Ido Schimmel says:

====================
selftests: forwarding: Add tests for VXLAN routing

VXLAN routing allows hosts in different overlay networks (i.e.,
different VNIs) to communicate with one another.

Two popular routing models are asymmetric and symmetric routing.

In asymmetric routing the ingress VTEP routes the packet into the
correct VXLAN tunnel, whereas the egress VTEP only bridges the packet to
the correct host. Therefore, packets in different directions use
different VNIs - the target VNI.

In symmetric routing both the ingress and egress VTEPs perform routing
in the overlay network into / from the VXLAN tunnel. Packets in
different directions use the same VNI - the L3 VNI. Different tenants
(VRFs) use different L3 VNIs.

Patch #1 adds a test for asymmetric routing. Patches #2-#3 reuse the
topology and add test cases for ARP decapsulation and suppression.

Patch #4 adds a test for symmetric routing.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

ARP suppression allows the Linux bridge to answer ARP requests on behalf
of remote hosts. It reduces the amount of packets a VTEP needs to flood.

This test verifies that ARP suppression on / off works when a neighbour
exists and when it does not exist. It does so by sending an ARP request
from a host connected to one VTEP and checking whether it was received
by a second VTEP.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In a similar fashion to the asymmetric test, add a test for symmetric
routing. In symmetric routing both the ingress and egress VTEPs perform
routing in the overlay network into / from the VXLAN tunnel. Packets in
different directions use the same VNI - the L3 VNI. Different tenants
(VRFs) use different L3 VNIs.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Verify that ARP packets are correctly decapsulated by the ingress VTEP
by removing the neighbours configured on both VLAN interfaces and
running a ping test.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

In asymmetric routing the ingress VTEP routes the packet into the
correct VXLAN tunnel, whereas the egress VTEP only bridges the packet to
the correct host. Therefore, packets in different directions use
different VNIs - the target VNI.

The test uses a simple topology with two VTEPs and two VNIs and verifies
that ping passes between hosts (local / remote) in the same VLAN (VNI)
and in different VLANs belonging to the same tenant (VRF).

While the test does not check VM mobility, it does configure an anycast
gateway using a macvlan device on both VTEPs.
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Petr Machata <petrm@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Yangbo Lu says:

====================
External trigger stamp fifo support for ptp_qoriq

This patch-set is to add external trigger stamp fifo support by a new
binding "fsl,extts-fifo", and to add fiper pulse loopback support which
is very useful for validating trigger without external hardware.
Also fixed issues in interrupt enabling/handling.

"fsl,extts-fifo" is required to be added into 1588 timer dts node whose
hardware supports it. The work will be done for some QorIQ platforms dts in
the near future.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Added drivers/ptp/ptp_qoriq_debugfs.c into QorIQ PTP clock driver list.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add debugfs support for ptp_qoriq. Current debugfs
supports to control fiper1/fiper2 loopback mode. If the loopback mode
is enabled, the fiper1/fiper2 pulse is looped back into trigger1/
trigger2 input. This is very useful for validating hardware and driver
without external hardware. Below is an example to enable fiper1 loopback.

echo 1 > /sys/kernel/debug/2d10e00.ptp_clock/fiper1-loopback
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch is to add external trigger stamp fifo support
for 1588 timer.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Documented "fsl,extts-fifo" property.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The external trigger stamp FIFO was introduced as a new feature
for QorIQ 1588 timer IP block. This patch is to support it by
adding a new dts property "fsl,extts-fifo". Any QorIQ 1588 timer
supporting this feature is required to add this property in its
dts node.

In addition, the FIFO should be cleaned up before enabling external
trigger interrupts. Otherwise, there will be interrupts immediately
just after enabling external trigger interrupts.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The tmr_tevent register would update event bits
no matter tmr_temask bits were set or not. So we
should get interrupts by tmr_tevent & tmr_temask,
and clean up interrupts in tmr_tevent before
enabling them.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

nlmsg_put may fail, this fix add a check of its return value.
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Vinod Koul says:

====================
net: Add support for Qualcomm ethqos

Some Qualcomm SoCs sport a ethqos controller which use DW ip, so add
the glue driver which uses stmmac driver along with DT bindings for
this device.

This controller supports rgmii mode and doesn't work with existing
phy drivers as they do not remove the phy delay delay in this mode,
so fix the two phy drivers tested with this.

Changes in v3:
 - Add description in DT and rename the file and compatible as suggested by
   Rob
 - Update changelog for QCA8K driver
 - Update AT803x phy disable delay for all RGMxx modes

Changes in v2:
 - Fix the example in dt-binding
 - Remove DT property for disable the delay and disable delay for RGMII mode
   in AT803x and QCA8K PHY drivers
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

In RGMII mode we should not have any delay in port MAC, so disable
the delay.
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

For RGMII mode, phy delay should be disabled. Add this case along
with disable delay routines.
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add myself and Niklas as maintainers for this driver
Signed-off-by: Niklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add glue driver to support Qualcomm ETHQOS using stmmac driver.

This is based on downstream driver written by Siddarth Gupta, Sunil
Kumar Paidimarri, Rahul Ankushrao Kawadgave, Nisha Menon, Jagadeesh
Babu Challagundla, Chaitanya Pratapa, Lakshit Tyagi, Suraj Jaiswal,
Sneh Shah and Ventrapragada Ravi Kanth
Co-developed-by: Niklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: Niklas Cassel <niklas.cassel@linaro.org>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for Qualcomm ethqos found in some SoCs like QCS404.
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds support for the SIOCGHWTSTAMP ioctl which enables user
processes to read the current hwtstamp_config settings
non-destructively.
Signed-off-by: Artem Panfilov <panfilov.artyom@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Linus Lüssing says:

====================
bridge: implement Multicast Router Discovery (RFC4286)

This patchset adds initial Multicast Router Discovery support to
the Linux bridge (RFC4286). With MRD it is possible to detect multicast
routers and mark bridge ports and forward multicast packets to such routers
accordingly.

So far, multicast routers are detected via IGMP/MLD queries and PIM
messages in the Linux bridge. As there is only one active, selected
querier at a time RFC4541 ("Considerations for Internet Group Management
Protocol (IGMP) and Multicast Listener Discovery (MLD) Snooping
Switches") section 2.1.1.a) recommends snooping Multicast Router
Advertisements as provided by MRD (RFC4286).

The first two patches are refactoring some existing code which is reused
for parsing the Multicast Router Advertisements later in the fourth
patch. The third patch lets the bridge join the all-snoopers multicast
address to be able to reliably receive the Multicast Router
Advertisements.

What is not implemented yet from RFC4286 yet:

* Sending Multicast Router Solicitations:
  -> RFC4286: "[...] may be sent when [...] an interface is
     (re-)initialized [or] MRD is enabled"
* Snooping Multicast Router Terminations:
  -> currently this only relies on our own timeouts
* Adjusting timeouts with the values provided in the announcements

Changes in v2:

* rebased to current net-next/master (no conflicts/changes)
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

When multiple multicast routers are present in a broadcast domain then
only one of them will be detectable via IGMP/MLD query snooping. The
multicast router with the lowest IP address will become the selected and
active querier while all other multicast routers will then refrain from
sending queries.

To detect such rather silent multicast routers, too, RFC4286
("Multicast Router Discovery") provides a standardized protocol to
detect multicast routers for multicast snooping switches.

This patch implements the necessary MRD Advertisement message parsing
and after successful processing adds such routers to the internal
multicast router list.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>

Next to snooping IGMP/MLD queries RFC4541, section 2.1.1.a) recommends
to snoop multicast router advertisements to detect multicast routers.

Multicast router advertisements are sent to an "all-snoopers"
multicast address. To be able to receive them reliably, we need to
join this group.

Otherwise other snooping switches might refrain from forwarding these
advertisements to us.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>

With this patch the internal use of the skb_trimmed is reduced to
the ICMPv6/IGMP checksum verification. And for the length checks
the newly introduced helper functions are used instead of calculating
and checking with skb->len directly.

These changes should hopefully make it easier to verify that length
checks are performed properly.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch refactors ip_mc_check_igmp(), ipv6_mc_check_mld() and
their callers (more precisely, the Linux bridge) to not rely on
the skb_trimmed parameter anymore.

An skb with its tail trimmed to the IP packet length was initially
introduced for the following three reasons:

1) To be able to verify the ICMPv6 checksum.
2) To be able to distinguish the version of an IGMP or MLD query.
   They are distinguishable only by their size.
3) To avoid parsing data for an IGMPv3 or MLDv2 report that is
   beyond the IP packet but still within the skb.

The first case still uses a cloned and potentially trimmed skb to
verfiy. However, there is no need to propagate it to the caller.
For the second and third case explicit IP packet length checks were
added.

This hopefully makes ip_mc_check_igmp() and ipv6_mc_check_mld() easier
to read and verfiy, as well as easier to use.
Signed-off-by: Linus Lüssing <linus.luessing@c0d3.blue>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds PCI device IDs to support following cards:

1. Add device id 0x0205 for HINIC 100GE dual port mezz card.
2. Add device id 0x0210 for HINIC 25GE quad port mezz card.
3. Delete device id 0x0201 for HINIC 100GE dual port card, because
this is used by other product.
4. Macro of device id 0x200 is modified for HINIC 100GE dual port card.
Signed-off-by: Xue Chaojing <xuechaojing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The old non-PCIe chip versions support PCI DAC, however this feature
seems to be fragile, see comment in the PCI error handler. Therefore
it's disabled per default. I think meanwhile it's time remove support
for this legacy feature. This helps to reduce complexity of the driver.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8 years ago, as part of 6f0333b8 ("r8169: use 50% less ram for RX
ring"), the alignment requirement for rx buffers was silently changed
from 8 bytes to 16 bytes. I found nothing explaining this, also the
chip specs I have only mention an 8 byte requirement.
AFAICS kmalloc_node() guarantees allocated memory to be at least
"long long" aligned, what is 8 bytes on a 32 bit machine.
So we can take this memory as-is and avoid some overhead by changing
the alignment requirement back to 8 bytes.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently the soft reset (if defined) is done only if the driver also
implements the config_init callback. I think this dependency is a
mistake, so let's remove it.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

It was reported that on a system with nfsboot and w/o initramfs network
fails because trying to load the PHY driver returns -ENOENT. Reason was
that due to missing initramfs the modprobe binary isn't available.
So we have to ignore error code -ENOENT.

Fixes: 13d0ab67 ("net: phy: check return code when requesting PHY driver module")
Reported-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Tested-by: Krzysztof Kozlowski <krzk@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2019-01-22

This series contains updates to i40e and xsk.

Jan exports xdp_get_umem_from_qid() for other drivers/modules to use.
Refactored the code use the netdev provided umems, instead of containing
them inside our i40e_vsi.

Aleksandr fixes an issue where RSS queues were misconfigured, so limit
the RSS queue number to the online CPU number.

Damian adds support for ethtool's setting and getting the FEC
configuration.

Grzegorz fixes a type mismatch, where the return value was not matching
the function declaration.

Sergey adds checks in the queue configuration handler to ensure the
number of queue pairs requested by the VF is less than maximum possible.

Lihong cleans up code left around from earlier silicon validation in the
i40e debugfs code.

Julia Lawall and Colin Ian King clean up white space indentation issues
found.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Nikolay Aleksandrov says:

====================
bonding: add 3ad stats and export them via xstats

This set adds support for counting some 3ad-specific packet types and
exports the new stats via the xstats API. atomic64 counters are used
since these are not fastpaths and we can avoid the per-cpu allocations.
Each 3ad counter is exported as a separate attribute to be easily
extensible since we plan to add more later. The stats are per-slave and
when the master stats are requested the slaves' stats are summed up.
Patches 01 and 02 do minor cleanups in preparation for the new stats
API. Patch 03 adds the new stats and patch 04 adds xstats support to
export them.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds support for extended statistics (xstats) call to the
bonding. The first user would be the 3ad code which counts the following
events:
 - LACPDU Rx/Tx
 - LACPDU unknown type Rx
 - LACPDU illegal Rx
 - Marker Rx/Tx
 - Marker response Rx/Tx
 - Marker unknown type Rx

All of these are exported via netlink as separate attributes to be
easily extensible as we plan to add more in the future.
Similar to how the bridge and other xstats exports, the structure
inside is:
 [ IFLA_STATS_LINK_XSTATS ]
   -> [ LINK_XSTATS_TYPE_BOND ]
        -> [ BOND_XSTATS_3AD ]
             -> [ 3ad stats attributes ]

With this structure it's easy to add more stat types later.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Count the following types of 3ad packets per slave:
 - rx/tx lacpdu
 - rx/tx marker
 - rx/tx marker response
 - rx illegal lacpdus (right now counted on wrong length)
 - rx unknown lacpdu type
 - rx unknown marker type

The counters are using atomic64 since this is not fast path.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Since the received lacpdu is accessed via skb_header_pointer() in
bond_3ad_lacpdu_recv() we no longer need to check for skb->len's length.
If the returned lacpdu pointer is not null that should be enough.
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

No functional changes, adjust the style of bond_3ad_rx_indication to
prepare it for the stats changes:
 - reduce indentation by returning early on wrong length
 - remove extra new lines between switch cases
 - add marker local variable and use it to reduce line length
 - rearrange local variables in reverse xmas tree
 - separate final return
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Enable Inline TLS record by default
Signed-off-by: Atul Gupta <atul.gupta@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

free tls context in sock destruct. close may not be the last
call to free sock but force releasing the ctx in close
will result in GPF when ctx referred again in tcp_done

[  515.330477] general protection fault: 0000 [#1] SMP PTI
[  515.330539] CPU: 5 PID: 0 Comm: swapper/5 Not tainted 4.20.0-rc7+ #10
[  515.330657] Hardware name: Supermicro X8ST3/X8ST3, BIOS 2.0b
11/07/2013
[  515.330844] RIP: 0010:tls_hw_unhash+0xbf/0xd0
[
[  515.332220] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  515.332340] CR2: 00007fab32c55000 CR3: 000000009261e000 CR4:
00000000000006e0
[  515.332519] Call Trace:
[  515.332632]  <IRQ>
[  515.332793]  tcp_set_state+0x5a/0x190
[  515.332907]  ? tcp_update_metrics+0xe3/0x350
[  515.333023]  tcp_done+0x31/0xd0
[  515.333130]  tcp_rcv_state_process+0xc27/0x111a
[  515.333242]  ? __lock_is_held+0x4f/0x90
[  515.333350]  ? tcp_v4_do_rcv+0xaf/0x1e0
[  515.333456]  tcp_v4_do_rcv+0xaf/0x1e0
Signed-off-by: Atul Gupta <atul.gupta@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

build protos is required for tls_hw_prot also hence moved to
'tls_build_proto' and called as required from tls_init
and tls_hw_proto. This is required since build_protos
for v4 is moved from tls_register to tls_init in
commit <28cb6f1e>
Signed-off-by: Atul Gupta <atul.gupta@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

There have been many people complaining about the inconsistent
behaviors of IPv4 and IPv6 devconf when creating new network
namespaces.  Currently, for IPv4, we inherit all current settings
from init_net, but for IPv6 we reset all setting to default.

This patch introduces a new /proc file
/proc/sys/net/core/devconf_inherit_init_net to control the
behavior of whether to inhert sysctl current settings from init_net.
This file itself is only available in init_net.

As demonstrated below:

Initial setup in init_net:
 # cat /proc/sys/net/ipv4/conf/all/rp_filter
 2
 # cat /proc/sys/net/ipv6/conf/all/accept_dad
 1

Default value 0 (current behavior):
 # ip netns del test
 # ip netns add test
 # ip netns exec test cat /proc/sys/net/ipv4/conf/all/rp_filter
 2
 # ip netns exec test cat /proc/sys/net/ipv6/conf/all/accept_dad
 0

Set to 1 (inherit from init_net):
 # echo 1 > /proc/sys/net/core/devconf_inherit_init_net
 # ip netns del test
 # ip netns add test
 # ip netns exec test cat /proc/sys/net/ipv4/conf/all/rp_filter
 2
 # ip netns exec test cat /proc/sys/net/ipv6/conf/all/accept_dad
 1

Set to 2 (reset to default):
 # echo 2 > /proc/sys/net/core/devconf_inherit_init_net
 # ip netns del test
 # ip netns add test
 # ip netns exec test cat /proc/sys/net/ipv4/conf/all/rp_filter
 0
 # ip netns exec test cat /proc/sys/net/ipv6/conf/all/accept_dad
 0

Set to a value out of range (invalid):
 # echo 3 > /proc/sys/net/core/devconf_inherit_init_net
 -bash: echo: write error: Invalid argument
 # echo -1 > /proc/sys/net/core/devconf_inherit_init_net
 -bash: echo: write error: Invalid argument
Reported-by: Zhu Yanjun <Yanjun.Zhu@windriver.com>
Reported-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Cc: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Acked-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>