netns: let net_generic take pointer-to-const args

This commit is same in nature as v2.6.37-rc1-755-g3654654; the network
namespace itself is not modified when calling net_generic, so the
parameter can be const.

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

packet: use vzalloc()

alloc_one_pg_vec_page() is supposed to return zeroed memory, so use
vzalloc() instead of vmalloc()

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: update version number for ixgbe

This will reflect addition of new X540 hardware

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net-next: Add multiqueue support to vmxnet3 driver

Add multiqueue support to vmxnet3 driver

This change adds multiqueue and thus receive side scaling support
to vmxnet3 device driver. Number of rx queues is limited to 1 in cases
where MSI is not configured or one MSIx vector is not available per rx
queue

Signed-off-by: Shreyas Bhatewara <sbhatewara@vmware.com>
Reviewed-by: Bhavesh Davda <bhavesh@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

Conflicts:
drivers/net/bonding/bond_main.c
net/core/net-sysfs.c
net/ipv6/addrconf.c

X25: remove bkl in routing ioctls

Routing doesn't use the socket data and is protected by x25_route_list_lock

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: remove bkl in inq and outq ioctls

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: remove bkl in timestamp ioctls

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: pushdown bkl in ioctls

Push down the bkl in the ioctls so they can be removed one at a time.

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

filter: use reciprocal divide

At compile time, we can replace the DIV_K instruction (divide by a
constant value) by a reciprocal divide.

At exec time, the expensive divide is replaced by a multiply, a less
expensive operation on most processors.

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

filter: cleanup codes[] init

Starting the translated instruction to 1 instead of 0 allows us to
remove one descrement at check time and makes codes[] array init
cleaner.

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

filter: optimize sk_run_filter

Remove pc variable to avoid arithmetic to compute fentry at each filter
instruction. Jumps directly manipulate fentry pointer.

As the last instruction of filter[] is guaranteed to be a RETURN, and
all jumps are before the last instruction, we dont need to check filter
bounds (number of instructions in filter array) at each iteration, so we
remove it from sk_run_filter() params.

On x86_32 remove f_k var introduced in commit 57fe93b374a6b871
(filter: make sure filters dont read uninitialized memory)

Note : We could use a CONFIG_ARCH_HAS_{FEW|MANY}_REGISTERS in order to
avoid too many ifdefs in this code.

This helps compiler to use cpu registers to hold fentry and A
accumulator.

On x86_32, this saves 401 bytes, and more important, sk_run_filter()
runs much faster because less register pressure (One less conditional
branch per BPF instruction)

# size net/core/filter.o net/core/filter_pre.o
   text    data     bss     dec     hex filename
   2948       0       0    2948     b84 net/core/filter.o
   3349       0       0    3349     d15 net/core/filter_pre.o

on x86_64 :
# size net/core/filter.o net/core/filter_pre.o
   text    data     bss     dec     hex filename
   5173       0       0    5173    1435 net/core/filter.o
   5224       0       0    5224    1468 net/core/filter_pre.o

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

net: fix kernel-doc for sk_filter_rcu_release

Fix kernel-doc warning for sk_filter_rcu_release():

Warning(net/core/filter.c:586): missing initial short description on line:
 *  sk_filter_rcu_release: Release a socket filter by rcu_head

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Fix to avoid firmware update when interface is not open.

Since interrupts are enabled only when open is called on the interface,
Attempting a firmware update operation when interface is down could lead to
partial success or failure of operation. This fix fails the request if
netif_running is false.

Signed-off-by: Sarveshwar Bandi <Sarveshwar.Bandi@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: fix IP_VS dependencies

When NF_CONNTRACK is enabled, IP_VS uses conntrack symbols.
Therefore IP_VS can't be linked statically when conntrack
is built modular.

Reported-by: Justin P. Mattock <justinmattock@gmail.com>
Tested-by: Justin P. Mattock <justinmattock@gmail.com>
Signed-off-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: irda: irttp: sync error paths of data- and udata-requests

irttp_data_request() returns meaningful errorcodes, while irttp_udata_request()
just returns -1 in similar situations. Sync the two and the loglevels of the
accompanying output.

Signed-off-by: Wolfram Sang <w.sang@pengutronix.de>
Cc: Samuel Ortiz <sameo@linux.intel.com>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: Expose reachable and retrans timer values as msecs

Expose reachable and retrans timer values in msecs instead of jiffies.
Both timer values are already exposed as msecs in the neighbour table
netlink interface.

The creation timestamp format with increased precision is kept but
cleaned up.

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

can: EG20T PCH: use BIT(X)

Replace bit assignment value to BIT(X).
For easy to readable/identifiable, replace all bit assigned macros to BIT(X)

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.okisemi.com>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

can: EG20T PCH: add prefix to macro

For easy to readable/identifiable, add prefix "PCH_" to all of #define macros.

Signed-off-by: Tomoya MORINAGA <tomoya-linux@dsn.okisemi.com>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

atm: fore200e: Fix build warning.

GCC (rightfully) complains that:

drivers/atm/fore200e.c:614:5: warning: operation on 'cmdq->head' may be undefined

This is due to the FORE200E_NEXT_ENTRY macro, which essentially
evaluates to:

i = ++i % m

Make it what's explicitly intended here which is:

i = (i + 1) % m

and the warning goes away.

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

ipv6: Expose IFLA_PROTINFO timer values in msecs instead of jiffies

IFLA_PROTINFO exposes timer related per device settings in jiffies.
Change it to expose these values in msecs like the sysctl interface
does.

I did not find any users of IFLA_PROTINFO which rely on any of these
values and even if there are, they are likely already broken because
there is no way for them to reliably convert such a value to another
time format.

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

igmp: refine skb allocations

IGMP allocates MTU sized skbs. This may fail for large MTU (order-2
allocations), so add a fallback to try lower sizes.

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

net: move definitions of BPF_S_* to net/core/filter.c

BPF_S_* are used internally, should not be exposed to the others.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Hagen Paul Pfeifer <hagen@jauu.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

filter: Optimize instruction revalidation code.

Since repeating u16 value to u8 value conversion using switch() clause's
case statement is wasteful, this patch introduces u16 to u8 mapping table
and removes most of case statements. As a result, the size of net/core/filter.o
is reduced by about 29% on x86.

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

3c59x: fix build failure on !CONFIG_PCI

VORTEX_PCI() could return NULL so it needs to be casted before
accessing any member of struct pci_dev. This fixes following
build failure. Likewise VORTEX_EISA() was changed also.

  CC [M]  drivers/net/3c59x.o
drivers/net/3c59x.c: In function 'acpi_set_WOL':
drivers/net/3c59x.c:3211:39: warning: dereferencing 'void *' pointer
drivers/net/3c59x.c:3211:39: error: request for member 'current_state' in something not a structure or union
make[3]: *** [drivers/net/3c59x.o] Error 1
make[2]: *** [drivers/net/3c59x.o] Error 2
make[1]: *** [sub-make] Error 2
make: *** [all] Error 2

Signed-off-by: Namhyung Kim <namhyung@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipg.c: remove id [SUNDANCE, 0x1021]

ipg.c:
  The id [SUNDANCE, 0x1021] (=[0x13f0, 0x1021]) is defined
  at dl2k.h and ipg.c.
  But this device works better with dl2k driver.

  This problem is similar with the commit
  [25cca5352712561fba97bd37c495593d641c1d39
  ipg: Remove device claimed by dl2k from pci id table]
  at 11 Feb 2010.

Signed-off-by: Ken Kawasaki <ken_kawasaki@spring.nifty.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add priority field to pktgen

Add option to set skb priority to pktgen. Useful for testing
QOS features. Also by running pktgen on the vlan device the
qdisc on the real device can be tested.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: caif: spi: fix potential NULL dereference

alloc_netdev() is not checked here for NULL return value.  dev is
check instead.  It might lead to NULL dereference of ndev.

Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ath9k_htc: Avoid setting QoS control for non-QoS frames

Setting tid information in the TX header is required only for QoS
frames. Not handling this case causes severe data loss with some APs.

Cc: stable@kernel.org
Signed-off-by: Rajkumar Manoharan <rmanoharan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

net: zero kobject in rx_queue_release

netif_set_real_num_rx_queues() can decrement and increment
the number of rx queues. For example ixgbe does this as
features and offloads are toggled. Presumably this could
also happen across down/up on most devices if the available
resources changed (cpu offlined).

The kobject needs to be zero'd in this case so that the
state is not preserved across kobject_put()/kobject_init_and_add().

This resolves the following error report.

ixgbe 0000:03:00.0: eth2: NIC Link is Up 10 Gbps, Flow Control: RX/TX
kobject (ffff880324b83210): tried to init an initialized object, something is seriously wrong.
Pid: 1972, comm: lldpad Not tainted 2.6.37-rc18021qaz+ #169
Call Trace:
 [<ffffffff8121c940>] kobject_init+0x3a/0x83
 [<ffffffff8121cf77>] kobject_init_and_add+0x23/0x57
 [<ffffffff8107b800>] ? mark_lock+0x21/0x267
 [<ffffffff813c6d11>] net_rx_queue_update_kobjects+0x63/0xc6
 [<ffffffff813b5e0e>] netif_set_real_num_rx_queues+0x5f/0x78
 [<ffffffffa0261d49>] ixgbe_set_num_queues+0x1c6/0x1ca [ixgbe]
 [<ffffffffa0262509>] ixgbe_init_interrupt_scheme+0x1e/0x79c [ixgbe]
 [<ffffffffa0274596>] ixgbe_dcbnl_set_state+0x167/0x189 [ixgbe]

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Fix duplicate volatile warning.

jiffies is defined as "volatile".

  extern unsigned long volatile __jiffy_data jiffies;

ACCESS_ONCE() uses "volatile".
As a result, some compilers warn duplicate `volatile' for ACCESS_ONCE(jiffies).

Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: David S. Miller <davem@davemloft.net>

dccp ccid-2: whitespace fix-up

This fixes whitespace noise introduced in commit "dccp ccid-2: Algorithm to
update buffer state", 5753fdfe8bd8e9a2ff9e5af19b0ffc78bfcd502a, 14 Nov 2010.

Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

bonding: IGMP handling cleanup

Instead of iterating in_dev->mc_list from bonding driver, its better
to call a helper function provided by igmp.c
Details of implementation (locking) are private to igmp code.

ip_mc_rejoin_group(struct ip_mc_list *im) becomes
ip_mc_rejoin_groups(struct in_device *in_dev);

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

MAINTAINERS: Add stmmac maintainer

Add STMMAC to the list of supported Ethernet drivers
and myself as maintainer.

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

bonding: fix a race in IGMP handling

RCU conversion in IGMP code done in net-next-2.6 raised a race in
__bond_resend_igmp_join_requests().

It iterates in_dev->mc_list without appropriate protection (RTNL, or
read_lock on in_dev->mc_list_lock).

Another cpu might delete an entry while we use it and trigger a fault.

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

cfg80211: fix can_beacon_sec_chan, reenable HT40

This follows wireless-testing 9236d838c920e90708570d9bbd7bb82d30a38130
("cfg80211: fix extension channel checks to initiate communication") and
fixes accidental case fall-through. Without this fix, HT40 is entirely
blocked.

Signed-off-by: Mark Mentovai <mark@moxienet.com>
Cc: stable@kernel.org
Acked-by: Luis R. Rodriguez <lrodriguez@atheros.com
Signed-off-by: John W. Linville <linville@tuxdriver.com>

gianfar: fix signedness issue

irq_of_parse_and_map() has an unsigned return type.
Testing for a negative error value doesn't work here.

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

net: ipv4: tcp_probe: cleanup snprintf() use

snprintf() returns number of bytes that were copied if there is no overflow.
This code uses return value as number of copied bytes.  Theoretically format
string '%lu.%09lu %pI4:%u %pI4:%u %d %#x %#x %u %u %u %u\n' may be expanded
up to 163 bytes.  In reality tv.tv_sec is just few bytes instead of 20, 2 ports
are just 5 bytes each instead of 10, length is 5 bytes instead of 10.  The rest
is an unstrusted input.  Theoretically if tv_sec is big then copy_to_user() would
overflow tbuf.

tbuf was increased to fit in 163 bytes.  snprintf() is used to follow return
value semantic.

Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: zero kobject in rx_queue_release

netif_set_real_num_rx_queues() can decrement and increment
the number of rx queues. For example ixgbe does this as
features and offloads are toggled. Presumably this could
also happen across down/up on most devices if the available
resources changed (cpu offlined).

The kobject needs to be zero'd in this case so that the
state is not preserved across kobject_put()/kobject_init_and_add().

This resolves the following error report.

ixgbe 0000:03:00.0: eth2: NIC Link is Up 10 Gbps, Flow Control: RX/TX
kobject (ffff880324b83210): tried to init an initialized object, something is seriously wrong.
Pid: 1972, comm: lldpad Not tainted 2.6.37-rc18021qaz+ #169
Call Trace:
 [<ffffffff8121c940>] kobject_init+0x3a/0x83
 [<ffffffff8121cf77>] kobject_init_and_add+0x23/0x57
 [<ffffffff8107b800>] ? mark_lock+0x21/0x267
 [<ffffffff813c6d11>] net_rx_queue_update_kobjects+0x63/0xc6
 [<ffffffff813b5e0e>] netif_set_real_num_rx_queues+0x5f/0x78
 [<ffffffffa0261d49>] ixgbe_set_num_queues+0x1c6/0x1ca [ixgbe]
 [<ffffffffa0262509>] ixgbe_init_interrupt_scheme+0x1e/0x79c [ixgbe]
 [<ffffffffa0274596>] ixgbe_dcbnl_set_state+0x167/0x189 [ixgbe]

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

net: use the macros defined for the members of flowi

Use the macros defined for the members of flowi to clean the code up.

Signed-off-by: Changli Gao <xiaosuo@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: more Kconfig whitespace cleanup

indentation for TSI108_ETH entry was too big.

Signed-off-by: Philippe De Muyter <phdm@macqel.be>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/can/sja1000: Use printf extension %pR for struct resource

Using %pR standardizes the struct resource output.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/vxge/vxge-main.c: Remove unnecessary casts of pci_get_drvdata

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Jon Mason <jon.mason@exar.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/s2io.c: Remove unnecessary casts of pci_get_drvdata

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: bnx2x: fix error value sign

bnx2x_init_one() should return negative value on error.
By mistake it returns ENODEV instead of -ENODEV.

Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Acked-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

8139cp: fix checksum broken

I am not family with RealTek RTL-8139C+ series 10/100 PCI Ethernet driver.
I try to guess the meaning of RxProtoIP and IPFail.
RxProtoIP stands for received IPv4 packet that upper protocol is not tcp and udp.
!(status & IPFail) is true means that driver correctly to check checksum in IPv4 header.

If these are right, driver will set ip_summed with CHECKSUM_UNNECESSARY for other
upper protocol, e.g. sctp, igmp protocol. This will cause protocol stack ignores
checksum check for packets with invalid checksum.

This patch is only compile-test.

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

r8169: fix checksum broken

If r8196 received packets with invalid sctp/igmp(not tcp, udp) checksum, r8196 set skb->ip_summed
wit CHECKSUM_UNNECESSARY. This cause that upper protocol don't check checksum field.

I am not family with r8196 driver. I try to guess the meaning of RxProtoIP and IPFail.
RxProtoIP stands for received IPv4 packet that upper protocol is not tcp and udp.
!(opts1 & IPFail) is true means that driver correctly to check checksum in IPv4 header.

If it's right, I think we should not set ip_summed wit CHECKSUM_UNNECESSARY for my sctp packets
with invalid checksum.

If it's not right, please tell me.

Signed-off-by: Shan Wei <shanwei@cn.fujitsu.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rds: Integer overflow in RDS cmsg handling

In rds_cmsg_rdma_args(), the user-provided args->nr_local value is
restricted to less than UINT_MAX.  This seems to need a tighter upper
bound, since the calculation of total iov_size can overflow, resulting
in a small sock_kmalloc() allocation.  This would probably just result
in walking off the heap and crashing when calling rds_rdma_pages() with
a high count value.  If it somehow doesn't crash here, then memory
corruption could occur soon after.

Signed-off-by: Dan Rosenberg <drosenberg@vsecurity.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: Bumped up driver version to 5.0.12

Signed-off-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: lro off message log from set rx checsum

Log LRO off message while disabling rx checksum only when
LRO is already enabled.

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: Add description for CN1000Q adapter

Add description for CN1000Q adapter

Signed-off-by: Sritej Velaga <sritej.velaga@qlogic.com>
Signed-off-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: Fix for kdump

In case of kdump environment the function may be in unknown state.
Reset the function using PCI FLR before initializing it.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: Allow minimum bandwidth of zero

Allow minimum bandwidth to be set zero

Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Anirban Chakraborty <anirban.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: AF_INET6 link address family

IPv6 already exposes some address family data via netlink in the
IFLA_PROTINFO attribute if RTM_GETLINK request is sent with the
address family set to AF_INET6. We take over this format and
reuse all the code.

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Cc: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: AF_INET link address family

Implements the AF_INET link address family exposing the per
device configuration settings via netlink using the attribute
IFLA_INET_CONF.

The format of IFLA_INET_CONF differs depending on the direction
the attribute is sent. The attribute sent by the kernel consists
of a u32 array, basically a 1:1 copy of in_device->cnf.data[].
The attribute expected by the kernel must consist of a sequence
of nested u32 attributes, each representing a change request,
e.g.
[IFLA_INET_CONF] = {
[IPV4_DEVCONF_FORWARDING] = 1,
[IPV4_DEVCONF_NOXFRM] = 0,
}

libnl userspace API documentation and example available from:
http://www.infradead.org/~tgr/libnl/doc-git/group__link__inet.html

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

inet: Define IPV4_DEVCONF_MAX

Define IPV4_DEVCONF_MAX to get rid of MAX - 1 notation.

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

rtnetlink: Link address family API

Each net_device contains address family specific data such as
per device settings and statistics. We already expose this data
via procfs/sysfs and partially netlink.

The netlink method requires the requester to send one RTM_GETLINK
request for each address family it wishes to receive data of
and then merge this data itself.

This patch implements a new API which combines all address family
specific link data in a new netlink attribute IFLA_AF_SPEC.
IFLA_AF_SPEC contains a sequence of nested attributes, one for each
address family which in turn defines the structure of its own
attribute. Example:

   [IFLA_AF_SPEC] = {
       [AF_INET] = {
           [IFLA_INET_CONF] = ...,
       },
       [AF_INET6] = {
           [IFLA_INET6_FLAGS] = ...,
           [IFLA_INET6_CONF] = ...,
       }
   }

The API also allows for address families to implement a function
which parses the IFLA_AF_SPEC attribute sent by userspace to
implement address family specific link options.

Signed-off-by: Thomas Graf <tgraf@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

clarify documentation for net.ipv4.igmp_max_memberships

This patch helps clarify documentation for
net.ipv4.igmp_max_memberships by providing a formula for
calculating the maximum number of multicast groups that can be
subscribed to, plus defining the theoretical limit.

Signed-off-by: Jiri Pirko <jpirko@redhat.com>
Signed-off-by: Jeremy Eder <jeder@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/isdn/hisax: Add printf format/argument verification and fix fallout

Add __attribute__((format... to several functins
Make formats and arguments match.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

SELinux: return -ECONNREFUSED from ip_postroute to signal fatal error

The SELinux netfilter hooks just return NF_DROP if they drop a packet.  We
want to signal that a drop in this hook is a permanant fatal error and is not
transient.  If we do this the error will be passed back up the stack in some
places and applications will get a faster interaction that something went
wrong.

Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

network: tcp_connect should return certain errors up the stack

The current tcp_connect code completely ignores errors from sending an skb.
This makes sense in many situations (like -ENOBUFFS) but I want to be able to
immediately fail connections if they are denied by the SELinux netfilter hook.
Netfilter does not normally return ECONNREFUSED when it drops a packet so we
respect that error code as a final and fatal error that can not be recovered.

Based-on-patch-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: allow hooks to pass error code back up the stack

SELinux would like to pass certain fatal errors back up the stack.  This patch
implements the generic netfilter support for this functionality.

Based-on-patch-by: Patrick McHardy <kaber@trash.net>
Signed-off-by: Eric Paris <eparis@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net/atm: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/vxge: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Acked-by: Jon Mason <jon.mason@exar.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/usb: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/qlge: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/qla3xxx.c: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/pcmcia: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/bonding: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/isdn/i4l: Remove unnecessary casts of netdev_priv

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

e1000e: add netpoll support for MSI/MSI-X IRQ modes

With enabling CONFIG_PCI_MSI, e1000e could work in MSI/MSI-X IRQ mode,
and netpoll controller didn't deal with those IRQ modes on e1000e.

This patch add the handling MSI/MSI-X IRQ modes to netpoll controller,
so that netconsole could work with those IRQ modes.

Signed-off-by: Dongdong Deng <dongdong.deng@windriver.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: 82574 intermittently fails to initialize with manageability f/w

The driver can fail initializing the hardware when manageability firmware
is performing concurrent MDIO operations because the hardware semaphore
scheme to prevent concurrent operations between software and firmware is
incorrect for 82574/82583.  Instead of using the SWSM register, the driver
should be using the EXTCNF_CTRL register.  A software mutex is also added
to prevent simultaneous software threads from performing similar concurrent
accesses.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

e1000e: 82571 SerDes link handle null code word from partner

SerDes Link detection on certain 82571 mezzanine cards can fail when the
link is forced, the link partner does not recognize forced link and the
link partner sends null code words.  Detect the null code words and return
to auto-negotiation state which causes the link partner to begin responding
with valid code words.  Within a reasonable interval the link will finally
settle as forced by both partners.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Remove extra struct page member from the buffer info structure

declaration.

Reported-by: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igbvf: Remove some dead code in igbvf

Removed unused variable in igbvf.

Signed-off-by: Julian Stecklina <jsteckli@os.inf.tu-dresden.de>
Acked-by: Greg Rose <greg.v.rose@intel.com>
Tested-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

igbvf: Update version and Copyright

Update version string and copyright notice

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbevf: Fix Oops

The driver is calling netif_carrier_off and netif_tx_stop_all_queues
before the netdevice is registered which causes an Oops.  Move call
to netif_carrier_off after the netdevice is registered and remove
call to netif_tx_stop_all_queues because there aren't any TX
queues yet.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Emil Tantilov <emil.s.tantilov@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: refactor ixgbe_alloc_queues()

I noticed ring variable was initialized before allocations, and that
memory node management was a bit ugly. We also leak memory in case of
ring allocations error.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: add support for x540 MAC

This patch adds support for the x540 MAC which is the next MAC in the
82598/82599 line.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: add MAC and PHY support for x540

Adds the new x540.c file and Aquantia 1202 PHY for X540 support.

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: make silicon specific functions generic

The new MAC type X540 shares much of the same functionality of
some silicon specific functions.  To reduce duplicate code,
made these functions generic.

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: make sure FCoE DDP user buffers are really released by the HW

When invalidating the DDP context is invalidated, the HW may not be done
with the user buffer right away. In which case, we poll the FCBUFF register
to check if the buffer valid bit is cleared or not, if not, we wait for max
100us that is guaranteed by the HW.

Signed-off-by: Yi Zou <yi.zou@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: invalidate FCoE DDP context when no error status is available

The hw automatically invalidates the context if DDP is successful or there is
error detected. In case there is no error status available from the hw,
initializing the per context error status to be 1 allows the DDP context to be
still invalidated via the upper layer call to ddp_put().

Signed-off-by: Yi Zou <yi.zou@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: avoid doing FCoE DDP when adapter is DOWN or RESETTING

There is no point to allow incoming DDP requests from the upper layer stack if
the adapter is going down or being reset.

Signed-off-by: Yi Zou <yi.zou@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: rework Tx hang detection to fix reoccurring false Tx hangs

The Tx hang logic has been known to detect false hangs when
the device is receiving pause frames or has delayed processing
for some other reason.

This patch makes the logic more robust and resolves these
known issues. The old logic checked to see if the device
was paused by querying the HW then the hang logic was
aborted if the device was currently paused. This check was
racy because the device could have been in the pause state
any time up to this check. The other operation of the
hang logic is to verify the Tx ring is still advancing
the old logic checked the EOP timestamp. This is not
sufficient to determine the ring is not advancing but
only infers that it may be moving slowly.

Here we add logic to track the number of completed Tx
descriptors and use the adapter stats to check if any
pause frames have been received since the previous Tx
hang check. This way we avoid racing with the HW
register and do not detect false hangs if the ring is
advancing slowly.

This patch is primarily the work of Jesse Brandeburg. I
clean it up some and fixed the PFC checking.

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: Resolve null function pointer accesses on 82598 w/ multi-speed fiber

This change resolves some null function pointer accesses on 82598 when a
multi-speed fiber module is inserted into the adapter.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: populate the ring->q_vector pointer during ring mapping

The q_vector back pointer was not being set in the rings so it would not
have been possible to determine the parent q_vector of the ring.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup ixgbe_map_rings_to_vectors

This change cleans up some of the items in ixgbe_map_rings_to_vectors.
Specifically it merges the two for loops and drops the unnecessary vectors
parameter.

It also moves the vector names into the q_vectors themselves.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: simplify math and improve stack use of ixgbe_set_itr functions

This change is meant to improve the stack utilization and simplify the math
used in ixgbe_set_itr_msix.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup unclear references to reg_idx

There are a number of places where we use the variable j to contain the
register index of the ring.  Instead of using such a non-descriptive
variable name it is better that we name it reg_idx so that it is clear what
the variable contains.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup unnecessary return value in ixgbe_cache_ring_rss

This change is just to cleanup some confusing logic in ixgbe_cache_ring_rss
which can be simplified by adding a conditional with return to the start of
the call.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: Cleanup DCB logic, whitespace, and comments in ixgbe_ethtool.c

This change address a few whitespace issues in DCB #ifdefs, adds a comment
calling out the DCB specific registers, and nests an if statement inline
with a number of if statements related to flow control.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: add WOL support for backplane adapters

This change adds support for certain 82599 based Mezzanine adapters.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup ixgbe_set_tx_csum ethtool flags configuration

This change makes it so that we always disable SCTP regardless of mac type
since we shouldn't need to check mac type before disabling a feature that
isn't supported on a given piece of hardware.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: change mac_type if statements to switch statements

This change replaces a number of if/elseif/else statements with switch
statements to support the addition of future devices to the ixgbe driver.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup use of ixgbe_rsc_count and RSC_CB

This change cleans up the use of rsc_count and changes it to a boolean since
the actual numerical value is used nowhere in the Rx cleanup path.  I am
also moving the skb count into the RSC_CB path since it is much easier to
track it there than when it is passed as a parameter to various function
calls.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

ixgbe: cleanup ATR filter setup function

This change cleans up the ixgbe_atr filter setup function so that it uses
fewer items from the stack.  Since the code is only applicable to IPv4 w/
TCP it makes sense to just use the pointers based on the headers themselves
instead of copying them to temp variables and then writing those to the
filters.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>