GRO on IPv4 doesn't aggregate frames if they don't have DF bit set.

Some servers use IP_MTU_DISCOVER/IP_PMTUDISC_PROBE, so linux receivers
are unable to aggregate this kind of traffic.

The right thing to do is to allow aggregation as long as the DF bit has
same value on all segments.

bnx2x LRO does this correctly.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Jerry Chu <hkchu@google.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The current state of affairs is that read()/write() will setup
RFS (Receive Flow Steering) for internet protocol sockets while
poll()/epoll() does not.

When poll() gets called with a TCP or UDP socket, we should update
the flow target.

This permits to RFS (if enabled) to select the appropriate CPU for
following incoming packets.

Note: Only connected UDP sockets can benefit from RFS.
Signed-off-by: David Majnemer <majnemer@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Paul Turner <pjt@google.com>
Cc: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Corrects an byte order conflict introduced by "sctp: Correct access to
skb->{network, transport}_header". All the values in question are host
byte order.
Reported-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

pci_enable_device() will set device power state to D0,
so it's no need to do it again in tulip_init_one().
Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

pci_enable_device() will set device power state to D0,
so it's no need to do it again in qlcnic_attach_func().
Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

pci_enable_device() will set device power state to D0,
so it's no need to do it again in jme_init_one().
Signed-off-by: Yijing Wang <wangyijing@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add support for this new phy ID.
Signed-off-by: Rick Hoover <RHoover@digilentinc.com>
Signed-off-by: Steven Wang <steven.wang@digilentinc.com>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This phy is on Xilinx ZC702 zynq development board.
Signed-off-by: Anirudha Sarangi <anirudh@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use macros from linux/marvell_phy.h instead of duplicate
magic phy ID in the driver.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Phy is compatible with Vitesse 8221.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

- Remove trailing white space
- Remove spaces before tag
- Fix comments
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If the receiver supports DSACK, sender can detect false recoveries and
revert cwnd reductions triggered by either severe network reordering or
concurrent reordering and loss event.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Upon detecting spurious fast retransmit via timestamps during recovery,
use PRR to clock out new data packet instead of retransmission. Once
all retransmission are proven spurious, the sender then reverts the
cwnd reduction and congestion state to open or disorder.

The current code does the opposite: it undoes cwnd as soon as any
retransmission is spurious and continues to retransmit until all
data are acked. This nullifies the point to undo the cwnd because
the sender is still retransmistting spuriously. This patch fixes
it. The undo_ssthresh argument of tcp_undo_cwnd_reductiuon() is no
longer needed and is removed.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Refactor and relocate various functions or variables to prepare the
undo fix.  Remove some unused function arguments. Rename tcp_undo_cwr
to tcp_undo_cwnd_reduction to be consistent with the rest of
CWR related function names.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch series fixes an undo bug in fast recovery: the sender
mistakenly undos the cwnd too early but continues fast retransmits
until all pending data are acked. This also multiplies the SNMP
stat PARTIALUNDO events by the degree of the network reordering.

The first patch prepares the fix by consolidating the accounting
of newly_acked_sacked in tcp_cwnd_reduction(), instead of updating
newly_acked_sacked everytime sacked_out is adjusted.  Also pass
acked and prior_unsacked as const type because they are readonly
in the rest of recovery processing.
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This adds device tree parsing support for the shared driver of mv643xx_eth.
As the bindings are slightly different from current PPC bindings new binding
documentation is also added. Following PPC-style device setup, the shared
driver now also adds port platform_devices and sets up port platform_data.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Ethernet controllers found on Kirkwood SoCs not only suffer from loosing
MAC address register contents on clock gating but also some important
registers are reset to values that would break ethernet. This patch
clears the CLK125_BYPASS_EN bit for DT enabled Kirkwood only by using
of_device_is_compatible() instead of #ifdefs. Non-DT Kirkwood is not
affected as it installs a clock gating workaround because of the MAC
address issue above. Other Orion SoCs do not suffer from register reset,
do not have the bit in question, or do not have the register at all.
Moreover, system controllers on PPC using this driver should also be
protected from clearing that bit.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This connects to a phy node passed to the port device instead of probing
the phy by phy_addr.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This adds a struct device_node pointer for a phy passed by phandle
to mv643xx_eth node.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Make use of managed devm_ioremap and remove corresponding iounmap.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Using a separated mdio bus driver with mvmdio, phy_detach on network device
removal will not stop the phy and finally lead to NULL pointer dereference
in mvmdio due to non-existent network device. Use phy_disconnect instead
to properly stop phy device from accessing network device prior removal of
the network device.
Signed-off-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Correct email address and years.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Enable emaclite for Xilinx ARM Zynq platform.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Emaclite can be used on ARM zynq where in_be32/out_be32 IO
functions are not present. Use standard __raw_readl/__raw_writel
IO functions instead.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

xemaclite_adjust_link is used locally.
It removes sparse warning:
drivers/net/ethernet/xilinx/xilinx_emaclite.c:916:6: warning:
symbol 'xemaclite_adjust_link' was not declared. Should it be static?
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

For system which contains at least two ethernet IP where
one IP manage MDIO bus with several PHYs.

Example dts node:
ethernet_mac0: ethernet@81000000 {
	compatible = "xlnx,xps-ethernetlite-1.00.a";
	device_type = "network";
	interrupt-parent = <&xps_intc_0>;
	interrupts = < 1 0 >;
		local-mac-address = [ 00 0a 35 00 db bb ];
	phy-handle = <&ethernet_mac0_phy0>;
	reg = < 0x81000000 0x10000 >;
	xlnx,duplex = <0x1>;
	xlnx,family = "spartan3e";
	xlnx,include-internal-loopback = <0x0>;
	xlnx,include-mdio = <0x1>;
	xlnx,rx-ping-pong = <0x0>;
	xlnx,tx-ping-pong = <0x0>;
	ethernet_mac0_mdio {
		#address-cells = <1>;
		#size-cells = <0>;
		ethernet_mac0_phy0: phy@1 {
			reg = <0x1>;
		} ;
		ethernet_mac0_phy1: phy@3 {
			reg = <0x3>;
		} ;
	} ;
} ;
ethernet_mac2: ethernet@81040000 {
	compatible = "xlnx,xps-ethernetlite-1.00.a";
	device_type = "network";
	interrupt-parent = <&xps_intc_0>;
	interrupts = < 11 0 >;
		local-mac-address = [ 00 0a 35 00 db bb ];
	phy-handle = <&ethernet_mac0_phy1>;
	reg = < 0x81040000 0x10000 >;
	xlnx,duplex = <0x1>;
	xlnx,family = "spartan3e";
	xlnx,include-internal-loopback = <0x0>;
	xlnx,include-mdio = <0x0>;
	xlnx,rx-ping-pong = <0x0>;
	xlnx,tx-ping-pong = <0x0>;
} ;
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Be more verbose when any problem happens.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Manish Chopra says:

====================
This patch series includes following changes:

* Log driver version information with firmware version information.
* Log "Unknown" adapter name and serial number in case of
  mismatched adapter type.
* netxen_setup_intr() function code cleanup and changes to avoid attaching
  of multiple pci functions on an adapter with mixed interrupts mode
  [MSI-X/MSI + INTx].
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o Adapters do not support co-existence of INTx Interrupt with
  MSI-X or MSI among multiple functions. Prevent attaching
  of a function during normal load, if adapter gets into mixed
  mode of interrupts
Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o log "Unknown" board name and "Unknown" serial number in case
  of mismatched adapter type found. This will avoid weird characters
  logs when an adapter is in bad state or corrupted.
Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Included changes:
- reduce broadcast overhead on non-lossy wired links
- fix typos in kernel doc
- use eth_hdr() when possible
- use netdev_allock_skb_ip_align() and don't deal with NET_IP_ALIGN
- change VID semantic in the BLA component
- other minor cleanups and code refactoring
Signed-off-by: David S. Miller <davem@davemloft.net>

Instead of dereferencing pointer and put values on stack we could use nice
%*phC specifier.
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

After b924551b ("bonding: fix enslaving in alb mode when link down") we
don't need the bond parameter in alb_swap_mac_addr(), so remove it.
Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When vxlan is compiled as builtin, its init code
runs before IPv6 init, this could cause problems
if we create IPv6 socket in the latter patch.

Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

'unsigned int' is slightly better.

Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <amwang@redhat.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>