Shannon Nelson says:

====================
ionic txrx updates

These are a few patches to do some cleanup in the packet
handling and give us more flexibility in tuning performance
by allowing us to put Tx handling on separate interrupts
when it makes sense for particular traffic loads.

v3: simplified queue count change logging, removed unnecessary
    check for no count change
v2: dropped the original patch 2 for ringsize change
    changed the separated tx/rx interrupts to use ethtool -L
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Add the capability to split the Tx queues onto their own
interrupts with their own napi contexts.  This gives the
opportunity for more direct control of Tx interrupt
handling, such as CPU affinity and interrupt coalescing,
useful for some traffic loads.

v2: use ethtool -L, not a vendor specific priv-flag
v3: simplify logging, drop unnecessary "no-change" tests
Reviewed-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>

We give the tx clean path its own budget and service routine in
order to give a little more leeway to be more aggressive, and
in preparation for coming changes.  We've found this gives us
a little better performance in some packet processing scenarios
without hurting other scenarios.
Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>

We really don't need to hit the Rx queue doorbell so many times,
we can wait to the end and cause a little less thrash.
Signed-off-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>

The GRE tunnel can be used to transport traffic that does not rely on a
Internet checksum (e.g. SCTP). The issue can be triggered creating a GRE
or GRETAP tunnel and transmitting SCTP traffic ontop of it where CRC
offload has been disabled. In order to fix the issue we need to
recompute the GRE csum in gre_gso_segment() not relying on the inner
checksum.
The issue is still present when we have the CRC offload enabled.
In this case we need to disable the CRC offload if we require GRE
checksum since otherwise skb_checksum() will report a wrong value.

Fixes: 90017acc ("sctp: Add GSO support")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We need to clear all of the bridge private skb variables as they can be
stale due to the packet being recirculated through the stack and then
transmitted through the bridge device. Similar memset is already done on
bridge's input. We've seen cases where proxyarp_replied was 1 on routed
multicast packets transmitted through the bridge to ports with neigh
suppress which were getting dropped. Same thing can in theory happen with
the port isolation bit as well.

Fixes: 821f1b21 ("bridge: add new BR_NEIGH_SUPPRESS port flag to suppress arp and nd flood")
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

"how" was used as a boolean. Change the type to bool, and improve
variable name
Signed-off-by: Florent Fourcot <florent.fourcot@wifirst.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

Second parameter of addrconf_ifdown "how" is used as a boolean
internally. It does not make sense to call it with something different
of 0 or 1.

This value is set to 2 in all git history.
Signed-off-by: Florent Fourcot <florent.fourcot@wifirst.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

Eelco Chaudron says:

====================
net: openvswitch: masks cache enhancements

This patchset adds two enhancements to the Open vSwitch masks cache.

Changes in v4 [patch 2/2 only]:
 - Remove null check before calling free_percpu()
 - Make ovs_dp_change() return appropriate error codes

Changes in v3 [patch 2/2 only]:
 - Use is_power_of_2() function
 - Use array_size() function
 - Fix remaining sparse errors

Changes in v2 [patch 2/2 only]:
 - Fix sparse warnings
 - Fix netlink policy items reported by Florian Westphal
====================
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch makes the masks cache size configurable, or with
a size of 0, disable it.
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add a counter that counts the number of masks cache hits, and
export it through the megaflow netlink statistics.
Reviewed-by: Paolo Abeni <pabeni@redhat.com>
Reviewed-by: Tonghao Zhang <xiangxia.m.yue@gmail.com>
Signed-off-by: Eelco Chaudron <echaudro@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Release skb memory in mvpp2_rx() if mvpp2_rx_refill routine fails

Fixes: b5015854 ("net: mvpp2: fix refilling BM pools in RX path")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Acked-by: Matteo Croce <mcroce@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

info cannot be NULL here since its being accessed earlier
in the function: nlmsg_parse(info->nlhdr...). Remove this
redundant NULL check.
Signed-off-by: Gaurav Singh <gaurav1086@gmail.com>
Reviewed-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: David S. Miller <davem@davemloft.net>

ovs_ct_put_key() is potentially copying uninitialized kernel stack memory
into socket buffers, since the compiler may leave a 3-byte hole at the end
of `struct ovs_key_ct_tuple_ipv4` and `struct ovs_key_ct_tuple_ipv6`. Fix
it by initializing `orig` with memset().

Fixes: 9dd7f890 ("openvswitch: Add original direction conntrack tuple to sw_flow_key.")
Suggested-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Peilin Ye <yepeilin.cs@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When openvswitch conntrack offload with act_ct action. Fragment packets
defrag in the ingress tc act_ct action and miss the next chain. Then the
packet pass to the openvswitch datapath without the mru. The over
mtu packet will be dropped in output action in openvswitch for over mtu.

"kernel: net2: dropped over-mtu packet: 1528 > 1500"

This patch add mru in the tc_skb_ext for adefrag and miss next chain
situation. And also add mru in the qdisc_skb_cb. The act_ct set the mru
to the qdisc_skb_cb when the packet defrag. And When the chain miss,
The mru is set to tc_skb_ext which can be got by ovs datapath.

Fixes: b57dc7c1 ("net/sched: Introduce action ct")
Signed-off-by: wenxu <wenxu@ucloud.cn>
Reviewed-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Bruno Thomsen says:

====================
Improve MDIO Ethernet PHY reset

This patch series is a result of trying to upstream a new device
tree for a TQMa7D based board[1][2]. Initial this DTS used some
deprecated PHY reset properties on the FEC device; NXP Ethernet
MAC also known as Freescale Fast Ethernet Controller.

When switching from FEC properties[3]:
"phy-reset-gpios"
"phy-reset-duration"
"phy-reset-post-delay"

To MDIO PHY properties[4]:
"reset-gpios"
"reset-assert-us"
"reset-deassert-us"

The result was that no Ethernet PHY device was detected on boot.

This issue could be worked around by disabling PHY type ID auto-
detection by using "ethernet-phy-id0022.1560" as compatible
string and not "ethernet-phy-ieee802.3-c22".

Upstreaming a DTS with this workaround was not accepted, so I
digged into the MDIO reset flow and found that it had a few
missing parts compared to the deprecated FEC reset function.
After some more testing and logic analyzer traces it was
revealed that the failed PHY communication was due to missing
initial device reset.

I was suggested[5] in a earlier mail thread to use MDIO bus
reset as that was performed before auto-detection, but current
device tree binding was limited to reset assert in usec.
Microchip/Micrel Ethernet PHYs recommended reset circuit[8],
figure 7-12, is a little "slow" after reset deassert as that
is left to a RC circuit with a tau of ~100ms; using a 10k PU
resistor together with a 10uF decoupling capacitor. The diode
in serie of the reset signal converts the GPIO push-pull output
into a open-drain output. So a post reset delay in the range
of 500-1000ms is needed, depending on component tolerances
and general hardware design margins.

In the first version of this patch series[6] I reused the
"reset-delay-us" property for reset deassert in usec as that
would cause 50/50% duty-cycle, but that would always apply.
The solution in this patch series is to add a new MDIO bus
property, so post reset delay is optional and configured
separately.

MDIO bus properties[7]:
"reset-delay-us"
"reset-post-delay-us" (new)

I have not marked this with "Fixes:" as no single commit is the
cause and historically this code has only supported MDIO devices
that need reset after auto-detection. The patch series also uses
a new flexible sleep helper function that was introduced in
5.8-rc1, so the driver uses the optimal sleep function depending
on value loaded from device tree.

Future work in this area could add new properties on the MDIO
device, so reset points are configurable, e.g. no reset,
before/after auto-detection or both.

[1] https://lore.kernel.org/linux-devicetree/20200629114927.17379-2-bruno.thomsen@gmail.com/
[2] https://lore.kernel.org/linux-devicetree/20200716172611.5349-2-bruno.thomsen@gmail.com/
[3] https://elixir.bootlin.com/linux/v5.7.8/source/Documentation/devicetree/bindings/net/fsl-fec.txt#L44
[4] https://elixir.bootlin.com/linux/v5.8-rc4/source/Documentation/devicetree/bindings/net/mdio.yaml#L78
[5] https://lore.kernel.org/netdev/CAOMZO5DtYDomD8FDCZDwYCSr2AwNT81Ay4==aDxXyBxtyvPiJA@mail.gmail.com/
[6] https://lore.kernel.org/netdev/20200728090203.17313-1-bruno.thomsen@gmail.com/
[7] https://elixir.bootlin.com/linux/v5.8-rc4/source/Documentation/devicetree/bindings/net/mdio.yaml#L36
[8] http://ww1.microchip.com/downloads/en/DeviceDoc/00002202C.pdf
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

MDIO device reset assert and deassert length was created by
usleep_range() but that does not ensure optimal handling of
all the different values from device tree properties.
By switching to the new flexible sleeping helper function,
fsleep(), the correct delay function is called depending on
delay length, e.g. udelay(), usleep_range() or msleep().
Signed-off-by: Bruno Thomsen <bruno.thomsen@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Load new "reset-post-delay-us" value from MDIO properties,
and if configured to a greater then zero delay do a
flexible sleeping delay after MDIO bus reset deassert.
This allows devices to exit reset state before start
bus communication.
Signed-off-by: Bruno Thomsen <bruno.thomsen@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

MDIO bus reset pulse width is created by using udelay()
and that function might not be optimal depending on
device tree value. By switching to the new fsleep() helper
the correct delay function is called depending on
delay length, e.g. udelay(), usleep_range() or msleep().
Signed-off-by: Bruno Thomsen <bruno.thomsen@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add "reset-post-delay-us" parameter to MDIO bus properties,
so it's possible to add a delay after reset deassert.
This is optional in case external hardware slows down
release of the reset signal.
Signed-off-by: Bruno Thomsen <bruno.thomsen@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Resolved kernel/bpf/btf.c using instructions from merge commit
69138b34Signed-off-by: David S. Miller <davem@davemloft.net>

Pull networking fixes from David Miller:

 1) Encap offset calculation is incorrect in esp6, from Sabrina Dubroca.

 2) Better parameter validation in pfkey_dump(), from Mark Salyzyn.

 3) Fix several clang issues on powerpc in selftests, from Tanner Love.

 4) cmsghdr_from_user_compat_to_kern() uses the wrong length, from Al
    Viro.

 5) Out of bounds access in mlx5e driver, from Raed Salem.

 6) Fix transfer buffer memleak in lan78xx, from Johan Havold.

 7) RCU fixups in rhashtable, from Herbert Xu.

 8) Fix ipv6 nexthop refcnt leak, from Xiyu Yang.

 9) vxlan FDB dump must be done under RCU, from Ido Schimmel.

10) Fix use after free in mlxsw, from Ido Schimmel.

11) Fix map leak in HASH_OF_MAPS bpf code, from Andrii Nakryiko.

12) Fix bug in mac80211 Tx ack status reporting, from Vasanthakumar
    Thiagarajan.

13) Fix memory leaks in IPV6_ADDRFORM code, from Cong Wang.

14) Fix bpf program reference count leaks in mlx5 during
    mlx5e_alloc_rq(), from Xin Xiong.

* git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (86 commits)
  vxlan: fix memleak of fdb
  rds: Prevent kernel-infoleak in rds_notify_queue_get()
  net/sched: The error lable position is corrected in ct_init_module
  net/mlx5e: fix bpf_prog reference count leaks in mlx5e_alloc_rq
  net/mlx5e: E-Switch, Specify flow_source for rule with no in_port
  net/mlx5e: E-Switch, Add misc bit when misc fields changed for mirroring
  net/mlx5e: CT: Support restore ipv6 tunnel
  net: gemini: Fix missing clk_disable_unprepare() in error path of gemini_ethernet_port_probe()
  ionic: unlock queue mutex in error path
  atm: fix atm_dev refcnt leaks in atmtcp_remove_persistent
  net: ethernet: mtk_eth_soc: fix MTU warnings
  net: nixge: fix potential memory leak in nixge_probe()
  devlink: ignore -EOPNOTSUPP errors on dumpit
  rxrpc: Fix race between recvmsg and sendmsg on immediate call failure
  MAINTAINERS: Replace Thor Thayer as Altera Triple Speed Ethernet maintainer
  selftests/bpf: fix netdevsim trap_flow_action_cookie read
  ipv6: fix memory leaks on IPV6_ADDRFORM path
  net/bpfilter: Initialize pos in __bpfilter_process_sockopt
  igb: reinit_locked() should be called with rtnl_lock
  e1000e: continue to init PHY even when failed to disable ULP
  ...

Pull thread fix from Christian Brauner:
 "A simple spelling fix for dequeue_synchronous_signal()"

* tag 'for-linus-2020-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
  signal: fix typo in dequeue_synchronous_signal()

Pull perf tooling fixes from Arnaldo Carvalho de Melo:

 - Fix libtraceevent build with binutils 2.35

 - Fix memory leak in process_dynamic_array_len in libtraceevent

 - Fix 'perf test 68' zstd compression for s390

 - Fix record failure when mixed with ARM SPE event

* tag 'perf-tools-fixes-2020-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux:
  libtraceevent: Fix build with binutils 2.35
  perf tools: Fix record failure when mixed with ARM SPE event
  perf tests: Fix test 68 zstd compression for s390
  tools lib traceevent: Fix memory leak in process_dynamic_array_len

kernel test robot says:
net/mptcp/syncookies.c: In function 'mptcp_join_cookie_init':
include/linux/kernel.h:47:38: warning: division by zero [-Wdiv-by-zero]
 #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) + __must_be_array(arr))

I forgot that spinock_t size is 0 on UP, so ARRAY_SIZE cannot be used.

Fixes: 9466a1cc ("mptcp: enable JOIN requests even if cookies are in use")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

When vxlan interface is deleted, all fdbs are deleted by vxlan_flush().
vxlan_flush() flushes fdbs but it doesn't delete fdb, which contains
all-zeros-mac because it is deleted by vxlan_uninit().
But vxlan_uninit() deletes only the fdb, which contains both all-zeros-mac
and default vni.
So, the fdb, which contains both all-zeros-mac and non-default vni
will not be deleted.

Test commands:
    ip link add vxlan0 type vxlan dstport 4789 external
    ip link set vxlan0 up
    bridge fdb add to 00:00:00:00:00:00 dst 172.0.0.1 dev vxlan0 via lo \
	    src_vni 10000 self permanent
    ip link del vxlan0

kmemleak reports as follows:
unreferenced object 0xffff9486b25ced88 (size 96):
  comm "bridge", pid 2151, jiffies 4294701712 (age 35506.901s)
  hex dump (first 32 bytes):
    02 00 00 00 ac 00 00 01 40 00 09 b1 86 94 ff ff  ........@.......
    46 02 00 00 00 00 00 00 a7 03 00 00 12 b5 6a 6b  F.............jk
  backtrace:
    [<00000000c10cf651>] vxlan_fdb_append.part.51+0x3c/0xf0 [vxlan]
    [<000000006b31a8d9>] vxlan_fdb_create+0x184/0x1a0 [vxlan]
    [<0000000049399045>] vxlan_fdb_update+0x12f/0x220 [vxlan]
    [<0000000090b1ef00>] vxlan_fdb_add+0x12a/0x1b0 [vxlan]
    [<0000000056633c2c>] rtnl_fdb_add+0x187/0x270
    [<00000000dd5dfb6b>] rtnetlink_rcv_msg+0x264/0x490
    [<00000000fc44dd54>] netlink_rcv_skb+0x4a/0x110
    [<00000000dff433e7>] netlink_unicast+0x18e/0x250
    [<00000000b87fb421>] netlink_sendmsg+0x2e9/0x400
    [<000000002ed55153>] ____sys_sendmsg+0x237/0x260
    [<00000000faa51c66>] ___sys_sendmsg+0x88/0xd0
    [<000000006c3982f1>] __sys_sendmsg+0x4e/0x80
    [<00000000a8f875d2>] do_syscall_64+0x56/0xe0
    [<000000003610eefa>] entry_SYSCALL_64_after_hwframe+0x44/0xa9
unreferenced object 0xffff9486b1c40080 (size 128):
  comm "bridge", pid 2157, jiffies 4294701754 (age 35506.866s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 f8 dc 42 b2 86 94 ff ff  ..........B.....
    6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b 6b  kkkkkkkkkkkkkkkk
  backtrace:
    [<00000000a2981b60>] vxlan_fdb_create+0x67/0x1a0 [vxlan]
    [<0000000049399045>] vxlan_fdb_update+0x12f/0x220 [vxlan]
    [<0000000090b1ef00>] vxlan_fdb_add+0x12a/0x1b0 [vxlan]
    [<0000000056633c2c>] rtnl_fdb_add+0x187/0x270
    [<00000000dd5dfb6b>] rtnetlink_rcv_msg+0x264/0x490
    [<00000000fc44dd54>] netlink_rcv_skb+0x4a/0x110
    [<00000000dff433e7>] netlink_unicast+0x18e/0x250
    [<00000000b87fb421>] netlink_sendmsg+0x2e9/0x400
    [<000000002ed55153>] ____sys_sendmsg+0x237/0x260
    [<00000000faa51c66>] ___sys_sendmsg+0x88/0xd0
    [<000000006c3982f1>] __sys_sendmsg+0x4e/0x80
    [<00000000a8f875d2>] do_syscall_64+0x56/0xe0
    [<000000003610eefa>] entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 3ad7a4b1 ("vxlan: support fdb and learning in COLLECT_METADATA mode")
Signed-off-by: Taehee Yoo <ap420073@gmail.com>
Acked-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

It turns out that on commit 41d707b7 ("fib: fix fib_rules_ops
indirect calls wrappers") I forgot to include the case when
CONFIG_IP_MULTIPLE_TABLES is not set.

Fixes: 41d707b7 ("fib: fix fib_rules_ops indirect calls wrappers")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Brian Vazquez <brianvv@google.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
Signed-off-by: David S. Miller <davem@davemloft.net>

Fixes these errors:

net/ipv4/syncookies.c: In function 'tcp_get_cookie_sock':
net/ipv4/syncookies.c:216:19: error: 'struct tcp_request_sock' has no
member named 'drop_req'
  216 |   if (tcp_rsk(req)->drop_req) {
      |                   ^~
net/ipv4/syncookies.c: In function 'cookie_tcp_reqsk_alloc':
net/ipv4/syncookies.c:289:27: warning: unused variable 'treq'
[-Wunused-variable]
  289 |  struct tcp_request_sock *treq;
      |                           ^~~~
make[3]: *** [scripts/Makefile.build:280: net/ipv4/syncookies.o] Error 1
make[3]: *** Waiting for unfinished jobs....

Fixes: 9466a1cc ("mptcp: enable JOIN requests even if cookies are in use")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Florian Westphal <fw@strlen.de>
Acked-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull pin control fix from Linus Walleij:
 "A single last minute pin control fix to the Qualcomm driver fixing
  missing dual edge PCH interrupts"

* tag 'pinctrl-v5.8-4' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: qcom: Handle broken/missing PDC dual edge IRQs on sc7180

Merge tag 'mac80211-next-for-davem-2020-07-31' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac80211-next

Johannes Berg says:

====================
We have a number of changes
 * code cleanups and fixups as usual
 * AQL & internal TXQ improvements from Felix
 * some mesh 802.1X support bits
 * some injection improvements from Mathy of KRACK
   fame, so we'll see what this results in ;-)
 * some more initial S1G supports bits, this time
   (some of?) the userspace APIs
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

netdev protodown is a mechanism that allows protocols to
hold an interface down. It was initially introduced in
the kernel to hold links down by a multihoming protocol.
There was also an attempt to introduce protodown
reason at the time but was rejected. protodown and protodown reason
is supported by almost every switching and routing platform.
It was ok for a while to live without a protodown reason.
But, its become more critical now given more than
one protocol may need to keep a link down on a system
at the same time. eg: vrrp peer node, port security,
multihoming protocol. Its common for Network operators and
protocol developers to look for such a reason on a networking
box (Its also known as errDisable by most networking operators)

This patch adds support for link protodown reason
attribute. There are two ways to maintain protodown
reasons.
(a) enumerate every possible reason code in kernel
    - A protocol developer has to make a request and
      have that appear in a certain kernel version
(b) provide the bits in the kernel, and allow user-space
(sysadmin or NOS distributions) to manage the bit-to-reasonname
map.
	- This makes extending reason codes easier (kind of like
      the iproute2 table to vrf-name map /etc/iproute2/rt_tables.d/)

This patch takes approach (b).

a few things about the patch:
- It treats the protodown reason bits as counter to indicate
active protodown users
- Since protodown attribute is already an exposed UAPI,
the reason is not enforced on a protodown set. Its a no-op
if not used.
the patch follows the below algorithm:
  - presence of reason bits set indicates protodown
    is in use
  - user can set protodown and protodown reason in a
    single or multiple setlink operations
  - setlink operation to clear protodown, will return -EBUSY
    if there are active protodown reason bits
  - reason is not included in link dumps if not used

example with patched iproute2:
$cat /etc/iproute2/protodown_reasons.d/r.conf
0 mlag
1 evpn
2 vrrp
3 psecurity

$ip link set dev vxlan0 protodown on protodown_reason vrrp on
$ip link set dev vxlan0 protodown_reason mlag on
$ip link show
14: vxlan0: <BROADCAST,MULTICAST> mtu 1500 qdisc noop state DOWN mode
DEFAULT group default qlen 1000
    link/ether f6:06:be:17:91:e7 brd ff:ff:ff:ff:ff:ff protodown on <mlag,vrrp>

$ip link set dev vxlan0 protodown_reason mlag off
$ip link set dev vxlan0 protodown off protodown_reason vrrp off
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
pull-request: bpf 2020-07-31

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

We've added 5 non-merge commits during the last 21 day(s) which contain
a total of 5 files changed, 126 insertions(+), 18 deletions(-).

The main changes are:

1) Fix a map element leak in HASH_OF_MAPS map type, from Andrii Nakryiko.

2) Fix a NULL pointer dereference in __btf_resolve_helper_id() when no
   btf_vmlinux is available, from Peilin Ye.

3) Init pos variable in __bpfilter_process_sockopt(), from Christoph Hellwig.

4) Fix a cgroup sockopt verifier test by specifying expected attach type,
   from Jean-Philippe Brucker.

Note that when net gets merged into net-next later on, there is a small
merge conflict in kernel/bpf/btf.c between commit 5b801dfb ("bpf: Fix
NULL pointer dereference in __btf_resolve_helper_id()") from the bpf tree
and commit 138b9a05 ("bpf: Remove btf_id helpers resolving") from the
net-next tree.

Resolve as follows: remove the old hunk with the __btf_resolve_helper_id()
function. Change the btf_resolve_helper_id() so it actually tests for a
NULL btf_vmlinux and bails out:

int btf_resolve_helper_id(struct bpf_verifier_log *log,
                          const struct bpf_func_proto *fn, int arg)
{
        int id;

        if (fn->arg_type[arg] != ARG_PTR_TO_BTF_ID || !btf_vmlinux)
                return -EINVAL;
        id = fn->btf_id[arg];
        if (!id || id > btf_vmlinux->nr_types)
                return -EINVAL;
        return id;
}

Let me know if you run into any others issues (CC'ing Jiri Olsa so he's in
the loop with regards to merge conflict resolution).
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

We expecte prog_p to be protected by rcu, so adding the rcu annotation
to fix the following sparse warning:

drivers/net/tun.c:3003:36: warning: incorrect type in argument 2 (different address spaces)
drivers/net/tun.c:3003:36:    expected struct tun_prog [noderef] __rcu **prog_p
drivers/net/tun.c:3003:36:    got struct tun_prog **prog_p
drivers/net/tun.c:3292:42: warning: incorrect type in argument 2 (different address spaces)
drivers/net/tun.c:3292:42:    expected struct tun_prog **prog_p
drivers/net/tun.c:3292:42:    got struct tun_prog [noderef] __rcu **
drivers/net/tun.c:3296:42: warning: incorrect type in argument 2 (different address spaces)
drivers/net/tun.c:3296:42:    expected struct tun_prog **prog_p
drivers/net/tun.c:3296:42:    got struct tun_prog [noderef] __rcu **
Reported-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Steffen Klassert says:

====================
pull request (net): ipsec 2020-07-31

1) Fix policy matching with mark and mask on userspace interfaces.
   From Xin Long.

2) Several fixes for the new ESP in TCP encapsulation.
   From Sabrina Dubroca.

3) Fix crash when the hold queue is used. The assumption that
   xdst->path and dst->child are not a NULL pointer only if dst->xfrm
   is not a NULL pointer is true with the exception of using the
   hold queue. Fix this by checking for hold queue usage before
   dereferencing xdst->path or dst->child.

4) Validate pfkey_dump parameter before sending them.
   From Mark Salyzyn.

5) Fix the location of the transport header with ESP in UDPv6
   encapsulation. From Sabrina Dubroca.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Saeed Mahameed says:

====================
Mellanox, mlx5 fixes 2020-07-30

This small patchset introduces some fixes to mlx5 driver.

Please pull and let me know if there is any problem.

For -stable v4.18:
 ('net/mlx5e: fix bpf_prog reference count leaks in mlx5e_alloc_rq')

For -stable v5.7:
 ('net/mlx5e: E-Switch, Add misc bit when misc fields changed for mirroring')
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This change adds TCP_NLA_EDT to SCM_TIMESTAMPING_OPT_STATS that reports
the earliest departure time(EDT) of the timestamped skb. By tracking EDT
values of the skb from different timestamps, we can observe when and how
much the value changed. This allows to measure the precise delay
injected on the sender host e.g. by a bpf-base throttler.
Signed-off-by: Yousuk Seung <ysseung@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Tony Nguyen says:

====================
1GbE Intel Wired LAN Driver Updates 2020-07-30

This series contains updates to e100, e1000, e1000e, igb, igbvf, ixgbe,
ixgbevf, iavf, and driver documentation.

Vaibhav Gupta converts legacy .suspend() and .resume() to generic PM
callbacks for e100, igbvf, ixgbe, ixgbevf, and iavf.

Suraj Upadhyay replaces 1 byte memsets with assignments for e1000,
e1000e, igb, and ixgbe.

Alexander Klimov replaces http links with https.

Miaohe Lin replaces uses of memset to clear MAC addresses with
eth_zero_addr().
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Florian Westphal says:

====================
mptcp: add syncookie support

Changes in v2:
- first patch renames req->ts_cookie to req->syncookie instead of
  removing ts_cookie member.
- patch to add 'want_cookie' arg to init_req() functions has been dropped.
  All users of that arg were changed to check 'req->syncookie' instead.

v1 cover letter:

When syn-cookies are used the SYN?ACK never contains a MPTCP option,
because the code path that creates a request socket based on a valid
cookie ACK lacks the needed changes to construct MPTCP request sockets.

After this series, if SYN carries MP_CAPABLE option, the option is not
cleared anymore and request socket will be reconstructed using the
MP_CAPABLE option data that is re-sent with the ACK.

This means that no additional state gets encoded into the syn cookie or
the TCP timestamp.

There are two caveats for SYN-Cookies with MPTCP:

1. When syn-cookies are used, the server-generated key is not stored.
The drawback is that the next connection request that comes in before
the cookie-ACK has a small chance that it will generate the same local_key.

If this happens, the cookie ACK that comes in second will (re)compute the
token hash and then detects that this is already in use.
Unlike normal case, where the server will pick a new key value and then
re-tries, we can't do that because we already committed to the key value
(it was sent to peer already).

Im this case, MPTCP cannot be used and late TCP fallback happens.

2). SYN packets with a MP_JOIN requests cannot be handled without storing
    state. This is because the SYN contains a nonce value that is needed to
    verify the HMAC of the MP_JOIN ACK that completes the three-way
    handshake.  Also, a local nonce is generated and used in the cookie
    SYN/ACK.

There are only 2 ways to solve this:
 a) Do not support JOINs when cookies are in effect.
 b) Store the nonces somewhere.

The approach chosen here is b).
Patch 8 adds a fixed-size (1024 entries) state table to store the
information required to validate the MP_JOIN ACK and re-build the
request socket.

State gets stored when syn-cookies are active and the token in the JOIN
request referred to an established MPTCP connection that can also accept
a new subflow.

State is restored if the ACK cookie is valid, an MP_JOIN option is present
and the state slot contains valid data from a previous SYN.

After the request socket has been re-build, normal HMAC check is done just
as without syn cookies.

Largely identical to last RFC, except patch #8 which follows Paolos
suggestion to use a private table storage area rather than keeping
request sockets around.  This also means I dropped the patch to remove
const qualifier from sk_listener pointers.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Also add test cases with MP_JOIN when tcp_syncookies sysctl is 2 (i.e.,
syncookies are always-on).

While at it, also print the test number and add the test number
to the pcap files that can be generated optionally.

This makes it easier to match the pcap to the test case.
Signed-off-by: Florian Westphal <fw@strlen.de>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

check we can establish connections also when syn cookies are in use.

Check that
MPTcpExtMPCapableSYNRX and MPTcpExtMPCapableACKRX increase for each
MPTCP test.

Check TcpExtSyncookiesSent and TcpExtSyncookiesRecv increase in netns2.
Signed-off-by: Florian Westphal <fw@strlen.de>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>