VNIC testing on multi-core Power systems showed SAR stats drift
and packet rate inconsistencies under load.
Implements ndo_get_stats64 to provide safe aggregation of queue-level
atomic64 counters into rtnl_link_stats64 for use by tools like 'ip -s',
'ifconfig', and 'sar'. Switch to ndo_get_stats64 to align SAR reporting
with the standard kernel interface for retrieving netdev stats.
This removes redundant per-adapter stat updates, reduces overhead,
eliminates cacheline bouncing from hot path updates, and improves
the accuracy of reported packet rates.
Signed-off-by: Mingming Cao <mmc@linux.ibm.com>
Reviewed-by: Brian King <bjking1@linux.ibm.com>
Reviewed-by: Dave Marquardt <davemarq@linux.ibm.com>
Reviewed-by: Simon Horman <horms@kernel.org>
----
Changes since v3:
link to v3: https://www.spinics.net/lists/netdev/msg1107999.html
-- keep per queue counters as u64 (this patch) and drop off patch 1 in v3
Link: https://patch.msgid.link/20250716152115.61143-1-mmc@linux.ibm.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
qdisc_sleeping variable is declared as "struct Qdisc __rcu" and
as such needs proper annotation while accessing it.
Without rtnl_dereference(), the following error is generated by sparse:
drivers/net/ethernet/mellanox/mlx5/core/en/qos.c:377:40: warning:
incorrect type in initializer (different address spaces)
drivers/net/ethernet/mellanox/mlx5/core/en/qos.c:377:40: expected
struct Qdisc *qdisc
drivers/net/ethernet/mellanox/mlx5/core/en/qos.c:377:40: got struct
Qdisc [noderef] __rcu *qdisc_sleeping
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Link: https://patch.msgid.link/1752675472-201445-4-git-send-email-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add PHY counter functionality to the shared library. The implementation
is identical for the current QCA807X and QCA808X PHYs.
The PHY counter can be configured to perform CRC checking for both received
and transmitted packets. Additionally, the packet counter can be set to
automatically clear after it is read.
The PHY counter includes 32-bit packet counters for both RX (received) and
TX (transmitted) packets, as well as 16-bit counters for recording CRC
error packets for both RX and TX.
Signed-off-by: Luo Jie <quic_luoj@quicinc.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20250715-qcom_phy_counter-v3-1-8b0e460a527b@quicinc.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
net_iovs should have the dma address set to 0 so that
netmem_dma_unmap_page_attrs() correctly skips the unmap. This was
not done in mlx5 when support for devmem tx was added and resulted
in the splat below when the platform iommu was enabled.
This patch addresses the issue by using netmem_dma_unmap_addr_set()
which handles the net_iov case when setting the dma address. A small
refactoring of mlx5e_dma_push() was required to be able to use this API.
The function was split in two versions and each version called
accordingly. Note that netmem_dma_unmap_addr_set() introduces an
additional if case.
Splat:
WARNING: CPU: 14 PID: 2587 at drivers/iommu/dma-iommu.c:1228 iommu_dma_unmap_page+0x7d/0x90
Modules linked in: [...]
Unloaded tainted modules: i10nm_edac(E):1 fjes(E):1
CPU: 14 UID: 0 PID: 2587 Comm: ncdevmem Tainted: G S E 6.15.0+ #3 PREEMPT(voluntary)
Tainted: [S]=CPU_OUT_OF_SPEC, [E]=UNSIGNED_MODULE
Hardware name: HPE ProLiant DL380 Gen10 Plus/ProLiant DL380 Gen10 Plus, BIOS U46 06/01/2022
RIP: 0010:iommu_dma_unmap_page+0x7d/0x90
Code: [...]
RSP: 0000:ff6b1e3ea0b2fc58 EFLAGS: 00010246
RAX: 0000000000000000 RBX: ff46ef2d0a2340c8 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000001
RBP: 0000000000000001 R08: 0000000000000000 R09: ffffffff8827a120
R10: 0000000000000000 R11: 0000000000000000 R12: 00000000d8000000
R13: 0000000000000008 R14: 0000000000000001 R15: 0000000000000000
FS: 00007feb69adf740(0000) GS:ff46ef2c779f1000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007feb69cca000 CR3: 0000000154b97006 CR4: 0000000000773ef0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
PKRU: 55555554
Call Trace:
<TASK>
dma_unmap_page_attrs+0x227/0x250
mlx5e_poll_tx_cq+0x163/0x510 [mlx5_core]
mlx5e_napi_poll+0x94/0x720 [mlx5_core]
__napi_poll+0x28/0x1f0
net_rx_action+0x33a/0x420
? mlx5e_completion_event+0x3d/0x40 [mlx5_core]
handle_softirqs+0xe8/0x2f0
__irq_exit_rcu+0xcd/0xf0
common_interrupt+0x47/0xa0
asm_common_interrupt+0x26/0x40
RIP: 0033:0x7feb69cd08ec
Code: [...]
RSP: 002b:00007ffc01b8c880 EFLAGS: 00000246
RAX: 00000000c3a60cf7 RBX: 0000000000045e12 RCX: 000000000000000e
RDX: 00000000000035b4 RSI: 0000000000000000 RDI: 00007ffc01b8c8c0
RBP: 00007ffc01b8c8b0 R08: 0000000000000000 R09: 0000000000000064
R10: 00007ffc01b8c8c0 R11: 0000000000000000 R12: 00007feb69cca000
R13: 00007ffc01b90e48 R14: 0000000000427e18 R15: 00007feb69d07000
</TASK>
Cc: Mina Almasry <almasrymina@google.com>
Reported-by: Stanislav Fomichev <stfomichev@gmail.com>
Closes: https://lore.kernel.org/all/aFM6r9kFHeTdj-25@mini-arch/
Fixes: 5a842c288c ("net/mlx5e: Add TX support for netmems")
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Reviewed-by: Carolina Jubran <cjubran@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Acked-by: Stanislav Fomichev <sdf@fomichev.me>
Link: https://patch.msgid.link/1752649242-147678-1-git-send-email-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Antonio Quartulli says:
====================
This bugfix batch includes the following changes:
* properly propagate sk mark to skb->mark field
* reject unexpected incoming netlink attributes
* reset GSO state when moving skb from transport to tunnel layer
* tag 'ovpn-net-20250716' of https://github.com/OpenVPN/ovpn-net-next:
ovpn: reset GSO metadata after decapsulation
ovpn: reject unexpected netlink attributes
ovpn: propagate socket mark to skb in UDP
====================
Link: https://patch.msgid.link/20250716115443.16763-1-antonio@openvpn.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The deadlock appears in a stack trace like:
virtnet_probe()
rtnl_lock()
virtio_config_changed_work()
netdev_notify_peers()
rtnl_lock()
It happens if the VMM sends a VIRTIO_NET_S_ANNOUNCE request while the
virtio-net driver is still probing.
The config_work in probe() will get scheduled until virtnet_open() enables
the config change notification via virtio_config_driver_enable().
Fixes: df28de7b00 ("virtio-net: synchronize operstate with admin state on up/down")
Signed-off-by: Zigit Zo <zuozhijie@bytedance.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Link: https://patch.msgid.link/20250716115717.1472430-1-zuozhijie@bytedance.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The per-CPU variable ppp::xmit_recursion is protecting against recursion
due to wrong configuration of the ppp unit. The per-CPU variable
relies on disabled BH for its locking. Without per-CPU locking in
local_bh_disable() on PREEMPT_RT this data structure requires explicit
locking.
The ppp::xmit_recursion is used as a per-CPU boolean. The counter is
checked early in the send routing and the transmit path is only entered
if the counter is zero. Then the counter is incremented to avoid
recursion. It used to detect recursion on channel::downl and
ppp::wlock.
Create a struct ppp_xmit_recursion and move the counter into it.
Add local_lock_t to the struct and use local_lock_nested_bh() for
locking. Due to possible nesting, the lock cannot be acquired
unconditionally but it requires an owner field to identify recursion
before attempting to acquire the lock.
The counter is incremented and checked only after the lock is acquired.
Since it functions as a boolean rather than a count, and its role is now
superseded by the owner field, it can be safely removed.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20250715150806.700536-2-bigeasy@linutronix.de
Reviewed-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Johannes Berg says:
====================
Another set of changes, notably:
- cfg80211: fix double-free introduced earlier
- mac80211: fix RCU iteration in CSA
- iwlwifi: many cleanups (unused FW APIs, PCIe code, WoWLAN)
- mac80211: some work around how FIPS affects wifi, which was
wrong (RC4 is used by TKIP, not only WEP)
- cfg/mac80211: improvements for unsolicated probe response
handling
* tag 'wireless-next-2025-07-17' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless-next: (64 commits)
wifi: cfg80211: fix double free for link_sinfo in nl80211_station_dump()
wifi: cfg80211: fix off channel operation allowed check for MLO
wifi: mac80211: use RCU-safe iteration in ieee80211_csa_finish
wifi: mac80211_hwsim: Update comments in header
wifi: mac80211: parse unsolicited broadcast probe response data
wifi: cfg80211: parse attribute to update unsolicited probe response template
wifi: mac80211: don't use TPE data from assoc response
wifi: mac80211: handle WLAN_HT_ACTION_NOTIFY_CHANWIDTH async
wifi: mac80211: simplify __ieee80211_rx_h_amsdu() loop
wifi: mac80211: don't mark keys for inactive links as uploaded
wifi: mac80211: only assign chanctx in reconfig
wifi: mac80211_hwsim: Declare support for AP scanning
wifi: mac80211: clean up cipher suite handling
wifi: mac80211: don't send keys to driver when fips_enabled
wifi: cfg80211: Fix interface type validation
wifi: mac80211: remove ieee80211_link_unreserve_chanctx() return value
wifi: mac80211: don't unreserve never reserved chanctx
mwl8k: Add missing check after DMA map
wifi: mac80211: make VHT opmode NSS ignore a debug message
wifi: iwlwifi: remove support of several iwl_ppag_table_cmd versions
...
====================
Link: https://patch.msgid.link/20250717094610.20106-47-johannes@sipsolutions.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Johannes Berg says:
====================
Couple of fixes:
- ath12k performance regression from -rc1
- cfg80211 counted_by() removal for scan request
as it doesn't match usage and keeps complaining
- iwlwifi crash with certain older devices
- iwlwifi missing an error path unlock
- iwlwifi compatibility with certain BIOS updates
* tag 'wireless-2025-07-17' of https://git.kernel.org/pub/scm/linux/kernel/git/wireless/wireless:
wifi: iwlwifi: Fix botched indexing conversion
wifi: cfg80211: remove scan request n_channels counted_by
wifi: ath12k: Fix packets received in WBM error ring with REO LUT enabled
wifi: iwlwifi: mask reserved bits in chan_state_active_bitmap
wifi: iwlwifi: pcie: fix locking on invalid TOP reset
====================
Link: https://patch.msgid.link/20250717091831.18787-5-johannes@sipsolutions.net
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tony Nguyen says:
====================
Add RDMA support for Intel IPU E2000 in idpf
Tatyana Nikolova says:
This idpf patch series is the second part of the staged submission for
introducing RDMA RoCEv2 support for the IPU E2000 line of products,
referred to as GEN3.
To support RDMA GEN3 devices, the idpf driver uses common definitions
of the IIDC interface and implements specific device functionality in
iidc_rdma_idpf.h.
The IPU model can host one or more logical network endpoints called
vPorts per PCI function that are flexibly associated with a physical
port or an internal communication port.
Other features as it pertains to GEN3 devices include:
* MMIO learning
* RDMA capability negotiation
* RDMA vectors discovery between idpf and control plane
These patches are split from the submission "Add RDMA support for Intel
IPU E2000 (GEN3)" [1]. The patches have been tested on a range of hosts
and platforms with a variety of general RDMA applications which include
standalone verbs (rping, perftest, etc.), storage and HPC applications.
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
[1] https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
This idpf patch series is the second part of the staged submission for
introducing RDMA RoCEv2 support for the IPU E2000 line of products,
referred to as GEN3.
To support RDMA GEN3 devices, the idpf driver uses common definitions
of the IIDC interface and implements specific device functionality in
iidc_rdma_idpf.h.
The IPU model can host one or more logical network endpoints called
vPorts per PCI function that are flexibly associated with a physical
port or an internal communication port.
Other features as it pertains to GEN3 devices include:
* MMIO learning
* RDMA capability negotiation
* RDMA vectors discovery between idpf and control plane
These patches are split from the submission "Add RDMA support for Intel
IPU E2000 (GEN3)" [1]. The patches have been tested on a range of hosts
and platforms with a variety of general RDMA applications which include
standalone verbs (rping, perftest, etc.), storage and HPC applications.
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
[1] https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
IWL reviews:
v3: https://lore.kernel.org/all/20250708210554.1662-1-tatyana.e.nikolova@intel.com/
v2: https://lore.kernel.org/all/20250612220002.1120-1-tatyana.e.nikolova@intel.com/
v1 (split from previous series):
https://lore.kernel.org/all/20250523170435.668-1-tatyana.e.nikolova@intel.com/
v3: https://lore.kernel.org/all/20250207194931.1569-1-tatyana.e.nikolova@intel.com/
RFC v2: https://lore.kernel.org/all/20240824031924.421-1-tatyana.e.nikolova@intel.com/
RFC: https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/linux:
idpf: implement get LAN MMIO memory regions
idpf: implement IDC vport aux driver MTU change handler
idpf: implement remaining IDC RDMA core callbacks and handlers
idpf: implement RDMA vport auxiliary dev create, init, and destroy
idpf: implement core RDMA auxiliary dev create, init, and destroy
idpf: use reserved RDMA vectors from control plane
====================
Link: https://patch.msgid.link/20250714181002.2865694-1-anthony.l.nguyen@intel.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Implement the PCIe Congestion Event notifier which triggers a work item
to query the PCIe Congestion Event object. The result of the congestion
state is reflected in the new ethtool stats:
* pci_bw_inbound_high: the device has crossed the high threshold for
inbound PCIe traffic.
* pci_bw_inbound_low: the device has crossed the low threshold for
inbound PCIe traffic
* pci_bw_outbound_high: the device has crossed the high threshold for
outbound PCIe traffic.
* pci_bw_outbound_low: the device has crossed the low threshold for
outbound PCIe traffic
The high and low thresholds are currently configured at 90% and 75%.
These are hysteresis thresholds which help to check if the
PCI bus on the device side is in a congested state.
If low + 1 = high then the device is in a congested state. If low == high
then the device is not in a congested state.
The counters are also documented.
A follow-up patch will make the thresholds configurable.
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/1752589821-145787-3-git-send-email-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add initial infrastructure to create and destroy the PCIe Congestion
Event object if the object is supported.
The verb for the object creation function is "set" instead of
"create" because the function will accommodate the modify operation
as well in a subsequent patch.
The next patches will hook it up to the event handler and will add
actual functionality.
Signed-off-by: Dragos Tatulea <dtatulea@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://patch.msgid.link/1752589821-145787-2-git-send-email-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tony Nguyen says:
====================
Intel Wired LAN Driver Updates 2025-07-15 (ixgbe, fm10k, i40e, ice)
Arnd Bergmann resolves compile issues with large NR_CPUS for ixgbe, fm10k,
and i40e.
For ice:
Dave adds a NULL check for LAG netdev.
Michal corrects a pointer check in debugfs initialization.
* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
ice: check correct pointer in fwlog debugfs
ice: add NULL check in eswitch lag check
ethernet: intel: fix building with large NR_CPUS
====================
Link: https://patch.msgid.link/20250715202948.3841437-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
np->name was being used after calling of_node_put(np), which
releases the node and can lead to a use-after-free bug.
Previously, of_node_put(np) was called unconditionally after
of_find_device_by_node(np), which could result in a use-after-free if
pdev is NULL.
This patch moves of_node_put(np) after the error check to ensure
the node is only released after both the error and success cases
are handled appropriately, preventing potential resource issues.
Fixes: 23290c7bc1 ("net: airoha: Introduce Airoha NPU support")
Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://patch.msgid.link/20250715143102.3458286-1-alok.a.tiwari@oracle.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
gso_size is expected by the networking stack to be the size of the
payload (thus, not including ethernet/IP/TCP-headers). However, cqe_bcnt
is the full sized frame (including the headers). Dividing cqe_bcnt by
lro_num_seg will then give incorrect results.
For example, running a bpftrace higher up in the TCP-stack
(tcp_event_data_recv), we commonly have gso_size set to 1450 or 1451 even
though in reality the payload was only 1448 bytes.
This can have unintended consequences:
- In tcp_measure_rcv_mss() len will be for example 1450, but. rcv_mss
will be 1448 (because tp->advmss is 1448). Thus, we will always
recompute scaling_ratio each time an LRO-packet is received.
- In tcp_gro_receive(), it will interfere with the decision whether or
not to flush and thus potentially result in less gro'ed packets.
So, we need to discount the protocol headers from cqe_bcnt so we can
actually divide the payload by lro_num_seg to get the real gso_size.
v2:
- Use "(unsigned char *)tcp + tcp->doff * 4 - skb->data)" to compute header-len
(Tariq Toukan <tariqt@nvidia.com>)
- Improve commit-message (Gal Pressman <gal@nvidia.com>)
Fixes: e586b3b0ba ("net/mlx5: Ethernet Datapath files")
Signed-off-by: Christoph Paasch <cpaasch@openai.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Link: https://patch.msgid.link/20250715-cpaasch-pf-925-investigate-incorrect-gso_size-on-cx-7-nic-v2-1-e06c3475f3ac@openai.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The ovpn_netdev_write() function is responsible for injecting
decapsulated and decrypted packets back into the local network stack.
Prior to this patch, the skb could retain GSO metadata from the outer,
encrypted tunnel packet. This original GSO metadata, relevant to the
sender's transport context, becomes invalid and misleading for the
tunnel/data path once the inner packet is exposed.
Leaving this stale metadata intact causes internal GSO validation checks
further down the kernel's network stack (validate_xmit_skb()) to fail,
leading to packet drops. The reasons for these failures vary by
protocol, for example:
- for ICMP, no offload handler is registered;
- for TCP and UDP, the respective offload handlers return errors when
comparing skb->len to the outdated skb_shinfo(skb)->gso_size.
By calling skb_gso_reset(skb) we ensure the inner packet is presented to
gro_cells_receive() with a clean state, correctly indicating it is an
individual packet from the perspective of the local stack.
This change eliminates the "Driver has suspect GRO implementation, TCP
performance may be compromised" warning and improves overall TCP
performance by allowing GSO/GRO to function as intended on the
decapsulated traffic.
Fixes: 11851cbd60 ("ovpn: implement TCP transport")
Reported-by: Gert Doering <gert@greenie.muc.de>
Closes: https://github.com/OpenVPN/ovpn-net-next/issues/4
Tested-by: Gert Doering <gert@greenie.muc.de>
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
Netlink ops do not expect all attributes to be always set, however
this condition is not explicitly coded any where, leading the user
to believe that all sent attributes are somewhat processed.
Fix this behaviour by introducing explicit checks.
For CMD_OVPN_PEER_GET and CMD_OVPN_KEY_GET directly open-code the
needed condition in the related ops handlers.
While for all other ops use attribute subsets in the ovpn.yaml spec file.
Fixes: b7a63391aa ("ovpn: add basic netlink support")
Reported-by: Ralf Lici <ralf@mandelbit.com>
Closes: https://github.com/OpenVPN/ovpn-net-next/issues/19
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
OpenVPN allows users to configure a FW mark on sockets used to
communicate with other peers. The mark is set by means of the
`SO_MARK` Linux socket option.
However, in the ovpn UDP code path, the socket's `sk_mark` value is
currently ignored and it is not propagated to outgoing `skbs`.
This commit ensures proper inheritance of the field by setting
`skb->mark` to `sk->sk_mark` before handing the `skb` to the network
stack for transmission.
Fixes: 08857b5ec5 ("ovpn: implement basic TX path (UDP)")
Reviewed-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: Ralf Lici <ralf@mandelbit.com>
Link: https://www.mail-archive.com/openvpn-devel@lists.sourceforge.net/msg31877.html
Signed-off-by: Antonio Quartulli <antonio@openvpn.net>
The wx_rx_buffer structure contained two DMA address fields: 'dma' and
'page_dma'. However, only 'page_dma' was actually initialized and used
to program the Rx descriptor. But 'dma' was uninitialized and used in
some paths.
This could lead to undefined behavior, including DMA errors or
use-after-free, if the uninitialized 'dma' was used. Althrough such
error has not yet occurred, it is worth fixing in the code.
Fixes: 3c47e8ae11 ("net: libwx: Support to receive packets in NAPI")
Cc: stable@vger.kernel.org
Signed-off-by: Jiawen Wu <jiawenwu@trustnetic.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Link: https://patch.msgid.link/20250714024755.17512-3-jiawenwu@trustnetic.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
