atp870u: Convert remaining in[bwl] and out[bwl] to wrappers

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Rename is885() to atp_is()

Now that all the is* functions except is885() are gone, rename is885() to
atp_is() to avoid confusion. Don't know what "is" means, though...

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove is870()

Now that is885() supports everything from is870() and the rest of the code
is almost identical, remove is870() and use is885() instead.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Move 870-specific code out of is870()

Move few remaining 870-specific code lines out of is870()

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Add remaining 870 support to is885()

Add remaining 870 support to is885():
 - different synw, no synuw
 - synu[4] = 0x0c
 - atp_writeb_io(dev, c, 0x04, 0x00); instead of
   atp_writeb_io(dev, c, 0x14, 0x00); (isn't that a bug?)
 - atp_writeb_io(dev, c, 0x14, 0xff); instead of
   atp_writeb_io(dev, c, 0x14, 0x06);
 - different mbuf[3] and mbuf[4] checks

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Add wide_chip parameter to is870() and is885()

Don't check chip_ver in is870() but add wide_chip parameter for that.
Then add the non-wide support to is885().

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove is880()

Now that is880() and is885() are almost identical (except for some cpu_relax()
calls and debug printks), remove is880() and use is885() instead.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Move chip-specific lines out of is880() and is885()

Move few chip-specifis lines out of is880() and is885() so they become
almost identical.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Add channel parameter to is870() and is880()

Add channel parameter to is870() and is880() functions to simplify comparing
them with is885().

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Unify code format in is870(), is880() and is885()

Unify code formatting in is870(), is880() and is885() functions to simplify
comparing them.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Convert is885() to use wrappers

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Convert is880() to use wrappers

Subtract 0x40 to use _io access wrappers.  Now it's obvious that is870()
and is880() are very similar.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Convert is870() to use wrappers

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Introduce HW access wrappers

Introduce *_read? and *_write? wrappers to improve code readability.
Also make sure that baseport is always initialized, not only for ATP880.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove ugly gotos #5

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove ugly gotos #4

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove ugly gotos #3

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove ugly gotos #2

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove ugly gotos

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmpcip #2

Untangle the tmpcip crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmpcip

Untangle the tmpcip crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #8

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #7

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #6

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #5

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #4

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #3

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport #2

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Untangle tmport

Untangle the tmport crap so it becomes obvious what ports are accessed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove tmport1

Remove tmport1 temporary variable to simplify the code.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

atp870u: Remove workport

Remove workport temporary variable to simplify the code.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Hannes Reinicke <hare@suse.de>
Acked-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

SCSI: initio: remove duplicate module device table

The initio driver has for many years had two copies of the
same module device table. One of them is also used for registering
the other driver, the other one is entirely useless after the
large scale cleanup that Alan Cox did back in 2007.

The compiler warns about this whenever the driver is built-in:

drivers/scsi/initio.c:131:29: warning: 'i91u_pci_devices' defined but not used [-Wunused-variable]

This removes the extraneous table and the warning.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Fixes: 72d39fea901 ("[SCSI] initio: Convert into a real Linux driver and update to modern style")
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

qla2xxx: Remove unavailable firmware files

Remove firmware binary names for the ISPs, which are not submitted to
linux-firmware.

Signed-off-by: Himanshu Madhani <himanshu.madhani@qlogic.com>
Signed-off-by: Giridhar Malavali <giridhar.malavali@qlogic.com>
Reviewed-by: Julian Calaby <julian.calaby@gmail.com>
Reviewed-by: Xose Vazquez Perez <xose.vazquez@gmail.com>
Cc: Hannes Reinecke <hare@suse.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

st: Remove obsolete scsi_tape.max_pfn

Its last user was removed 10 years ago, in commit
8b05b773b6030de5 ("[SCSI] convert st to use scsi_execute_async").

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reviewed-by: Ewan D. Milne <emilne@redhat.com>
Acked-by: Kai Mäkisara <kai.makisara@kolumbus.fi>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

Merge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Fix list tests in netfilter ingress support, from Florian Westphal.

 2) Fix reversal of input and output interfaces in ingress hook
    invocation, from Pablo Neira Ayuso.

 3) We have a use after free in r8169, caught by Dave Jones, fixed by
    Francois Romieu.

 4) Splice use-after-free fix in AF_UNIX frmo Hannes Frederic Sowa.

 5) Three ipv6 route handling bug fixes from Martin KaFai Lau:
    a) Don't create clone routes not managed by the fib6 tree
    b) Don't forget to check expiration of DST_NOCACHE routes.
    c) Handle rt->dst.from == NULL properly.

 6) Several AF_PACKET fixes wrt transport header setting and SKB
    protocol setting, from Daniel Borkmann.

 7) Fix thunder driver crash on shutdown, from Pavel Fedin.

 8) Several Mellanox driver fixes (max MTU calculations, use of correct
    DMA unmap in TX path, etc.) from Saeed Mahameed, Tariq Toukan, Doron
    Tsur, Achiad Shochat, Eran Ben Elisha, and Noa Osherovich.

 9) Several mv88e6060 DSA driver fixes (wrong bit definitions for
    certain registers, etc.) from Neil Armstrong.

10) Make sure to disable preemption while updating per-cpu stats of ip
    tunnels, from Jason A.  Donenfeld.

11) Various ARM64 bpf JIT fixes, from Yang Shi.

12) Flush icache properly in ARM JITs, from Daniel Borkmann.

13) Fix masking of RX and TX interrupts in ravb driver, from Masaru
    Nagai.

14) Fix netdev feature propagation for devices not implementing
    ->ndo_set_features().  From Nikolay Aleksandrov.

15) Big endian fix in vmxnet3 driver, from Shrikrishna Khare.

16) RAW socket code increments incorrect SNMP counters, fix from Ben
    Cartwright-Cox.

17) IPv6 multicast SNMP counters are bumped twice, fix from Neil Horman.

18) Fix handling of VLAN headers on stacked devices when REORDER is
    disabled.  From Vlad Yasevich.

19) Fix SKB leaks and use-after-free in ipvlan and macvlan drivers, from
    Sabrina Dubroca.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (83 commits)
  MAINTAINERS: Update Mellanox's Eth NIC driver entries
  net/core: revert "net: fix __netdev_update_features return.." and add comment
  af_unix: take receive queue lock while appending new skb
  rtnetlink: fix frame size warning in rtnl_fill_ifinfo
  net: use skb_clone to avoid alloc_pages failure.
  packet: Use PAGE_ALIGNED macro
  packet: Don't check frames_per_block against negative values
  net: phy: Use interrupts when available in NOLINK state
  phy: marvell: Add support for 88E1540 PHY
  arm64: bpf: make BPF prologue and epilogue align with ARM64 AAPCS
  macvlan: fix leak in macvlan_handle_frame
  ipvlan: fix use after free of skb
  ipvlan: fix leak in ipvlan_rcv_frame
  vlan: Do not put vlan headers back on bridge and macvlan ports
  vlan: Fix untag operations of stacked vlans with REORDER_HEADER off
  via-velocity: unconditionally drop frames with bad l2 length
  ipg: Remove ipg driver
  dl2k: Add support for IP1000A-based cards
  snmp: Remove duplicate OUTMCAST stat increment
  net: thunder: Check for driver data in nicvf_remove()
  ...

MAINTAINERS: Update Mellanox's Eth NIC driver entries

Eugenia (Jenny) Emantayev is replacing Amir Vadai as the
mlx4 Ethernet driver maintainer.

Saeed Mahameed is assigned to maintain mlx5 Eth functionality.

Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/core: revert "net: fix __netdev_update_features return.." and add comment

This reverts commit 00ee59271777 ("net: fix __netdev_update_features return
on ndo_set_features failure")
and adds a comment explaining why it's okay to return a value other than
0 upon error. Some drivers might actually change flags and return an
error so it's better to fire a spurious notification rather than miss
these.

CC: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

af_unix: take receive queue lock while appending new skb

While possibly in future we don't necessarily need to use
sk_buff_head.lock this is a rather larger change, as it affects the
af_unix fd garbage collector, diag and socket cleanups. This is too much
for a stable patch.

For the time being grab sk_buff_head.lock without disabling bh and irqs,
so don't use locked skb_queue_tail.

Fixes: 869e7c62486e ("net: af_unix: implement stream sendpage support")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Reported-by: Eric Dumazet <edumazet@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rtnetlink: fix frame size warning in rtnl_fill_ifinfo

Fix the following warning:

  CC      net/core/rtnetlink.o
net/core/rtnetlink.c: In function ‘rtnl_fill_ifinfo’:
net/core/rtnetlink.c:1308:1: warning: the frame size of 2864 bytes is larger than 2048 bytes [-Wframe-larger-than=]
 }
 ^
by splitting up the huge rtnl_fill_ifinfo into some smaller ones, so we
don't have the huge frame allocations at the same time.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: use skb_clone to avoid alloc_pages failure.

1. new skb only need dst and ip address(v4 or v6).
2. skb_copy may need high order pages, which is very rare on long running server.

Signed-off-by: Junwei Zhang <linggao.zjw@alibaba-inc.com>
Signed-off-by: Martin Zhang <martinbj2008@gmail.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

packet: Use PAGE_ALIGNED macro

Use PAGE_ALIGNED(...) instead of open-coding it.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

packet: Don't check frames_per_block against negative values

rb->frames_per_block is an unsigned int, thus can never be negative.

Also fix spacing in the calculation of frames_per_block.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: Use interrupts when available in NOLINK state

The NOLINK state will poll the phy once a second to see if the link
has come up. If the phy has an interrupt line, this polling can be
skipped, since the phy should interrupt when the link returns.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

phy: marvell: Add support for 88E1540 PHY

The 88E1540 can be found embedded in the Marvell 88E6352 switch.  It
is compatible with the 88E1510, so add support for it, using the
88E1510 specific functions.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

arm64: bpf: make BPF prologue and epilogue align with ARM64 AAPCS

Save and restore FP/LR in BPF prog prologue and epilogue, save SP to FP
in prologue in order to get the correct stack backtrace.

However, ARM64 JIT used FP (x29) as eBPF fp register, FP is subjected to
change during function call so it may cause the BPF prog stack base address
change too.

Use x25 to replace FP as BPF stack base register (fp). Since x25 is callee
saved register, so it will keep intact during function call.
It is initialized in BPF prog prologue when BPF prog is started to run
everytime. Save and restore x25/x26 in BPF prologue and epilogue to keep
them intact for the outside of BPF. Actually, x26 is unnecessary, but SP
requires 16 bytes alignment.

So, the BPF stack layout looks like:

                                 high
         original A64_SP =>   0:+-----+ BPF prologue
                                |FP/LR|
         current A64_FP =>  -16:+-----+
                                | ... | callee saved registers
                                +-----+
                                |     | x25/x26
         BPF fp register => -80:+-----+
                                |     |
                                | ... | BPF prog stack
                                |     |
                                |     |
         current A64_SP =>      +-----+
                                |     |
                                | ... | Function call stack
                                |     |
                                +-----+
                                  low

CC: Zi Shen Lim <zlim.lnx@gmail.com>
CC: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Yang Shi <yang.shi@linaro.org>
Acked-by: Zi Shen Lim <zlim.lnx@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

macvlan: fix leak in macvlan_handle_frame

Reset pskb in macvlan_handle_frame in case skb_share_check returned a
clone.

Fixes: 8a4eb5734e8d ("net: introduce rx_handler results and logic around that")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipvlan: fix use after free of skb

ipvlan_handle_frame is a rx_handler, and when it returns a value other
than RX_HANDLER_CONSUMED (here, NET_RX_DROP aka RX_HANDLER_ANOTHER),
__netif_receive_skb_core expects that the skb still exists and will
process it further, but we just freed it.

Fixes: 2ad7bf363841 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipvlan: fix leak in ipvlan_rcv_frame

Pass a **skb to ipvlan_rcv_frame so that if skb_share_check returns a
new skb, we actually use it during further processing.

It's safe to ignore the new skb in the ipvlan_xmit_* functions, because
they call ipvlan_rcv_frame with local == true, so that dev_forward_skb
is called and always takes ownership of the skb.

Fixes: 2ad7bf363841 ("ipvlan: Initial check-in of the IPVLAN driver.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'vlan-reorder'

Vladislav Yasevich says:

====================
Fix issues with vlans without REORDER_HEADER

A while ago Phil Sutter brought up an issue with vlans without
REORDER_HEADER and bridges.  The problem was that if a vlan
without REORDER_HEADER was a port in the bridge, the bridge ended
up forwarding corrupted packets that still contained the vlan header.
The same issue exists for bridge mode macvlan/macvtap devices.

An additional issue with vlans without REORDER_HEADER is that stacking
them also doesn't work.  The reason here is that skb_reorder_vlan_header()
function assumes that it on ETH_HLEN bytes deep into the packet.  That
is not the case, when you a vlan without REORRDER_HEADER flag set.

This series attempts to correct these 2 issues.

1) To solve the stacked vlans problem, the patch simply use
skb->mac_len as an offset to start copying mac addresses that
is part of header reordering.

2) To fix the issue with bridge/macvlan/macvtap, the second patch
simply doesn't write the vlan header back to the packet if the
vlan device is either a bridge or a macvlan port.  This ends up
being the simplest and least performance intrussive solution.

I've considered extending patch 2 to all stacked devices (essentially
checked for the presense of rx_handler), but that feels like a broader
restriction and _may_ break existing uses.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

vlan: Do not put vlan headers back on bridge and macvlan ports

When a vlan is configured with REORDER_HEADER set to 0, the vlan
header is put back into the packet and makes it appear that
the vlan header is still there even after it's been processed.
This posses a problem for bridge and macvlan ports.  The packets
passed to those device may be forwarded and at the time of the
forward, vlan headers end up being unexpectedly present.

With the patch, we make sure that we do not put the vlan header
back (when REORDER_HEADER is 0) if a bridge or macvlan has
been configured on top of the vlan device.

Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

vlan: Fix untag operations of stacked vlans with REORDER_HEADER off

When we have multiple stacked vlan devices all of which have
turned off REORDER_HEADER flag, the untag operation does not
locate the ethernet addresses correctly for nested vlans.
The reason is that in case of REORDER_HEADER flag being off,
the outer vlan headers are put back and the mac_len is adjusted
to account for the presense of the header.  Then, the subsequent
untag operation, for the next level vlan, always use VLAN_ETH_HLEN
to locate the begining of the ethernet header and that ends up
being a multiple of 4 bytes short of the actuall beginning
of the mac header (the multiple depending on the how many vlan
encapsulations ethere are).

As a reslult, if there are multiple levles of vlan devices
with REODER_HEADER being off, the recevied packets end up
being dropped.

To solve this, we use skb->mac_len as the offset.  The value
is always set on receive path and starts out as a ETH_HLEN.
The value is also updated when the vlan header manupations occur
so we know it will be correct.

Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

via-velocity: unconditionally drop frames with bad l2 length

By default the driver allowed incorrect frames to be received. What is
worse the code does not handle very short frames correctly. The FCS
length is unconditionally subtracted, and the underflow can cause
skb_put to be called with large number after implicit cast to unsigned.
And indeed, an skb_over_panic() was observed with via-velocity.

This removes the module parameter as it does not work in it's
current state, and should be implemented via NETIF_F_RXALL if needed.

Suggested-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: Timo Teräs <timo.teras@iki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "A fs-cache regression fix, and adding a warning about obnoxiou^W
  moderation of list given in MAINTAINERS"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  MAINTAINERS: linux-cachefs@redhat.com is moderated for non-subscribers
  FS-Cache: Add missing initialization of ret in cachefiles_write_page()

Merge branch 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

Pull crypto fix from Herbert Xu:
 "This fixes a bug in the qat driver where a user-space pointer is
  dereferenced"

* 'linus' of git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: qat - don't use userspace pointer

MAINTAINERS: linux-cachefs@redhat.com is moderated for non-subscribers

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

FS-Cache: Add missing initialization of ret in cachefiles_write_page()

fs/cachefiles/rdwr.c: In function ‘cachefiles_write_page’:
fs/cachefiles/rdwr.c:882: warning: ‘ret’ may be used uninitialized in
this function

If the jump to label "error" is taken, "ret" will indeed be
uninitialized, and random stack data may be printed by the debug code.

Fixes: 102f4d900c9c8f5e ("FS-Cache: Handle a write to the page immediately beyond the EOF marker")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

ipg: Remove ipg driver

Now that IP1000A chips are supported by dl2k driver, the buggy ipg
driver can be removed.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

dl2k: Add support for IP1000A-based cards

Add support for IP1000A chips to dl2k driver.
IP1000A chip looks like a TC9020 with integrated PHY.

This allows IP1000A chips to work reliably because the ipg driver is
buggy - it loses packets under load and then completely stops
transmitting data.

Tested with Asus NX1101 v2.0 at 10, 100 and 1000Mbps:
vendor=0x13f0 device=0x1023 (rev 0x41)
subsystem vendor=0x1043 device=0x8180

MAC address registers access needed to be changed from 8-bit to 16-bit
because 8-bit does not work on IP1000A. 8-bit access is not even
allowed in the TC9020 datasheet (although it worked). 16-bit access
works on both.

Tested that it does not break D-Link DGE-550T (DL-2000 chip, probably
a rebranded TC9020):
vendor=0x1186 device=0x4000 (rev 0x0c)
subsystem vendor=0x1186 device=0x4000

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

snmp: Remove duplicate OUTMCAST stat increment

the OUTMCAST stat is double incremented, getting bumped once in the mcast code
itself, and again in the common ip output path.  Remove the mcast bump, as its
not needed

Validated by the reporter, with good results

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Reported-by: Claus Jensen <claus.jensen@microsemi.com>
CC: Claus Jensen <claus.jensen@microsemi.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: thunder: Check for driver data in nicvf_remove()

In some cases the crash is caused by nicvf_remove() being called from
outside. For example, if we try to feed the device to vfio after the
probe has failed for some reason. So, move the check to better place.

Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/core: use netdev name in warning if no parent

A recent flaw in the netdev feature setting resulted in warnings
like this one from VLAN interfaces:

 WARNING: CPU: 1 PID: 4975 at net/core/dev.c:2419 skb_warn_bad_offload+0xbc/0xcb()
 : caps=(0x00000000001b5820, 0x00000000001b5829) len=2782 data_len=0 gso_size=1348 gso_type=16 ip_summed=3

The ":" is supposed to be preceded by a driver name, but in this
case it is an empty string since the device has no parent.

There are many types of network devices without a parent. The
anonymous warnings for these devices can be hard to debug.  Log
the network device name instead in these cases to assist further
debugging.

This is mostly similar to how __netdev_printk() handles orphan
devices.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

af_unix: don't append consumed skbs to sk_receive_queue

In case multiple writes to a unix stream socket race we could end up in a
situation where we pre-allocate a new skb for use in unix_stream_sendpage
but have to free it again in the locked section because another skb
has been appended meanwhile, which we must use. Accidentally we didn't
clear the pointer after consuming it and so we touched freed memory
while appending it to the sk_receive_queue. So, clear the pointer after
consuming the skb.

This bug has been found with syzkaller
(http://github.com/google/syzkaller) by Dmitry Vyukov.

Fixes: 869e7c62486e ("net: af_unix: implement stream sendpage support")
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: switchdev: fix return code of fdb_dump stub

rtnl_fdb_dump always expects an index to be returned by the ndo_fdb_dump op,
but when CONFIG_NET_SWITCHDEV is off, it returns an error.

Fix that by returning the given unmodified idx.

A similar fix was 0890cf6cb6ab ("switchdev: fix return value of
switchdev_port_fdb_dump in case of error") but for the CONFIG_NET_SWITCHDEV=y
case.

Fixes: 45d4122ca7cd ("switchdev: add support for fdb add/del/dump via switchdev_port_obj ops.")
Signed-off-by: Dragos Tatulea <dragos@endocode.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Fix VLANs null-pointer for 57710, 57711

Commit 05cc5a39ddb7 "bnx2x: add vlan filtering offload" introduced
a regression in regard for vlans for 57710, 57711 adapters -
Loading 8021q module on a machine with such an adapter would cause
a null pointer dereference, as the driver mistakenly publishes it
has capabilities for vlan CTAG filtering.

Reported-by: Otto Sabart <osabart@redhat.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ravb: remove unhandle int cause

This driver does not handle the AVB-DMAC Receive FIFO Warning interrupt
now, so the interrupt should not be enabled.

Signed-off-by: Masaru Nagai <masaru.nagai.vx@renesas.com>
Signed-off-by: Yoshihiro Kaneko <ykaneko0929@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

raw: increment correct SNMP counters for ICMP messages

Sending ICMP packets with raw sockets ends up in the SNMP counters
logging the type as the first byte of the IPv4 header rather than
the ICMP header. This is fixed by adding the IP Header Length to
the casting into a icmphdr struct.

Signed-off-by: Ben Cartwright-Cox <ben@benjojo.co.uk>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sfc: constify pci_error_handlers structures

This pci_error_handlers structure is never modified, like all the other
pci_error_handlers structures, so declare it as const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: cavium: liquidio: constify pci_error_handlers structures

This pci_error_handlers structure is never modified, like all the other
pci_error_handlers structures, so declare it as const.

Done with the help of Coccinelle.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>

Driver: Vmxnet3: Fix use of mfTableLen for big endian architectures

Signed-off-by: Shrikrishna Khare <skhare@vmware.com>
Reported-by: Masao Uebayashi <uebayasi@gmail.com>
Signed-off-by: Bhavesh Davda <bhavesh@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: usb: cdc_ether: add Dell DW5580 as a mobile broadband adapter

Since Dell DW5580 is a 3G modem, this patch adds the device as a
mobile broadband adapter

Signed-off-by: Daniele Palmas <dnlplm@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix __netdev_update_features return on ndo_set_features failure

If ndo_set_features fails __netdev_update_features() will return -1 but
this is wrong because it is expected to return 0 if no features were
changed (see netdev_update_features()), which will cause a netdev
notifier to be called without any actual changes. Fix this by returning
0 if ndo_set_features fails.

Fixes: 6cb6a27c45ce ("net: Call netdev_features_change() from netdev_update_features()")
CC: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix feature changes on devices without ndo_set_features

When __netdev_update_features() was updated to ensure some features are
disabled on new lower devices, an error was introduced for devices which
don't have the ndo_set_features() method set. Before we'll just set the
new features, but now we return an error and don't set them. Fix this by
returning the old behaviour and setting err to 0 when ndo_set_features
is not present.

Fixes: e7868a85e1b2 ("net/core: ensure features get disabled on new lower devs")
CC: Jarod Wilson <jarod@redhat.com>
CC: Jiri Pirko <jiri@resnulli.us>
CC: Ido Schimmel <idosch@mellanox.com>
CC: Sander Eikelenboom <linux@eikelenboom.it>
CC: Andy Gospodarek <gospo@cumulusnetworks.com>
CC: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Andy Gospodarek <gospo@cumulusnetworks.com>
Reviewed-by: Jarod Wilson <jarod@redhat.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Dave Young <dyoung@redhat.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

switchdev: bridge: Check return code is not EOPNOTSUPP

When NET_SWITCHDEV=n, switchdev_port_attr_set simply returns EOPNOTSUPP.
In this case we should not emit errors and warnings to the kernel log.

Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Fixes: 0bc05d585d38 ("switchdev: allow caller to explicitly request
attr_set as deferred")
Fixes: 6ac311ae8bfb ("Adding switchdev ageing notification on port
bridged")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: replace hardcoded values with existing define

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: remove unused local rsstable array

Remove rsstable array and its initialization from be_set_rss_hash_opts().
The array became unused after "e255787 be2net: Support for configurable
RSS hash key". The initial RSS table is now filled and stored for later
usage during Rx queue creation.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Acked-by: Sathya Perla <sathya.perla@avagotech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ravb: Fix int mask value overwritten issue

When RX/TX interrupt for Network Control queue and Best Effort queue
is issued at the same time, the interrupt mask of Network Control
queue will be reset when the mask of Best Effort queue is set.
This patch fixes this problem.

Signed-off-by: Masaru Nagai <masaru.nagai.vx@renesas.com>
Signed-off-by: Yoshihiro Kaneko <ykaneko0929@gmail.com>
Acked-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: smsc911x: Reset PHY during initialization

On certain hardware after software reboot the chip may get stuck and fail
to reinitialize during reset. This can be fixed by ensuring that PHY is
reset too.

Old PHY resetting method required operational MDIO interface, therefore
the chip should have been already set up. In order to be able to function
during probe, it is changed to use PMT_CTRL register.

The problem could be observed on SMDK5410 board.

Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bpf, arm64: start flushing icache range from header

While recently going over ARM64's BPF code, I noticed that the icache
range we're flushing should start at header already and not at ctx.image.

Reason is that after b569c1c622c5 ("net: bpf: arm64: address randomize
and write protect JIT code"), we also want to make sure to flush the
random-sized trap in front of the start of the actual program (analogous
to x86). No operational differences from user side.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Zi Shen Lim <zlim.lnx@gmail.com>
Cc: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

bpf, arm: start flushing icache range from header

During review I noticed that the icache range we're flushing should
start at header already and not at ctx.image.

Reason is that after 55309dd3d4cd ("net: bpf: arm: address randomize
and write protect JIT code"), we also want to make sure to flush the
random-sized trap in front of the start of the actual program (analogous
to x86). No operational differences from user side.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Nicolas Schichan <nschichan@freebox.fr>
Cc: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

bpf: samples: exclude asm/sysreg.h for arm64

commit 338d4f49d6f7114a017d294ccf7374df4f998edc
("arm64: kernel: Add support for Privileged Access Never") includes sysreg.h
into futex.h and uaccess.h. But, the inline assembly used by asm/sysreg.h is
incompatible with llvm so it will cause BPF samples build failure for ARM64.
Since sysreg.h is useless for BPF samples, just exclude it from Makefile via
defining __ASM_SYSREG_H.

Signed-off-by: Yang Shi <yang.shi@linaro.org>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

arm64: bpf: fix JIT frame pointer setup

BPF fp should point to the top of the BPF prog stack. The original
implementation made it point to the bottom incorrectly.
Move A64_SP to fp before reserve BPF prog stack space.

CC: Zi Shen Lim <zlim.lnx@gmail.com>
CC: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Yang Shi <yang.shi@linaro.org>
Reviewed-by: Zi Shen Lim <zlim.lnx@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: vitesse: add support for VSC8601

This adds support for the Vitesse VSC8601 PHY. Generic functions are
used for everything except interrupt handling.

Signed-off-by: Mans Rullgard <mans@mansr.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: phy: at803x: support interrupt on 8030 and 8035

Commit 77a993942 "phy/at8031: enable at8031 to work on interrupt mode"
added interrupt support for the 8031 PHY but left out the other two
chips supported by this driver.

This patch sets the .ack_interrupt and .config_intr functions for the
8030 and 8035 drivers as well.

Signed-off-by: Mans Rullgard <mans@mansr.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ip_tunnel: disable preemption when updating per-cpu tstats

Drivers like vxlan use the recently introduced
udp_tunnel_xmit_skb/udp_tunnel6_xmit_skb APIs. udp_tunnel6_xmit_skb
makes use of ip6tunnel_xmit, and ip6tunnel_xmit, after sending the
packet, updates the struct stats using the usual
u64_stats_update_begin/end calls on this_cpu_ptr(dev->tstats).
udp_tunnel_xmit_skb makes use of iptunnel_xmit, which doesn't touch
tstats, so drivers like vxlan, immediately after, call
iptunnel_xmit_stats, which does the same thing - calls
u64_stats_update_begin/end on this_cpu_ptr(dev->tstats).

While vxlan is probably fine (I don't know?), calling a similar function
from, say, an unbound workqueue, on a fully preemptable kernel causes
real issues:

[  188.434537] BUG: using smp_processor_id() in preemptible [00000000] code: kworker/u8:0/6
[  188.435579] caller is debug_smp_processor_id+0x17/0x20
[  188.435583] CPU: 0 PID: 6 Comm: kworker/u8:0 Not tainted 4.2.6 #2
[  188.435607] Call Trace:
[  188.435611]  [<ffffffff8234e936>] dump_stack+0x4f/0x7b
[  188.435615]  [<ffffffff81915f3d>] check_preemption_disabled+0x19d/0x1c0
[  188.435619]  [<ffffffff81915f77>] debug_smp_processor_id+0x17/0x20

The solution would be to protect the whole
this_cpu_ptr(dev->tstats)/u64_stats_update_begin/end blocks with
disabling preemption and then reenabling it.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'mv88e6060-fixes'

Neil Armstrong says:

====================
net: dsa: mv88e6060: cleanup and fix setup

This patchset introduces some fixes and a registers addressing cleanup for
the mv88e6060 DSA driver.

The first patch removes the poll_link as mv88e6xxx.
The 3 following patches fixes the setup in regards of the datasheet.
The 2 last patches introduces a clean header and replaces all magic values.

v2: cleanup InitReady patch, add missing Acked-by and fix header copyright notice
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: replace magic values with register defines

To align with the mv88e6xxx code, use the register defines to
access all the register addresses and bit fields.

Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: add register defines header file

To align with the mv88e6xxx code, add a similar header file
with all the register defines.
The file is based on the mv88e6xxx header for coherency.

Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: use the correct bit shift for mac0

According to the mv88e6060 datasheet, the first mac byte must
be at position 9 instead of 8 since the bit 8 is used to select
if the mac address must differ for each port for Pause frames.
Use the correct shift and set the same mac address for all port.

Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: use the correct MaxFrameSize bit

According to the mv88e6060 datasheet, the MaxFrameSize bit position
is 10 instead of 11 which is reserved.
Use the bit correctly to setup max frame size to 1536.

Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: use the correct InitReady bit

According to the mv88e6060 datasheet, the InitReady bit position
is 11 and the polarity is inverted.
Use the bit correctly to detect the end of initialization.

Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dsa: mv88e6060: remove poll_link callback

As of mv88e6xxx remove the poll_link callback since the link
state change polling is now handled by the phylib.

Tested on a mv88e6060 B0 device with a TI DM816X SoC.

Suggested-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Linux 4.4-rc1

Merge branch 'mellanox-net-fixes'

Or Gerlitz says:

====================
Mellanox NIC driver update, Nov 12, 2015

Few small mlx5 and mlx4 fixes from the team... done over
net commit c5a3788 "Merge branch 'akpm' (patches from Andrew)"

Eran's patch needs to go to 4.2 and 4.3 stable kernels.

Tariq's patch need to go to 4.3 stable too.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx4_core: Avoid returning success in case of an error flow

The err variable wasn't set with the correct error value in some cases.

Fixes: 47605df95398 ('mlx4: Modify proxy/tunnel QP mechanism [..]')
Signed-off-by: Noa Osherovich <noaos@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx4_core: Fix sleeping while holding spinlock at rem_slave_counters

When cleaning slave's counter resources, we hold a spinlock that
protects the slave's counters list. As part of the clean, we call
__mlx4_clear_if_stat which calls mlx4_alloc_cmd_mailbox which is a
sleepable function.

In order to fix this issue, hold the spinlock, and copy all counter
indices into a temporary array, and release the spinlock. Afterwards,
iterate over this array and free every counter. Repeat this scenario
until the original list is empty (a new counter might have been added
while releasing the counters from the temporary array).

Fixes: b72ca7e96acf ("net/mlx4_core: Reset counters data when freed")
Reported-by: Moni Shoua <monis@mellanox.com>
Tested-by: Moni Shoua <monis@mellanox.com>
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx5e: Use the right DMA free function on TX path

On xmit path we use skb_frag_dma_map() which is using dma_map_page(),
while upon completion we dma-unmap the skb fragments using
dma_unmap_single() rather than dma_unmap_page().

To fix this, we now save the dma map type on xmit path and use this
info to call the right dma unmap method upon TX completion.

Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx5e: Max mtu comparison fix

On change mtu the driver compares between hardware queried mtu and
software requested mtu. We need to compare between software
representation of the queried mtu and the requested mtu.

Fixes: facc9699f0fe ('net/mlx5e: Fix HW MTU settings')
Signed-off-by: Doron Tsur <doront@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx5e: Added self loopback prevention

Prevent outgoing multicast frames from looping back to the RX queue.

By introducing new HW capability self_lb_en_modifiable, which indicates
the support to modify self_lb_en bit in modify_tir command.

When this capability is set we can prevent TIRs from sending back
loopback multicast traffic to their own RQs, by "refreshing TIRs" with
modify_tir command, on every time new channels (SQs/RQs) are created at
device open.
This is needed since TIRs are static and only allocated once on driver
load, and the loopback decision is under their responsibility.

Fixes issues of the kind:
"IPv6: eth2: IPv6 duplicate address fe80::e61d:2dff:fe5c:f2e9 detected!"
The issue is seen since the IPv6 solicitations multicast messages are
loopedback and the network stack thinks they are coming from another host.

Fixes: 5c50368f3831 ("net/mlx5e: Light-weight netdev open/stop")
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/mlx5e: Fix inline header size calculation

mlx5e_get_inline_hdr_size didn't take into account the vlan insertion
into the inline WQE segment.
This could lead to max inline violation in cases where
skb_headlen(skb) + VLAN_HLEN >= sq->max_inline.

Fixes: 3ea4891db8d0 ("net/mlx5e: Fix LSO vlan insertion")
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipvs: use skb_to_full_sk() helper

SYNACK packets might be attached to request sockets.

Use skb_to_full_sk() helper to avoid illegal accesses to
inet_sk(skb->sk)

Fixes: ca6fb0651883 ("tcp: attach SYNACK messages to request sockets instead of listener")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: Sander Eikelenboom <linux@eikelenboom.it>
Acked-by: Julian Anastasov <ja@ssi.bg>
Acked-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>