affinity
to its IRQs. On each netdev reset, the driver should re-assign the user's
settings to the IRQs. Unify this task across all drivers by moving the CPU
affinity to napi->config.
However, to move the CPU affinity to core, we also need to move aRFS
rmap management since aRFS uses its own IRQ notifiers.
For the aRFS, add a new netdev flag "rx_cpu_rmap_auto". Drivers supporting
aRFS should set the flag via netif_enable_cpu_rmap() and core will allocate
and manage the aRFS rmaps. Freeing the rmap is also done by core when the
netdev is freed. For better IRQ affinity management, move the IRQ rmap
notifier inside the napi_struct and add new notify.notify and
notify.release functions: netif_irq_cpu_rmap_notify() and
netif_napi_affinity_release().
Now we have the aRFS rmap management in core, add CPU affinity mask to
napi_config. To delegate the CPU affinity management to the core, drivers
must:
1 - set the new netdev flag "irq_affinity_auto":
netif_enable_irq_affinity(netdev)
2 - create the napi with persistent config:
netif_napi_add_config()
3 - bind an IRQ to the napi instance: netif_napi_set_irq()
the core will then make sure to use re-assign affinity to the napi's
IRQ.
The default IRQ mask is set to one cpu starting from the closest NUMA.
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Link: https://patch.msgid.link/20250224232228.990783-2-ahmed.zaki@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
After the previous commit is finally safe to revert commit dbae2b0628
("net: skb: introduce and use a single page frag cache"): do it here.
The intended goal of such change was to counter a performance regression
introduced by commit 3226b158e6 ("net: avoid 32 x truesize
under-estimation for tiny skbs").
Unfortunately, the blamed commit introduces another regression for the
virtio_net driver. Such a driver calls napi_alloc_skb() with a tiny
size, so that the whole head frag could fit a 512-byte block.
The single page frag cache uses a 1K fragment for such allocation, and
the additional overhead, under small UDP packets flood, makes the page
allocator a bottleneck.
Thanks to commit bf9f1baa27 ("net: add dedicated kmem_cache for
typical/small skb->head"), this revert does not re-introduce the
original regression. Actually, in the relevant test on top of this
revert, I measure a small but noticeable positive delta, just above
noise level.
The revert itself required some additional mangling due to recent updates
in the affected code.
Suggested-by: Eric Dumazet <edumazet@google.com>
Fixes: dbae2b0628 ("net: skb: introduce and use a single page frag cache")
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This reverts commit 011b033590.
Sabrina reports that the revert may trigger warnings due to intervening
changes, especially the ability to rise MAX_SKB_FRAGS. Let's drop it
and revisit once that part is also ironed out.
Fixes: 011b033590 ("Revert "net: skb: introduce and use a single page frag cache"")
Reported-by: Sabrina Dubroca <sd@queasysnail.net>
Link: https://lore.kernel.org/6bf54579233038bc0e76056c5ea459872ce362ab.1739375933.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This reverts commit dbae2b0628 ("net: skb: introduce and use a single
page frag cache"). The intended goal of such change was to counter a
performance regression introduced by commit 3226b158e6 ("net: avoid
32 x truesize under-estimation for tiny skbs").
Unfortunately, the blamed commit introduces another regression for the
virtio_net driver. Such a driver calls napi_alloc_skb() with a tiny
size, so that the whole head frag could fit a 512-byte block.
The single page frag cache uses a 1K fragment for such allocation, and
the additional overhead, under small UDP packets flood, makes the page
allocator a bottleneck.
Thanks to commit bf9f1baa27 ("net: add dedicated kmem_cache for
typical/small skb->head"), this revert does not re-introduce the
original regression. Actually, in the relevant test on top of this
revert, I measure a small but noticeable positive delta, just above
noise level.
The revert itself required some additional mangling due to the
introduction of the SKB_HEAD_ALIGN() helper and local lock infra in the
affected code.
Suggested-by: Eric Dumazet <edumazet@google.com>
Fixes: dbae2b0628 ("net: skb: introduce and use a single page frag cache")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Link: https://patch.msgid.link/e649212fde9f0fdee23909ca0d14158d32bb7425.1738877290.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
dev->nd_net can change, readers should either
use rcu_read_lock() or RTNL.
We currently use a generic helper, dev_net() with
no debugging support. We probably have many hidden bugs.
Add dev_net_rcu() helper for callers using rcu_read_lock()
protection.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20250205155120.1676781-2-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Rx/tx queues embed their own kobject for registering their per-queue
sysfs files. The issue is they're using the kobject default groups for
this and entirely rely on the kobject refcounting for releasing their
sysfs paths.
In order to remove rtnl_trylock calls we need sysfs files not to rely on
their associated kobject refcounting for their release. Thus we here
move queues sysfs files from the kobject default groups to their own
groups which can be removed separately.
Signed-off-by: Antoine Tenart <atenart@kernel.org>
Link: https://patch.msgid.link/20250204170314.146022-3-atenart@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
After the blamed commits below, some UDP tunnel use dstats for
accounting. On the xmit path, all the UDP-base tunnels ends up
using iptunnel_xmit_stats() for stats accounting, and the latter
assumes the relevant (tunnel) network device uses tstats.
The end result is some 'funny' stat report for the mentioned UDP
tunnel, e.g. when no packet is actually dropped and a bunch of
packets are transmitted:
gnv2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1450 qdisc noqueue \
state UNKNOWN mode DEFAULT group default qlen 1000
link/ether ee:7d:09:87:90:ea brd ff:ff:ff:ff:ff:ff
RX: bytes packets errors dropped missed mcast
14916 23 0 15 0 0
TX: bytes packets errors dropped carrier collsns
0 1566 0 0 0 0
Address the issue ensuring the same binary layout for the overlapping
fields of dstats and tstats. While this solution is a bit hackish, is
smaller and with no performance pitfall compared to other alternatives
i.e. supporting both dstat and tstat in iptunnel_xmit_stats() or
reverting the blamed commit.
With time we should possibly move all the IP-based tunnel (and virtual
devices) to dstats.
Fixes: c77200c074 ("bareudp: Handle stats using NETDEV_PCPU_STAT_DSTATS.")
Fixes: 6fa6de3022 ("geneve: Handle stats using NETDEV_PCPU_STAT_DSTATS.")
Fixes: be226352e8 ("vxlan: Handle stats using NETDEV_PCPU_STAT_DSTATS.")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Guillaume Nault <gnault@redhat.com>
Link: https://patch.msgid.link/2e1c444cf0f63ae472baff29862c4c869be17031.1738432804.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
ndo_do_ioctl is no longer used by the appletalk subsystem after commit
45bd1c5ba7 ("net: appletalk: Drop aarp_send_probe_phase1()").
Signed-off-by: 谢致邦 (XIE Zhibang) <Yeking@Red54.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/tencent_4AC6ED413FEA8116B4253D3ED6947FDBCF08@qq.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Record the pending configuration in net_device struct.
ethtool core duplicates the current config and the specific
handlers (for now just ringparam) can modify it.
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250119020518.1962249-4-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Separate the HDS config from the ethtool state struct.
The HDS config contains just simple parameters, not state.
Having it as a separate struct will make it easier to clone / copy
and also long term potentially make it per-queue.
Reviewed-by: Michael Chan <michael.chan@broadcom.com>
Link: https://patch.msgid.link/20250119020518.1962249-2-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>
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>
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>
Add helpers for locking the netdev instance, use it in drivers
and the shaper code. This will make grepping for the lock usage
much easier, as we extend the lock to cover more fields.
Reviewed-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Link: https://patch.msgid.link/20250115035319.559603-2-kuba@kernel.org
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>
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() 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>
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>
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>
Introduce the description of a hwtstamp provider, mainly defined with a
the hwtstamp source and the phydev pointer.
Add a hwtstamp provider description within the netdev structure to
allow saving the hwtstamp we want to use. This prepares for future
support of an ethtool netlink command to select the desired hwtstamp
provider. By default, the old API that does not support hwtstamp
selectability is used, meaning the hwtstamp provider pointer is unset.
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
kernel-doc -Wall warns about missing Return: statement for non-void
functions. We have a number of kdocs in our headers which are missing
the colon, IOW they use
* Return some value
or
* Returns some value
Having the colon makes some sense, it should help kdoc parser avoid
false positives. So add them. This is mostly done with a sed script,
and removing the unnecessary cases (mostly the comments which aren't
kdoc).
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Acked-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Reviewed-by: Edward Cree <ecree.xilinx@gmail.com>
Acked-by: Alexandra Winter <wintera@linux.ibm.com>
Acked-by: Pablo Neira Ayuso <pablo@netfilter.org>
Link: https://patch.msgid.link/20241205165914.1071102-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently vrf is the only module that uses NETDEV_PCPU_STAT_DSTATS.
In order to make this kind of statistics available to other modules,
we need to define the update functions in netdevice.h.
Therefore, let's define dev_dstats_*() functions for RX and TX packet
updates (packets, bytes and drops). Use these new functions in vrf.c
instead of vrf_rx_stats() and the other manual counter updates.
While there, update the type of the "len" variables to "unsigned int",
so that there're aligned with both skb->len and the new dstats update
functions.
Signed-off-by: Guillaume Nault <gnault@redhat.com>
Link: https://patch.msgid.link/d7a552ee382c79f4854e7fcc224cf176cd21150d.1733313925.git.gnault@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
To make the system page pool usable as a source for allocating XDP
frames, we need to register it with xdp_reg_mem_model(), so that page
return works correctly. This is done in preparation for using the system
page_pool to convert XDP_PASS XSk frames to skbs; for the same reason,
make the per-cpu variable non-static so we can access it from other
source files as well (but w/o exporting).
Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Link: https://patch.msgid.link/20241203173733.3181246-7-aleksander.lobakin@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In lots of places, bpf_prog pointer is used only for tracing or other
stuff that doesn't modify the structure itself. Same for net_device.
Address at least some of them and add `const` attributes there. The
object code didn't change, but that may prevent unwanted data
modifications and also allow more helpers to have const arguments.
Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In a similar fashion to ndo_fdb_add, which was covered in the previous
patch, add the bool *notified argument to ndo_fdb_del. Callees that send a
notification on their own set the flag to true.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/06b1acf4953ef0a5ed153ef1f32d7292044f2be6.1731589511.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Currently when FDB entries are added to or deleted from a VXLAN netdevice,
the VXLAN driver emits one notification, including the VXLAN-specific
attributes. The core however always sends a notification as well, a generic
one. Thus two notifications are unnecessarily sent for these operations. A
similar situation comes up with bridge driver, which also emits
notifications on its own:
# ip link add name vx type vxlan id 1000 dstport 4789
# bridge monitor fdb &
[1] 1981693
# bridge fdb add de:ad:be:ef:13:37 dev vx self dst 192.0.2.1
de:ad:be:ef:13:37 dev vx dst 192.0.2.1 self permanent
de:ad:be:ef:13:37 dev vx self permanent
In order to prevent this duplicity, add a paremeter to ndo_fdb_add,
bool *notified. The flag is primed to false, and if the callee sends a
notification on its own, it sets it to true, thus informing the core that
it should not generate another notification.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Nikolay Aleksandrov <razor@blackwall.org>
Link: https://patch.msgid.link/cbf6ae8195e85cbf922f8058ce4eba770f3b71ed.1731589511.git.petrm@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a per-NAPI IRQ suspension parameter, which can be get/set with
netdev-genl.
This patch doesn't change any behavior but prepares the code for other
changes in the following commits which use irq_suspend_timeout as a
timeout for IRQ suspension.
Signed-off-by: Martin Karsten <mkarsten@uwaterloo.ca>
Co-developed-by: Joe Damato <jdamato@fastly.com>
Signed-off-by: Joe Damato <jdamato@fastly.com>
Tested-by: Joe Damato <jdamato@fastly.com>
Tested-by: Martin Karsten <mkarsten@uwaterloo.ca>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Link: https://patch.msgid.link/20241109050245.191288-2-jdamato@fastly.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Create a mapping between a netdev and its neighoburs,
allowing for much cheaper flushes.
Signed-off-by: Gilad Naaman <gnaaman@drivenets.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://patch.msgid.link/20241107160444.2913124-7-gnaaman@drivenets.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The first -next "new features" pull request for v6.13. This is a big
one as we have not been able to send one earlier. We have also some
patches affecting other subsystems: in staging we deleted the rtl8192e
driver and in debugfs added a new interface to save struct
file_operations memory; both were acked by GregKH.
Because of the lib80211/libipw move there were quite a lot of
conflicts and to solve those we decided to merge net-next into
wireless-next.
Currently there's one conflict in
Documentation/networking/net_cachelines/net_device.rst. To fix that
just remove the iw_public_data line:
https://lore.kernel.org/all/20241011121014.674661a0@canb.auug.org.au/
And when net is merged to net-next there will be another simple
conflict in in net/mac80211/cfg.c:
https://lore.kernel.org/all/20241024115523.4cd35dde@canb.auug.org.au/
Major changes:
cfg80211/mac80211
* stop exporting wext symbols
* new mac80211 op to indicate that a new interface is to be added
* support radio separation of multi-band devices
Wireless Extensions
* move wext spy implementation to libiw
* remove iw_public_data from struct net_device
brcmfmac
* optional LPO clock support
ipw2x00
* move remaining lib80211 code into libiw
wilc1000
* WILC3000 support
rtw89
* RTL8852BE and RTL8852BE-VT BT-coexistence improvements
-----BEGIN PGP SIGNATURE-----
iQFFBAABCgAvFiEEiBjanGPFTz4PRfLobhckVSbrbZsFAmcbz9YRHGt2YWxvQGtl
cm5lbC5vcmcACgkQbhckVSbrbZsabQf8CWJ/kyonw/Z8hRxgfE/7D6Jiqoq7R+ML
8W8lbc6F5wra4eCBq/oo6UVV36Ss6mxQYcRcmLq+nCkXa4qdMpg/z55QECMHxx5Z
YnIBbD2vBrIj7W21gfCKH1WJ+b5IQFZl3zuxuCgXjxD9TJM2CjUfOkvrhrqqzrPn
clfUx5f01vfv2jdvClPR5977gFE5One/ANeRQNs7uDS0TeeD2P+61DEB1//htIJo
7GwwCyUJCeOcfWRMzQwhpoppWKcPAV70kSVJrl/fRstS68vQGSQbcx9yiNeWkSFw
JXjQGdc8eYLPzLqECwS0KwFkta6AXbafAYYXe1wdlAzr+kmJ9x5oqA==
=x+mr
-----END PGP SIGNATURE-----
Merge tag 'wireless-next-2024-10-25' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next
Kalle Valo says:
====================
wireless-next patches for v6.13
The first -next "new features" pull request for v6.13. This is a big
one as we have not been able to send one earlier. We have also some
patches affecting other subsystems: in staging we deleted the rtl8192e
driver and in debugfs added a new interface to save struct
file_operations memory; both were acked by GregKH.
Because of the lib80211/libipw move there were quite a lot of
conflicts and to solve those we decided to merge net-next into
wireless-next.
Major changes:
cfg80211/mac80211
* stop exporting wext symbols
* new mac80211 op to indicate that a new interface is to be added
* support radio separation of multi-band devices
Wireless Extensions
* move wext spy implementation to libiw
* remove iw_public_data from struct net_device
brcmfmac
* optional LPO clock support
ipw2x00
* move remaining lib80211 code into libiw
wilc1000
* WILC3000 support
rtw89
* RTL8852BE and RTL8852BE-VT BT-coexistence improvements
* tag 'wireless-next-2024-10-25' of git://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next: (126 commits)
mac80211: Remove NOP call to ieee80211_hw_config
wifi: iwlwifi: work around -Wenum-compare-conditional warning
wifi: mac80211: re-order assigning channel in activate links
wifi: mac80211: convert debugfs files to short fops
debugfs: add small file operations for most files
wifi: mac80211: remove misleading j_0 construction parts
wifi: mac80211_hwsim: use hrtimer_active()
wifi: mac80211: refactor BW limitation check for CSA parsing
wifi: mac80211: filter on monitor interfaces based on configured channel
wifi: mac80211: refactor ieee80211_rx_monitor
wifi: mac80211: add support for the monitor SKIP_TX flag
wifi: cfg80211: add monitor SKIP_TX flag
wifi: mac80211: add flag to opt out of virtual monitor support
wifi: cfg80211: pass net_device to .set_monitor_channel
wifi: mac80211: remove status->ampdu_delimiter_crc
wifi: cfg80211: report per wiphy radio antenna mask
wifi: mac80211: use vif radio mask to limit creating chanctx
wifi: mac80211: use vif radio mask to limit ibss scan frequencies
wifi: cfg80211: add option for vif allowed radios
wifi: iwlwifi: allow IWL_FW_CHECK() with just a string
...
====================
Link: https://patch.msgid.link/20241025170705.5F6B2C4CEC3@smtp.kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
npinfo is not used in any of the ndo_netpoll_setup() methods.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20241018052108.2610827-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Some workloads hit the infamous dev_watchdog() message:
"NETDEV WATCHDOG: eth0 (xxxx): transmit queue XX timed out"
It seems possible to hit this even for perfectly normal
BQL enabled drivers:
1) Assume a TX queue was idle for more than dev->watchdog_timeo
(5 seconds unless changed by the driver)
2) Assume a big packet is sent, exceeding current BQL limit.
3) Driver ndo_start_xmit() puts the packet in TX ring,
and netdev_tx_sent_queue() is called.
4) QUEUE_STATE_STACK_XOFF could be set from netdev_tx_sent_queue()
before txq->trans_start has been written.
5) txq->trans_start is written later, from netdev_start_xmit()
if (rc == NETDEV_TX_OK)
txq_trans_update(txq)
dev_watchdog() running on another cpu could read the old
txq->trans_start, and then see QUEUE_STATE_STACK_XOFF, because 5)
did not happen yet.
To solve the issue, write txq->trans_start right before one XOFF bit
is set :
- _QUEUE_STATE_DRV_XOFF from netif_tx_stop_queue()
- __QUEUE_STATE_STACK_XOFF from netdev_tx_sent_queue()
From dev_watchdog(), we have to read txq->state before txq->trans_start.
Add memory barriers to enforce correct ordering.
In the future, we could avoid writing over txq->trans_start for normal
operations, and rename this field to txq->xoff_start_time.
Fixes: bec251bc8b ("net: no longer stop all TX queues in dev_watchdog()")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: Toke Høiland-Jørgensen <toke@redhat.com>
Link: https://patch.msgid.link/20241015194118.3951657-1-edumazet@google.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add a persistent NAPI config area for NAPI configuration to the core.
Drivers opt-in to setting the persistent config for a NAPI by passing an
index when calling netif_napi_add_config.
napi_config is allocated in alloc_netdev_mqs, freed in free_netdev
(after the NAPIs are deleted).
Drivers which call netif_napi_add_config will have persistent per-NAPI
settings: NAPI IDs, gro_flush_timeout, and defer_hard_irq settings.
Per-NAPI settings are saved in napi_disable and restored in napi_enable.
Co-developed-by: Martin Karsten <mkarsten@uwaterloo.ca>
Signed-off-by: Martin Karsten <mkarsten@uwaterloo.ca>
Signed-off-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20241011184527.16393-6-jdamato@fastly.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Allow per-NAPI gro_flush_timeout setting.
The existing sysfs parameter is respected; writes to sysfs will write to
all NAPI structs for the device and the net_device gro_flush_timeout
field. Reads from sysfs will read from the net_device field.
The ability to set gro_flush_timeout on specific NAPI instances will be
added in a later commit, via netdev-genl.
Signed-off-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://patch.msgid.link/20241011184527.16393-4-jdamato@fastly.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add defer_hard_irqs to napi_struct in preparation for per-NAPI
settings.
The existing sysfs parameter is respected; writes to sysfs will write to
all NAPI structs for the device and the net_device defer_hard_irq field.
Reads from sysfs show the net_device field.
The ability to set defer_hard_irqs on specific NAPI instances will be
added in a later commit, via netdev-genl.
Signed-off-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://patch.msgid.link/20241011184527.16393-2-jdamato@fastly.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Introduce the basic infrastructure to implement the net-shaper
core functionality. Each network devices carries a net-shaper cache,
the NL get() operation fetches the data from such cache.
The cache is initially empty, will be fill by the set()/group()
operation implemented later and is destroyed at device cleanup time.
The net_shaper_fill_handle(), net_shaper_ctx_init(), and
net_shaper_generic_pre() implementations handle generic index type
attributes, despite the current caller always pass a constant value
to avoid more noise in later patches using them with different
attributes.
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Link: https://patch.msgid.link/ddd10fd645a9367803ad02fca4a5664ea5ace170.1728460186.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The wireless-next tree was based on something older, and there
are now conflicts between -rc2 and work here. Merge net-next,
which has enough of -rc2 for the conflicts to happen, resolving
them in the process.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Upcoming per-netns RTNL conversion needs to get rid
of shared hash tables.
fib_info_devhash[] is one of them.
It is unclear why we used a hash table, because
a single hlist_head per net device was cheaper and scalable.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://patch.msgid.link/20241004134720.579244-5-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Commit 24ab059d2e ("net: check dev->gso_max_size in gso_features_check()")
added a dev->gso_max_size test to gso_features_check() in order to fall
back to GSO when needed.
This was added as it was noticed that some drivers could misbehave if TSO
packets get too big. However, the check doesn't respect dev->gso_ipv4_max_size
limit. For instance, a device could be configured with BIG TCP for IPv4,
but not IPv6.
Therefore, add a netif_get_gso_max_size() equivalent to netif_get_gro_max_size()
and use the helper to respect both limits before falling back to GSO engine.
Fixes: 24ab059d2e ("net: check dev->gso_max_size in gso_features_check()")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240923212242.15669-2-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add a small netif_get_gro_max_size() helper which returns the maximum IPv4
or IPv6 GRO size of the netdevice.
We later add a netif_get_gso_max_size() equivalent as well for GSO, so that
these helpers can be used consistently instead of open-coded checks.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240923212242.15669-1-daniel@iogearbox.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Add a netdev_dmabuf_binding struct which represents the
dma-buf-to-netdevice binding. The netlink API will bind the dma-buf to
rx queues on the netdevice. On the binding, the dma_buf_attach
& dma_buf_map_attachment will occur. The entries in the sg_table from
mapping will be inserted into a genpool to make it ready
for allocation.
The chunks in the genpool are owned by a dmabuf_chunk_owner struct which
holds the dma-buf offset of the base of the chunk and the dma_addr of
the chunk. Both are needed to use allocations that come from this chunk.
We create a new type that represents an allocation from the genpool:
net_iov. We setup the net_iov allocation size in the
genpool to PAGE_SIZE for simplicity: to match the PAGE_SIZE normally
allocated by the page pool and given to the drivers.
The user can unbind the dmabuf from the netdevice by closing the netlink
socket that established the binding. We do this so that the binding is
automatically unbound even if the userspace process crashes.
The binding and unbinding leaves an indicator in struct netdev_rx_queue
that the given queue is bound, and the binding is actuated by resetting
the rx queue using the queue API.
The netdev_dmabuf_binding struct is refcounted, and releases its
resources only when all the refs are released.
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Kaiyuan Zhang <kaiyuanz@google.com>
Signed-off-by: Mina Almasry <almasrymina@google.com>
Reviewed-by: Pavel Begunkov <asml.silence@gmail.com> # excluding netlink
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Link: https://patch.msgid.link/20240910171458.219195-4-almasrymina@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
idpf: XDP chapter II: convert Tx completion to libeth
Alexander Lobakin says:
XDP for idpf is currently 5 chapters:
* convert Rx to libeth;
* convert Tx completion to libeth (this);
* generic XDP and XSk code changes;
* actual XDP for idpf via libeth_xdp;
* XSk for idpf (^).
Part II does the following:
* adds generic libeth Tx completion routines;
* converts idpf to use generic libeth Tx comp routines;
* fixes Tx queue timeouts and robustifies Tx completion in general;
* fixes Tx event/descriptor flushes (writebacks).
Most idpf patches again remove more lines than adds.
Generic Tx completion helpers and structs are needed as libeth_xdp
(Ch. III) makes use of them. WB_ON_ITR is needed since XDPSQs don't
want to work without it at all. Tx queue timeouts fixes are needed
since without them, it's way easier to catch a Tx timeout event when
WB_ON_ITR is enabled.
* '200GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
idpf: enable WB_ON_ITR
idpf: fix netdev Tx queue stop/wake
idpf: refactor Tx completion routines
netdevice: add netdev_tx_reset_subqueue() shorthand
idpf: convert to libeth Tx buffer completion
libeth: add Tx buffer completion helpers
====================
Link: https://patch.msgid.link/20240909205323.3110312-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
dev_pick_tx_cpu_id() has been introduced with two users by
commit a4ea8a3dac ("net: Add generic ndo_select_queue functions").
The use in AF_PACKET has been removed in 2019 by
commit b71b5837f8 ("packet: rework packet_pick_tx_queue() to use common code selection")
The other user was a Netlogic XLP driver, removed in 2021 by
commit 47ac6f567c ("staging: Remove Netlogic XLP network driver").
It's relatively unlikely that any modern driver will need an
.ndo_select_queue implementation which picks purely based on CPU ID
and skips XPS, delete dev_pick_tx_cpu_id()
Found by code inspection.
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240906161059.715546-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a shorthand similar to other net*_subqueue() helpers for resetting
the queue by its index w/o obtaining &netdev_tx_queue beforehand
manually.
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In commit 6f8b12d661 ("net: napi: add hard irqs deferral feature")
napi_defer_irqs was added to net_device and napi_defer_irqs_count was
added to napi_struct, both as type int.
This value never goes below zero, so there is not reason for it to be a
signed int. Change the type for both from int to u32, and add an
overflow check to sysfs to limit the value to S32_MAX.
The limit of S32_MAX was chosen because the practical limit before this
patch was S32_MAX (anything larger was an overflow) and thus there are
no behavioral changes introduced. If the extra bit is needed in the
future, the limit can be raised.
Before this patch:
$ sudo bash -c 'echo 2147483649 > /sys/class/net/eth4/napi_defer_hard_irqs'
$ cat /sys/class/net/eth4/napi_defer_hard_irqs
-2147483647
After this patch:
$ sudo bash -c 'echo 2147483649 > /sys/class/net/eth4/napi_defer_hard_irqs'
bash: line 0: echo: write error: Numerical result out of range
Similarly, /sys/class/net/XXXXX/tx_queue_len is defined as unsigned:
include/linux/netdevice.h: unsigned int tx_queue_len;
And has an overflow check:
dev_change_tx_queue_len(..., unsigned long new_len):
if (new_len != (unsigned int)new_len)
return -ERANGE;
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Joe Damato <jdamato@fastly.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240904153431.307932-1-jdamato@fastly.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Ability to handle maximum FCoE frames of 2158 bytes can never be changed
and thus more of an attribute, not a toggleable feature.
Move it from netdev_features_t to "cold" priv flags (bitfield bool) and
free yet another feature bit.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
"Interface can't change network namespaces" is rather an attribute,
not a feature, and it can't be changed via Ethtool.
Make it a "cold" private flag instead of a netdev_feature and free
one more bit.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
NETIF_F_LLTX can't be changed via Ethtool and is not a feature,
rather an attribute, very similar to IFF_NO_QUEUE (and hot).
Free one netdev_features_t bit and make it a "hot" private flag.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Make dev->priv_flags `u32` back and define bits higher than 31 as
bitfield booleans as per Jakub's suggestion. This simplifies code
which accesses these bits with no optimization loss (testb both
before/after), allows to not extend &netdev_priv_flags each time,
but also scales better as bits > 63 in the future would only add
a new u64 to the structure with no complications, comparing to
that extending ::priv_flags would require converting it to a bitmap.
Note that I picked `unsigned long :1` to not lose any potential
optimizations comparing to `bool :1` etc.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
When net devices propagate xdp configurations to slave devices,
we will need to perform a memory provider check to ensure we're
not binding xdp to a device using unreadable netmem.
Currently the ->ndo_bpf calls in a few places. Adding checks to all
these places would not be ideal.
Refactor all the ->ndo_bpf calls into one place where we can add this
check in the future.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Link topologies containing multiple network PHYs attached to the same
net_device can be found when using a PHY as a media converter for use
with an SFP connector, on which an SFP transceiver containing a PHY can
be used.
With the current model, the transceiver's PHY can't be used for
operations such as cable testing, timestamping, macsec offload, etc.
The reason being that most of the logic for these configuration, coming
from either ethtool netlink or ioctls tend to use netdev->phydev, which
in multi-phy systems will reference the PHY closest to the MAC.
Introduce a numbering scheme allowing to enumerate PHY devices that
belong to any netdev, which can in turn allow userspace to take more
precise decisions with regard to each PHY's configuration.
The numbering is maintained per-netdev, in a phy_device_list.
The numbering works similarly to a netdevice's ifindex, with
identifiers that are only recycled once INT_MAX has been reached.
This prevents races that could occur between PHY listing and SFP
transceiver removal/insertion.
The identifiers are assigned at phy_attach time, as the numbering
depends on the netdevice the phy is attached to. The PHY index can be
re-used for PHYs that are persistent.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Tested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Adding the NAPI pointer to struct netdev_queue made it grow into another
cacheline, even though there was 44 bytes of padding available.
The struct was historically grouped as follows:
/* read-mostly stuff (align) */
/* ... random control path fields ... */
/* write-mostly stuff (align) */
/* ... 40 byte hole ... */
/* struct dql (align) */
It seems that people want to add control path fields after
the read only fields. struct dql looks pretty innocent
but it forces its own alignment and nothing indicates that
there is a lot of empty space above it.
Move dql above the xmit_lock. This shifts the empty space
to the end of the struct rather than in the middle of it.
Move two example fields there to set an example.
Hopefully people will now add new fields at the end of
the struct. A lot of the read-only stuff is also control
path-only, but if we move it all we'll have another hole
in the middle.
Before:
/* size: 384, cachelines: 6, members: 16 */
/* sum members: 284, holes: 3, sum holes: 100 */
After:
/* size: 320, cachelines: 5, members: 16 */
/* sum members: 284, holes: 1, sum holes: 8 */
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20240820205119.1321322-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
We re-define values of enum netdev_priv_flags as preprocessor
macros with the same name. I guess this was done to avoid breaking
out of tree modules which may use #ifdef X for kernel compatibility?
Commit 7aa98047df ("net: move net_device priv_flags out from UAPI")
which added the enum doesn't say. In any case, the flags with defines
are quite old now, and defines for new flags don't get added.
OOT drivers have to resort to code greps for compat detection, anyway.
Let's delete these defines, save LoC, help LXR link to the right place.
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Link: https://patch.msgid.link/20240801163401.378723-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In fact, this structure contains a flexible array at the end, but
historically its size, alignment etc., is calculated manually.
There are several instances of the structure embedded into other
structures, but also there's ongoing effort to remove them and we
could in the meantime declare &net_device properly.
Declare the array explicitly, use struct_size() and store the array
size inside the structure, so that __counted_by() can be applied.
Don't use PTR_ALIGN(), as SLUB itself tries its best to ensure the
allocated buffer is aligned to what the user expects.
Also, change its alignment from %NETDEV_ALIGN to the cacheline size
as per several suggestions on the netdev ML.
bloat-o-meter for vmlinux:
free_netdev 445 440 -5
netdev_freemem 24 - -24
alloc_netdev_mqs 1481 1450 -31
On x86_64 with several NICs of different vendors, I was never able to
get a &net_device pointer not aligned to the cacheline size after the
change.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Kees Cook <kees@kernel.org>
Link: https://patch.msgid.link/20240710113036.2125584-1-leitao@debian.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Commit 31e0aa99dc ("ethtool: Veto some operations during firmware flashing process")
added a flag module_fw_flash_in_progress to struct net_device. As
this is ethtool related state, move it to the recently created
struct ethtool_netdev_state, accessed via the 'ethtool' member of
struct net_device.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Edward Cree <ecree.xilinx@gmail.com>
Reviewed-by: Michal Kubiak <michal.kubiak@intel.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Link: https://patch.msgid.link/20240703121849.652893-1-edward.cree@amd.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net_dev->ethtool is a pointer to new struct ethtool_netdev_state, which
currently contains only the wol_enabled field.
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Edward Cree <ecree.xilinx@gmail.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Link: https://patch.msgid.link/293a562278371de7534ed1eb17531838ca090633.1719502239.git.ecree.xilinx@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Some operations cannot be performed during the firmware flashing
process.
For example:
- Port must be down during the whole flashing process to avoid packet loss
while committing reset for example.
- Writing to EEPROM interrupts the flashing process, so operations like
ethtool dump, module reset, get and set power mode should be vetoed.
- Split port firmware flashing should be vetoed.
In order to veto those scenarios, add a flag in 'struct net_device' that
indicates when a firmware flash is taking place on the module and use it
to prevent interruptions during the process.
Signed-off-by: Danielle Ratson <danieller@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
The NetDIM library, currently leveraged by an array of NICs, delivers
excellent acceleration benefits. Nevertheless, NICs vary significantly
in their dim profile list prerequisites.
Specifically, virtio-net backends may present diverse sw or hw device
implementation, making a one-size-fits-all parameter list impractical.
On Alibaba Cloud, the virtio DPU's performance under the default DIM
profile falls short of expectations, partly due to a mismatch in
parameter configuration.
I also noticed that ice/idpf/ena and other NICs have customized
profilelist or placed some restrictions on dim capabilities.
Motivated by this, I tried adding new params for "ethtool -C" that provides
a per-device control to modify and access a device's interrupt parameters.
Usage
========
The target NIC is named ethx.
Assume that ethx only declares support for rx profile setting
(with DIM_PROFILE_RX flag set in profile_flags) and supports modification
of usec and pkt fields.
1. Query the currently customized list of the device
$ ethtool -c ethx
...
rx-profile:
{.usec = 1, .pkts = 256, .comps = n/a,},
{.usec = 8, .pkts = 256, .comps = n/a,},
{.usec = 64, .pkts = 256, .comps = n/a,},
{.usec = 128, .pkts = 256, .comps = n/a,},
{.usec = 256, .pkts = 256, .comps = n/a,}
tx-profile: n/a
2. Tune
$ ethtool -C ethx rx-profile 1,1,n_2,n,n_3,3,n_4,4,n_n,5,n
"n" means do not modify this field.
$ ethtool -c ethx
...
rx-profile:
{.usec = 1, .pkts = 1, .comps = n/a,},
{.usec = 2, .pkts = 256, .comps = n/a,},
{.usec = 3, .pkts = 3, .comps = n/a,},
{.usec = 4, .pkts = 4, .comps = n/a,},
{.usec = 256, .pkts = 5, .comps = n/a,}
tx-profile: n/a
3. Hint
If the device does not support some type of customized dim profiles,
the corresponding "n/a" will display.
If the "n/a" field is being modified, -EOPNOTSUPP will be reported.
Signed-off-by: Heng Qi <hengqi@linux.alibaba.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20240621101353.107425-4-hengqi@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
softnet_data::process_queue 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.
softnet_data::input_queue_head can be updated lockless. This is fine
because this value is only update CPU local by the local backlog_napi
thread.
Add a local_lock_t to softnet_data and use local_lock_nested_bh() for locking
of process_queue. This change adds only lockdep coverage and does not
alter the functional behaviour for !PREEMPT_RT.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20240620132727.660738-11-bigeasy@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Softirq is preemptible on PREEMPT_RT. Without a per-CPU lock in
local_bh_disable() there is no guarantee that only one device is
transmitting at a time.
With preemption and multiple senders it is possible that the per-CPU
`recursion' counter gets incremented by different threads and exceeds
XMIT_RECURSION_LIMIT leading to a false positive recursion alert.
The `more' member is subject to similar problems if set by one thread
for one driver and wrongly used by another driver within another thread.
Instead of adding a lock to protect the per-CPU variable it is simpler
to make xmit per-task. Sending and receiving skbs happens always
in thread context anyway.
Having a lock to protected the per-CPU counter would block/ serialize two
sending threads needlessly. It would also require a recursive lock to
ensure that the owner can increment the counter further.
Make the softnet_data.xmit a task_struct member on PREEMPT_RT. Add
needed wrapper.
Cc: Ben Segall <bsegall@google.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: Juri Lelli <juri.lelli@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Valentin Schneider <vschneid@redhat.com>
Cc: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20240620132727.660738-9-bigeasy@linutronix.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This declaration was added to the header to be called from ethtool.
ethtool is separated from core for code organization but it is not really
a separate entity, it controls very core things.
As ethtool is an internal stuff it is not wise to have it in netdevice.h.
Move the declaration to net/core/dev.h instead.
Remove the EXPORT_SYMBOL_GPL call as ethtool can not be built as a module.
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Link: https://lore.kernel.org/r/20240612-feature_ptp_netnext-v15-2-b2a086257b63@bootlin.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
I find the behavior of xa_for_each_start() slightly counter-intuitive.
It doesn't end the iteration by making the index point after the last
element. IOW calling xa_for_each_start() again after it "finished"
will run the body of the loop for the last valid element, instead
of doing nothing.
This works fine for netlink dumps if they terminate correctly
(i.e. coalesce or carefully handle NLM_DONE), but as we keep getting
reminded legacy dumps are unlikely to go away.
Fixing this generically at the xa_for_each_start() level seems hard -
there is no index reserved for "end of iteration".
ifindexes are 31b wide, tho, and iterator is ulong so for
for_each_netdev_dump() it's safe to go to the next element.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The pcpu_sw_netstats and pcpu_lstats structs both contain a set of
u64_stats_t fields for individual stats, but pcpu_dstats uses u64s
instead.
Make this consistent by using u64_stats_t across all stats types.
The per-cpu dstats are only used by the vrf driver at present, so update
that driver as part of this change.
Signed-off-by: Jeremy Kerr <jk@codeconstruct.com.au>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: David Ahern <dsahern@kernel.org>
Link: https://lore.kernel.org/r/20240607-dstats-v3-1-cc781fe116f7@codeconstruct.com.au
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This API enables the net stack to reset the queues used for devmem TCP.
Signed-off-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: Shailend Chand <shailend@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We will convert ioctl(SIOCGARP) to RCU, and then we need to copy
dev->name which is currently protected by rtnl_lock().
This patch does the following:
1) Add seqlock netdev_rename_lock to protect dev->name
2) Add netdev_copy_name() that copies dev->name to buffer
under netdev_rename_lock
3) Use netdev_copy_name() in netdev_get_name() and drop
devnet_rename_sem
Suggested-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/netdev/CANn89iJEWs7AYSJqGCUABeVqOCTkErponfZdT5kV-iD=-SajnQ@mail.gmail.com/
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Link: https://lore.kernel.org/r/20240430015813.71143-7-kuniyu@amazon.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
It is impossible to use init_dummy_netdev together with alloc_netdev()
as the 'setup' argument.
This is because alloc_netdev() initializes some fields in the net_device
structure, and later init_dummy_netdev() memzero them all. This causes
some problems as reported here:
https://lore.kernel.org/all/20240322082336.49f110cc@kernel.org/
Split the init_dummy_netdev() function in two. Create a new function called
init_dummy_netdev_core() that does not memzero the net_device structure.
Then have init_dummy_netdev() memzero-ing and calling
init_dummy_netdev_core(), keeping the old behaviour.
init_dummy_netdev_core() is the new function that could be called as an
argument for alloc_netdev().
Also, create a helper to allocate and initialize dummy net devices,
leveraging init_dummy_netdev_core() as the setup argument. This function
basically simplify the allocation of dummy devices, by allocating and
initializing it. Freeing the device continue to be done through
free_netdev()
Suggested-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Link topologies containing multiple network PHYs attached to the same
net_device can be found when using a PHY as a media converter for use
with an SFP connector, on which an SFP transceiver containing a PHY can
be used.
With the current model, the transceiver's PHY can't be used for
operations such as cable testing, timestamping, macsec offload, etc.
The reason being that most of the logic for these configuration, coming
from either ethtool netlink or ioctls tend to use netdev->phydev, which
in multi-phy systems will reference the PHY closest to the MAC.
Introduce a numbering scheme allowing to enumerate PHY devices that
belong to any netdev, which can in turn allow userspace to take more
precise decisions with regard to each PHY's configuration.
The numbering is maintained per-netdev, in a phy_device_list.
The numbering works similarly to a netdevice's ifindex, with
identifiers that are only recycled once INT_MAX has been reached.
This prevents races that could occur between PHY listing and SFP
transceiver removal/insertion.
The identifiers are assigned at phy_attach time, as the numbering
depends on the netdevice the phy is attached to. The PHY index can be
re-used for PHYs that are persistent.
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
For short netdev holds within a function there are still a lot of
users of dev_put() rather than netdev_put(). Add DEFINE_FREE() to
allow making those safer.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 14d898f3c1 ("dev: Move received_rps counter next
to RPS members in softnet data") was unfortunate:
received_rps is dirtied by a cpu and never read by other
cpus in fast path.
Its presence in the hot RPS cache line (shared by many cpus)
is hurting RPS/RFS performance.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
input_queue_tail_incr_save() is incrementing the sd queue_tail
and save it in the flow last_qtail.
Two issues here :
- no lock protects the write on last_qtail, we should use appropriate
annotations.
- We can perform this write after releasing the per-cpu backlog lock,
to decrease this lock hold duration (move away the cache line miss)
Also move input_queue_head_incr() and rps helpers to include/net/rps.h,
while adding rps_ prefix to better reflect their role.
v2: Fixed a build issue (Jakub and kernel build bots)
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If under extreme cpu backlog pressure enqueue_to_backlog() has
to drop a packet, it could do this without dirtying a cache line
and potentially slowing down the target cpu.
Move sd->dropped into a separate cache line, and make it atomic.
In non pressure mode, this field is not touched, no need to consume
valuable space in a hot cache line.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Move dev_xmit_recursion() and friends to net/core/dev.h
They are only used from net/core/dev.c and net/core/filter.c.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
kick_defer_list_purge() is defined in net/core/dev.c
and used from net/core/skubff.c
Because we need softnet_data, include <linux/netdevice.h>
from net/core/dev.h
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Unlike IPv6 tunnels which use purely-kernel __ip6_tnl_parm structure
to store params inside the kernel, IPv4 tunnel code uses the same
ip_tunnel_parm which is being used to talk with the userspace.
This makes it difficult to alter or add any fields or use a
different format for whatever data.
Define struct ip_tunnel_parm_kern, a 1:1 copy of ip_tunnel_parm for
now, and use it throughout the code. Define the pieces, where the copy
user <-> kernel happens, as standalone functions, and copy the data
there field-by-field, so that the kernel-side structure could be easily
modified later on and the users wouldn't have to care about this.
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The defer_list is a per-CPU list which is used to free skbs outside of
the socket lock and on the CPU on which they have been allocated.
The list is processed during NAPI callbacks so ideally the list is
cleaned up.
Should the amount of skbs on the list exceed a certain water mark then
the softirq is triggered remotely on the target CPU by invoking a remote
function call. The raise of the softirqs via a remote function call
leads to waking the ksoftirqd on PREEMPT_RT which is undesired.
The backlog-NAPI threads already provide the infrastructure which can be
utilized to perform the cleanup of the defer_list.
The NAPI state is updated with the input_pkt_queue.lock acquired. It
order not to break the state, it is needed to also wake the backlog-NAPI
thread with the lock held. This requires to acquire the use the lock in
rps_lock_irq*() if the backlog-NAPI threads are used even with RPS
disabled.
Move the logic of remotely starting softirqs to clean up the defer_list
into kick_defer_list_purge(). Make sure a lock is held in
rps_lock_irq*() if backlog-NAPI threads are used. Schedule backlog-NAPI
for defer_list cleanup if backlog-NAPI is available.
Acked-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stanislav Fomichev