- set min/max_mtu in all hdlc drivers, remove hdlc_change_mtu
- sent max_mtu in lec driver, remove lec_change_mtu
- set min/max_mtu in x25_asy driver

CC: netdev@vger.kernel.org
CC: Krzysztof Halasa <khc@pm.waw.pl>
CC: Krzysztof Halasa <khalasa@piap.pl>
CC: Jan "Yenya" Kasprzak <kas@fi.muni.cz>
CC: Francois Romieu <romieu@fr.zoreil.com>
CC: Kevin Curtis <kevin.curtis@farsite.co.uk>
CC: Zhao Qiang <qiang.zhao@nxp.com>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

- set max_mtu in wil6210 driver
- set max_mtu in atmel driver
- set min/max_mtu in cisco airo driver, remove airo_change_mtu
- set min/max_mtu in ipw2100/ipw2200 drivers, remove libipw_change_mtu
- set min/max_mtu in p80211netdev, remove wlan_change_mtu
- set min/max_mtu in net/mac80211/iface.c and remove ieee80211_change_mtu
- set min/max_mtu in wimax/i2400m and remove i2400m_change_mtu
- set min/max_mtu in intersil/hostap and remove prism2_change_mtu
- set min/max_mtu in intersil/orinoco
- set min/max_mtu in tty/n_gsm and remove gsm_change_mtu

CC: netdev@vger.kernel.org
CC: linux-wireless@vger.kernel.org
CC: Maya Erez <qca_merez@qca.qualcomm.com>
CC: Simon Kelley <simon@thekelleys.org.uk>
CC: Stanislav Yakovlev <stas.yakovlev@gmail.com>
CC: Johannes Berg <johannes@sipsolutions.net>
CC: Inaky Perez-Gonzalez <inaky.perez-gonzalez@intel.com>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Acked-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

usbnet:
- Remove stale new_mtu <= 0 check in usbnet.c
- Set min_mtu = 0, max_mtu = 65535 (sub-drivers must set their own
  max_mtu and/or min_mtu as needed)

r8152:
- Set appropriate max_mtu for different variants (1500 or 9194)

lan78xx:
- Set max_mtu = 9000

asix_driver:
- max_mtu = 16384 for ax88178 variant

ax88179:
- max_mtu = 4088

cdc_ncm:
- max_mtu from hardware

cdc-phonet:
- min_mtu = 6, max_mtu = 65541

sierra_net:
- max_mtu = 1500, call usbnet_change_mtu directly
- sierra_net_change_mtu checked for MTU > 1500, then called
  usbnet_change_mtu, but if we set max_mtu to let the network core handle
  the range check, then we can simply call usbnet_change_mtu directly

smsc75xx:
- max_mtu = 9000

CC: netdev@vger.kernel.org
CC: Woojung Huh <woojung.huh@microchip.com>
CC: Microchip Linux Driver Support <UNGLinuxDriver@microchip.com>
CC: Hayes Wang <hayeswang@realtek.com>
CC: Oliver Neukum <oneukum@suse.com>
CC: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Somehow, I missed a healthy number of ethernet drivers in the last pass.
Most of these drivers either were in need of an updated max_mtu to make
jumbo frames possible to enable again. In a few cases, also setting a
different min_mtu to match previous lower bounds. There are also a few
drivers that had no upper bounds checking, so they're getting a brand new
ETH_MAX_MTU that is identical to IP_MAX_MTU, but accessible by includes
all ethernet and ethernet-like drivers all have already.

acenic:
- min_mtu = 0, max_mtu = 9000

amazon/ena:
- min_mtu = 128, max_mtu = adapter->max_mtu

amd/xgbe:
- min_mtu = 0, max_mtu = 9000

sb1250:
- min_mtu = 0, max_mtu = 1518

cxgb3:
- min_mtu = 81, max_mtu = 65535

cxgb4:
- min_mtu = 81, max_mtu = 9600

cxgb4vf:
- min_mtu = 81, max_mtu = 65535

benet:
- min_mtu = 256, max_mtu = 9000

ibmveth:
- min_mtu = 68, max_mtu = 65535

ibmvnic:
- min_mtu = adapter->min_mtu, max_mtu = adapter->max_mtu
- remove now redundant ibmvnic_change_mtu

jme:
- min_mtu = 1280, max_mtu = 9202

mv643xx_eth:
- min_mtu = 64, max_mtu = 9500

mlxsw:
- min_mtu = 0, max_mtu = 65535
- Basically bypassing the core checks, and instead relying on dynamic
  checks in the respective switch drivers' ndo_change_mtu functions

ns83820:
- min_mtu = 0
- remove redundant ns83820_change_mtu, only checked for mtu > 1500

netxen:
- min_mtu = 0, max_mtu = 8000 (P2), max_mtu = 9600 (P3)

qlge:
- min_mtu = 1500, max_mtu = 9000
- driver only supports setting mtu to 1500 or 9000, so the core check only
  rules out < 1500 and > 9000, qlge_change_mtu still needs to check that
  the value is 1500 or 9000

qualcomm/emac:
- min_mtu = 46, max_mtu = 9194

xilinx_axienet:
- min_mtu = 64, max_mtu = 9000

Fixes: 61e84623 ("net: centralize net_device min/max MTU checking")
CC: netdev@vger.kernel.org
CC: Jes Sorensen <jes@trained-monkey.org>
CC: Netanel Belgazal <netanel@annapurnalabs.com>
CC: Tom Lendacky <thomas.lendacky@amd.com>
CC: Santosh Raspatur <santosh@chelsio.com>
CC: Hariprasad S <hariprasad@chelsio.com>
CC: Sathya Perla <sathya.perla@broadcom.com>
CC: Ajit Khaparde <ajit.khaparde@broadcom.com>
CC: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com>
CC: Somnath Kotur <somnath.kotur@broadcom.com>
CC: Thomas Falcon <tlfalcon@linux.vnet.ibm.com>
CC: John Allen <jallen@linux.vnet.ibm.com>
CC: Guo-Fu Tseng <cooldavid@cooldavid.org>
CC: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
CC: Jiri Pirko <jiri@mellanox.com>
CC: Ido Schimmel <idosch@mellanox.com>
CC: Manish Chopra <manish.chopra@qlogic.com>
CC: Sony Chacko <sony.chacko@qlogic.com>
CC: Rajesh Borundia <rajesh.borundia@qlogic.com>
CC: Timur Tabi <timur@codeaurora.org>
CC: Anirudha Sarangi <anirudh@xilinx.com>
CC: John Linn <John.Linn@xilinx.com>
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix typo in parameter description.
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace dev_kfree_skb with dev_kfree_skb_any in mtk_start_xmit()
which can be called from hard irq context (netpoll) and from
other contexts. mtk_start_xmit() only frees skbs that it has
dropped.

This is detected by Coccinelle semantic patch.
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace dev_kfree_skb with dev_kfree_skb_any in dwceqos_start_xmit()
which can be called from hard irq context (netpoll) and from
other contexts. dwceqos_start_xmit() only frees skbs that it has
dropped.

This is detected by Coccinelle semantic patch.
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The AQR106 and AQR107 can use the existing driver.
Signed-off-by: NShaohui Xie <Shaohui.Xie@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

clk_prepare, clk_enable and their counterparts (at least the common clk
ones, but also most others) do check for the clk being NULL anyhow (and
return 0 then), so there is no gain when the caller checks, too.
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Acked-by: NFugang Duan <fugang.duan@nxp.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

softirq handlers use RCU protection to lookup listeners,
and write operations all happen from process context.
We do not need to block BH for dump operations.

Also SYN_RECV since request sockets are stored in the ehash table :

 1) inet_diag_dump_icsk() no longer need to clear
    cb->args[3] and cb->args[4] that were used as cursors while
    iterating the old per listener hash table.

 2) Also factorize a test : No need to scan listening_hash[]
    if r->id.idiag_dport is not zero.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The patch isolating the u16 writes for pxa assumed all machine_is_*()
calls were removed, and therefore removed the mach-types.h include which
provided them.

Unfortunately 2 machine_is_*() remained in smc91x.c file including
smc91x.h from which the include was removed, triggering the error:
 drivers/net/ethernet/smsc/smc91x.c: In function ‘smc_drv_probe’:
 drivers/net/ethernet/smsc/smc91x.c:2380:2: error: implicit declaration
 of function ‘machine_is_assabet’
 [-Werror=implicit-function-declaration]
	if (machine_is_assabet() && machine_has_neponset())

This adds back the wrongly removed include.

Fixes: d09d747a ("net: smc91x: isolate u16 writes alignment workaround")
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Tailroom is supposed to be of length sizeof(struct ila_lwt) but
sizeof(struct ila_params) is currently allocated.

This leads to the dst_cache and connected member of ila_lwt being
referenced out of bounds.

struct ila_lwt {
	struct ila_params p;
	struct dst_cache dst_cache;
	u32 connected : 1;
};

Fixes: 65d7ab8d ("net: Identifier Locator Addressing module")
Signed-off-by: NThomas Graf <tgraf@suug.ch>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Zach Brown says:

====================
macb: Add ethtool get_ringparam and set_ringparam to cadence

There are use cases like RT that would benefit from being able to tune the
macb rx/tx ring sizes. The ethtool set_ringparam function is the standard way
of doing so.

The first patch changes the hardcoded tx/rx ring sizes to variables that are
set to a hardcoded default.

The second patch implements the get_ringparam and set_ringparam fucntions.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some applications want to tune the size of the macb rx/tx ring buffers.
The ethtool set_ringparam function is the standard way of doing it.
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The macb driver hardcoded the tx/rx ring sizes. This made it
impossible to change the sizes at run time.

Add tx_ring_size, and rx_ring_size variables to macb object, which
are initilized with default vales during macb_init. Change all
references to RX_RING_SIZE and TX_RING_SIZE to their respective
replacements.
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace dev_kfree_skb with dev_kfree_skb_any in arc_emac_tx()
which can be called from hard irq context (netpoll) and from
other contexts. arc_emac_tx() only frees skbs that it has
dropped.

This is detected by Coccinelle semantic patch.
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jiri Benc says:

====================
openvswitch: remove unused code

Removed unused functions and unnecessary EXPORT_SYMBOLs from openvswitch.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some symbols exported to other modules are really used only by
openvswitch.ko. Remove the exports.

Tested by loading all 4 openvswitch modules, nothing breaks.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

ovs_vport_deferred_free is not used anywhere. It's the only caller of
free_vport_rcu thus this one can be removed, too.
Signed-off-by: NJiri Benc <jbenc@redhat.com>
Acked-by: NPravin B Shelar <pshelar@ovn.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A BPF program is required to check the return register of a
map_elem_lookup() call before accessing memory. The verifier keeps
track of this by converting the type of the result register from
PTR_TO_MAP_VALUE_OR_NULL to PTR_TO_MAP_VALUE after a conditional
jump ensures safety. This check is currently exclusively performed
for the result register 0.

In the event the compiler reorders instructions, BPF_MOV64_REG
instructions may be moved before the conditional jump which causes
them to keep their type PTR_TO_MAP_VALUE_OR_NULL to which the
verifier objects when the register is accessed:

0: (b7) r1 = 10
1: (7b) *(u64 *)(r10 -8) = r1
2: (bf) r2 = r10
3: (07) r2 += -8
4: (18) r1 = 0x59c00000
6: (85) call 1
7: (bf) r4 = r0
8: (15) if r0 == 0x0 goto pc+1
 R0=map_value(ks=8,vs=8) R4=map_value_or_null(ks=8,vs=8) R10=fp
9: (7a) *(u64 *)(r4 +0) = 0
R4 invalid mem access 'map_value_or_null'

This commit extends the verifier to keep track of all identical
PTR_TO_MAP_VALUE_OR_NULL registers after a map_elem_lookup() by
assigning them an ID and then marking them all when the conditional
jump is observed.
Signed-off-by: NThomas Graf <tgraf@suug.ch>
Reviewed-by: NJosef Bacik <jbacik@fb.com>
Acked-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of using a private copy of struct net_device_stats in struct
fs_enet_private, use stats from struct net_device. Also remove the now
unnecessary .ndo_get_stats function.
Signed-off-by: NTobias Klauser <tklauser@distanz.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The memory return by kzalloc() has already be set to zero, so
remove useless memset(0).
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes the following sparse warning:

drivers/net/dsa/mv88e6xxx/chip.c:2866:5: warning:
 symbol 'mv88e6xxx_g1_set_switch_mac' was not declared. Should it be static?
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add the following four products of Lenovo and sort the order of the list.

	VID		PID
	0x17ef		0x3062
	0x17ef		0x3069
	0x17ef		0x720c
	0x17ef		0x7214
Signed-off-by: NHayes Wang <hayeswang@realtek.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use sizeof variable instead of literal number to enhance the readability.
Signed-off-by: NGao Feng <fgao@ikuai8.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Robert Jarzmik says:

====================
support smc91x on mainstone and devicetree

This series aims at bringing support to mainstone board on a
device-tree based build, as what is already in place for legacy
mainstone.

The bulk of the mainstone "specific" behavior is that a u16 write
doesn't work on a address of the form 4*n + 2, while it works on 4*n.

The legacy workaround was in SMC_outw(), with calls to
machine_is_mainstone(). These calls don't work with a pxa27x-dt
machine type, which is used when a generic device-tree pxa27x machine
is used to boot the mainstone board.

Therefore, this series enables the smc91c111 adapter of the mainstone
board to work on a device-tree build, exaclty as it's been working for
years with the legacy arch/arm/mach-pxa/mainstone.c definition.

As a sum up, this extends an existing mechanism to device-tree based
pxa platforms.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add a workaround for mainstone, idp and stargate2 boards, for u16 writes
which must be aligned on 32 bits addresses.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Cc: Jeremy Linton <jeremy.linton@arm.com>
Acked-by: NRob Herring <robh@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For device-tree builds, platforms such as mainstone, idp and stargate2
must have their u16 writes all aligned on 32 bit boundaries. This is
already enabled in platform data builds, and this patch adds it to
device-tree builds.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Writes to u16 has a special handling on 3 PXA platforms, where the
hardware wiring forces these writes to be u32 aligned.

This patch isolates this handling for PXA platforms as before, but
enables this "workaround" to be set up dynamically, which will be the
case in device-tree build types.

This patch was tested on 2 PXA platforms : mainstone, which relies on
the workaround, and lubbock, which doesn't.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Instead of having the smc91x driver relying on machine_is_*() calls,
provide this data through platform data, ie. idp, mainstone and
stargate.

This way, the driver doesn't need anymore machine_is_*() calls, which
wouldn't work anymore with a device-tree build.
Signed-off-by: NRobert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes: 61e84623 ("net: centralize net_device min/max MTU checking")
Signed-off-by: NBert Kenward <bkenward@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Zach Brown says:

====================
Add support for led triggers on phy link state change

Fix skge driver that declared enum contants that conflicted with enum
constants in linux/leds.h

Create function that encapsulates actions taken during the adjust phy link step
of phy state changes.

Create function that provides list of speeds currently supported by the phy.

Add support for led triggers on phy link state changes by adding
a config option. When set the config option will create a set of led triggers
for each phy device. Users can use the led triggers to represent link state
changes on the phy.

v2:
 * New patch that creates phy_adjust_link function to encapsulate actions taken
   when adjusting phy link during phy state changes
 * led trigger speed strings changed to match existing phy speed strings
 * New function that maps speeds to led triggers
 * Replace magic constants with definitions when declaring trigger name
   buffer and number of triggers.
v3:
 * Changed LED_ON to LED_REG_ON in skge driver to avoid possible future
   conflict and improve consistency.
 * Dropped rtl8712 patch that was accepted separately.
v4:
 * tweaked commit message
v5
 * Changed commit message to explain relationship between the new triggers and
   leds driven by phys.
 * Added new patch that creates phy_supported_speeds function.
 * Moved phy_leds_triggers_register and phy_leds_triggers_unregister to
   phy_attach and phy_detach respectively. This change is so the
   phydev->supported field will be filled by the time the triggers are
   registered.
 * Changed hardcoded list of triggers to dynamic list determined by speeds
   return by phy_supported_speeds.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Create an option CONFIG_LED_TRIGGER_PHY (default n), which will create a
set of led triggers for each instantiated PHY device. There is one LED
trigger per link-speed, per-phy.
The triggers are registered during phy_attach and unregistered during
phy_detach.

This allows for a user to configure their system to allow a set of LEDs
not controlled by the phy to represent link state changes on the phy.
LEDS controlled by the phy are unaffected.

For example, we have a board where some of the leds in the
RJ45 socket are controlled by the phy, but others are not. Using the
triggers provided by this patch the leds not controlled by the phy can
be configured to show the current speed of the ethernet connection. The
leds controlled by the phy are unaffected.
Signed-off-by: NJosh Cartwright <josh.cartwright@ni.com>
Signed-off-by: NNathan Sullivan <nathan.sullivan@ni.com>
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

net: phy: Create phy_supported_speeds function which lists speeds currently supported by a phydevice

phy_supported_speeds provides a means to get a list of all the speeds a
phy device currently supports.
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

During phy state machine state transitions some set of actions should
occur whenever the link state changes. These actions should be
encapsulated into a single function

This patch adds the phy_adjust_link function, which is called whenever
phydev->adjust_link would have been called before. Actions that should
occur whenever the phy link is adjusted can now be added to the
phy_adjust_link function.
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Adding led support for phy causes namespace conflicts for some
phy drivers.

The marvel skge driver declared an enum for representing the states of
Link LED Register. The enum contained constant LED_OFF which conflicted
with declartation found in linux/leds.h.
LED_OFF changed to LED_REG_OFF
Also changed LED_ON to LED_REG_ON to avoid possible future conflict and
for consistency.
Signed-off-by: NZach Brown <zach.brown@ni.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

David Ahern says:

====================
net: Fix netdev adjacency tracking

The netdev adjacency tracking is failing to create proper dependencies
for some topologies. For example this topology

        +--------+
        |  myvrf |
        +--------+
          |    |
          |  +---------+
          |  | macvlan |
          |  +---------+
          |    |
      +----------+
      |  bridge  |
      +----------+
          |
      +--------+
      | bond1  |
      +--------+
          |
      +--------+
      |  eth3  |
      +--------+

hits 1 of 2 problems depending on the order of enslavement. The base set of
commands for both cases:

    ip link add bond1 type bond
    ip link set bond1 up
    ip link set eth3 down
    ip link set eth3 master bond1
    ip link set eth3 up

    ip link add bridge type bridge
    ip link set bridge up
    ip link add macvlan link bridge type macvlan
    ip link set macvlan up

    ip link add myvrf type vrf table 1234
    ip link set myvrf up

    ip link set bridge master myvrf

Case 1 enslave macvlan to the vrf before enslaving the bond to the bridge:

    ip link set macvlan master myvrf
    ip link set bond1 master bridge

Attempts to delete the VRF:
    ip link delete myvrf

trigger the BUG in __netdev_adjacent_dev_remove:

[  587.405260] tried to remove device eth3 from myvrf
[  587.407269] ------------[ cut here ]------------
[  587.408918] kernel BUG at /home/dsa/kernel.git/net/core/dev.c:5661!
[  587.411113] invalid opcode: 0000 [#1] SMP
[  587.412454] Modules linked in: macvlan bridge stp llc bonding vrf
[  587.414765] CPU: 0 PID: 726 Comm: ip Not tainted 4.8.0+ #109
[  587.416766] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.7.5-20140531_083030-gandalf 04/01/2014
[  587.420241] task: ffff88013ab6eec0 task.stack: ffffc90000628000
[  587.422163] RIP: 0010:[<ffffffff813cef03>]  [<ffffffff813cef03>] __netdev_adjacent_dev_remove+0x40/0x12c
...
[  587.446053] Call Trace:
[  587.446424]  [<ffffffff813d1542>] __netdev_adjacent_dev_unlink+0x20/0x3c
[  587.447390]  [<ffffffff813d16a3>] netdev_upper_dev_unlink+0xfa/0x15e
[  587.448297]  [<ffffffffa00003a3>] vrf_del_slave+0x13/0x2a [vrf]
[  587.449153]  [<ffffffffa00004a4>] vrf_dev_uninit+0xea/0x114 [vrf]
[  587.450036]  [<ffffffff813d19b0>] rollback_registered_many+0x22b/0x2da
[  587.450974]  [<ffffffff813d1aac>] unregister_netdevice_many+0x17/0x48
[  587.451903]  [<ffffffff813de444>] rtnl_delete_link+0x3c/0x43
[  587.452719]  [<ffffffff813dedcd>] rtnl_dellink+0x180/0x194

When the BUG is converted to a WARN_ON it shows 4 missing adjacencies:
  eth3 - myvrf, mvrf - eth3, bond1 - myvrf and myvrf - bond1

All of those are because the __netdev_upper_dev_link function does not
properly link macvlan lower devices to myvrf when it is enslaved.

The second case just flips the ordering of the enslavements:
    ip link set bond1 master bridge
    ip link set macvlan master myvrf

Then run:
    ip link delete bond1
    ip link delete myvrf

The vrf delete command hangs because myvrf has a reference that has not
been released. In this case the removal code does not account for 2 paths
between eth3 and myvrf - one from bridge to vrf and the other through the
macvlan.

Rather than try to maintain a linked list of all upper and lower devices
per netdevice, only track the direct neighbors. The remaining stack can
be determined by recursively walking the neighbors.

The existing netdev_for_each_all_upper_dev_rcu,
netdev_for_each_all_lower_dev and netdev_for_each_all_lower_dev_rcu macros
are replaced with APIs that walk the upper and lower device lists. The
new APIs take a callback function and a data arg that is passed to the
callback for each device in the list. Drivers using the old macros are
converted in separate patches to make it easier on reviewers. It is an
API conversion only; no functional change is intended.

v3
- address Stephen's comment to simplify logic and remove typecasts

v2
- fixed bond0 references in cover-letter
- fixed definition of netdev_next_lower_dev_rcu to mirror the upper_dev
  version.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Adjacency code only has debugs for the insert case. Add debugs for
the remove path and make both consistently worded to make it easier
to follow the insert and removal with reference counts.

In addition, change the BUG to a WARN_ON. A missing adjacency at
removal time is not cause for a panic.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Lower list should be empty just like upper.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Only direct adjacencies are maintained. All upper or lower devices can
be learned via the new walk API which recursively walks the adj_list for
upper devices or lower devices.
Signed-off-by: NDavid Ahern <dsa@cumulusnetworks.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>