The new net.mptcp.pm_type sysctl determines which path manager will be
used by each newly-created MPTCP socket.

v2: Handle builds without CONFIG_SYSCTL
v3: Clarify logic for type-specific PM init (Geliang Tang and Paolo Abeni)
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Userspace-managed sockets should not have their subflows or
advertisements changed by the kernel path manager.

v3: Use helper function for PM mode (Paolo Abeni)
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

When a MPTCP connection is managed by a userspace PM, bypass the kernel
PM for incoming advertisements and subflow events. Netlink events are
still sent to userspace.

v2: Remove unneeded check in mptcp_pm_rm_addr_received() (Kishen Maloor)
v3: Add and use helper function for PM mode (Paolo Abeni)
Acked-by: Paolo Abeni <pabeni@redhat.com>
Co-developed-by: Kishen Maloor <kishen.maloor@intel.com>
Signed-off-by: Kishen Maloor <kishen.maloor@intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

When adding support for netlink path management commands, the kernel
needs to know whether paths are being controlled by the in-kernel path
manager or a userspace PM.
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

A few members of the mptcp_pm_data struct were assigned to hard-coded
values in mptcp_pm_data_reset(), and then immediately changed in
mptcp_pm_nl_data_init().

Instead, flatten all the assignments in to mptcp_pm_data_reset().

v2: Resolve conflicts due to rename of mptcp_pm_data_reset()
v4: Resolve conflict in mptcp_pm_data_reset()
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Michael Walle says:

====================
net: phy: micrel: add coma mode support

Add support to disable coma mode by a GPIO line.
====================

Link: https://lore.kernel.org/r/20220427214406.1348872-1-michael@walle.ccSigned-off-by: Jakub Kicinski <kuba@kernel.org>

The LAN8814 has a coma mode pin which puts the PHY into isolate and
power-dowm mode. Unfortunately, the mode cannot be disabled by a
register. Usually, the input pin has a pull-up and connected to a GPIO
which can then be used to disable the mode. Try to get the GPIO and
deassert it.
Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Both lan8814_ptp_init() and lan8814_ptp_probe_once() are only used if
PTP and PHY timestamping is enabed. Up until now the probe function just
returns early, if they are not needed. But we need the
phy_package_init_once() functionality for the coma mode GPIO setup. Move
the check into the functions itself.
Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

The LAN8814 has a coma mode pin which is used to put the PHY into
isolate and power-down mode. Usually strapped to be asserted by default.
A GPIO is then used to take the PHY out of this mode.
Signed-off-by: Michael Walle <michael@walle.cc>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Merge branch 'remove-NAPI_POLL_WEIGHT-copies'

Jakub Kicinski says:

====================
remove copies of the NAPI_POLL_WEIGHT define

netif_napi_add() takes weight as the last argument. The value of
that parameter is hard to come up with and depends on many factors,
so driver authors are encouraged to use NAPI_POLL_WEIGHT.

We should probably move weight to an "advanced" version of the API
(__netif_napi_add()?) and simplify the life of most driver authors.

In preparation for such API changes this series removes local
defines equivalent to NAPI_POLL_WEIGHT from drivers, so that a simple
coccinelle / spatch script does not get thrown off by them.

v2:
 - drop staging bits (patch 2)
 - fix subject (patch 8)
 - add qeth change (patch 15)
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Acked-by: Alexandra Winter <wintera@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Acked-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Acked-by: Lino Sanfilippo <LinoSanfilippo@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Defining local versions of NAPI_POLL_WEIGHT with the same
values in the drivers just makes refactoring harder.

Drop the special defines in a bunch of drivers where the
removal is relatively simple so grouping into one patch
does not impact reviewability.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Paul Durrant <paul@xen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The mv88e6xxx driver expects switches that are configured in single chip
addressing mode to have the MDIO address configured as 0. This is due to
the switch ADDR pins representing the single chip addressing mode as 0.
However depending on the device (e.g. MV88E6*41) the switch does not
respond on address 0 or any other address below 16 (the first port
address) in single chip addressing mode. This allows for other devices
to be on the same shared MDIO bus despite the switch being in single
chip addressing mode.

When using a switch that works this way it is not possible to configure
switch driver as single chip addressing via device tree, along with
another MDIO device on the same bus with address 0, as both devices
would have the same address of 0 resulting in mdiobus_register_device
-EBUSY errors for one of the devices with address 0.

In order to support this configuration the switch node can have its MDIO
address configured as 16 (the first address that the device responds
to). During initialization the driver will treat this address similar to
how address 0 is, however because this address is also a valid
multi-chip address (in certain switch models, but not all) the driver
will configure the SMI in single chip addressing mode and attempt to
detect the switch model. If the device is configured in single chip
addressing mode this will succeed and the initialization process can
continue. If it fails to detect a valid model this is because the switch
model register is not a valid register when in multi-chip mode, it will
then fall back to the existing SMI initialization process using the MDIO
address as the multi-chip mode address.

This detection method is safe if the device is in either mode because
the single chip addressing mode read is a direct SMI/MDIO read operation
and has no side effects compared to the SMI writes required for the
multi-chip addressing mode.

In order to implement this change, the reset gpio configuration is moved
to occur before any SMI initialization. This ensures that the device has
the same/correct reset gpio state for both mv88e6xxx_smi_init calls.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220427130928.540007-1-nathan@nathanrossi.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

- Add HW api to configure policer:
  - SR TCM policer mode is only supported for now.
  - Policer ingress/egress direction support.
- Add police action support into flower
Signed-off-by: Volodymyr Mytnyk <volodymyr.mytnyk@plvision.eu>
Link: https://lore.kernel.org/r/1651061148-21321-1-git-send-email-volodymyr.mytnyk@plvision.euSigned-off-by: Jakub Kicinski <kuba@kernel.org>

phy_attach_direct() first calls phy_init_hw() (which restores interrupt
settings through ->config_intr()), then calls phy_disable_interrupts().

So if phydev->interrupts was previously set to 1, interrupts are briefly
enabled, then disabled, which seems nonsensical.

If it was previously set to 0, interrupts are disabled twice, which is
equally nonsensical.

Deduplicate interrupt disablement.
Signed-off-by: Lukas Wunner <lukas@wunner.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/805ccdc606bd8898d59931bd4c7c68537ed6e550.1651040826.git.lukas@wunner.deSigned-off-by: Jakub Kicinski <kuba@kernel.org>

Adding a new socket option, SO_RCVMARK, to indicate that SO_MARK
should be included in the ancillary data returned by recvmsg().

Renamed the sock_recv_ts_and_drops() function to sock_recv_cmsgs().
Signed-off-by: Erin MacNeil <lnx.erin@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Link: https://lore.kernel.org/r/20220427200259.2564-1-lnx.erin@gmail.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

include/linux/netdevice.h
net/core/dev.c
  6510ea97 ("net: Use this_cpu_inc() to increment net->core_stats")
  794c24e9 ("net-core: rx_otherhost_dropped to core_stats")
https://lore.kernel.org/all/20220428111903.5f4304e0@canb.auug.org.au/

drivers/net/wan/cosa.c
  d48fea84 ("net: cosa: fix error check return value of register_chrdev()")
  89fbca33 ("net: wan: remove support for COSA and SRP synchronous serial boards")
https://lore.kernel.org/all/20220428112130.1f689e5e@canb.auug.org.au/Signed-off-by: Jakub Kicinski <kuba@kernel.org>

Pull networking fixes from Jakub Kicinski:
 "Including fixes from bluetooth, bpf and netfilter.

  Current release - new code bugs:

   - bridge: switchdev: check br_vlan_group() return value

   - use this_cpu_inc() to increment net->core_stats, fix preempt-rt

  Previous releases - regressions:

   - eth: stmmac: fix write to sgmii_adapter_base

  Previous releases - always broken:

   - netfilter: nf_conntrack_tcp: re-init for syn packets only,
     resolving issues with TCP fastopen

   - tcp: md5: fix incorrect tcp_header_len for incoming connections

   - tcp: fix F-RTO may not work correctly when receiving DSACK

   - tcp: ensure use of most recently sent skb when filling rate samples

   - tcp: fix potential xmit stalls caused by TCP_NOTSENT_LOWAT

   - virtio_net: fix wrong buf address calculation when using xdp

   - xsk: fix forwarding when combining copy mode with busy poll

   - xsk: fix possible crash when multiple sockets are created

   - bpf: lwt: fix crash when using bpf_skb_set_tunnel_key() from
     bpf_xmit lwt hook

   - sctp: null-check asoc strreset_chunk in sctp_generate_reconf_event

   - wireguard: device: check for metadata_dst with skb_valid_dst()

   - netfilter: update ip6_route_me_harder to consider L3 domain

   - gre: make o_seqno start from 0 in native mode

   - gre: switch o_seqno to atomic to prevent races in collect_md mode

  Misc:

   - add Eric Dumazet to networking maintainers

   - dt: dsa: realtek: remove realtek,rtl8367s string

   - netfilter: flowtable: Remove the empty file"

* tag 'net-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (65 commits)
  tcp: fix F-RTO may not work correctly when receiving DSACK
  Revert "ibmvnic: Add ethtool private flag for driver-defined queue limits"
  net: enetc: allow tc-etf offload even with NETIF_F_CSUM_MASK
  ixgbe: ensure IPsec VF<->PF compatibility
  MAINTAINERS: Update BNXT entry with firmware files
  netfilter: nft_socket: only do sk lookups when indev is available
  net: fec: add missing of_node_put() in fec_enet_init_stop_mode()
  bnx2x: fix napi API usage sequence
  tls: Skip tls_append_frag on zero copy size
  Add Eric Dumazet to networking maintainers
  netfilter: conntrack: fix udp offload timeout sysctl
  netfilter: nf_conntrack_tcp: re-init for syn packets only
  net: dsa: lantiq_gswip: Don't set GSWIP_MII_CFG_RMII_CLK
  net: Use this_cpu_inc() to increment net->core_stats
  Bluetooth: hci_sync: Cleanup hci_conn if it cannot be aborted
  Bluetooth: hci_event: Fix creating hci_conn object on error status
  Bluetooth: hci_event: Fix checking for invalid handle on error status
  ice: fix use-after-free when deinitializing mailbox snapshot
  ice: wait 5 s for EMP reset after firmware flash
  ice: Protect vf_state check by cfg_lock in ice_vc_process_vf_msg()
  ...

Pull thermal control fixes from Rafael Wysocki:
 "These take back recent chages that started to confuse users and fix up
  an attr.show callback prototype in a driver.

  Specifics:

   - Stop warning about deprecation of the userspace thermal governor
     and cooling device status interface, because there are cases in
     which user space has to drive thermal management with the help of
     them (Daniel Lezcano)

   - Fix attr.show callback prototype in the int340x thermal driver
     (Kees Cook)"

* tag 'thermal-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  thermal/governor: Remove deprecated information
  Revert "thermal/core: Deprecate changing cooling device state from userspace"
  thermal: int340x: Fix attr.show callback prototype

Pull power management fixes from Rafael Wysocki:
 "These fix up recent intel_idle driver changes and fix some ARM cpufreq
  driver issues.

  Specifics:

   - Fix issues with the Qualcomm's cpufreq driver (Dmitry Baryshkov,
     Vladimir Zapolskiy).

   - Fix memory leak with the Sun501 driver (Xiaobing Luo).

   - Make intel_idle enable C1E promotion on all CPUs when C1E is
     preferred to C1 (Artem Bityutskiy).

   - Make C6 optimization on Sapphire Rapids added recently work as
     expected if both C1E and C1 are "preferred" (Artem Bityutskiy)"

* tag 'pm-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  intel_idle: Fix SPR C6 optimization
  intel_idle: Fix the 'preferred_cstates' module parameter
  cpufreq: qcom-cpufreq-hw: Clear dcvs interrupts
  cpufreq: fix memory leak in sun50i_cpufreq_nvmem_probe
  cpufreq: qcom-cpufreq-hw: Fix throttle frequency value on EPSS platforms
  cpufreq: qcom-hw: provide online/offline operations
  cpufreq: qcom-hw: fix the opp entries refcounting
  cpufreq: qcom-hw: fix the race between LMH worker and cpuhp
  cpufreq: qcom-hw: drop affinity hint before freeing the IRQ

Pull ACPI fixes from Rafael WysockiL
 "These fix up the ACPI processor driver after a change made during the
  5.16 cycle that inadvertently broke falling back to shallower C-states
  when C3 cannot be used.

  Specifics:

   - Make the ACPI processor driver avoid falling back to C3 type of
     C-states when C3 cannot be requested (Ville Syrjälä)

   - Revert a quirk that is not necessary any more after fixing the
     underlying issue properly (Ville Syrjälä)"

* tag 'acpi-5.18-rc5' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm:
  Revert "ACPI: processor: idle: fix lockup regression on 32-bit ThinkPad T40"
  ACPI: processor: idle: Avoid falling back to C3 type C-states

Merge tag 'platform-drivers-x86-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86

Pull x86 platform driver fixes from Hans de Goede:
 "Highlights:

   - asus-wmi bug-fixes

   - intel-sdsu bug-fixes

   - build (warning) fixes

   - couple of hw-id additions"

* tag 'platform-drivers-x86-v5.18-3' of git://git.kernel.org/pub/scm/linux/kernel/git/pdx86/platform-drivers-x86:
  platform/x86/intel: pmc/core: change pmc_lpm_modes to static
  platform/x86/intel/sdsi: Fix bug in multi packet reads
  platform/x86/intel/sdsi: Poll on ready bit for writes
  platform/x86/intel/sdsi: Handle leaky bucket
  platform/x86: intel-uncore-freq: Prevent driver loading in guests
  platform/x86: gigabyte-wmi: added support for B660 GAMING X DDR4 motherboard
  platform/x86: dell-laptop: Add quirk entry for Latitude 7520
  platform/x86: asus-wmi: Fix driver not binding when fan curve control probe fails
  platform/x86: asus-wmi: Potential buffer overflow in asus_wmi_evaluate_method_buf()
  tools/power/x86/intel-speed-select: fix build failure when using -Wl,--as-needed

Merge tag 'regulator-fix-v5.18-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator

Pull regulator fix from Mark Brown:
 "A minor fix for the DT binding documentation of the rt5190a driver"

* tag 'regulator-fix-v5.18-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/regulator:
  regulator: dt-bindings: Revise the rt5190a buck/ldo description

Currently DSACK is regarded as a dupack, which may cause
F-RTO to incorrectly enter "loss was real" when receiving
DSACK.

Packetdrill to demonstrate:

// Enable F-RTO and TLP
    0 `sysctl -q net.ipv4.tcp_frto=2`
    0 `sysctl -q net.ipv4.tcp_early_retrans=3`
    0 `sysctl -q net.ipv4.tcp_congestion_control=cubic`

// Establish a connection
   +0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3
   +0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
   +0 bind(3, ..., ...) = 0
   +0 listen(3, 1) = 0

// RTT 10ms, RTO 210ms
  +.1 < S 0:0(0) win 32792 <mss 1000,sackOK,nop,nop,nop,wscale 7>
   +0 > S. 0:0(0) ack 1 <...>
 +.01 < . 1:1(0) ack 1 win 257
   +0 accept(3, ..., ...) = 4

// Send 2 data segments
   +0 write(4, ..., 2000) = 2000
   +0 > P. 1:2001(2000) ack 1

// TLP
+.022 > P. 1001:2001(1000) ack 1

// Continue to send 8 data segments
   +0 write(4, ..., 10000) = 10000
   +0 > P. 2001:10001(8000) ack 1

// RTO
+.188 > . 1:1001(1000) ack 1

// The original data is acked and new data is sent(F-RTO step 2.b)
   +0 < . 1:1(0) ack 2001 win 257
   +0 > P. 10001:12001(2000) ack 1

// D-SACK caused by TLP is regarded as a dupack, this results in
// the incorrect judgment of "loss was real"(F-RTO step 3.a)
+.022 < . 1:1(0) ack 2001 win 257 <sack 1001:2001,nop,nop>

// Never-retransmitted data(3001:4001) are acked and
// expect to switch to open state(F-RTO step 3.b)
   +0 < . 1:1(0) ack 4001 win 257
+0 %{ assert tcpi_ca_state == 0, tcpi_ca_state }%

Fixes: e33099f9 ("tcp: implement RFC5682 F-RTO")
Signed-off-by: Pengcheng Yang <yangpc@wangsu.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/1650967419-2150-1-git-send-email-yangpc@wangsu.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>

Pablo Neira Ayuso says:

====================
Netfilter fixes for net

1) Fix incorrect TCP connection tracking window reset for non-syn
   packets, from Florian Westphal.

2) Incorrect dependency on CONFIG_NFT_FLOW_OFFLOAD, from Volodymyr Mytnyk.

3) Fix nft_socket from the output path, from Florian Westphal.

* git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf:
  netfilter: nft_socket: only do sk lookups when indev is available
  netfilter: conntrack: fix udp offload timeout sysctl
  netfilter: nf_conntrack_tcp: re-init for syn packets only
====================

Link: https://lore.kernel.org/r/20220428142109.38726-1-pablo@netfilter.orgSigned-off-by: Jakub Kicinski <kuba@kernel.org>

Pull gfs2 fix from Andreas Gruenbacher:

 - No short reads or writes upon glock contention

* tag 'gfs2-v5.18-rc4-fix2' of git://git.kernel.org/pub/scm/linux/kernel/git/gfs2/linux-gfs2:
  gfs2: No short reads or writes upon glock contention

This reverts commit 723ad916

When client requests channel or ring size larger than what the server
can support the server will cap the request to the supported max. So,
the client would not be able to successfully request resources that
exceed the server limit.

Fixes: 723ad916 ("ibmvnic: Add ethtool private flag for driver-defined queue limits")
Signed-off-by: Dany Madden <drt@linux.ibm.com>
Link: https://lore.kernel.org/r/20220427235146.23189-1-drt@linux.ibm.comSigned-off-by: Jakub Kicinski <kuba@kernel.org>