This isn't all that relevant for RX right now, but TX can be concurrent
due to multi-queue and the accounting is therefore broken.

Use the standard per-CPU statistics to avoid this.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This isn't necessary any more as the stack will automatically
update the TXQ's trans_start after calling ndo_start_xmit().
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The OCB input path already checked that the BSSID is the broadcast
address, so the later check can never fail.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The function really shouldn't be called prepare_for_handlers(),
all it does is check if the frame should be dropped. Rename it
to ieee80211_accept_frame() and clean it up a bit.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

With promisc support gone, only AP and P2P-Device type interfaces
still clear IEEE80211_RX_RA_MATCH. In both cases this isn't really
necessary though, so we can remove that flag and the code.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

This support is essentially useless as typically networks are encrypted,
frames will be filtered by hardware, and rate scaling will be done with
the intended recipient in mind. For real monitoring of the network, the
monitor mode support should be used instead.

Removing it removes a lot of corner cases.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Allow debug builds to configure the station hash table maximum
size in order to run with hash collisions in limited scenarios
such as hwsim testing. The default remains 0 which effectively
means no limit.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Implement the necessary software segmentation on the normal
TX path so that fast-xmit can use segmentation offload if
the hardware (or driver) supports it.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

If drivers want to support S/G (really just gather DMA on TX) then
we can now easily support this on the fast-xmit path since it just
needs to write to the ethernet header (and already has a check for
that being possible.)

However, disallow this on the regular TX path (which has to handle
fragmentation, software crypto, etc.) by calling skb_linearize().

Also allow the related HIGHDMA since that's not interesting to the
code in mac80211 at all anyway.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When we go through the complete TX processing, there are a number
of things like fragmentation and software crypto that require the
checksum to be calculated already.

In favour of maintainability, instead of adding the necessary call
to skb_checksum_help() in all the places that need it, just do it
once before the regular TX processing.

Right now this only affects the TI wlcore and QCA ath10k drivers
since they're the only ones using checksum offload. The previous
commits enabled fast-xmit for them in almost all cases.

For wlcore this even fixes a corner case: when a key fails to be
programmed to hardware software encryption gets used, encrypting
frames with a bad checksum.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The driver can clearly enable fast-xmit since it does rate
control in the device and thus must do duration calculation
there as well.
Acked-by: NKalle Valo <kvalo@qca.qualcomm.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The driver can clearly enable fast-xmit since it does rate
control in the device and thus must do duration calculation
there as well.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

IBSS can be supported very easily since it uses the standard station
authorization state etc. so it just needs to be covered by the header
building switch statement.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

When crypto is offloaded then in some cases it's all handled
by the device, and in others only some space for the IV must
be reserved in the frame. Handle both of these cases in the
fast-xmit path, up to a limit of 18 bytes of space for IVs.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

If the driver handles fragmentation then it wouldn't
be done in software so we can still use the fast-xmit
path in that case.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

For hwsim, the duration field in frames is already not valid for
the common case of HT/VHT MCSes, so there's little point in trying
to keep it accurate for the legacy rates. Enable the fast-xmit code
to allow testing that, although given the dependency on hardware
crypto it will only be enabled in open network configurations.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

In order to speed up mac80211's TX path, add the "fast-xmit" cache
that will cache the data frame 802.11 header and other data to be
able to build the frame more quickly. This cache is rebuilt when
external triggers imply changes, but a lot of the checks done per
packet today are simplified away to the check for the cache.

There's also a more detailed description in the code.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

A couple of enums in mac80211.h became structures recently, but the
comments didn't follow suit, leading to errors like:

  Error(.//include/net/mac80211.h:367): Cannot parse enum!
  Documentation/DocBook/Makefile:93: recipe for target 'Documentation/DocBook/80211.xml' failed
  make[1]: *** [Documentation/DocBook/80211.xml] Error 1
  Makefile:1361: recipe for target 'mandocs' failed
  make: *** [mandocs] Error 2

Fix the comments comments accordingly.  Added a couple of other small
comment fixes while I was there to silence other recently-added docbook
warnings.
Reported-by: NJim Davis <jim.epost@gmail.com>
Signed-off-by: NJonathan Corbet <corbet@lwn.net>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Both minstrel (reported by Sven Eckelmann) and the iwlwifi rate
control aren't properly taking concurrency into account. It's
likely that the same is true for other rate control algorithms.

In the case of minstrel this manifests itself in crashes when an
update and other data access are run concurrently, for example
when the stations change bandwidth or similar. In iwlwifi, this
can cause firmware crashes.

Since fixing all rate control algorithms will be very difficult,
just provide locking for invocations. This protects the internal
data structures the algorithms maintain.

I've manipulated hostapd to test this, by having it change its
advertised bandwidth roughly ever 150ms. At the same time, I'm
running a flood ping between the client and the AP, which causes
this race of update vs. get_rate/status to easily happen on the
client. With this change, the system survives this test.
Reported-by: NSven Eckelmann <sven@open-mesh.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The mesh plink code uses sta->lock to serialize access to the
plink state fields between the peer link state machine and the
peer link timer.  Some paths (e.g. those involving
mps_qos_null_tx()) unfortunately hold this spinlock across
frame tx, which is soon to be disallowed.  Add a new spinlock
just for plink access.
Signed-off-by: NBob Copeland <me@bobcopeland.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Pull networking fixes from David Miller:

 1) Fix verifier memory corruption and other bugs in BPF layer, from
    Alexei Starovoitov.

 2) Add a conservative fix for doing BPF properly in the BPF classifier
    of the packet scheduler on ingress.  Also from Alexei.

 3) The SKB scrubber should not clear out the packet MARK and security
    label, from Herbert Xu.

 4) Fix oops on rmmod in stmmac driver, from Bryan O'Donoghue.

 5) Pause handling is not correct in the stmmac driver because it
    doesn't take into consideration the RX and TX fifo sizes.  From
    Vince Bridgers.

 6) Failure path missing unlock in FOU driver, from Wang Cong.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (44 commits)
  net: dsa: use DEVICE_ATTR_RW to declare temp1_max
  netns: remove BUG_ONs from net_generic()
  IB/ipoib: Fix ndo_get_iflink
  sfc: Fix memcpy() with const destination compiler warning.
  altera tse: Fix network-delays and -retransmissions after high throughput.
  net: remove unused 'dev' argument from netif_needs_gso()
  act_mirred: Fix bogus header when redirecting from VLAN
  inet_diag: fix access to tcp cc information
  tcp: tcp_get_info() should fetch socket fields once
  net: dsa: mv88e6xxx: Add missing initialization in mv88e6xxx_set_port_state()
  skbuff: Do not scrub skb mark within the same name space
  Revert "net: Reset secmark when scrubbing packet"
  bpf: fix two bugs in verification logic when accessing 'ctx' pointer
  bpf: fix bpf helpers to use skb->mac_header relative offsets
  stmmac: Configure Flow Control to work correctly based on rxfifo size
  stmmac: Enable unicast pause frame detect in GMAC Register 6
  stmmac: Read tx-fifo-depth and rx-fifo-depth from the devicetree
  stmmac: Add defines and documentation for enabling flow control
  stmmac: Add properties for transmit and receive fifo sizes
  stmmac: fix oops on rmmod after assigning ip addr
  ...

Pull sparc updates from David Miller:
 "The PowerPC folks have a really nice scalable IOMMU pool allocator
  that we wanted to make use of for sparc.  So here we have a series
  that abstracts out their code into a common layer that anyone can make
  use of.

  Sparc is converted, and the PowerPC folks have reviewed and ACK'd this
  series and plan to convert PowerPC over as well"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  iommu-common: Fix PARISC compile-time warnings
  sparc: Make LDC use common iommu poll management functions
  sparc: Make sparc64 use scalable lib/iommu-common.c functions
  sparc: Break up monolithic iommu table/lock into finer graularity pools and lock

Pull arch/tile updates from Chris Metcalf:
 "These are mostly nohz_full changes, plus a smattering of minor fixes
  (notably a couple for ftrace)"

* git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  tile: nohz: warn if nohz_full uses hypervisor shared cores
  tile: ftrace: fix function_graph tracer issues
  tile: map data region shadow of kernel as R/W
  tile: support CONTEXT_TRACKING and thus NOHZ_FULL
  tile: support arch_irq_work_raise
  arch: tile: fix null pointer dereference on pt_regs pointer
  tile/elf: reorganize notify_exec()
  tile: use si_int instead of si_ptr for compat_siginfo

Since commit da4759c7 (sysfs: Use only return value from is_visible for
the file mode), it is possible to reduce the permissions of a file.

So declare temp1_max with the DEVICE_ATTR_RW macro and remove the write
permission in dsa_hwmon_attrs_visible if set_temp_limit isn't provided.
Signed-off-by: NVivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: NGuenter Roeck <linux@roeck-us.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pull MIPS updates from Ralf Baechle:
 "This is the main pull request for MIPS for Linux 4.1.  Most
  noteworthy:

   - Add more Octeon-optimized crypto functions
   - Octeon crypto preemption and locking fixes
   - Little endian support for Octeon
   - Use correct CSR to soft reset Octeons
   - Support LEDs on the Octeon-based DSR-1000N
   - Fix PCI interrupt mapping for the Octeon-based DSR-1000N
   - Mark prom_free_prom_memory() as __init for a number of systems
   - Support for Imagination's Pistachio SOC.  This includes arch and
     CLK bits.  I'd like to merge pinctrl bits later
   - Improve parallelism of csum_partial for certain pipelines
   - Organize DTB files in subdirs like other architectures
   - Implement read_sched_clock for all MIPS platforms other than
     Octeon
   - Massive series of 38 fixes and cleanups for the FPU emulator /
     kernel
   - Further FPU remulator work to support new features.  This sits on a
     separate branch which also has been pulled into the 4.1 KVM branch
   - Clean up and fixes for the SEAD3 eval board; remove unused file
   - Various updates for Netlogic platforms
   - A number of small updates for Loongson 3 platforms
   - Increase the memory limit for ATH79 platforms to 256MB
   - A fair number of fixes and updates for BCM47xx platforms
   - Finish the implementation of XPA support
   - MIPS FDC support.  No, not floppy controller but Fast Debug Channel :)
   - Detect the R16000 used in SGI legacy platforms
   - Fix Kconfig dependencies for the SSB bus support"

* 'upstream' of git://git.linux-mips.org/pub/scm/ralf/upstream-linus: (265 commits)
  MIPS: Makefile: Fix MIPS ASE detection code
  MIPS: asm: elf: Set O32 default FPU flags
  MIPS: BCM47XX: Fix detecting Microsoft MN-700 & Asus WL500G
  MIPS: Kconfig: Disable SMP/CPS for 64-bit
  MIPS: Hibernate: flush TLB entries earlier
  MIPS: smp-cps: cpu_set FPU mask if FPU present
  MIPS: lose_fpu(): Disable FPU when MSA enabled
  MIPS: ralink: add missing symbol for RALINK_ILL_ACC
  MIPS: ralink: Fix bad config symbol in PCI makefile.
  SSB: fix Kconfig dependencies
  MIPS: Malta: Detect and fix bad memsize values
  Revert "MIPS: Avoid pipeline stalls on some MIPS32R2 cores."
  MIPS: Octeon: Delete override of cpu_has_mips_r2_exec_hazard.
  MIPS: Fix cpu_has_mips_r2_exec_hazard.
  MIPS: kernel: entry.S: Set correct ISA level for mips_ihb
  MIPS: asm: spinlock: Fix addiu instruction for R10000_LLSC_WAR case
  MIPS: r4kcache: Use correct base register for MIPS R6 cache flushes
  MIPS: Kconfig: Fix typo for the r2-to-r6 emulator kernel parameter
  MIPS: unaligned: Fix regular load/store instruction emulation for EVA
  MIPS: unaligned: Surround load/store macros in do {} while statements
  ...

Pull Xtensa updates from Chris Zankel:

 - fix linker script transformation for .text / .text.fixup

 - wire bpf and execveat syscalls

 - provide __NR_sync_file_range2 instead of __NR_sync_file_range, as
   that's what xtensa uses.

 - make xtfpgs LCD driver functional and configurable.  This fixes
   hardware lockup on KC705/ML605 boot

 - add audio subsystem bits to xtfpga DTS and provide sample KC705
   config with audio features enabled

 - add CY7C67300 USB controller support to XTFPGA

 - fix locking issues in ISS network driver

 - document PIC and MX interrupt distributor device tree bindings

* tag 'xtensa-20150416' of git://github.com/czankel/xtensa-linux:
  xtensa: xtfpga: add CY7C67300 USB controller support
  irqchip: xtensa-pic: xtensa-mx: document DT bindings
  xtensa: ISS: fix locking in TAP network adapter
  xtensa: Fix fix linker script transformation for .text / .text.fixup
  xtensa: provide __NR_sync_file_range2 instead of __NR_sync_file_range
  xtensa: wire bpf and execveat syscalls
  xtensa: xtfpga: fix hardware lockup caused by LCD driver
  xtensa: xtfpga: provide defconfig with audio subsystem
  xtensa: xtfpga: add audio card to xtfpga DTS

Fixes warnings due to
- no DMA_ERROR_CODE on PARISC,
- sizeof (unsigned long) == 4 bytes on PARISC.
Signed-off-by: NSowmini Varadhan <sowmini.varadhan@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This inline has ~500 callsites.

On 04/14/2015 08:37 PM, David Miller wrote:
> That BUG_ON() was added 7 years ago, and I don't remember it ever
> triggering or helping us diagnose something, so just remove it and
> keep the function inlined.

On x86 allyesconfig build:

    text     data      bss       dec     hex filename
82447071 22255384 20627456 125329911 77861f7 vmlinux4
82441375 22255384 20627456 125324215 7784bb7 vmlinux5prime
Signed-off-by: NDenys Vlasenko <dvlasenk@redhat.com>
CC: Eric W. Biederman <ebiederm@xmission.com>
CC: David S. Miller <davem@davemloft.net>
CC: Jan Engelhardt <jengelh@medozas.de>
CC: Jiri Pirko <jpirko@redhat.com>
CC: linux-kernel@vger.kernel.org
CC: netdev@vger.kernel.org
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, iflink of the parent interface was always accessed, even
when interface didn't have a parent and hence we crashed there.

Handle the interface types properly: for a child interface, return
the ifindex of the parent, for parent interface, return its ifindex.

For child devices, make sure to set the parent pointer prior to
invoking register_netdevice(), this allows the new ndo to be called
by the stack immediately after the child device is registered.

Fixes: 5aa7add8 ('infiniband/ipoib: implement ndo_get_iflink')
Reported-by: NHonggang Li <honli@redhat.com>
Signed-off-by: NErez Shitrit <erezsh@mellanox.com>
Signed-off-by: NHonggang Li <honli@redhat.com>
Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com>+
Acked-by: NNicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/ethernet/sfc/selftest.c: In function ‘efx_iterate_state’:
drivers/net/ethernet/sfc/selftest.c:388:9: warning: passing argument 1 of ‘memcpy’ discards ‘const’ qualifier from pointer target type [-Wdiscarded-array-qualifiers]

This is because the msg[] member of struct efx_loopback_payload
is marked as 'const'.  Remove that.
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix bug which occurs when more than <limit> packets are available during
napi-poll, leading to "delays" and retransmissions on the network.

Check for (count < limit) before checking the get_rx_status in tse_rx-function.
Function get_rx_status is reading from the response-fifo.
If there is currently a response in the fifo,
reading the last byte of the response pops the value from the fifo.
If the limit is checked as second condition
and the limit is reached the fifo is popped but the packet is not processed.
Signed-off-by: NAndreas Oetken <ennoerlangen@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pull block core fix from Jens Axboe:
 "A commit in the previous pull request introduce a regression.  So far
  only observed on qemu-sparc64, but it's a general bug.  Please pull
  this single fix to rectify that, thanks"

[ And it turns out that it's been seen outside of that qemu-sparc64
  case, and is easy to trigger with small number of CPUs and blk-mq
  enabled by default - Linus ]

* 'for-linus' of git://git.kernel.dk/linux-block:
  blk-mq: fix iteration of busy bitmap

Pull ACPICA updates from Rafael Wysocki:
 "This updates the kernel's ACPICA code to upstream revision 20150410
  and adds a fix for a GPE handling regression introduced during the
  3.19 cycle on top of that.

  Included are two stable-candidate bug fixes (one of them fixing a 3.16
  regression), multiple other fixes and a bunch of cleanups.

  Specifics:

   - Fix for a GPE handling regression on Dell Latitude D600 that caused
     GPE signaling to stop working on that machine, which appears to be
     due to a hardware glitch, but it used to work and it can be made
     work again in a relativly straightforward way (Rafael J Wysocki).

   - Fix for a mutex unlock regression related to the handling of ACPI
     tables introduced during the 3.16 development cycle (Octavian
     Purdila).

   - _REV modification to always return 2 which has been done by all
     versions of Windows since NT and the firmware people started to use
     it to distinguish between OSes in their AML and do some silly and
     wrong things on that basis (Bob Moore).

   - Fixes and cleanups related to the acpi_physicall_address data type
     including one stable-candidate fix for an issue occasionally
     occuring on 64-bit machines running 32-bit kernels where using
     offsets provided by the firmware may lead to address overflows (Lv
     Zheng).

   - External() opcode support infrastructure needed for recompiling
     disassembled ACPI tables in some cases including interpreter
     modification to ignore that opcode (Bob Moore).

   - Support for the "Windows 2015" string in _OSI (Bob Moore).

   - GPE debug interface change to return values read from hardware
     registers (Lv Zheng).

   - Removal of the __DATE__ macro usage in tools (Rasmus Villemoes).

   - Assorted minor fixes and cleanups (Lv Zheng, Rickard Strandqvist,
     Bob Moore)"

* tag 'acpica-4.1-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm: (32 commits)
  ACPICA: Store GPE register enable masks upfront
  ACPICA: Update version to 20150410.
  ACPICA: Fix a couple issues with the local printf module.
  ACPICA: Disassembler: Some cleanup of the table dump module.
  ACPICA: iASL: Add support for MSDM ACPI table.
  ACPICA: Update for SLIC ACPI table.
  ACPICA: Add "//" before ascii output of buffers.
  ACPICA: Remove unused internal AML opcode.
  ACPICA: Permanently set _REV to the value '2'.
  ACPICA: Add "Windows 2015" string to _OSI support.
  ACPICA: Add infrastructure for External() opcode.
  ACPICA: iASL: Enhancement for constant folding.
  ACPICA: iASL/Disassembler: Add option to assume table contains valid AML.
  ACPICA: Update AML Debugger global variables.
  ACPICA: Update Resource descriptor dump module.
  ACPICA: Fix a sscanf format string.
  ACPICA: Casting changes around acpi_physical_address/acpi_size.
  ACPICA: Resources: Correct conditional compilation definitions.
  ACPICA: Utilities: Correct conditional compilation definitions.
  ACPICA: Tables: Move an iasl specific table function to iasl source file.
  ...

The "hypervisor shared" cores are ones that the Tilera hypervisor
uses to receive interrupts to manage hypervisor-owned devices.
It's a bad idea to try to use those cores with nohz_full, since
they will get interrupted unpredictably -- and invisibly to Linux
tracing tools, since the interrupts are delivered at a higher
privilege level to the Tilera hypervisor.

Generate a clear warning at boot up that this doesn't end well
for the nohz_full cores in question.
Signed-off-by: NChris Metcalf <cmetcalf@ezchip.com>

- Add support for ARCH_SUPPORTS_FTRACE_OPS
- Replace the instruction in ftrace_call with the bundle {move r10, lr;
jal ftrace_stub}, so that the lr contains the right value after returning
from ftrace_stub.  An alternative fix might be to leave the instruction
in ftrace_call alone when it is being updated with ftrace_stub.
Signed-off-by: NTony Lu <zlu@ezchip.com>
Signed-off-by: NChris Metcalf <cmetcalf@ezchip.com>

This is necessary for things like reading /proc/kcore, doing ftrace,
etc.  It happens by default when using huge pages to map the kernel
data, but not when using small pages.
Signed-off-by: NChris Metcalf <cmetcalf@ezchip.com>

Add the TIF_NOHZ flag appropriately.

Add call to user_exit() on entry to do_work_pending() and on entry
to syscalls via do_syscall_trace_enter(), and also the top of
do_syscall_trace_exit() just because it's done in x86.

Add call to user_enter() at the bottom of do_work_pending() once we
have no more work to do before returning to userspace.

Wrap all the trap code in exception_enter() / exception_exit().
Signed-off-by: NChris Metcalf <cmetcalf@ezchip.com>
Acked-by: NFrederic Weisbecker <fweisbec@gmail.com>

In commit 04ffcb25 ("net: Add ndo_gso_check") Tom originally
added the 'dev' argument to be able to call ndo_gso_check().

Then later, when generalizing this in commit 5f35227e
("net: Generalize ndo_gso_check to ndo_features_check")
Jesse removed the call to ndo_gso_check() in netif_needs_gso()
by calling the new ndo_features_check() in a different place.
This made the 'dev' argument unused.

Remove the unused argument and go back to the code as before.

Cc: Tom Herbert <therbert@google.com>
Cc: Jesse Gross <jesse@nicira.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When you redirect a VLAN device to any device, you end up with
crap in af_packet on the xmit path because hard_header_len is
not equal to skb->mac_len.  So the redirected packet contains
four extra bytes at the start which then gets interpreted as
part of the MAC address.

This patch fixes this by only pushing skb->mac_len.  We also
need to fix ifb because it tries to undo the pushing done by
act_mirred.
Signed-off-by: NHerbert Xu <herbert@gondor.apana.org.au>
Acked-by: NAlexei Starovoitov <ast@plumgrid.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Two different problems are fixed here :

1) inet_sk_diag_fill() might be called without socket lock held.
   icsk->icsk_ca_ops can change under us and module be unloaded.
   -> Access to freed memory.
   Fix this using rcu_read_lock() to prevent module unload.

2) Some TCP Congestion Control modules provide information
   but again this is not safe against icsk->icsk_ca_ops
   change and nla_put() errors were ignored. Some sockets
   could not get the additional info if skb was almost full.

Fix this by returning a status from get_info() handlers and
using rcu protection as well.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>