net: dm9000: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: spider_net: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: minor cleanup to net_namespace.c.

Inline a small static function that's only ever called from one place.

Signed-off-by: Rob Landley <rlandley@parallels.com>
Reviewed-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: RTA_PREFSRC support for ipv6 route source address selection

[ipv6] Add support for RTA_PREFSRC

This patch allows a user to select the preferred source address
for a specific IPv6-Route. It can be set via a netlink message
setting RTA_PREFSRC to a valid IPv6 address which must be
up on the device the route will be bound to.

Signed-off-by: Daniel Walter <dwalter@barracuda.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: ignore looped-back NA while dad is running

[ipv6] Ignore looped-back NAs while in Duplicate Address Detection

If we send an unsolicited NA shortly after bringing up an
IPv6 address, the duplicate address detection algorithm
fails and the ip stays in tentative mode forever.
This is due a missing check if the NA is looped-back to us.

Signed-off-by: Daniel Walter <dwalter@barracuda.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rndis_host: Quirky devices are still 'point-to-point'

My changes in commit 4d42d417be75d750b82798922b6e775915e11bce were
written some time before the introduction of FLAG_POINTTOPOINT, so
didn't include that flag in the new driver_info.  Change the new
driver_info to be consistent.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-bonding: Adding support for throughputs larger than 65536 Mbps

This updates the bonding driver to support v2.6.27-rc3 enhancements
(b11f8d8c aka. "ethtool: Expand ethtool_cmd.speed to 32 bits") which
allow to encode the Mbps link speed on 32-bits (Max 4 Pbps) instead of
16 (Max 65536 Mbps).

This patch also attempts to compact struct slave by reordering its
fields.

Signed-off-by: David Decotigny <decot@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-bonding: Fix minor/cosmetic type inconsistencies

The __get_link_speed() function returns a u16 value which was stored
in a u32 local variable. This patch uses the return value directly,
thus fixing that minor type consistency.

The 'duplex' field in struct slave being encoded on 8 bits, to be more
consistent we use a u8 integer (instead of u16) whenever we copy it to
local variables.

Signed-off-by: David Decotigny <decot@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-bonding: Fix minor sparse complaints

This gets rid of minor sparse complaints:
drivers/net/bonding/bond_main.c:4361:4: warning: do-while statement is not a compound statement
drivers/net/bonding/bond_main.c:243:12: warning: symbol 'bond_mode_name' was not declared. Should it be static?

Signed-off-by: David Decotigny <decot@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlge: make nic_operations struct const

The struct nic_operations is just function pointers and should be
declared const for added security.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sfc: make function tables const

The phy, mac, and board information structures should be const.
Since tables contain function pointer this improves security
(at least theoretically).

Compile tested only.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Phonet: convert bound sockets hash list to RCU

This gets rid of the last spinlock in the Phonet stack proper.

Signed-off-by: Rémi Denis-Courmont <remi.denis-courmont@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ethtool: allow custom interval for physical identification

When physical identification of an adapter is done by toggling the
mechanism on and off through software utilizing the set_phys_id operation,
it is done with a fixed duration for both on and off states.  Some drivers
may want to set a custom duration for the on/off intervals.  This patch
changes the API so the return code from the driver's entry point when it
is called with ETHTOOL_ID_ACTIVE can specify the frequency at which to
cycle the on/off states, and updates the drivers that have already been
converted to use the new set_phys_id and use the synchronous method for
identifying an adapter.

The physical identification frequency set in the updated drivers is based
on how it was done prior to the introduction of set_phys_id.

Compile tested only.  Also fixes a compiler warning in sfc.

v2: drivers do not return -EINVAL for ETHOOL_ID_ACTIVE
v3: fold patchset into single patch and cleanup per Ben's feedback

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Cc: Sathya Perla <sathya.perla@emulex.com>
Cc: Subbu Seetharaman <subbu.seetharaman@emulex.com>
Cc: Ajit Khaparde <ajit.khaparde@emulex.com>
Cc: Michael Chan <mchan@broadcom.com>
Cc: Eilon Greenstein <eilong@broadcom.com>
Cc: Divy Le Ray <divy@chelsio.com>
Cc: Don Fry <pcnet32@frontier.com>
Cc: Jon Mason <jdmason@kudzu.us>
Cc: Solarflare linux maintainers <linux-net-drivers@solarflare.com>
Cc: Steve Hodgson <shodgson@solarflare.com>
Cc: Stephen Hemminger <shemminger@linux-foundation.org>
Cc: Matt Carlson <mcarlson@broadcom.com>
Acked-by: Jon Mason <jdmason@kudzu.us>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: Call fib_select_default() only when actually necessary.

fib_select_default() is a complete NOP, and completely pointless
to invoke, when we have no more than 1 default route installed.

And this is far and away the common case.

So remember how many prefixlen==0 routes we have in the routing
table, and elide the call when we have no more than one of those.

This cuts output route creation time by 157 cycles on Niagara2+.

In order to add the new int to fib_table, we have to correct the type
of ->tb_data[] to unsigned long, otherwise the private area will be
unaligned on 64-bit systems.

Signed-off-by: David S. Miller <davem@davemloft.net>
Reviewed-by: Eric Dumazet <eric.dumazet@gmail.com>

NET: AX.25, NETROM, ROSE: Remove SOCK_DEBUG calls

Nobody alive seems to recall when they last were useful.

Signed-off-by: Ralf Baechle <ralf@linux-mips.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: DCB, X540 devices do not respond to pause frames

DCB enabled X540 devices are not responding to pause frames
due to a missing register set that was added for these
devices that did not exist in other devices.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: DCB, misallocated packet buffer size with X540 device

The X540 device has a smaller packet buffer but the DCB configuration
never took this into account. Under stress this can result in the DMA
engine hanging and TX Unit hang occurring to reset the device. This
patch reworks the packet buffer allocation routine used for DCB on
82599 and X540 devices to account for RX packet buffer sizes.

This fixes the immediate hang. We should consolidate the various
hardware specific routines for configuring features into a single
routine. This will make it much harder to miss feature cases like
this.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: make device_caps() generic

x540 has the same device capability word in the EEPROM as 82599.
This patch renames ixgbe_get_device_caps_82599 to
ixgbe_get_device_caps_generic, moves it to ixgbe_common.h and
sets up the function pointer for x540.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: explicitly disable 100H for x540

100H is not supported on this HW, but the bit is set on the PHY.
This can result in link at 100F when advertising only 1000F.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: add support for new HW

Add new device ID supported by ixgbe.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: fix semaphores in eeprom routines for x540

HW can upload EEPROM content from flash while
in a middle of checksum calculation. Take NVM ownership for the whole
process of checksum update.

Call ixgbe_read_eerd_generic() and ixgbe_write_eewr_generic() directly to
avoid double take of semaphores which leads to long loading times.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup short msleep's (<20ms) to use usleep_range

Since msleep might not sleep for the desired amount when less
than 20ms use usleep_range.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: fix 82599 KR downshift coexistence with LESM FW module

Disable KR to KX4/KX downshift on 82599 backplane devices when
LESM (Link Establishment State Machine) is enabled in FW. Those
features cannot co-exist as they both manipulate the same registers.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phillip Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: move disabling of relaxed ordering in start_hw()

Relaxed ordering can lead to issues with some chipsets.

This patch makes sure that it is disabled by default and
not only when DCA is on.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: refactor common start_hw code for 82599 and x540

Factored out the common start_hw code into a new function
ixgbe_start_hw_gen2() so that it can be used by x540 and 82599.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: update version string for Dell CEM use

Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>

ixgbe: fix namespacecheck issue

Set ixgbe_identify_82599() as static

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: correct function number for some 82598 parts

Some 82598 parts have LAN0 disabled and LAN1 enabled and the LAN ID bits in
Device Status register report the NIC as having only LAN1 as enabled. This
causes ixgbe_set_lan_id_multi_port_pcie() to set bus->func = 1 which is
incorrect.

Force bus->func to 0 when LAN0 is disabled in the EEPROM.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: fix return value checks

The value of status was incorrectly tested. Also whitespace cleanup.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Evan Swanson <evan.swanson@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: DCB, further cleanups to app configuration

With the app data on the kernel dcb_app list we no longer
need to specifically handle them in ixgbe for the CEE case.
So now we can remove app handling logic and check when the
hw is configured if the app data matches the hardware
configuration in set_hw_all().

If it does not match then we can reconfigure.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: If ASPM L0s needs to be disabled, do it prior to enabling device

Based on a patch from Naga Chumbalkar <nagananda.chumbalkar@hp.com>:

If ASPM L0s needs to be disabled due to HW errata, do it prior to
"enabling" the device. This way if the kernel ever defaults its
aspm_policy to POLICY_POWERSAVE, then the e1000e driver will get a
chance to disable ASPM on the misbehaving device *prior* to calling
pci_enable_device_mem(). This will be useful in situations
where the BIOS indicates ASPM support on the server by clearing the
ACPI FADT "ASPM Controls" bit.

Note:
The kernel (2.6.38) currently uses the BIOS "default" as its aspm_policy.
However, Linux distros can diverge from that and set the default to
"powersave".

v2: o cleanup namespace pollution of e1000e_disable_aspm(),
    o fix type and initialization of the new aspm_disable_flag in a few
      functions, and
    o redefine FLAG2_DISABLE_ASPM_L0S to the first unused bit in
      adapter->flags2.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Cc: Naga Chumbalkar <nagananda.chumbalkar@hp.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: fix stats locking in e1000_watchdog_task

Just move the unlock down a bit because it unlocks too
early leaving a chance for get_stats64() run in parallel
while it is still accessing the stats.

Signed-off-by: Flavio Leitner <fleitner@redhat.com>
Acked-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: convert to new VLAN model

This switches the e1000e driver to use the new VLAN interfaces.

CC: Jesse Gross <jesse@nicira.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>

e1000e: PCIe link speed in GT/s, not GB/s

Correct the log message when driver loads.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: convert short duration msleep() to usleep_range()

With durations less than 20ms, the jiffies or legacy timer backed msleep()
may sleep ~20ms which might not be what the caller expects.  Instead, it
is recommended to use the hrtimers backed usleep_range().  For more, see
Documentation/timers/timers-howto.txt.  Issues reported by checkpatch.

In addition, remove unnecessary sleep in e1000e_write_nvm_spi().

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb: introduce igb_thermal_sensor_event for sensor checking

The code for thermal sensor checking should be wrapped into a function.

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb: fix typo in igb_validate_nvm_checksum_82580

Comment spelling fix.

Signed-off-by: Stefan Assmann <sassmann@kpanic.de>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igb: Add anti-spoofing feature documentation

Add the documentation for the anti-spoofing feature in the HW.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

net: allow shifted access in smsc911x V2

This is a revised patch that permits a shifted access to the
LAN9221 registers.  More specifically:

 It adds a shift parameter in the platform_data.
 It introduces an ops in smsc911x_data.
 A choice of access function to use at run-time.
 Four new shifted access function.

Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Alessandro Rubini <rubini@gnudd.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Add support for extended VPD blocks

In some devices, the VPD block is relocated to a different area in
NVRAM.  The original location can still contain old, but still valid VPD
data.  This patch changes the code to look for an extended VPD block in
NVRAM.  If one is found, that block is used for all VPD operations
instead.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Add jumbo frame loopback tests to selftest

This patch adds jumbo frame loopback test support to the ethtool
selftest.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Automatically size stat/test string arrays

This patch reimplements the size preprocessor constants of the stats and
ethtool test string arrays.  The size is calculated at compile time
rather than using static constants.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Benjamin Li <benli@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Dump registers when status block shows errors

This patch monitors the error bit of the status word within the status
block.  If it is set, the driver will dump the driver state after
validating the error and then reset the chip.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Provide full regdump on tx timeout

The current amount of information provided in the output of a tx timeout
is insufficient to determine a root cause.  This patch replaces the
terse, four-register status output with a more complete body of
information.  For PCIe devices, the full register space is dumped.  For
other devices, select registers are dumped instead.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rndis_host: Poll status before control channel where necessary

Some RNDIS devices don't respond on the control channel until polled
on the status channel.  In particular, this was reported to be the
case for the 2Wire HomePortal 1000SW and for some Windows Mobile
devices.

This is roughly based on a patch by John Carr <john.carr@unrouted.co.uk>
which is currently applied by Mandriva.

Reported-by: Mark Glassberg <vzeeaxwl@myfairpoint.net>
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: review Wol and enable the Unicast support

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

macb: Add rx overrun counter

Signed-off-by: Alexander Stein <alexander.stein@systec-electronic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ethtool support to configure number of channels

Ethtool support to configure RX, TX and other channels. combined field
in struct ethtool_channels to reflect set of channel (RX, TX or other).
Other channel can be link interrupts, SR-IOV coordination etc.

ETHTOOL_GCHANNELS will report max and current number of RX channels,
max and current number of TX channels, max and current number of other channel
or max and current number of combined channel.

Number of channel can be modify upto max number of channel through
ETHTOOL_SCHANNELS command.

Ben Hutchings:
o define 'combined' and 'other' types.  Most multiqueue drivers pair up RX and TX
  queues so that most channels combine RX and TX work.
o Please could you use a kernel-doc comment to describe the structure.

Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Missing 'inline' in vlan-disabled vlan_untag()

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-davem' of git://git./linux/kernel/git/linville/wireless-next-2.6

atm: iphase: Fix set-but-not-used warnings.

The "iavcc" and "iadev" cases are obvious.

The intr_status and frmr_intr cases are reading a register to clear
the chip status.  This driver is pretty old and creaky, and uses
volatile pointer dereferences to do register I/O when it should be
using readl() and friends.  However that it outside of the scope of
these changes.

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

9p: Kill set but unused variable in 9p_client_{read,write}() and p9_client_readdir()

Fixes the following warnings:

net/9p/client.c:1305:18: warning: variable ‘total’ set but not used [-Wunused-but-set-variable]
net/9p/client.c:1370:18: warning: variable ‘total’ set but not used [-Wunused-but-set-variable]
net/9p/client.c:1769:18: warning: variable ‘total’ set but not used [-Wunused-but-set-variable]

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

net: vlan_features comment clarification

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ioc3: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bnx2x: convert to hw_features

Since ndo_fix_features callback is postponing features change when
bp->recovery_state != BNX2X_RECOVERY_DONE, netdev_update_features()
has to be called again when this condition changes. Previously,
ethtool_ops->set_flags callback returned -EBUSY in that case
(it's not possible in the new model).

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
v5: - don't delay set_features, as it's rtnl_locked - same as recovery process
v4: - complete bp->rx_csum -> NETIF_F_RXCSUM conversion
    - add check for failed ndo_set_features in ndo_open callback
v3: - include NETIF_F_LRO in hw_features
    - don't call netdev_update_features() if bnx2x_nic_load() failed
v2: - comment in ndo_fix_features callback
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bna: convert to hw_features

Note: looks like bnad->conf_mutex is duplicating rtnl_lock.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ps3_gelic: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: sky2: convert to hw_features

Caveats:
 - driver modifies vlan_features on HW VLAN TX changes
 - broken RX checksum will be reenabled on features change

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add RTNL_ASSERT in __netdev_update_features()

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: vlan: make non-hw-accel rx path similar to hw-accel

Now there are 2 paths for rx vlan frames. When rx-vlan-hw-accel is
enabled, skb is untagged by NIC, vlan_tci is set and the skb gets into
vlan code in __netif_receive_skb - vlan_hwaccel_do_receive.

For non-rx-vlan-hw-accel however, tagged skb goes thru whole
__netif_receive_skb, it's untagged in ptype_base hander and reinjected

This incosistency is fixed by this patch. Vlan untagging happens early in
__netif_receive_skb so the rest of code (ptype_all handlers, rx_handlers)
see the skb like it was untagged by hw.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
v1->v2:
remove "inline" from vlan_core.c functions
Signed-off-by: David S. Miller <davem@davemloft.net>

ethtool: time to blink provided in seconds not jiffies

When blinking for a duration set by the user, the value specified is in
seconds but it is used as the number of jiffies in the timeout after which
the Physical ID indicator is deactivated.  Fix by converting the timeout
to seconds.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next-2.6 into for-davem

Conflicts:
drivers/net/wireless/ath/ar9170/main.c
drivers/net/wireless/ath/ar9170/phy.c
drivers/net/wireless/zd1211rw/zd_rf_rf2959.c

vxge: update driver version

Update vxge driver version to 2.5.3

Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

vxge: spin-lock issue

In vxge_hw_vpath_close, __vxge_hw_vp_terminate memsets the vpath which
clobbers the spin lock state, then the driver attempts to acquire the
spin lock.  Resolve this by not zeroing the lock part of vpath struct,
clean-up vpath locking in init, close, and fix locking hole in fw_api
call.

Issue found by Bob Picco <bpicco@redhat.com>

Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

vxge: always enable hardware time stamp

Hardware time stamp calculation can only be enabled by the privileged
function. Enable it always by default and simply use the ethtool
interface to set a flag to indicate whether or not the respective
function driver should indicate the timestamp along with the received
packet.

Also, make certain fields in vxge_hw_device_config bit-fields to reduce
the size of the struct.

Signed-off-by: Jon Mason <jdmason@kudzu.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

s2io: Fix warnings due to -Wunused-but-set-variable.

Most of these are cases where we are trying to read back a register
after a write to ensure completion.

Simply pre-fixing the readl() or readq() with "(void)" is sufficient
because these are volatile operations and the compiler cannot eliminate
them just because no real assignment takes place.

The case of free_rxd_blk()'s assignments to "struct buffAdd *ba" is a
real spurious assignment as this variable is completely otherwise
unused.

Signed-off-by: David S. Miller <davem@davemloft.net>
Acked-by: Jon Mason <jdmason@kudzu.us>

niu: Fix warnings due to -Wunused-but-set-variable

Most of these were legitimate, and once case was a real bug
(not propagating errors from ->xcvr_init() methods).

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

caif: Add BUG_ON if dev_info is missing in packet

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

caif: Don't resend if dev_queue_xmit fails.

If CAIF Link Layer returns an error, we no longer try to re-build the
CAIF packet and resend it. Instead, we simply return any transmission
errors to the socket client.

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

caif: code cleanup

Cleanup of new CAIF code.
  * make local functions static
  * remove code that is never used
  * expand get_caif_conf() since wrapper is no longer needed
  * make args to comparison functions const
  * rename connect_req_to_link_param to keep exported names
    consistent

Compile tested only.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of /linux/kernel/git/davem/net-2.6

Conflicts:
drivers/net/smsc911x.c

niu: Recognise original ethtool class code for AH/ESP flow hashing

When the RX network flow classification interface was originally
defined for reporting and controlling of flow hashing, AH and ESP were
not given distinct flow class codes (apparently because the Sun
Neptune hardware treats them very similarly).

For flow steering, they must be distinguished, so new and separate
flow class codes were added for AH and ESP.  But for backward-
compatibility, flow hash operations should continue to support the
original class codes.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

gianfar: Clean up implementation of RX network flow classification

This code was cribbed from niu, so gfar_set_hash_opts() begins by
converting the ethtool flow class code into a class code for Sun
Neptune hardware, then does the same thing again for the hardware it's
really dealing with.  It may also return -1 (-EPERM) for some
unhandled ethtool flow class codes.

Remove the useless code and definitions, and fix the error code.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlge: use ethtool set_phys_id

This is a stab at replacing old ethtool phys_id with set_phys_id
on the Qlogic 10Gb driver. Compile tested only.

Not sure if set_led_cfg will flash continuously, or needs
to be replaced by ETHTOOL_ID_ON/ETHTOOL_ID_OFF

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

v3 ethtool: add ntuple flow specifier data to network flow classifier

This change is meant to add an ntuple data extensions to the rx network flow
classification specifiers.  The idea is to allow ntuple to be displayed via
the network flow classification interface.

The first patch had some left over stuff from the original flow extension
flags I had added.  That bit is removed in this patch.

The second had some left over comments that stated we ignored bits in the
masks when we actually match them.

This work is based on input from Ben Hutchings.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ethtool: prevent null pointer dereference with NTUPLE set but no set_rx_ntuple

This change is meant to prevent a possible null pointer dereference if
NETIF_F_NTUPLE is defined but the set_rx_ntuple function pointer is not.

The main motivation behind this patch is to eventually replace the ntuple
interfaces entirely with the network flow classifier interfaces.  This
allows the device drivers to maintain the ntuple check internally while
using the network flow classifier interface for setting up and displaying
rules.
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

caif: performance bugfix - allow radio stack to prioritize packets.

    In the CAIF Payload message the Packet Type indication must be set to
    UNCLASSIFIED in order to allow packet prioritization in the modem's
    network stack. Otherwise TCP-Ack is not prioritized in the modems
    transmit queue.

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

caif: Bugfix use for_each_safe when removing list nodes.

Signed-off-by: Sjur Brændeland <sjur.brandeland@stericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netxen: limit skb frags for non tso packet

Machines are getting deadlock in four node cluster environment.
All nodes are accessing (find /gfs2 -depth -print|cpio -ocv > /dev/null)
200 GB storage on a GFS2 filesystem.
This result in memory fragmentation and driver receives 18 frags for
1448 byte packets.
For non tso packet, fw drops the tx request, if it has >14 frags.

Fixing it by pulling extra frags.

Cc: stable@kernel.org
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding:fix two typos

replace relpy with reply.
replace premanent with permanent.

Signed-off-by: Weiping Pan(潘卫平) <panweiping3@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix tranmitted/tranmitting typo

replace tranmitted with transmitted.
replace tranmitting with transmitting.

Signed-off-by: Weiping Pan(潘卫平) <panweiping3@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding:delete unused rlb_interval_counter

Now, alb_bond_info uses rx_ntt,rlb_update_delay_counter and
rlb_update_retry_counter to decide when to call rlb_update_rx_clients().

Signed-off-by: Weiping Pan(潘卫平) <panweiping3@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding:delete unused alb_timer

Now bonding-alb uses delayed_work instead of timer_list.

Signed-off-by: Weiping Pan(潘卫平) <panweiping3@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding:set save_load to 0 when initializing

It is unnecessary to set save_load to 1 here,
as the tx_hashtbl is just kzalloced.

Signed-off-by: Weiping Pan(潘卫平) <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: fix Transmit Underflow error

On some old MAC chips without COE sometime the
Transmit Underflow error is issued.

The driver aborted all the transmission process
and initialized it from scratch.
This breaks the network activity as raised by Nachiketa
on a SPEAr board.

The patch is to fix this rare underflow event.
The driver will only clear the interrupt and the Tx
DMA will go out the Suspend state as soon as the
descriptor is fetched again.
The driver will continue to bump-up the DMA FIFO threshold
that, indeed, helped somebody to prevent this kind of error
in the past as well.

Reported-by: Nachiketa Prachanda <nprachanda@ncomputing.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: fix open funct when exit on error

This patch reviews the open function and fixes some
errors when exit with an error state.
It also moves the request_irq after core is initialized
when interrupts are properly masked.

Signed-off-by: Shiraz Hashim <shiraz.hashim@st.com>
Hacked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: fixed dma lib build when turn-on the debug option

This patch fixes a compilation error when build the
dwmac_lib with the DEBUG option enabled.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

iwlegacy: make iwl3945 and iwl4965 select IWLWIFI_LEGACY

Otherwise, IWLWIFI_LEGACY has to be selected independently before the
drivers are made available.

Reported-by: Dave Airlie <airlied@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
Cc: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: Wey-Yi Guy <wey-yi.w.guy@intel.com>

pci: fix PCI bus allocation alignment handling

In commit 13583b16592a ("PCI: refactor io size calculation code") Ram
had a thinko in the refactorization of the code: the end result used the
variable 'align' for the bus alignment, but the original code used
'min_align'.

Since then, another use of that 'align' variable got introduced by
commit c8adf9a3e873 ("PCI: pre-allocate additional resources to devices
only after successful allocation of essential resources.")

Fix both of those uses to use 'min_align' as they should.

Daniel Hellstrom <daniel@gaisler.com>
Acked-by: Ram Pai <linuxram@us.ibm.com>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (34 commits)
  net: Add support for SMSC LAN9530, LAN9730 and LAN89530
  mlx4_en: Restoring RX buffer pointer in case of failure
  mlx4: Sensing link type at device initialization
  ipv4: Fix "Set rt->rt_iif more sanely on output routes."
  MAINTAINERS: add entry for Xen network backend
  be2net: Fix suspend/resume operation
  be2net: Rename some struct members for clarity
  pppoe: drop PPPOX_ZOMBIEs in pppoe_flush_dev
  dsa/mv88e6131: add support for mv88e6085 switch
  ipv6: Enable RFS sk_rxhash tracking for ipv6 sockets (v2)
  be2net: Fix a potential crash during shutdown.
  bna: Fix for handling firmware heartbeat failure
  can: mcp251x: Allow pass IRQ flags through platform data.
  smsc911x: fix mac_lock acquision before calling smsc911x_mac_read
  iwlwifi: accept EEPROM version 0x423 for iwl6000
  rt2x00: fix cancelling uninitialized work
  rtlwifi: Fix some warnings/bugs
  p54usb: IDs for two new devices
  wl12xx: fix potential buffer overflow in testmode nvs push
  zd1211rw: reset rx idle timer from tasklet
  ...

net: Add support for SMSC LAN9530, LAN9730 and LAN89530

This patch adds support for SMSC's LAN9530, LAN9730 and LAN89530 USB
ethernet controllers to the existing smsc95xx driver by adding
their new USB VID/PID pairs.

Signed-off-by: Steve Glendinning <steve.glendinning@smsc.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bnx2: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: via-velocity: convert to hw_features

Trivial conversion.

This also enables toggling TX VLAN offload and fixes TX checksumming
race with offload changes.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ksz884x: convert to hw_features

This also fixes possible race when changing receive checksum state
and removes IPV6_CSUM_GEN_HACK as it's always set.

BTW, The claim about fake IPV6 checksum looks dubious. If that were true,
then there's a problem in networking core and should be fixed there and not
in random drivers.

BTW#2, there's no MAINTAINERS entry for this driver. I assume this driver
is supported by Micrel?

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: stmmac: convert to hw_features

This also removes TSO as it's made fully in software --- better to leave this
to networking core.

If the MAC features can be detected at probe time and not at open, then
stmmac_fix_features could be simplified by limiting hw_features. That's
also better for users as they don't see offloads being togglable but
never turned on.

Redundant fallbacks for TX csum are removed as it's already handled
by network core.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: 8139cp: convert to hw_features

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: r8169: convert to hw_features

Simple conversion with a bit of needed cleanup.

This also fixes:
 - confusion around vlan_features in rtl8169_vlan_mode(),
 - problem with broken TSO for too big MTU (the limit is set
   at 0xFFF --- max MSS field value).

SG+IP_CSUM+TSO is left disabled by default, based on suggestion by
David Dillow.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: benet: convert to hw_features - fixup

Fix up after merge with NETIF_F_RXHASH implementation.

This allows to toggle NETIF_F_RXHASH and NETIF_F_HW_VLAN_TX.

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Remove invalid offloads

Remove offload changing ethtool ops which drivers don't really support:

 - fs_enet
 - ucc_geth

Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>