net: Make flow cache paths use a const struct flowi.

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

xfrm: Mark flowi arg to xfrm_resolve_and_create_bundle() const.

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

xfrm: Mark flowi arg to xfrm_dst_{alloc_copy,update_origin}() const.

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

xfrm: Mark flowi arg to xfrm_bundle_create() const.

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

xfrm: Mark flowi arg to xfrm_tmpl_resolve{,_one}() const.

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

xfrm: Mark flowi arg to xfrm_expand_policies() const.

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

xfrm: Mark flowi arg to xfrm_policy_{lookup_by_type,match}() const.

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

xfrm: Kill strict arg to xfrm_bundle_ok().

Always set to "0".

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

net: Mark flowi arg to flow_cache_uli_match() const.

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

xfrm: Mark flowi arg to xfrm_state_find() const.

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

xfrm: Mark flowi arg to xfrm_init_tempstate() const.

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

xfrm: Mark flowi arg to xfrm_state_look_at() const.

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

xfrm: Mark flowi arg to security_xfrm_state_pol_flow_match() const.

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

xfrm: Mark flowi arg to xfrm_selector_match() const.

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

xfrm: Mark token args to addr_match() const.

Also, make it return a real bool.

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

xfrm: Mark flowi arg to xfrm_type->reject() const.

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

xfrm: Mark flowi arg to ->init_tempsel() const.

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

xfrm: Mark flowi arg to ->fill_dst() const.

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

xfrm: Mark flowi arg to ->get_tos() const.

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

xfrm: Mark flowi arg const in key extraction helpers.

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

cls_u32: fix sparse warnings

The variable _data is used in asm-generic to define sections
which causes sparse warnings, so just rename the variable.

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

Merge branch 'net/ax88796' of git://git.pengutronix.de/git/mkl/linux-2.6

be2net: use hba_port_num instead of port_num

Use hba_port_num for phy loopback and ethtool phy identification.

From: Suresh R <suresh.reddy@emulex.com>
Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: add code to display temperature of ASIC

Add support to display temperature of ASIC via ethtool -S

From: Somnath K <somnath.kotur@emulex.com>
Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: fix to ignore transparent vlan ids wrongly indicated by NIC

With transparent VLAN tagging, the ASIC wrongly indicates packets with VLAN ID.
Strip them off in the driver. The VLAN Tag to be stripped will be given to the host
as an async message.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: variable name change

change occurances of stats_ioctl_sent to stats_cmd_sent

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: fixes in ethtool selftest

> add missing separator between items in ethtool self_test array
> fix reporting of test resluts when link is down and
  when selftest command fails.

From: Suresh R <suresh.reddy@emulex.com>
Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: add new counters to display via ethtool stats

New counters:
> jabber frame stats
> red drop stats

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'for-davem' of git://git./linux/kernel/git/bwh/sfc-next-2.6

net: add __rcu annotations to sk_wq and wq

Add proper RCU annotations/verbs to sk_wq and wq members

Fix __sctp_write_space() sk_sleep() abuse (and sock->wq access)

Fix sunrpc sk_sleep() abuse too

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

Merge branch 'fec' of git://git.pengutronix.de/git/ukl/linux-2.6

sh: sh_eth: Add support ethtool

This commit supports following functions.
  - get_settings
  - set_settings
  - nway_reset
  - get_msglevel
  - set_msglevel
  - get_link
  - get_strings
  - get_ethtool_stats
  - get_sset_count

About other function, the device does not support.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Signed-off-by: Nobuhiro Iwamatsu <nobuhiro.iwamatsu.yj@renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ax88796: use generic mdio_bitbang driver

..instead of using hand-crafted and not proper working version.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: clean up probe and remove function

This way we can remove the struct resource pointers from the private data.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: make pointer to platform data const

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: remove platform_device member from struct ax_device

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: use netdev_<LEVEL> instead of dev_<LEVEL> and pr_<LEVEL>

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: remove first_init parameter from ax_init_dev()

ax_init_dev() is always called with first_init=1.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: remove memset of private data

It's allocated via alloc_netdev, thus already zeroed.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: don't use magic ei_status to acces private data

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

ax88796: fix codingstyle and checkpatch warnings

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>

sctp: fix compile warnings in sctp_tsnmap_num_gabs

net/sctp/tsnmap.c: In function ‘sctp_tsnmap_num_gabs’:
net/sctp/tsnmap.c:347: warning: ‘start’ may be used uninitialized in this function
net/sctp/tsnmap.c:347: warning: ‘end’ may be used uninitialized in this function

Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: Remove debug macro of TCP_CHECK_TIMER

Now, TCP_CHECK_TIMER is not used for debuging, it does nothing.
And, it has been there for several years, maybe 6 years.

Remove it to keep code clearer.

Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: document tcp_max_ssthresh (Limited Slow-Start)

Base on Ilpo's patch about documenting tcp_max_ssthresh.
(see http://marc.info/?l=linux-netdev&m=117950581307310&w=2)

According to errata of RFC3742, fix the number of segments increased
during RTT time.

Just to state the occasion to use this parameter, But
about how to set parameter value, maybe some others can do it.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

Conflicts:
Documentation/feature-removal-schedule.txt
drivers/net/e1000e/netdev.c
net/xfrm/xfrm_policy.c

sctp: fix reporting of unknown parameters

commit 5fa782c2f5ef6c2e4f04d3e228412c9b4a4c8809 re-worked the
handling of unknown parameters. sctp_init_cause_fixed() can now
return -ENOSPC if there is not enough tailroom in the error
chunk skb. When this happens, the error header is not appended to
the error chunk. In that case, the payload of the unknown parameter
should not be appended either.

Signed-off-by: Jiri Bohac <jbohac@suse.cz>
Acked-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dcb: match dcb_app protocol field with 802.1Qaz spec

The dcb_app protocol field is a __u32 however the 802.1Qaz
specification defines it as a 16 bit field. This patch brings
the structure inline with the spec making it a __u16.

CC: Shmulik Ravid <shmulikr@broadcom.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: fix inet_twsk_deschedule()

Eric W. Biederman reported a lockdep splat in inet_twsk_deschedule()

This is caused by inet_twsk_purge(), run from process context,
and commit 575f4cd5a5b6394577 (net: Use rcu lookups in inet_twsk_purge.)
removed the BH disabling that was necessary.

Add the BH disabling but fine grained, right before calling
inet_twsk_deschedule(), instead of whole function.

With help from Linus Torvalds and Eric W. Biederman

Reported-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Daniel Lezcano <daniel.lezcano@free.fr>
CC: Pavel Emelyanov <xemul@openvz.org>
CC: Arnaldo Carvalho de Melo <acme@redhat.com>
CC: stable <stable@kernel.org> (# 2.6.33+)
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' of git://git./linux/kernel/git/kaber/nf-2.6

Merge branch 'rtc-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'rtc-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  RTC: Re-enable UIE timer/polling emulation
  RTC: Revert UIE emulation removal
  RTC: Release mutex in error path of rtc_alarm_irq_enable

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

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (37 commits)
  net: deinit automatic LIST_HEAD
  net: dont leave active on stack LIST_HEAD
  net: provide default_advmss() methods to blackhole dst_ops
  tg3: Restrict phy ioctl access
  drivers/net: Call netif_carrier_off at the end of the probe
  ixgbe: work around for DDP last buffer size
  ixgbe: fix panic due to uninitialised pointer
  e1000e: flush all writebacks before unload
  e1000e: check down flag in tasks
  isdn: hisax: Use l2headersize() instead of dup (and buggy) func.
  arp_notify: unconditionally send gratuitous ARP for NETDEV_NOTIFY_PEERS.
  cxgb4vf: Use defined Mailbox Timeout
  cxgb4vf: Quiesce Virtual Interfaces on shutdown ...
  cxgb4vf: Behave properly when CONFIG_DEBUG_FS isn't defined ...
  cxgb4vf: Check driver parameters in the right place ...
  pch_gbe: Fix the MAC Address load issue.
  iwlwifi: Delete iwl3945_good_plcp_health.
  net/can/softing: make CAN_SOFTING_CS depend on CAN_SOFTING
  netfilter: nf_iterate: fix incorrect RCU usage
  pch_gbe: Fix the issue that the receiving data is not normal.
  ...

Merge branch 'for-linus/bugfixes' of git://xenbits.xen.org/people/ianc/linux-2.6

* 'for-linus/bugfixes' of git://xenbits.xen.org/people/ianc/linux-2.6:
  xen: suspend and resume system devices when running PVHVM

ipv4: Implement __ip_dev_find using new interface address hash.

Much quicker than going through the FIB tables.

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

ipv4: Add hash table of interface addresses.

This will be used to optimize __ip_dev_find() and friends.

With help from Eric Dumazet.

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

Merge branch 'fixes-2.6.38' of git://git./linux/kernel/git/tj/wq

* 'fixes-2.6.38' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
  workqueue: make sure MAYDAY_INITIAL_TIMEOUT is at least 2 jiffies long
  workqueue, freezer: unify spelling of 'freeze' + 'able' to 'freezable'
  workqueue: wake up a worker when a rescuer is leaving a gcwq

net: deinit automatic LIST_HEAD

commit 9b5e383c11b08784 (net: Introduce
unregister_netdevice_many()) left an active LIST_HEAD() in
rollback_registered(), with possible memory corruption.

Even if device is freed without touching its unreg_list (and therefore
touching the previous memory location holding LISTE_HEAD(single), better
close the bug for good, since its really subtle.

(Same fix for default_device_exit_batch() for completeness)

Reported-by: Michal Hocko <mhocko@suse.cz>
Tested-by: Michal Hocko <mhocko@suse.cz>
Reported-by: Eric W. Biderman <ebiderman@xmission.com>
Tested-by: Eric W. Biderman <ebiderman@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Ingo Molnar <mingo@elte.hu>
CC: Octavian Purdila <opurdila@ixiacom.com>
CC: stable <stable@kernel.org> [.33+]
Signed-off-by: David S. Miller <davem@davemloft.net>

net: dont leave active on stack LIST_HEAD

Eric W. Biderman and Michal Hocko reported various memory corruptions
that we suspected to be related to a LIST head located on stack, that
was manipulated after thread left function frame (and eventually exited,
so its stack was freed and reused).

Eric Dumazet suggested the problem was probably coming from commit
443457242beb (net: factorize
sync-rcu call in unregister_netdevice_many)

This patch fixes __dev_close() and dev_close() to properly deinit their
respective LIST_HEAD(single) before exiting.

References: https://lkml.org/lkml/2011/2/16/304
References: https://lkml.org/lkml/2011/2/14/223

Reported-by: Michal Hocko <mhocko@suse.cz>
Tested-by: Michal Hocko <mhocko@suse.cz>
Reported-by: Eric W. Biderman <ebiderman@xmission.com>
Tested-by: Eric W. Biderman <ebiderman@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Ingo Molnar <mingo@elte.hu>
CC: Octavian Purdila <opurdila@ixiacom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: provide default_advmss() methods to blackhole dst_ops

Commit 0dbaee3b37e118a (net: Abstract default ADVMSS behind an
accessor.) introduced a possible crash in tcp_connect_init(), when
dst->default_advmss() is called from dst_metric_advmss()

Reported-by: George Spelvin <linux@horizon.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Expand CONFIG_DEBUG_LIST to several other list operations

When list debugging is enabled, we aim to readably show list corruption
errors, and the basic list_add/list_del operations end up having extra
debugging code in them to do some basic validation of the list entries.

However, "list_del_init()" and "list_move[_tail]()" ended up avoiding
the debug code due to how they were written. This fixes that.

So the _next_ time we have list_move() problems with stale list entries,
we'll hopefully have an easier time finding them..

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Merge branch 'pm-fixes' of git://git./linux/kernel/git/rafael/suspend-2.6

* 'pm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/suspend-2.6:
  PM / Hibernate: Return error code when alloc_image_page() fails

Merge branch 'drm-fixes' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/radeon/kms: add missing frac fb div flag for dce4+
  drm/radeon/kms: do not reject X16 and Y16X16 floating-point formats on r300
  drm/nouveau: fix suspend/resume on GPUs that don't have PM support
  drm/nouveau: flips/flipd need to always set 'evict' for move_accel_cleanup()
  drm/nv40: fix tiling-related setup for a number of chipsets
  drm/nouveau: fix non-EDIDful native mode selection
  drm/nouveau: Fix detection of DDC-based LVDS on DCB15 boards.
  drm/nv04-nv40: Fix NULL dereference when we fail to find an LVDS native mode.
  drm/nv10: Fix crash when allocating a BO larger than half the available VRAM.

Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IB/qib: Prevent double completions after a timeout or RNR error
  IB/qib: Fix double add_timer()
  RDMA/nes: Don't generate async events for unregistered devices

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

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc-2.6:
  sparc64: Fix NMI startup bug which also breaks perf.
  sparc: fix size argument to find_next_zero_bit()
  sparc: use bitmap_set()
  sparc32: unaligned memory access (MNA) trap handler bug

fs/partitions: Validate map_count in Mac partition tables

Validate number of blocks in map and remove redundant variable.

Signed-off-by: Timo Warns <warns@pre-sense.de>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

enic: Always use single transmit and single receive hardware queues per device

We believe that our earlier patch for supporting multiple hardware
receive queues per enic device requires more internal testing. At this
point, we think that it's best to disable the use of multiple receive
queues. The current patch provides an effective means for the same.

Also, we continue to disallow multiple hardware transmit queues per
device. But change the way we enforce this in order to maintain
consistency with the way receive queues are handled.

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: Use const'ify fib_result deep in the route call chains.

The only troublesome bit here is __mkroute_output which wants
to override res->fi and res->type, compute those in local
variables instead.

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

ipv4: Mark fib_combine_itag()'s 'res' arg as const.

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

ipv4: Avoid use of signed integers in fib_trie code.

GCC emits all kinds of crazy zero extensions when we go from signed
int, to unsigned short, etc. etc.

This transformation has to be legal because:

1) In tkey_extract_bits() in mask_pfx(), the values are used to
   perform shifts, on which negative values are undefined by C.

2) In fib_table_lookup() we perform comparisons with unsigned
   values, constants, and additions.  None of which should
   encounter negative values.

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

net: Add initial_ref arg to dst_alloc().

This allows avoiding multiple writes to the initial __refcnt.

The most simplest cases of wanting an initial reference of "1"
in ipv4 and ipv6 have been converted, the rest have been left
along and kept at the existing "0".

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

ipv4: Consolidate ipv4 dst allocation logic.

This also allows us to combine all the dst->flags settings and avoid
read/modify/write sequences to this struct member.

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

ipv4: Move rcu_read_{lock,unlock}() into ip_route_output_slow().

Simplifies tail of __ip_route_output_key().

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

ipv4: Simplify output route creation call sequence.

There's a lot of redundancy and unnecessary stack frames
in the output route creation path.

1) Make __mkroute_output() return error pointers.

2) Eliminate ip_mkroute_output() entirely, made possible by #1.

3) Call __mkroute_output() directly and handling the returning error
   pointers in ip_route_output_slow().

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

RTC: Re-enable UIE timer/polling emulation

This patch re-enables UIE timer/polling emulation for rtc devices
that do not support alarm irqs.

CC: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
CC: Thomas Gleixner <tglx@linutronix.de>
Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Tested-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>

RTC: Revert UIE emulation removal

Uwe pointed out that my alarm based UIE emulation is not sufficient
to replace the older timer/polling based UIE emulation on devices
where there is no alarm irq. This causes rtc devices without alarms
to return -EINVAL to UIE ioctls. The fix is to re-instate the old
timer/polling method for devices without alarm irqs.

This patch reverts the following commits:
042620a018afcfba1d678062b62e46 - Remove UIE emulation
1daeddd5962acad1bea55e524fc0fa - Cleanup removed UIE emulation declaration
b5cc8ca1c9c3a37eaddf709b2fd3e1 - Remove Kconfig symbol for UIE emulation

The emulation mode will still need to be wired-in with a following
patch before it will work.

CC: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
CC: Thomas Gleixner <tglx@linutronix.de>
Reported-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>

RTC: Release mutex in error path of rtc_alarm_irq_enable

On hardware that doesn't support alarm interrupts, rtc_alarm_irq_enable
could return without releasing the ops_lock mutex.

This was introduced in
aa0be0f (RTC: Propagate error handling via rtc_timer_enqueue properly)

This patch corrects the issue by only returning once the mutex is
released.

[john.stultz: Reworded the commit log]

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: John Stultz <john.stultz@linaro.org>

loopback: convert to hw_features

This also enables TSOv6, TSO-ECN, and UFO as loopback clearly can handle them.

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

net: introduce NETIF_F_RXCSUM

Introduce NETIF_F_RXCSUM to replace device-private flags for RX checksum
offload. Integrate it with ndo_fix_features.

ethtool_op_get_rx_csum() is removed altogether as nothing in-tree uses it.

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

net: use ndo_fix_features for ethtool_ops->set_flags

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

net: ethtool: use ndo_fix_features for offload setting

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

net: Introduce new feature setting ops

This introduces a new framework to handle device features setting.
It consists of:
  - new fields in struct net_device:
+ hw_features - features that hw/driver supports toggling
+ wanted_features - features that user wants enabled, when possible
  - new netdev_ops:
+ feat = ndo_fix_features(dev, feat) - API checking constraints for
enabling features or their combinations
+ ndo_set_features(dev) - API updating hardware state to match
changed dev->features
  - new ethtool commands:
+ ETHTOOL_GFEATURES/ETHTOOL_SFEATURES: get/set dev->wanted_features
and trigger device reconfiguration if resulting dev->features
changed
+ ETHTOOL_GSTRINGS(ETH_SS_FEATURES): get feature bits names (meaning)

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

ethtool: factorize get/set_one_feature

This allows to enable GRO even if RX csum is disabled. GRO will not
be used for packets without hardware checksum anyway.

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

ethtool: factorize ethtool_get_strings() and ethtool_get_sset_count()

This is needed for unified offloads patch.

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

ethtool: enable GSO and GRO by default

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

ethtool: move EXPORT_SYMBOL(ethtool_op_set_tx_csum) to correct place

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

enic: Clean up: Remove a not needed #ifdef

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

enic: Bug fix: Reset driver count of registered unicast addresses to zero during device reset

During a device reset, clear the counter for the no. of unicast addresses registered.
Also, rename the routines that update unicast and multicast address lists.

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tg3: Restrict phy ioctl access

If management firmware is present and the device is down, the firmware
will assume control of the phy.  If a phy access were allowed from the
host, it will collide with firmware phy accesses, resulting in
unpredictable behavior.  This patch fixes the problem by disallowing phy
accesses during the problematic condition.

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

drivers/net: Call netif_carrier_off at the end of the probe

Without calling of netif_carrier_off at the end of the probe the operstate
is unknown when the device is initially opened. By default the carrier is
on so when the device is opened and netif_carrier_on is called the link
watch event is not fired and operstate remains zero (unknown).

This patch fixes this behavior in forcedeth and r8169.

Signed-off-by: Ivan Vecera <ivecera@redhat.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branches 'nes' and 'qib' into for-next

IB/qib: Prevent double completions after a timeout or RNR error

There is a double completion associated with error handling for RC QPs.

The sequence is:

 - The do_rc_ack() routine fields an RNR nack and there are 0
   rnr_retries configured on the QP.
 - qib_error_qp() stops the pending timer
 - qib_rc_send_complete() is called from sdma_complete()
 - qib_rc_send_complete() starts the timer because the msb of the psn
   just completed says an ack is needed.
 - a bunch of flushes occur as ipoib posts WQEs to an error'ed QP
 - rc_timeout() calls qib_restart_rc()
 - qib_restart_rc() calls qib_send_complete() with a
   IB_WC_RETRY_EXC_ERR on a wqe that has already been completed in the
   past

The fix avoids starting the timer since another packet will never
arrive.

Signed-off-by: Mike Marciniszyn <mike.marciniszyn@qlogic.com>
Signed-off-by: Roland Dreier <roland@purestorage.com>

sfc: Implement hardware acceleration of RFS

Use the existing filter management functions to insert TCP/IPv4 and
UDP/IPv4 4-tuple filters for Receive Flow Steering.

For each channel, track how many RFS filters are being added during
processing of received packets and scan the corresponding number of
table entries for filters that may be reclaimed.  Do this in batches
to reduce lock overhead.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

bnx2x: Support for managing RX indirection table

Support fetching and retrieving RX indirection table via ethtool.

Signed-off-by: Tom Herbert <therbert@google.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/fec: remove unused driver data

Apart from not being used the first argument isn't even a struct
platform_device *.

Reported-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>

netfilter: ip6t_LOG: fix a flaw in printing the MAC

The flaw was in skipping the second byte in MAC header due to increasing
the pointer AND indexed access starting at '1'.

Signed-off-by: Joerg Marx <joerg.marx@secunet.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: tproxy: do not assign timewait sockets to skb->sk

Assigning a socket in timewait state to skb->sk can trigger
kernel oops, e.g. in nfnetlink_log, which does:

if (skb->sk) {
        read_lock_bh(&skb->sk->sk_callback_lock);
        if (skb->sk->sk_socket && skb->sk->sk_socket->file) ...

in the timewait case, accessing sk->sk_callback_lock and sk->sk_socket
is invalid.

Either all of these spots will need to add a test for sk->sk_state != TCP_TIME_WAIT,
or xt_TPROXY must not assign a timewait socket to skb->sk.

This does the latter.

If a TW socket is found, assign the tproxy nfmark, but skip the skb->sk assignment,
thus mimicking behaviour of a '-m socket .. -j MARK/ACCEPT' re-routing rule.

The 'SYN to TW socket' case is left unchanged -- we try to redirect to the
listener socket.

Cc: Balazs Scheidler <bazsi@balabit.hu>
Cc: KOVACS Krisztian <hidden@balabit.hu>
Signed-off-by: Florian Westphal <fwestphal@astaro.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>

xen: suspend and resume system devices when running PVHVM

Otherwise we fail to properly suspend/resume all of the emulated devices.

Something between 2.6.38-rc2 and rc3 appears to have exposed this
issue, but it's always been wrong not to do this.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Stefano Stabellini <stefano.stabellini@eu.citrix.com>
Acked-by: Jeremy Fitzhardinge <jeremy@goop.org>

ixgbe: work around for DDP last buffer size

A HW limitation was recently discovered where the last buffer in a DDP offload
cannot be a full buffer size in length. Fix the issue with a work around by
adding another buffer with size = 1.

Signed-off-by: Amir Hanania <amir.hanania@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: fix panic due to uninitialised pointer

Systems containing an 82599EB and running a backported driver from
upstream were panicing on boot.  It turns out hw->mac.ops.setup_sfp is
only set for 82599, so one should check to be sure that pointer is set
before continuing in ixgbe_sfp_config_module_task.  I verified by
inspection that the upstream driver has the same issue and also added a
check before the call in ixgbe_sfp_link_config.

Signed-off-by: Andy Gospodarek <andy@greyhouse.net>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: flush all writebacks before unload

The driver was not flushing all writebacks before unloading, possibly
causing memory to be written by the hardware after the driver had
reinitialized the rings.

This adds missing functionality to flush any pending writebacks and is
called in all spots where descriptors should be completed before the driver
begins processing.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: check down flag in tasks

This change is part of a fix to avoid any tasks running while the driver is
exiting and deinitializing resources.

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