This patch moves the udp_rmem_min, udp_wmem_min
to namespace and init the udp_l3mdev_accept explicitly.

The udp_rmem_min/udp_wmem_min affect udp rx/tx queue,
with this patch namespaces can set them differently.
Signed-off-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

David Ahern says:

====================
net/ipv6: Address checks need to consider the L3 domain

IPv6 prohibits a local address from being used as a gateway for a route.
However, it is ok for the gateway to be a local address in a different L3
domain (e.g., VRF). This allows, for example, veth pairs to connect VRFs.

ip6_route_info_create calls ipv6_chk_addr_and_flags for gateway addresses
to determine if the address is a local one, but ipv6_chk_addr_and_flags
does not currently consider L3 domains. As a result routes can not be
added in one VRF with a nexthop that points to a local address in a
second VRF.

Resolve by comparing the l3mdev for the passed in device and requiring an
l3mdev match with the device containing an address. The intent of checking
for an address on the specified device versus any device in the domain is
mantained by a new argument to skip the check between the passed in device
and the device with the address.

Patch 1 moves the gateway validation from ip6_route_info_create into a
helper; the function is long enough and refactoring drops the indent
level.

Patch 2 adds a skip_dev_check argument to ipv6_chk_addr_and_flags to
allow a device to always be passed yet skip the device check when
looking at addresses and fixes up a few ipv6_chk_addr callers that
pass a NULL device.

Patch 3 adds l3mdev checks to ipv6_chk_addr_and_flags.

Patches 4 and 5 do some refactoring to the fib_tests script and then
patch 6 adds nexthop validation tests.

v4
- separated l3mdev check into a separate patch (patch 3 of this set)
  as suggested by Kirill
- consolidated dev and ipv6_chk_addr_and_flags call into 1 if (Kirill)
- added a temp variable for gw type (Kirill)

v3
- set skip_dev_check in ipv6_chk_addr based on dev == NULL (per
  comment from Ido)

v2
- handle 2 variations of route spec with sane error path
- add test cases
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Add series of tests for valid and invalid nexthop specs for IPv6.

$ TEST=fib_nexthop_test ./fib_tests.sh
...
IPv6 nexthop tests
    TEST: Directly connected nexthop, unicast address              [ OK ]
    TEST: Directly connected nexthop, unicast address with device  [ OK ]
    TEST: Gateway is linklocal address                             [ OK ]
    TEST: Gateway is linklocal address, no device                  [ OK ]
    TEST: Gateway can not be local unicast address                 [ OK ]
    TEST: Gateway can not be local unicast address, with device    [ OK ]
    TEST: Gateway can not be a local linklocal address             [ OK ]
    TEST: Gateway can be local address in a VRF                    [ OK ]
    TEST: Gateway can be local address in a VRF, with device       [ OK ]
    TEST: Gateway can be local linklocal address in a VRF          [ OK ]
    TEST: Redirect to VRF lookup                                   [ OK ]
    TEST: VRF route, gateway can be local address in default VRF   [ OK ]
    TEST: VRF route, gateway can not be a local address            [ OK ]
    TEST: VRF route, gateway can not be a local addr with device   [ OK ]
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Allow a user to run just a specific fib test by setting the TEST
environment variable.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Replace 'ip -netns testns' with the alias IP. Shortens the line lengths
and makes running the commands manually a bit easier.
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Lookup the L3 master device for the passed in device. Only consider
addresses on netdev's with the same master device. If the device is
not enslaved or is NULL, then the l3mdev is NULL which means only
devices not enslaved (ie, in the default domain) are considered.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6_chk_addr_and_flags determines if an address is a local address and
optionally if it is an address on a specific device. For example, it is
called by ip6_route_info_create to determine if a given gateway address
is a local address. The address check currently does not consider L3
domains and as a result does not allow a route to be added in one VRF
if the nexthop points to an address in a second VRF. e.g.,

    $ ip route add 2001:db8:1::/64 vrf r2 via 2001:db8:102::23
    Error: Invalid gateway address.

where 2001:db8:102::23 is an address on an interface in vrf r1.

ipv6_chk_addr_and_flags needs to allow callers to always pass in a device
with a separate argument to not limit the address to the specific device.
The device is used used to determine the L3 domain of interest.

To that end add an argument to skip the device check and update callers
to always pass a device where possible and use the new argument to mean
any address in the domain.

Update a handful of users of ipv6_chk_addr with a NULL dev argument. This
patch handles the change to these callers without adding the domain check.

ip6_validate_gw needs to handle 2 cases - one where the device is given
as part of the nexthop spec and the other where the device is resolved.
There is at least 1 VRF case where deferring the check to only after
the route lookup has resolved the device fails with an unintuitive error
"RTNETLINK answers: No route to host" as opposed to the preferred
"Error: Gateway can not be a local address." The 'no route to host'
error is because of the fallback to a full lookup. The check is done
twice to avoid this error.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Move gateway validation code from ip6_route_info_create into
ip6_validate_gw. Code move plus adjustments to handle the potential
reset of dev and idev and to make checkpatch happy.
Signed-off-by: David Ahern <dsahern@gmail.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Brad Mouring says:

====================
net: macb: Introduce phy-handle DT functionality

Consider the situation where a macb netdev is connected through
a phydev that sits on a mii bus other than the one provided to
this particular netdev. This situation is what this patchset aims
to accomplish through the existing phy-handle optional binding.

This optional binding (as described in the ethernet DT bindings doc)
directs the netdev to the phydev to use. This is precisely the
situation this patchset aims to solve, so it makes sense to introduce
the functionality to this driver (where the physical layout discussed
was encountered).

The devicetree snippet would look something like this:

...
   ethernet@feedf00d {
           ...
           phy-handle = <&phy0> // the first netdev is physically wired to phy0
           ...
           phy0: phy@0 {
                   ...
                   reg = <0x0> // MDIO address 0
                   ...
           }
           phy1: phy@1 {
                   ...
                   reg = <0x1> // MDIO address 1
                   ...
           }
           ...
   }

   ethernet@deadbeef {
           ...
           phy-handle = <&phy1> // tells the driver to use phy1 on the
                                // first mac's mdio bus (it's wired thusly)
           ...
   }
...

The work done to add the phy_node in the first place (dacdbb4d:
"net: macb: add fixed-link node support") will consume the
device_node (if found).

v2: Reorganization of mii probe/init functions, suggested by Andrew Lunn
v3: Moved some of the bus init code back into init (erroneously moved to probe)
    some style issues, and an unintialized variable warning addressed.
v4: Add Reviewed-by: tags
    Skip fallback code if phy-handle phandle is found
v5: Cleanup formatting issues
    Fix compile failure introduced in 1/4 "net: macb: Reorganize macb_mii
        bringup"
    Fix typo in "Documentation: macb: Document phy-handle binding"
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Document the existence of the optional binding, directing to the
general ethernet document that describes this binding.
Signed-off-by: Brad Mouring <brad.mouring@ni.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This optional binding (as described in the ethernet DT bindings doc)
directs the netdev to the phydev to use. This is useful for a phy
chip that has >1 phy in it, and two netdevs are using the same phy
chip (i.e. the second mac's phy lives on the first mac's MDIO bus)

The devicetree snippet would look something like this:

ethernet@feedf00d {
	...
	phy-handle = <&phy0> // the first netdev is physically wired to phy0
	...
	phy0: phy@0 {
		...
		reg = <0x0> // MDIO address 0
		...
	}
	phy1: phy@1 {
		...
		reg = <0x1> // MDIO address 1
		...
	}
...
}

ethernet@deadbeef {
	...
	phy-handle = <&phy1> // tells the driver to use phy1 on the
						 // first mac's mdio bus (it's wired thusly)
	...
}

The work done to add the phy_node in the first place (dacdbb4d:
"net: macb: add fixed-link node support") will consume the
device_node (if found).
Signed-off-by: Brad Mouring <brad.mouring@ni.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In phy_device's general probe, this device will already be set for
phy register polling, rendering this code redundant.
Signed-off-by: Brad Mouring <brad.mouring@ni.com>
Suggested-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The macb mii setup (mii_probe() and mii_init()) previously was
somewhat interspersed, likely a result of organic growth and hacking.

This change moves mii bus registration into mii_init and probing the
bus for devices into mii_probe.
Signed-off-by: Brad Mouring <brad.mouring@ni.com>
Suggested-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The packet_mmap documentation had links to no longer existing web
sites; replace with other site which has similar example.

Support for packet mmap has been in mainline versions of libpcap
for several years.
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit d1179094 ("enic: Add vxlan offload support for IPv6 pkts")
added vxlan offload support for IPv6 pkts. Required change in
enic_udp_tunnel_del was not made. This creates a bug where once user
adds IPv6 tunnel, hw offload for that cannot be deleted.

This patch removes check for IP proto in tunnel delete path. Driver need
not check for IP proto since same UDP port cannot be used to create two
tunnels.

Fixes: d1179094 ("enic: Add vxlan offload support for IPv6 pkts")
Signed-off-by: Govindarajulu Varadarajan <gvaradar@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The variable 'len' is being initialized with a value that is never
read and it is re-assigned later, hence the initialization is redundant
and can be removed.

Cleans up clang warning:
net/rxrpc/recvmsg.c:275:15: warning: Value stored to 'len' during its
initialization is never read
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

syzbot/kasan detected a double free in sctp_sendmsg_to_asoc:
BUG: KASAN: use-after-free in sctp_association_free+0x7b7/0x930
net/sctp/associola.c:332
Read of size 8 at addr ffff8801d8006ae0 by task syzkaller914861/4202

CPU: 1 PID: 4202 Comm: syzkaller914861 Not tainted 4.16.0-rc4+ #258
Hardware name: Google Google Compute Engine/Google Compute Engine
01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:17 [inline]
 dump_stack+0x194/0x24d lib/dump_stack.c:53
 print_address_description+0x73/0x250 mm/kasan/report.c:256
 kasan_report_error mm/kasan/report.c:354 [inline]
 kasan_report+0x23c/0x360 mm/kasan/report.c:412
 __asan_report_load8_noabort+0x14/0x20 mm/kasan/report.c:433
 sctp_association_free+0x7b7/0x930 net/sctp/associola.c:332
 sctp_sendmsg+0xc67/0x1a80 net/sctp/socket.c:2075
 inet_sendmsg+0x11f/0x5e0 net/ipv4/af_inet.c:763
 sock_sendmsg_nosec net/socket.c:629 [inline]
 sock_sendmsg+0xca/0x110 net/socket.c:639
 SYSC_sendto+0x361/0x5c0 net/socket.c:1748
 SyS_sendto+0x40/0x50 net/socket.c:1716
 do_syscall_64+0x281/0x940 arch/x86/entry/common.c:287
 entry_SYSCALL_64_after_hwframe+0x42/0xb7

This was introduced by commit:
f84af331 sctp: factor out sctp_sendmsg_to_asoc from sctp_sendmsg

As the newly refactored function moved the wait_for_sndbuf call to a
point after the association was connected, allowing for peeloff events
to occur, which in turn caused wait_for_sndbuf to return -EPIPE which
was not caught by the logic that determines if an association should be
freed or not.

Fix it the easy way by returning the ordering of
sctp_primitive_ASSOCIATE and sctp_wait_for_sndbuf to the old order, to
ensure that EPIPE will not happen.

Tested by myself using the syzbot reproducers with positive results
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
CC: davem@davemloft.net
CC: Xin Long <lucien.xin@gmail.com>
Reported-by: syzbot+a4e4112c3aff00c8cfd8@syzkaller.appspotmail.com
Reviewed-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

skb_copy_expand without __GFP_NOWARN already does a dump_stack
on OOM so these messages are redundant.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2018-03-14

This series contains updates to i40e and i40evf only.

Corentin Labbe cleans up the left over FCoE files in the i40e driver.

Gustavo A R Silva fixes a cut and paste error.

Paweł fixes a race condition when the VF driver is loaded on a host and
virsh is trying to attach it to the virtual machine and set a MAC
address.  Resolve the issue by adding polling in i40e_ndo_set_vf_mac()
when the VF is in reset mode.

Jake cleans up i40e_vlan_rx_register() since this only used in a single
location, so just inline the contents of the function.  Created a helper
function to proper update the per-filter statistics when we delete it.
Factored out the re-enabling ATR and SB rules.  Fixed an issue when
re-enabling ATR after the last TCPv4 filter is removed and ntuple is
still active, we were not restoring the TCPv4 filter input set.

Filip modifies the permission check function to ensure that it knows how
many filters are being requested, which allows the check to ensure that
the total number of filters in a single request does not cause us to go
over the limit.

Mariusz fixed an issue where the wrong calculation of partition id was
being done on OCP PHY mezzanine cards, which in turn caused wake on LAN
to be disabled on certain ports.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

When we re-enable ATR we need to restore the input set for TCPv4
filters, in order for ATR to function correctly. We already do this for
the normal case of re-enabling ATR when disabling ntuple support.
However, when re-enabling ATR after the last TCPv4 filter is removed (but
when ntuple support is still active), we did not restore the TCPv4
filter input set.

This can cause problems if the TCPv4 filters from FDir had changed the
input set, as ATR will no longer behave as expected.

When clearing the ATR auto-disable flag, make sure we restore the TCPv4
input set to avoid this.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch overwrites number of ports for X722 devices with support
for OCP PHY mezzanine.
The old method with checking if port is disabled in the PRTGEN_CNF
register cannot be used in this case. When the OCP is removed, ports
were seen as disabled, which resulted in wrong calculation of partition
id, that caused WoL to be disabled on certain ports.
Signed-off-by: Mariusz Stachura <mariusz.stachura@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

A future patch needs to expand on the logic for re-enabling ATR. Doing
so would cause some code to break the 80-character line limit.

To reduce the level of indentation, factor out helper functions for
re-enabling ATR and SB rules.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

When hardware has trouble with a particular filter, we delete it from
the list. Unfortunately, we did not properly update the per-filter
statistic when doing so.

Create a helper function to handle this, and properly reduce the
necessary counter so that it tracks the number of active filters
properly.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

When VF requests adding of MAC filters the checking is done against number
of already present MAC filters not adding them at the same time. It makes
it possible to add a bunch of filters at once possibly exceeding
acceptable limit of I40E_VC_MAX_MAC_ADDR_PER_VF filters.

This happens because when checking vf->num_mac, we do not check how many
filters are being requested at once. Modify the check function to ensure
that it knows how many filters are being requested. This allows the
check to ensure that the total number of filters in a single request
does not cause us to go over the limit.

Additionally, move the check to within the lock to ensure that the
vf->num_mac is checked while holding the lock to maintain consistency.
We could have simply moved the call to i40e_vf_check_permission to
within the loop, but this could cause a request to be non-atomic, and
add some but not all the addresses, while reporting an error code. We
want to avoid this behavior so that users are not confused about which
filters have or have not been added.
Signed-off-by: Filip Sadowski <filip.sadowski@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

We used to use the function i40e_vlan_rx_register as a way to hook
into the now defunct .ndo_vlan_rx_register netdev hook. This was
removed but we kept the function around because we still used it
internally to control enabling or disabling of VLAN stripping.

As pointed out in upstream review, VLAN stripping is only used in a
single location and the previous function is quite small, just inline
it into i40e_restore_vlan() rather than carrying the function
separately.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix for "Resource temporarily unavailable" problem when virsh is
trying to attach a device to VM. When the VF driver is loaded on
host and virsh is trying to attach it to the VM and set a MAC
address, it ends with a race condition between i40e_reset_vf and
i40e_ndo_set_vf_mac functions. The bug is fixed by adding polling
in i40e_ndo_set_vf_mac function For when the VF is in Reset mode.
Signed-off-by: Paweł Jabłoński <pawel.jablonski@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

It seems this is a copy-paste error and that the proper variable to use
in this particular case is _src_ instead of _dst_.

Addresses-Coverity-ID: 1465282 ("Copy-paste error")
Fixes: 0075fa0f ("i40evf: Add support to apply cloud filters")
Signed-off-by: Gustavo A R Silva <garsilva@embeddedor.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

i40e_fcoe support was removed via commit 9eed69a9 ("i40e: Drop FCoE code from core driver files")
But this left files in place but un-compilable.
Let's finish the cleaning.
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Xin Long says:

====================
sctp: add support for some sctp auth APIs from RFC6458

This patchset mainly adds support for SCTP AUTH Information for sendmsg,
described in RFC6458:

    5.3.8.  SCTP AUTH Information Structure (SCTP_AUTHINFO)

and also adds a sockopt described in RFC6458:

    8.3.4.  Deactivate a Shared Key (SCTP_AUTH_DEACTIVATE_KEY)

and two types of events for AUTHENTICATION_EVENT described in RFC6458:

    6.1.8.  SCTP_AUTHENTICATION_EVENT:
             - SCTP_AUTH_NO_AUTH
             - SCTP_AUTH_FREE_KEY

After this patchset, we have fully support for sctp_sendv in kernel.

Note that this patchset won't touch that sctp options merge conflict.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add SCTP_AUTH_NO_AUTH type for AUTHENTICATION_EVENT,
as described in section 6.1.8 of RFC6458.

      SCTP_AUTH_NO_AUTH:  This report indicates that the peer does not
         support SCTP authentication as defined in [RFC4895].

Note that the implementation is quite similar as that of
SCTP_ADAPTATION_INDICATION.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add SCTP_AUTH_FREE_KEY type for AUTHENTICATION_EVENT,
as described in section 6.1.8 of RFC6458.

      SCTP_AUTH_FREE_KEY:  This report indicates that the SCTP
         implementation will no longer use the key identifier specified
         in auth_keynumber.

After deactivating a key, it would never be used again, which means
it's refcnt can't be held/increased by new chunks. But there may be
some chunks in out queue still using it. So only when refcnt is 1,
which means no chunk in outqueue is using/holding this key either,
this EVENT would be sent.

When users receive this notification, they could do DEL_KEY sockopt to
remove this shkey, and also tell the peer that this key won't be used
in any chunk thoroughly from now on, then the peer can remove it as
well safely.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add sockopt SCTP_AUTH_DEACTIVATE_KEY, as described in
section 8.3.4 of RFC6458.

This set option indicates that the application will no longer send user
messages using the indicated key identifier.

Note that RFC requires that only deactivated keys that are no longer used
by an association can be deleted, but for the backward compatibility, it
is not to check deactivated when deleting or replacing one sh_key.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add support for SCTP AUTH Information for sendmsg,
as described in section 5.3.8 of RFC6458.

With this option, you can provide shared key identifier used for
sending the user message.

It's also a necessary send info for sctp_sendv.

Note that it reuses sinfo->sinfo_tsn to indicate if this option is
set and sinfo->sinfo_ssn to save the shkey ID which can be 0.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

With refcnt support for sh_key, chunks auth sh_keys can be decided
before enqueuing it. Changing the active key later will not affect
the chunks already enqueued.

Furthermore, this is necessary when adding the support for authinfo
for sendmsg in next patch.

Note that struct sctp_chunk can't be grown due to that performance
drop issue on slow cpu, so it just reuses head_skb memory for shkey
in sctp_chunk.
Signed-off-by: Xin Long <lucien.xin@gmail.com>
Acked-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Ursula Braun says:

====================
net/smc: fixes 2018-03-14

here are smc changes for the net-next tree.
The first patch enables SMC to work with mlx5-RoCE-devices.
Patches 2 and 3 deal with link group freeing.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The free_work worker must be scheduled when the link group is
abnormally terminated.
Signed-off-by: Karsten Graul <kgraul@linux.vnet.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Make sure there is no pending or running free_work worker for the link
group when freeing the link group.
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

smc allocates a certain number of CQ entries for used RoCE devices. For
mlx5 devices the chosen constant number results in a large allocation
causing this warning:

[13355.124656] WARNING: CPU: 3 PID: 16535 at mm/page_alloc.c:3883 __alloc_pages_nodemask+0x2be/0x10c0
[13355.124657] Modules linked in: smc_diag(O) smc(O) xt_CHECKSUM iptable_mangle ipt_MASQUERADE nf_nat_masquerade_ipv4 iptable_nat nf_nat_ipv4 nf_nat nf_conntrack_ipv4 nf_defrag_ipv4 xt_conntrack nf_conntrack ipt_REJECT nf_reject_ipv4 xt_tcpudp bridge stp llc ip6table_filter ip6_tables iptable_filter mlx5_ib ib_core sunrpc mlx5_core s390_trng rng_core ghash_s390 prng aes_s390 des_s390 des_generic sha512_s390 sha256_s390 sha1_s390 sha_common ptp pps_core eadm_sch dm_multipath dm_mod vhost_net tun vhost tap sch_fq_codel kvm ip_tables x_tables autofs4 [last unloaded: smc]
[13355.124672] CPU: 3 PID: 16535 Comm: kworker/3:0 Tainted: G           O    4.14.0uschi #1
[13355.124673] Hardware name: IBM 3906 M04 704 (LPAR)
[13355.124675] Workqueue: events smc_listen_work [smc]
[13355.124677] task: 00000000e2f22100 task.stack: 0000000084720000
[13355.124678] Krnl PSW : 0704c00180000000 000000000029da76 (__alloc_pages_nodemask+0x2be/0x10c0)
[13355.124681]            R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:0 PM:0 RI:0 EA:3
[13355.124682] Krnl GPRS: 0000000000000000 00550e00014080c0 0000000000000000 0000000000000001
[13355.124684]            000000000029d8b6 00000000f3bfd710 0000000000000000 00000000014080c0
[13355.124685]            0000000000000009 00000000ec277a00 0000000000200000 0000000000000000
[13355.124686]            0000000000000000 00000000000001ff 000000000029d8b6 0000000084723720
[13355.124708] Krnl Code: 000000000029da6a: a7110200		tmll	%r1,512
                          000000000029da6e: a774ff29		brc	7,29d8c0
                         #000000000029da72: a7f40001		brc	15,29da74
                         >000000000029da76: a7f4ff25		brc	15,29d8c0
                          000000000029da7a: a7380000		lhi	%r3,0
                          000000000029da7e: a7f4fef1		brc	15,29d860
                          000000000029da82: 5820f0c4		l	%r2,196(%r15)
                          000000000029da86: a53e0048		llilh	%r3,72
[13355.124720] Call Trace:
[13355.124722] ([<000000000029d8b6>] __alloc_pages_nodemask+0xfe/0x10c0)
[13355.124724]  [<000000000013bd1e>] s390_dma_alloc+0x6e/0x148
[13355.124733]  [<000003ff802eeba6>] mlx5_dma_zalloc_coherent_node+0x8e/0xe0 [mlx5_core]
[13355.124740]  [<000003ff802eee18>] mlx5_buf_alloc_node+0x70/0x108 [mlx5_core]
[13355.124744]  [<000003ff804eb410>] mlx5_ib_create_cq+0x558/0x898 [mlx5_ib]
[13355.124749]  [<000003ff80407d40>] ib_create_cq+0x48/0x88 [ib_core]
[13355.124751]  [<000003ff80109fba>] smc_ib_setup_per_ibdev+0x52/0x118 [smc]
[13355.124753]  [<000003ff8010bcb6>] smc_conn_create+0x65e/0x728 [smc]
[13355.124755]  [<000003ff801081a2>] smc_listen_work+0x2d2/0x540 [smc]
[13355.124756]  [<0000000000162c66>] process_one_work+0x1be/0x440
[13355.124758]  [<0000000000162f40>] worker_thread+0x58/0x458
[13355.124759]  [<0000000000169e7e>] kthread+0x14e/0x168
[13355.124760]  [<00000000009ce8be>] kernel_thread_starter+0x6/0xc
[13355.124762]  [<00000000009ce8b8>] kernel_thread_starter+0x0/0xc
[13355.124762] Last Breaking-Event-Address:
[13355.124764]  [<000000000029da72>] __alloc_pages_nodemask+0x2ba/0x10c0
[13355.124764] ---[ end trace 34be38b581c0b585 ]---

This patch reduces the smc constant for the maximum number of allocated
completion queue entries SMC_MAX_CQE by 2 to avoid high round up values
in the mlx5 code, and reduces the number of allocated completion queue
entries even more, if the final allocation for an mlx5 device hits the
MAX_ORDER limit.
Reported-by: Ihnken Menssen <menssen@de.ibm.com>
Signed-off-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Added ethtool changes to show port type as TP (Twisted Pair) for
10GBASE-T ports. Same driver and firmware works for liquidio NIC with
SFP+ ports or TP ports.
Signed-off-by: Veerasenareddy Burru <veerasenareddy.burru@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Now we have ndo_xdp_xmit, switch to use it instead of the slow generic
XDP TX routine. XDP_TX on TAP gets ~20% improvements from ~1.5Mpps to
~1.8Mpps on 2.60GHz Core(TM) i7-5600U.
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>