This patch cleans up the locking login in l2cap_sock_recvmsg by pairing
up each lock_sock call with a release_sock call. The function already
has a "done" label that handles releasing the socket and returning from
the function so the fix is rather simple.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The bt_sock_wait_state requires the sk lock to be held (through
lock_sock) so document it clearly in the code.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

Convert 0 to false and 1 to true when assigning values to bool
variables. Inspired by commit 3db1cd5c.

The simplified semantic patch that find this problem is as
follows (http://coccinelle.lip6.fr/):

@@
bool b;
@@
(
-b = 0
+b = false
|
-b = 1
+b = true
)
Signed-off-by: NPeter Senna Tschudin <peter.senna@gmail.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

For those controller that support the HCI_Set_Event_Mask_Page_2 command
we should include it in the init sequence. This patch implements sending
of the command and enables the events in it based on supported features
(currently only CSB is checked).
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

This patch adds support for reading the synchronization train parameters
for controllers that support the feature. Since the feature is
detectable through the local features page 2, which is retreived only in
stage 3 of the HCI init sequence, there is no other option than to add a
fourth stage to the init sequence.

For now the patch doesn't yet add storing of the parameters, but it is
nevertheless convenient to have around to see what kind of parameters
various controllers use by default (analyzable e.g. with the btmon user
space tool).
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

In the case of blocking sockets we should not proceed with sendmsg() if
the socket has the BT_SK_SUSPEND flag set. So far the code was only
ensuring that POLLOUT doesn't get set for non-blocking sockets using
poll() but there was no code in place to ensure that blocking sockets do
the right thing when writing to them.

This patch adds a new bt_sock_wait_ready helper function to sleep in the
sendmsg call if the BT_SK_SUSPEND flag is set, and wake up as soon as it
is unset. It also updates the L2CAP and RFCOMM sendmsg callbacks to take
advantage of this new helper function.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

When we have an LE link we should not respond to any data on the BR/EDR
L2CAP signaling channel (0x0001) and vice-versa when we have a BR/EDR
link we should not respond to LE L2CAP (CID 0x0005) signaling commands.
This patch fixes this issue by checking for a valid link type and
ignores data if it is wrong.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

When L2CAP packets return a non-zero error and the value is passed
onwards by l2cap_bredr_sig_cmd this will trigger a command reject packet
to be sent. However, the core specification (page 1416 in core 4.0) says
the following: "Command Reject packets should not be sent in response to
an identified Response packet.".

This patch ensures that a command reject packet is not sent for any
identified response packet by ignoring the error return value from the
response handler functions.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

There are several possible reason codes that can be sent in the command
reject L2CAP packet. Before this patch the code has used a hard-coded
single response code ("command not understood"). This patch adds a
helper function to map the return value of an L2CAP handler function to
the correct command reject reason.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

If we receive an L2CAP Disconnect Request for an unknown CID we should
not just silently drop it but reply with a proper Command Reject
response. This patch fixes this by ensuring that the disconnect handler
returns a proper error instead of 0 and will cause the function caller
to send the right response.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The EFAULT error should only be used for memory address related errors
and ENOENT might be needed for other purposes than invalid CID errors.
This patch fixes the l2cap_config_req, l2cap_connect_create_rsp and
l2cap_create_channel_req handlers to use the unique EBADSLT error to
indicate failed lookups on a given CID.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

When an L2CAP request handler returns non-zero the calling code will
send a command reject response. The l2cap_create_chan_req function will
in some cases send its own response but then still return a -EFAULT
error which would cause two responses to be sent. This patch fixes this
by making the function return 0 after sending its own response.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The struct for HCI_Set_Event_Mask is never used. Instead a local 8-byte
array is used for sending this command. Therefore, remove the
unnecessary struct definition.
Signed-off-by: NJohan Hedberg <johan.hedberg@intel.com>
Acked-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

When the user channel is set and an user application has full control
over the device, do not bother trying to schedule any queues except
the raw queue.

This is an optimization since with user channel, only the raw queue
is in use.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Acked-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

There is no need to use GFP_ATOMIC with skb_clone() when the code is
executed in a workqueue.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Acked-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

When the device has the user channel flag set, it means it is driven by
an user application. In that case do not allow any connections from
L2CAP or SCO sockets.

This is the same situation as when the device has the raw flag set and
it will then return EHOSTUNREACH.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Acked-by: NJohan Hedberg <johan.hedberg@intel.com>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

So far the only option to create a virtual AMP controller was by
setting a module parameter for the hci_vhci driver. This patch adds
the functionality to define inline to create either a BR/EDR or an
AMP controller.

In addition the client will be informed which HCI controller index
it got assigned. That is especially useful for automated end-to-end
testing.

To keep backwards compatibility with existing userspace, the command
for creating a controller type needs to be send right after opening
the device node. If the command is not send, it defaults back to
automatically creating a BR/EDR controller.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

To allow creating /dev/vhci device node, add the proper module alias for
this driver.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

This patch introcuces a new HCI socket channel that allows user
applications to take control over a specific HCI device. The application
gains exclusive access to this device and forces the kernel to stay away
and not manage it. In case of the management interface it will actually
hide the device.

Such operation is useful for security testing tools that need to operate
underneath the Bluetooth stack and need full control over a device. The
advantage here is that the kernel still provides the service of hardware
abstraction and HCI level access. The use of Bluetooth drivers for
hardware access also means that sniffing tools like btmon or hcidump
are still working and the whole set of transaction can be traced with
existing tools.

With the new channel it is possible to send HCI commands, ACL and SCO
data packets and receive HCI events, ACL and SCO packets from the
device. The format follows the well established H:4 protocol.

The new HCI user channel can only be established when a device has been
through its setup routine and is currently powered down. This is
enforced to not cause any problems with current operations. In addition
only one user channel per HCI device is allowed. It is exclusive access
for one user application. Access to this channel is limited to process
with CAP_NET_RAW capability.

Using this new facility does not require any external library or special
ioctl or socket filters. Just create the socket and bind it. After that
the file descriptor is ready to speak H:4 protocol.

        struct sockaddr_hci addr;
        int fd;

        fd = socket(AF_BLUETOOTH, SOCK_RAW, BTPROTO_HCI);

        memset(&addr, 0, sizeof(addr));
        addr.hci_family = AF_BLUETOOTH;
        addr.hci_dev = 0;
        addr.hci_channel = HCI_CHANNEL_USER;

        bind(fd, (struct sockaddr *) &addr, sizeof(addr));

The example shows on how to create a user channel for hci0 device. Error
handling has been left out of the example. However with the limitations
mentioned above it is advised to handle errors. Binding of the user
cahnnel socket can fail for various reasons. Specifically if the device
is currently activated by BlueZ or if the access permissions are not
present.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

This patch introduces a new user channel flag that allows to give full
control of a HCI device to a user application. The kernel will stay away
from the device and does not allow any further modifications of the
device states.

The existing raw flag is not used since it has a bit of unclear meaning
due to its legacy. Using a new flag makes the code clearer.

A device with the user channel flag set can still be enumerate using the
legacy API, but it does not longer enumerate using the new management
interface used by BlueZ 5 and beyond. This is intentional to not confuse
users of modern systems.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The various legacy ioctls used with HCI sockets are limited to raw
channel only. They are not used on the other channels and also have
no meaning there. So return an error if tried to use them.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The HCI sockets for monitor and control do not support any HCI specific
socket options and if tried, an error will be returned. However the
error used is EINVAL and that is not really descriptive. To make it
clear that these sockets are not handling HCI socket options, return
EBADFD instead.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

Even if this is legacy API, there is no reason to not report a proper
error when trying to reset a HCI device that is down.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The hci_dev check is not protected and so move it into the socket lock. In
addition return the HCI channel identifier instead of always 0 channel.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The HCI sockets do not have a peer associated with it and so make sure
that getpeername() returns EOPNOTSUPP since this operation is actually
not supported on HCI sockets.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

The handling of the raw socket filter is rather obscure code and it gets
in the way of future extensions. Instead of inline filtering in the raw
socket packet routine, refactor it into its own function.
Signed-off-by: NMarcel Holtmann <marcel@holtmann.org>
Signed-off-by: NGustavo Padovan <gustavo.padovan@collabora.co.uk>

Document force_mld_version parameter in ip-sysctl.txt.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We already have mld_{gq,ifc,dad}_start_timer() functions, so introduce
mld_{gq,ifc,dad}_stop_timer() functions to reduce code size and make it
more readable.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Make igmp6_event_query() a bit easier to read by refactoring code
parts into mld_process_v1() and mld_process_v2().
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Similarly as we do in MLDv2 queries, set a forged MLDv1 query with
0 ms mld_maxdelay to minimum timer shot time of 1 jiffies. This is
eventually done in igmp6_group_queried() anyway, so we can simplify
a check there.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RFC3810, 10. Security Considerations says under subsection 10.1.
Query Message:

  A forged Version 1 Query message will put MLDv2 listeners on that
  link in MLDv1 Host Compatibility Mode. This scenario can be avoided
  by providing MLDv2 hosts with a configuration option to ignore
  Version 1 messages completely.

Hence, implement a MLDv2-only mode that will ignore MLDv1 traffic:

  echo 2 > /proc/sys/net/ipv6/conf/ethX/force_mld_version  or
  echo 2 > /proc/sys/net/ipv6/conf/all/force_mld_version

Note that <all> device has a higher precedence as it was previously
also the case in the macro MLD_V1_SEEN() that would "short-circuit"
if condition on <all> case.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Get rid of MLDV2_MRC and use our new macros for mantisse and
exponent to calculate Maximum Response Delay out of the Maximum
Response Code.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace the macro with a function to make it more readable. GCC will
eventually decide whether to inline this or not (also, that's not
fast-path anyway).
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

i) RFC3810, 9.2. Query Interval [QI] says:

   The Query Interval variable denotes the interval between General
   Queries sent by the Querier. Default value: 125 seconds. [...]

ii) RFC3810, 9.3. Query Response Interval [QRI] says:

  The Maximum Response Delay used to calculate the Maximum Response
  Code inserted into the periodic General Queries. Default value:
  10000 (10 seconds) [...] The number of seconds represented by the
  [Query Response Interval] must be less than the [Query Interval].

iii) RFC3810, 9.12. Older Version Querier Present Timeout [OVQPT] says:

  The Older Version Querier Present Timeout is the time-out for
  transitioning a host back to MLDv2 Host Compatibility Mode. When an
  MLDv1 query is received, MLDv2 hosts set their Older Version Querier
  Present Timer to [Older Version Querier Present Timeout].

  This value MUST be ([Robustness Variable] times (the [Query Interval]
  in the last Query received)) plus ([Query Response Interval]).

Hence, on *default* the timeout results in:

  [RV] = 2, [QI] = 125sec, [QRI] = 10sec
  [OVQPT] = [RV] * [QI] + [QRI] = 260sec

Having that said, we currently calculate [OVQPT] (here given as 'switchback'
variable) as ...

  switchback = (idev->mc_qrv + 1) * max_delay

RFC3810, 9.12. says "the [Query Interval] in the last Query received". In
section "9.14. Configuring timers", it is said:

  This section is meant to provide advice to network administrators on
  how to tune these settings to their network. Ambitious router
  implementations might tune these settings dynamically based upon
  changing characteristics of the network. [...]

iv) RFC38010, 9.14.2. Query Interval:

  The overall level of periodic MLD traffic is inversely proportional
  to the Query Interval. A longer Query Interval results in a lower
  overall level of MLD traffic. The value of the Query Interval MUST
  be equal to or greater than the Maximum Response Delay used to
  calculate the Maximum Response Code inserted in General Query
  messages.

I assume that was why switchback is calculated as is (3 * max_delay), although
this setting seems to be meant for routers only to configure their [QI]
interval for non-default intervals. So usage here like this is clearly wrong.

Concluding, the current behaviour in IPv6's multicast code is not conform
to the RFC as switch back is calculated wrongly. That is, it has a too small
value, so MLDv2 hosts switch back again to MLDv2 way too early, i.e. ~30secs
instead of ~260secs on default.

Hence, introduce necessary helper functions and fix this up properly as it
should be.

Introduced in 06da92283 ("[IPV6]: Add MLDv2 support."). Credits to Hannes
Frederic Sowa who also had a hand in this as well. Also thanks to Hangbin Liu
who did initial testing.
Signed-off-by: NDaniel Borkmann <dborkman@redhat.com>
Cc: David Stevens <dlstevens@us.ibm.com>
Cc: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: NHannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 1b7fdd2a("tcp: do not use cached RTT for RTT estimation")
removes important comments on how RTO is initialized and updated.
Hopefully this patch puts those information back.
Signed-off-by: NYuchung Cheng <ycheng@google.com>
Acked-by: NNeal Cardwell <ncardwell@google.com>
Acked-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

vxlan-udp-recv function lookup vxlan_sock struct on every packet
recv by using udp-port number. we can use sk->sk_user_data to
store vxlan_sock and avoid lookup.
I have open coded rcu-api to store and read vxlan_sock from
sk_user_data to avoid sparse warning as sk_user_data is not
__rcu pointer.
Signed-off-by: NPravin B Shelar <pshelar@nicira.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is a nice helper to parse MAC. Let's use it and remove custom
implementation.
Signed-off-by: NAndy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

driver:stmmac: Adjust time stamp increase for 0.465 ns accurate only when Time stamp binary rollover is set.

The synopsys spec says When TSCRLSSR is cleard, the rollover value of
sub-second register is 0x7FFFFFFF(0.465 ns per clock).
Signed-off-by: NSonic Zhang <sonic.zhang@analog.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

net: sctp: Fix data chunk fragmentation for MTU values which are not multiple of 4

Initially the problem was observed with ipsec, but later it became clear that
SCTP data chunk fragmentation algorithm has problems with MTU values which are
not multiple of 4. Test program was used which just transmits 2000 bytes long
packets to other host. tcpdump was used to observe re-fragmentation in IP layer
after SCTP already fragmented data chunks.

With MTU 1500:
12:54:34.082904 IP (tos 0x2,ECT(0), ttl 64, id 0, offset 0, flags [DF], proto SCTP (132), length 1500)
    10.151.38.153.39303 > 10.151.24.91.54321: sctp (1) [DATA] (B) [TSN: 2366088589] [SID: 0] [SSEQ 1] [PPID 0x0]
12:54:34.082933 IP (tos 0x2,ECT(0), ttl 64, id 0, offset 0, flags [DF], proto SCTP (132), length 596)
    10.151.38.153.39303 > 10.151.24.91.54321: sctp (1) [DATA] (E) [TSN: 2366088590] [SID: 0] [SSEQ 1] [PPID 0x0]
12:54:34.090576 IP (tos 0x2,ECT(0), ttl 63, id 0, offset 0, flags [DF], proto SCTP (132), length 48)
    10.151.24.91.54321 > 10.151.38.153.39303: sctp (1) [SACK] [cum ack 2366088590] [a_rwnd 79920] [#gap acks 0] [#dup tsns 0]

With MTU 1499:
13:02:49.955220 IP (tos 0x2,ECT(0), ttl 64, id 48215, offset 0, flags [+], proto SCTP (132), length 1492)
    10.151.38.153.39084 > 10.151.24.91.54321: sctp[|sctp]
13:02:49.955249 IP (tos 0x2,ECT(0), ttl 64, id 48215, offset 1472, flags [none], proto SCTP (132), length 28)
    10.151.38.153 > 10.151.24.91: ip-proto-132
13:02:49.955262 IP (tos 0x2,ECT(0), ttl 64, id 0, offset 0, flags [DF], proto SCTP (132), length 600)
    10.151.38.153.39084 > 10.151.24.91.54321: sctp (1) [DATA] (E) [TSN: 404355346] [SID: 0] [SSEQ 1] [PPID 0x0]
13:02:49.956770 IP (tos 0x2,ECT(0), ttl 63, id 0, offset 0, flags [DF], proto SCTP (132), length 48)
    10.151.24.91.54321 > 10.151.38.153.39084: sctp (1) [SACK] [cum ack 404355346] [a_rwnd 79920] [#gap acks 0] [#dup tsns 0]

Here problem in data portion limit calculation leads to re-fragmentation in IP,
which is sub-optimal. The problem is max_data initial value, which doesn't take
into account the fact, that data chunk must be padded to 4-bytes boundary.
It's enough to correct max_data, because all later adjustments are correctly
aligned to 4-bytes boundary.

After the fix is applied, everything is fragmented correctly for uneven MTUs:
15:16:27.083881 IP (tos 0x2,ECT(0), ttl 64, id 0, offset 0, flags [DF], proto SCTP (132), length 1496)
    10.151.38.153.53417 > 10.151.24.91.54321: sctp (1) [DATA] (B) [TSN: 3077098183] [SID: 0] [SSEQ 1] [PPID 0x0]
15:16:27.083907 IP (tos 0x2,ECT(0), ttl 64, id 0, offset 0, flags [DF], proto SCTP (132), length 600)
    10.151.38.153.53417 > 10.151.24.91.54321: sctp (1) [DATA] (E) [TSN: 3077098184] [SID: 0] [SSEQ 1] [PPID 0x0]
15:16:27.085640 IP (tos 0x2,ECT(0), ttl 63, id 0, offset 0, flags [DF], proto SCTP (132), length 48)
    10.151.24.91.54321 > 10.151.38.153.53417: sctp (1) [SACK] [cum ack 3077098184] [a_rwnd 79920] [#gap acks 0] [#dup tsns 0]

The bug was there for years already, but
 - is a performance issue, the packets are still transmitted
 - doesn't show up with default MTU 1500, but possibly with ipsec (MTU 1438)
Signed-off-by: NAlexander Sverdlin <alexander.sverdlin@nsn.com>
Acked-by: NVlad Yasevich <vyasevich@gmail.com>
Acked-by: NNeil Horman <nhorman@tuxdriver.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>