As a preparation from removing rtnl lock dependency from rules update path,
use Qdisc rcu and reference counting capabilities instead of relying on
rtnl lock while working with Qdiscs. Create new tcf_block_release()
function, and use it to free resources taken by tcf_block_find().
Currently, this function only releases Qdisc and it is extended in next
patches in this series.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Implement function to take reference to Qdisc that relies on rcu read lock
instead of rtnl mutex. Function only takes reference to Qdisc if reference
counter isn't zero. Intended to be used by unlocked cls API.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, Qdisc API functions assume that users have rtnl lock taken. To
implement rtnl unlocked classifiers update interface, Qdisc API must be
extended with functions that do not require rtnl lock.

Extend Qdisc structure with rcu. Implement special version of put function
qdisc_put_unlocked() that is called without rtnl lock taken. This function
only takes rtnl lock if Qdisc reference counter reached zero and is
intended to be used as optimization.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Current implementation of qdisc_destroy() decrements Qdisc reference
counter and only actually destroy Qdisc if reference counter value reached
zero. Rename qdisc_destroy() to qdisc_put() in order for it to better
describe the way in which this function currently implemented and used.

Extract code that deallocates Qdisc into new private qdisc_destroy()
function. It is intended to be shared between regular qdisc_put() and its
unlocked version that is introduced in next patch in this series.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Rtnl lock is encapsulated in netlink and cannot be accessed by other
modules directly. This means that reference counted objects that rely on
rtnl lock cannot use it with refcounter helper function that atomically
releases decrements reference and obtains mutex.

This patch implements simple wrapper function around refcount_dec_and_lock
that obtains rtnl lock if reference counter value reached 0.
Signed-off-by: Vlad Buslov <vladbu@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

During socket close, if there is a open record with tx context, it needs
to be be freed apart from freeing up plaintext and encrypted scatter
lists. This patch frees up the open record if present in tx context.

Also tls_free_both_sg() has been renamed to tls_free_open_rec() to
indicate that the free record in tx context is being freed inside the
function.

Fixes: a42055e8 ("net/tls: Add support for async encryption")
Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Current async encryption implementation sometimes showed up socket
memory accounting error during socket close. This results in kernel
warning calltrace. The root cause of the problem is that socket var
sk_forward_alloc gets corrupted due to access in sk_mem_charge()
and sk_mem_uncharge() being invoked from multiple concurrent contexts
in multicore processor. The apis sk_mem_charge() and sk_mem_uncharge()
are called from functions alloc_plaintext_sg(), free_sg() etc. It is
required that memory accounting apis are called under a socket lock.

The plaintext sg data sent for encryption is freed using free_sg() in
tls_encryption_done(). It is wrong to call free_sg() from this function.
This is because this function may run in irq context. We cannot acquire
socket lock in this function.

We remove calling of function free_sg() for plaintext data from
tls_encryption_done() and defer freeing up of plaintext data to the time
when the record is picked up from tx_list and transmitted/freed. When
tls_tx_records() gets called, socket is already locked and thus there is
no concurrent access problem.

Fixes: a42055e8 ("net/tls: Add support for async encryption")
Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Version bump conflict in batman-adv, take what's in net-next.

iavf conflict, adjustment of netdev_ops in net-next conflicting
with poll controller method removal in net.
Signed-off-by: David S. Miller <davem@davemloft.net>

Clear ADDR64 dma bit in DMACFG register in case that HW_DMA_CAP_64B is
not detected on 64bit system.
The issue was observed when bootloader(u-boot) does not check macb
feature at DCFG6 register (DAW64_OFFSET) and enabling 64bit dma support
by default. Then macb driver is reading DMACFG register back and only
adding 64bit dma configuration but not cleaning it out.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Heiner Kallweit says:

====================
r8169: series with smaller improvements

This series includes smaller improvements, nothing exciting.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The check for pci_is_pcie() is redundant here because all
chip versions >=18 are PCIe only anyway. In addition use
dma_set_mask_and_coherent() instead of separate calls to
pci_set_dma_mask() and pci_set_consistent_dma_mask().
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Code can be slightly simplified by acking even events we're not
interested in. In addition add a comment making clear that the
read has no functional purpose and is just a PCI commit.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The networking core has a default watchdog timeout of 5s. I see no
need to define an own timeout of 6s which is basically the same.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

James writes:
  "SCSI fixes on 20180925

   Nine obvious bug fixes mostly in individual drivers.  The target fix
   is of particular importance because it's CVE related."

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: sd: don't crash the host on invalid commands
  scsi: ipr: System hung while dlpar adding primary ipr adapter back
  scsi: target: iscsi: Use bin2hex instead of a re-implementation
  scsi: target: iscsi: Use hex2bin instead of a re-implementation
  scsi: lpfc: Synchronize access to remoteport via rport
  scsi: ufs: Disable blk-mq for now
  scsi: sd: Contribute to randomness when running rotational device
  scsi: ibmvscsis: Ensure partition name is properly NUL terminated
  scsi: ibmvscsis: Fix a stringop-overflow warning

I wrote:
  "USB fixes for 4.19-rc6

   Here are some small USB core and driver fixes for reported issues for
   4.19-rc6.

   The most visible is the oops fix for when the USB core is built into the
   kernel that is present in 4.18.  Turns out not many people actually do
   that so it went unnoticed for a while.  The rest is some tiny typec,
   musb, and other core fixes.

   All have been in linux-next with no reported issues."

* tag 'usb-4.19-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: typec: mux: Take care of driver module reference counting
  usb: core: safely deal with the dynamic quirk lists
  usb: roles: Take care of driver module reference counting
  USB: handle NULL config in usb_find_alt_setting()
  USB: fix error handling in usb_driver_claim_interface()
  USB: remove LPM management from usb_driver_claim_interface()
  USB: usbdevfs: restore warning for nonsensical flags
  USB: usbdevfs: sanitize flags more
  Revert "usb: cdc-wdm: Fix a sleep-in-atomic-context bug in service_outstanding_interrupt()"
  usb: musb: dsps: do not disable CPPI41 irq in driver teardown

I wrote:
  "TTY/Serial driver fixes for 4.19-rc6

   Here are a number of small tty and serial driver fixes for reported
   issues for 4.19-rc6.

   One should hopefully resolve a much-reported issue that syzbot has found
   in the tty layer.  Although there are still more issues there, getting
   this fixed is nice to see finally happen.

   All of these have been in linux-next for a while with no reported
   issues."

* tag 'tty-4.19-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  serial: imx: restore handshaking irq for imx1
  tty: vt_ioctl: fix potential Spectre v1
  tty: Drop tty->count on tty_reopen() failure
  serial: cpm_uart: return immediately from console poll
  tty: serial: lpuart: avoid leaking struct tty_struct
  serial: mvebu-uart: Fix reporting of effective CSIZE to userspace

Greg (well I), wrote:
  "Char/Misc driver fixes for 4.19-rc6

   Here are some soundwire and intel_th (tracing) driver fixes for some
   reported issues.

   All of these have been in linux-next for a week with no reported issues."

* tag 'char-misc-4.19-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  intel_th: pci: Add Ice Lake PCH support
  intel_th: Fix resource handling for ACPI glue layer
  intel_th: Fix device removal logic
  soundwire: Fix acquiring bus lock twice during master release
  soundwire: Fix incorrect exit after configuring stream
  soundwire: Fix duplicate stream state assignment

This changes UAPI, breaking iwd and libell:

  ell/key.c: In function 'kernel_dh_compute':
  ell/key.c:205:38: error: 'struct keyctl_dh_params' has no member named 'private'; did you mean 'dh_private'?
    struct keyctl_dh_params params = { .private = private,
                                        ^~~~~~~
                                        dh_private

This reverts commit 8a2336e5.

Fixes: 8a2336e5 ("uapi/linux/keyctl.h: don't use C++ reserved keyword as a struct member name")
Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Randy Dunlap <rdunlap@infradead.org>
cc: Mat Martineau <mathew.j.martineau@linux.intel.com>
cc: Stephan Mueller <smueller@chronox.de>
cc: James Morris <jmorris@namei.org>
cc: "Serge E. Hallyn" <serge@hallyn.com>
cc: Mat Martineau <mathew.j.martineau@linux.intel.com>
cc: Andrew Morton <akpm@linux-foundation.org>
cc: Linus Torvalds <torvalds@linux-foundation.org>
cc: <stable@vger.kernel.org>
Signed-off-by: James Morris <james.morris@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Dave writes:
  "Networking fixes:

  1) Fix multiqueue handling of coalesce timer in stmmac, from Jose
     Abreu.

   2) Fix memory corruption in NFC, from Suren Baghdasaryan.

   3) Don't write reserved bits in ravb driver, from Kazuya Mizuguchi.

   4) SMC bug fixes from Karsten Graul, YueHaibing, and Ursula Braun.

   5) Fix TX done race in mvpp2, from Antoine Tenart.

   6) ipv6 metrics leak, from Wei Wang.

   7) Adjust firmware version requirements in mlxsw, from Petr Machata.

   8) Fix autonegotiation on resume in r8169, from Heiner Kallweit.

   9) Fixed missing entries when dumping /proc/net/if_inet6, from Jeff
      Barnhill.

   10) Fix double free in devlink, from Dan Carpenter.

   11) Fix ethtool regression from UFO feature removal, from Maciej
       Żenczykowski.

   12) Fix drivers that have a ndo_poll_controller() that captures the
       cpu entirely on loaded hosts by trying to drain all rx and tx
       queues, from Eric Dumazet.

   13) Fix memory corruption with jumbo frames in aquantia driver, from
       Friedemann Gerold."

* gitolite.kernel.org:/pub/scm/linux/kernel/git/davem/net: (79 commits)
  net: mvneta: fix the remaining Rx descriptor unmapping issues
  ip_tunnel: be careful when accessing the inner header
  mpls: allow routes on ip6gre devices
  net: aquantia: memory corruption on jumbo frames
  tun: remove ndo_poll_controller
  nfp: remove ndo_poll_controller
  bnxt: remove ndo_poll_controller
  bnx2x: remove ndo_poll_controller
  mlx5: remove ndo_poll_controller
  mlx4: remove ndo_poll_controller
  i40evf: remove ndo_poll_controller
  ice: remove ndo_poll_controller
  igb: remove ndo_poll_controller
  ixgb: remove ndo_poll_controller
  fm10k: remove ndo_poll_controller
  ixgbevf: remove ndo_poll_controller
  ixgbe: remove ndo_poll_controller
  bonding: use netpoll_poll_dev() helper
  netpoll: make ndo_poll_controller() optional
  rds: Fix build regression.
  ...

Until now, the Rx flow hash key was a 5-tuple (IP src, IP dst,
IP nextproto, L4 src port, L4 dst port) fixed value that we
configured at probe.

Add support for configuring this hash key at runtime.
We support all standard header fields configurable through ethtool,
but cannot differentiate between flow types, so the same hash key
is applied regardless of protocol.

We also don't support the discard option.
Signed-off-by: Ioana Radulescu <ruxandra.radulescu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

With CONFIG_DMA_API_DEBUG enabled we get DMA unmapping warning in
various places of the mvneta driver, for example when putting down an
interface while traffic is passing through.

The issue is when using s/w buffer management, the Rx buffers are mapped
using dma_map_page but unmapped with dma_unmap_single. This patch fixes
this by using the right unmapping function.

Fixes: 562e2f46 ("net: mvneta: Improve the buffer allocation method for SWBM")
Signed-off-by: Antoine Tenart <antoine.tenart@bootlin.com>
Reviewed-by: Gregory CLEMENT <gregory.clement@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Cong noted that we need the same checks introduced by commit 76c0ddd8
("ip6_tunnel: be careful when accessing the inner header")
even for ipv4 tunnels.

Fixes: c5441932 ("GRE: Refactor GRE tunneling code.")
Suggested-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The SPI protocol for the QCA7000 doesn't have any fault detection.
In order to increase the drivers reliability in noisy environments,
we could implement a write verification inspired by the enc28j60.
This should avoid situations were the driver wrongly assumes the
receive interrupt is enabled and miss all incoming packets.

This function is disabled per default and can be controlled via module
parameter wr_verify.
Signed-off-by: Michael Heimpold <michael.heimpold@i2se.com>
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In tls_sw_sendmsg() and tls_sw_sendpage(), it is possible that the
uninitialised variable 'ret' gets passed to sk_stream_error(). So
initialise local variable 'ret' to '0. The warnings were detected by
'smatch' tool.

Fixes: a42055e8 ("net/tls: Add support for async encryption")
Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

On processors with multi-engine crypto accelerators, it is possible that
multiple records get encrypted in parallel and their encryption
completion is notified to different cpus in multicore processor. This
leads to the situation where tls_encrypt_done() starts executing in
parallel on different cores. In current implementation, encrypted
records are queued to tx_ready_list in tls_encrypt_done(). This requires
addition to linked list 'tx_ready_list' to be protected. As
tls_decrypt_done() could be executing in irq content, it is not possible
to protect linked list addition operation using a lock.

To fix the problem, we remove linked list addition operation from the
irq context. We do tx_ready_list addition/removal operation from
application context only and get rid of possible multiple access to
the linked list. Before starting encryption on the record, we add it to
the tail of tx_ready_list. To prevent tls_tx_records() from transmitting
it, we mark the record with a new flag 'tx_ready' in 'struct tls_rec'.
When record encryption gets completed, tls_encrypt_done() has to only
update the 'tx_ready' flag to true & linked list add operation is not
required.

The changed logic brings some other side benefits. Since the records
are always submitted in tls sequence number order for encryption, the
tx_ready_list always remains sorted and addition of new records to it
does not have to traverse the linked list.

Lastly, we renamed tx_ready_list in 'struct tls_sw_context_tx' to
'tx_list'. This is because now, the some of the records at the tail are
not ready to transmit.

Fixes: a42055e8 ("net/tls: Add support for async encryption")
Signed-off-by: Vakul Garg <vakul.garg@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Roopa Prabhu says:

====================
few NTF_ROUTER related updates

This series allows setting of NTF_ROUTER by an external
entity (eg BGP E-VPN control plane). Also fixes missing
netlink notification on neigh NTF_ROUTER flag changes.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

send netlink notification if neigh_update results in NTF_ROUTER
change and if NEIGH_UPDATE_F_ISROUTER is on. Also move the
NTF_ROUTER change function into a helper.
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch allows admin setting of NTF_ROUTER flag
on a neighbour entry. This enables external control
plane (like bgp evpn) to manage neigh entries with
NTF_ROUTER flag.
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Summary:

This appears to be necessary and sufficient change to enable `MPLS` on
`ip6gre` tunnels (RFC4023).

This diff allows IP6GRE devices to be recognized by MPLS kernel module
and hence user can configure interface to accept packets with mpls
headers as well setup mpls routes on them.

Test Plan:

Test plan consists of multiple containers connected via GRE-V6 tunnel.
Then carrying out testing steps as below.

- Carry out necessary sysctl settings on all containers

```
sysctl -w net.mpls.platform_labels=65536
sysctl -w net.mpls.ip_ttl_propagate=1
sysctl -w net.mpls.conf.lo.input=1
```

- Establish IP6GRE tunnels

```
ip -6 tunnel add name if_1_2_1 mode ip6gre \
  local 2401:db00:21:6048:feed:0::1 \
  remote 2401:db00:21:6048:feed:0::2 key 1
ip link set dev if_1_2_1 up
sysctl -w net.mpls.conf.if_1_2_1.input=1
ip -4 addr add 169.254.0.2/31 dev if_1_2_1 scope link

ip -6 tunnel add name if_1_3_1 mode ip6gre \
  local 2401:db00:21:6048:feed:0::1 \
  remote 2401:db00:21:6048:feed:0::3 key 1
ip link set dev if_1_3_1 up
sysctl -w net.mpls.conf.if_1_3_1.input=1
ip -4 addr add 169.254.0.4/31 dev if_1_3_1 scope link
```

- Install MPLS encap rules on node-1 towards node-2

```
ip route add 192.168.0.11/32 nexthop encap mpls 32/64 \
  via inet 169.254.0.3 dev if_1_2_1
```

- Install MPLS forwarding rules on node-2 and node-3
```
// node2
ip -f mpls route add 32 via inet 169.254.0.7 dev if_2_4_1

// node3
ip -f mpls route add 64 via inet 169.254.0.12 dev if_4_3_1
```

- Ping 192.168.0.11 (node4) from 192.168.0.1 (node1) (where routing
  towards 192.168.0.1 is via IP route directly towards node1 from node4)
```
ping 192.168.0.11
```

- tcpdump on interface to capture ping packets wrapped within MPLS
  header which inturn wrapped within IP6GRE header

```
16:43:41.121073 IP6
  2401:db00:21:6048:feed::1 > 2401:db00:21:6048:feed::2:
  DSTOPT GREv0, key=0x1, length 100:
  MPLS (label 32, exp 0, ttl 255) (label 64, exp 0, [S], ttl 255)
  IP 192.168.0.1 > 192.168.0.11:
  ICMP echo request, id 1208, seq 45, length 64

0x0000:  6000 2cdb 006c 3c3f 2401 db00 0021 6048  `.,..l<?$....!`H
0x0010:  feed 0000 0000 0001 2401 db00 0021 6048  ........$....!`H
0x0020:  feed 0000 0000 0002 2f00 0401 0401 0100  ......../.......
0x0030:  2000 8847 0000 0001 0002 00ff 0004 01ff  ...G............
0x0040:  4500 0054 3280 4000 ff01 c7cb c0a8 0001  E..T2.@.........
0x0050:  c0a8 000b 0800 a8d7 04b8 002d 2d3c a05b  ...........--<.[
0x0060:  0000 0000 bcd8 0100 0000 0000 1011 1213  ................
0x0070:  1415 1617 1819 1a1b 1c1d 1e1f 2021 2223  .............!"#
0x0080:  2425 2627 2829 2a2b 2c2d 2e2f 3031 3233  $%&'()*+,-./0123
0x0090:  3435 3637                                4567
```
Signed-off-by: Saif Hasan <has@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Eelco Chaudron says:

====================
net/sched: Add hardware specific counters to TC actions

Add hardware specific counters to TC actions which will be exported
through the netlink API. This makes troubleshooting TC flower offload
easier, as it possible to differentiate the packets being offloaded.

v2 - Rebased on latest net-next
====================
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add additional counters that will store the bytes/packets processed by
hardware. These will be exported through the netlink interface for
displaying by the iproute2 tc tool
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add a new hardware specific basic counter, TCA_STATS_BASIC_HW. This can
be used to count packets/bytes processed by hardware offload.
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
pull-request: bpf 2018-09-24

The following pull-request contains BPF updates for your *net* tree.

The main changes are:

1) Several fixes for BPF sockmap to only allow sockets being attached in
   ESTABLISHED state, from John.

2) Fix up the license to LGPL/BSD for the libc compat header which contains
   fallback helpers that libbpf and bpftool is using, from Jakub.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Maxime Chevallier says:

====================
net: mvpp2: Add txq to CPU mapping

This short series adds XPS support to the mvpp2 driver, by mapping
txqs and CPUs. This comes with a patch using round-robin scheduling
for the HW to pick the next txq to transmit from, instead of the default
fixed-priority scheduling.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This commit allows each TXQ to be picked in a round-robin fashion by
the PPv2 transmit scheduling mechanism. This is opposed to the default
behaviour that prioritizes the highest numbered queues.
Suggested-by: Yan Markman <ymarkman@marvell.com>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since the PPv2 controller has multiple TX queues, we can spread traffic
by assining TX queues to CPUs, allowing to use XPS to balance egress
traffic between CPUs.

Suggested-by : Yan Markman <ymarkman@marvell.com>
Signed-off-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Mauro briefly writes:
  "media fixes for v4.19-rc5

   some drivers and Kbuild fixes"

* tag 'media/v4.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  media: platform: fix cros-ec-cec build error
  media: staging/media/mt9t031/Kconfig: remove bogus entry
  media: i2c: mt9v111: Fix v4l2-ctrl error handling
  media: camss: add missing includes
  media: camss: Use managed memory allocations
  media: camss: mark PM functions as __maybe_unused
  media: af9035: prevent buffer overflow on write
  media: video_function_calls.rst: drop obsolete video-set-attributes reference

This patch fixes skb_shared area, which will be corrupted
upon reception of 4K jumbo packets.

Originally build_skb usage purpose was to reuse page for skb to eliminate
needs of extra fragments. But that logic does not take into account that
skb_shared_info should be reserved at the end of skb data area.

In case packet data consumes all the page (4K), skb_shinfo location
overflows the page. As a consequence, __build_skb zeroed shinfo data above
the allocated page, corrupting next page.

The issue is rarely seen in real life because jumbo are normally larger
than 4K and that causes another code path to trigger.
But it 100% reproducible with simple scapy packet, like:

    sendp(IP(dst="192.168.100.3") / TCP(dport=443) \
          / Raw(RandString(size=(4096-40))), iface="enp1s0")

Fixes: 018423e9 ("net: ethernet: aquantia: Add ring support code")
Reported-by: Friedemann Gerold <f.gerold@b-c-s.de>
Reported-by: Michael Rauch <michael@rauch.be>
Signed-off-by: Friedemann Gerold <f.gerold@b-c-s.de>
Tested-by: Nikita Danilov <nikita.danilov@aquantia.com>
Signed-off-by: Igor Russkikh <igor.russkikh@aquantia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Eric Dumazet says:

====================
netpoll: avoid capture effects for NAPI drivers

As diagnosed by Song Liu, ndo_poll_controller() can
be very dangerous on loaded hosts, since the cpu
calling ndo_poll_controller() might steal all NAPI
contexts (for all RX/TX queues of the NIC).

This capture, showing one ksoftirqd eating all cycles
can last for unlimited amount of time, since one
cpu is generally not able to drain all the queues under load.

It seems that all networking drivers that do use NAPI
for their TX completions, should not provide a ndo_poll_controller() :

Most NAPI drivers have netpoll support already handled
in core networking stack, since netpoll_poll_dev()
uses poll_napi(dev) to iterate through registered
NAPI contexts for a device.

This patch series take care of the first round, we will
handle other drivers in future rounds.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

As diagnosed by Song Liu, ndo_poll_controller() can
be very dangerous on loaded hosts, since the cpu
calling ndo_poll_controller() might steal all NAPI
contexts (for all RX/TX queues of the NIC). This capture
can last for unlimited amount of time, since one
cpu is generally not able to drain all the queues under load.

tun uses NAPI for TX completions, so we better let core
networking stack call the napi->poll() to avoid the capture.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>