net: skbuff.c cleanup

(skb->data - skb->head) can be changed by skb_headroom(skb)

Remove some uses of NET_SKBUFF_DATA_USES_OFFSET, using
(skb_end_pointer(skb) - skb->head) or
(skb_tail_pointer(skb) - skb->head) : compiler does the right thing,
and this is more readable for us ;)

(struct skb_shared_info *) casts in pskb_expand_head() to help memcpy()
to use aligned moves.

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

skge: add GRO support

- napi_gro_flush() is exported from net/core/dev.c, to avoid
  an irq_save/irq_restore in the packet receive path.
- use napi_gro_receive() instead of netif_receive_skb()
- use napi_gro_flush() before calling __napi_complete()
- turn on NETIF_F_GRO by default
- Tested on a Marvell 88E8001 Gigabit NIC

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

drivers/net/can/mscan/mpc5xxx_can.c: Add of_node_put to avoid memory leak

Add a call to of_node_put in the error handling code following a call to
of_find_matching_node.

This patch also moves the existing call to of_node_put after the call to
iounmap in the error handling code, to make it possible to jump to
of_node_put without doing iounmap.  These appear to be disjoint operations,
so the ordering doesn't matter.

This patch furthermore changes the -ENODEV result in the error handling
code for of_find_matching_node to a return of 0, as found in the error
handling code for of_iomap, because the return type of the function is
unsigned.

The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
local idexpression x;
expression E,E1,E2;
statement S;
@@

*x =
(of_find_node_by_path
|of_find_node_by_name
|of_find_node_by_phandle
|of_get_parent
|of_get_next_parent
|of_get_next_child
|of_find_compatible_node
|of_match_node
|of_find_node_by_type
|of_find_node_with_property
|of_find_matching_node
|of_parse_phandle
)(...);
...
if (x == NULL) S
<... when != x = E
*if (...) {
  ... when != of_node_put(x)
      when != if (...) { ... of_node_put(x); ... }
(
  return <+...x...+>;
|
*  return ...;
)
}
...>
(
E2 = x;
|
of_node_put(x);
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Reviewed-by: Wolfram Sang <w.sang@pengutronix.de>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ns83820: Use helper to disable chip interrupts

Use helper routine to disable chip interrupts.

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: tunnels should use rcu_dereference

tunnel4_handlers, tunnel64_handlers, tunnel6_handlers and
tunnel46_handlers are protected by RCU, but we dont use appropriate rcu
primitives to scan them. rcu_lock() is already held by caller.

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

bnx2x: Update version to 1.52.53-5

Update version to 1.52.53-5

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Add BCM84823 to the supported PHYs

Add BCM84823 to the supported PHYs

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Change BCM848xx LED configuration

Change 848xx LED configuration according to the new microcode (Boards
were shipped with only with the new microcode)

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Remove unneeded setting of XAUI low power to BCM8727

Remove unneeded setting of XAUI low power to BCM8727. This was required
only in older microcode which is not in the field.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Change BCM848xx configuration according to IEEE

Change BCM848xx behavior to fit IEEE such that setting 10Mb/100Mb will
use force speed, and setting 1Gb/10Gb will use auto-negotiation with the
specific speed advertised

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Reset link before any new link settings

Reset link before any new link settings to avoid potential link issue
caused by previous link settings

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Fix potential link issue In BCM8727 based boards

In BCM8727 based boards, setting default 10G link speed after link was
set to 1G may lead to link down issue. The problem was setting the right
value, but to the wrong registers

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: Fix potential link issue of BCM8073/BCM8727

Fix potential link issue caused by insufficient delay time during SPIROM
load of BCM8073/BCM8727

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: add cksum flag

o New CKSUM flag added by fw to notify cksum is verified.
o Update version to 5.0.9

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

qlcnic: mac vlan learning support

Hypervisor allows, two VM's interfaces to have same mac address.
These VM's interfaces get differentiate with Vlan tag.
This patch add support to learn and configure mac+vlan filter on device.

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

qlcnic: support mac learning

Device eswitch need to configure with VM's mac address.
Hypervisor doesn't provide any utility/callbacks to get VM's mac address.
Unicast mac address filter improves performance and also provide
packet loopback capability i.e communication between VM.

Above features is by default off, can be turned on with module parameter
'mac_learn'.

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

qlcnic: fix mac override capability

o Rename mac_learning to mac_override
o Added check in set_mac to return error if mac override is disabled.
o Disabling mac_override only supported for Non priviledged functions.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: fix panic while using eth_hdr

o skb->mac_header is not set, so machine panics while using function eth_hdr.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: fix mac anti spoof policy

o Allow enabling/disabling mac anti spoof policy only for
  Non privilege functions.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: fix for setting default eswitch config

o Default eswitch config was set, even before eswitch capabilty get detected.
As a result setting default config was getting fail.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: fix mac addr read

o Mac addr was read from flash for every fw reset
for Non-priviledge function.It should be read only once.
o Remove unnecessary get_mac_addr callback

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: add api version in reg dump

o ethtool reg version bumped to 2

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qlcnic: backout firmware initialization update

o Backing out "firmware initialization update" changes, commit-id
  "d4066833bb1b35fefb1dd45eb2b10659d46bf151".
o fix heartbit spelling, it should be heartbeat

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: Fix another build warning.

drivers/net/stmmac/dwmac100_core.c: In function 'dwmac100_dump_mac_regs':
drivers/net/stmmac/dwmac100_core.c:47: warning: cast from pointer to integer of different size

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

gro: unexport tcp4_gro_receive and tcp4_gro_complete

tcp4_gro_receive() and tcp4_gro_complete() dont need to be exported.

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

net: add a comment on netdev->last_rx

As some driver authors seem to reintroduce dev->last_rx use,
add a comment to strongly discourage this.

Since commit 6cf3f41e6c0 (bonding, net: Move last_rx update into bonding
recv logic), network drivers dont need to update last_rx themselves,
unless they use this field to implement a timeout.

Not updating last_rx helps not dirtying a cache line, improving
performance in SMP.

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

pktgen: remove non used variable

remove non used variable "queue" in pg_cleanup

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

vlan: Use vlan_dev_real_dev in vlan_hwaccel_do_receive

[patch net-next-2.6] vlan: Use vlan_dev_real_dev in vlan_hwaccel_do_receive

Use helper as in other places.

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

Phonet: do not set POLLOUT in case of send buffer overflow

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

Phonet: correct sendmsg() error code from sock_alloc_send_skb()

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

Phonet: restore flow control credits when sending fails

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

stmmac: Fix build warnings.

In file included from drivers/net/stmmac/stmmac_ethtool.c:30:
drivers/net/stmmac/stmmac.h:111: warning: 'struct platform_device' declared inside parameter list
drivers/net/stmmac/stmmac.h:111: warning: its scope is only this definition or declaration, which is probably not what you want
drivers/net/stmmac/stmmac_main.c: In function 'stmmac_dvr_probe':
drivers/net/stmmac/stmmac_main.c:1744: warning: cast from pointer to integer of different size
In file included from drivers/net/stmmac/stmmac_mdio.c:31:
drivers/net/stmmac/stmmac.h:111: warning: 'struct platform_device' declared inside parameter list
drivers/net/stmmac/stmmac.h:111: warning: its scope is only this definition or declaration, which is probably not what you want
drivers/net/stmmac/dwmac1000_core.c: In function 'dwmac1000_dump_regs':
drivers/net/stmmac/dwmac1000_core.c:56: warning: cast from pointer to integer of different size

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

ipv6: struct xfrm6_tunnel in read_mostly section

tunnel6_handlers chain being scanned for each incoming packet,
make sure it doesnt share an often dirtied cache line.

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

net: struct xfrm_tunnel in read_mostly section

tunnel4_handlers chain being scanned for each incoming packet,
make sure it doesnt share an often dirtied cache line.

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

net/m68k: Hydra Ethernet - print whole resource instead of start address

resource_size_t changed from `unsigned long' to `phys_addr_t`, which is either
`u32' or `u64'.
Print the whole resource to remove the cast and to make it future-proof.

Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

TCP: update initial windows according to RFC 5681

This updates the use of larger initial windows, as originally specified in
RFC 3390, to use the newer IW values specified in RFC 5681, section 3.1.

The changes made in RFC 5681 are:
 a) the setting now is more clearly specified in units of segments (as the
    comments  by John Heffner emphasized, this was not very clear in RFC 3390);
 b) for connections with 1095 < SMSS <= 2190 there is now a change:
    - RFC 3390 says that IW <= 4380,
    - RFC 5681 says that IW = 3 * SMSS <= 6570.

Since RFC 3390 is older and "only" proposed standard, whereas the newer RFC 5681
is already draft standard, it seems preferable to use the newer IW variant.

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

dccp ccid-3: use per-route RTO or TCP RTO as fallback

This makes RTAX_RTO_MIN also available to CCID-3, replacing the compile-time
RTO lower bound with a per-route tunable value.

The original Kconfig option solved the problem that a very low RTT (in the
order of HZ) can trigger too frequent and unnecessary reductions of the
sending rate.

This tunable does not affect the initial RTO value of 2 seconds specified in
RFC 5348, section 4.2 and Appendix B. But like the hardcoded Kconfig value,
it allows to adapt to network conditions.

The same effect as the original Kconfig option of 100ms is now achieved by

> ip route replace to unicast 192.168.0.0/24 rto_min 100j dev eth0

(assuming HZ=1000).

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

dccp ccid-2: Share TCP's minimum RTO code

Using a fixed RTO_MIN of 0.2 seconds was found to cause problems for CCID-2
over 802.11g: at least once per session there was a spurious timeout. It
helped to then increase the the value of RTO_MIN over this link.

Since the problem is the same as in TCP, this patch makes the solution from
commit "05bb1fad1cde025a864a90cfeb98dcbefe78a44a"
       "[TCP]: Allow minimum RTO to be configurable via routing metrics."
available to DCCP.

This avoids reinventing the wheel, so that e.g. the following works in the
expected way now also for CCID-2:

> ip route change 10.0.0.2 rto_min 800 dev ath0

Luckily this useful rto_min function was recently moved to net/tcp.h,
which simplifies sharing code originating from TCP.

Documentation also updated (plus minor whitespace fixes).

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

tcp/dccp: Consolidate common code for RFC 3390 conversion

This patch consolidates initial-window code common to TCP and CCID-2:
 * TCP uses RFC 3390 in a packet-oriented manner (tcp_input.c) and
 * CCID-2 uses RFC 3390 in packet-oriented manner (RFC 4341).

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

dccp ccid-2: Remove wrappers around sk_{reset,stop}_timer()

This removes the wrappers around the sk timer functions, since not much is
gained from using them: the BUG_ON in start_rto_timer will never trigger
since that function is called only if:

 * the RTO timer expires (rto_expire, and then timer_pending() is false);
 * in tx_packet_sent only if !timer_pending() (BUG_ON is redundant here);
 * previously in new_ack, after stopping the timer (timer_pending() false).

Removing the wrappers also clears the way for eventually replacing the
RTO timer with the icsk-retransmission-timer, as it is already part of the
DCCP socket.

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

dccp ccid-2: Use u32 timestamps uniformly

Since CCID-2 is de facto a mini implementation of TCP, it makes sense to share
as much code as possible.

Hence this patch aligns CCID-2 timestamping with TCP timestamping.
This also halves the space consumption (on 64-bit systems).

The necessary include file <net/tcp.h> is already included by way of
net/dccp.h. Redundant includes have been removed.

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

tcp: Add TCP_USER_TIMEOUT socket option.

This patch provides a "user timeout" support as described in RFC793. The
socket option is also needed for the the local half of RFC5482 "TCP User
Timeout Option".

TCP_USER_TIMEOUT is a TCP level socket option that takes an unsigned int,
when > 0, to specify the maximum amount of time in ms that transmitted
data may remain unacknowledged before TCP will forcefully close the
corresponding connection and return ETIMEDOUT to the application. If
0 is given, TCP will continue to use the system default.

Increasing the user timeouts allows a TCP connection to survive extended
periods without end-to-end connectivity. Decreasing the user timeouts
allows applications to "fail fast" if so desired. Otherwise it may take
upto 20 minutes with the current system defaults in a normal WAN
environment.

The socket option can be made during any state of a TCP connection, but
is only effective during the synchronized states of a connection
(ESTABLISHED, FIN-WAIT-1, FIN-WAIT-2, CLOSE-WAIT, CLOSING, or LAST-ACK).
Moreover, when used with the TCP keepalive (SO_KEEPALIVE) option,
TCP_USER_TIMEOUT will overtake keepalive to determine when to close a
connection due to keepalive failure.

The option does not change in anyway when TCP retransmits a packet, nor
when a keepalive probe will be sent.

This option, like many others, will be inherited by an acceptor from its
listener.

Signed-off-by: H.K. Jerry Chu <hkchu@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix datapath typo

Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

IPVS: include net/ip6_checksum.h for csum_ipv6_magic

Fixes this build error:

net/netfilter/ipvs/ip_vs_core.c: In function 'ip_vs_nat_icmp_v6':
net/netfilter/ipvs/ip_vs_core.c:640: error: implicit declaration of function 'csum_ipv6_magic'

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

drivers/net/atl1c: Remove double test

The nic_type field is compared to athr_l2c twice.

The sematic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)

// <smpl>
@expression@
expression E;
@@

(
* E
  || ... || E
|
* E
  && ... && E
)
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>

starfire: use BUILD_BUG_ON for netdrv_addr_t

Detect size mismatch for netdrv_addr_t at build time rather than
checking at module load time.

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Ion Badulescu <ionut@badula.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

starfire: use PCI_VDEVICE

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Ion Badulescu <ionut@badula.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

phonet: use for_each_set_bit

Replace open-coded loop with for_each_set_bit().

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4: use bitmap library

Replace loops calling set_bit() and clear_bit() with bitmap_set() and
bitmap_clear().

Unlike loops calling set_bit() and clear_bit(), bitmap_set() and
bitmap_clear() are not atomic. But this is ok.
Because the bitmap operations are protected by bitmap->lock
except for initialization of the bitmap in mlx4_bitmap_init().

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Roland Dreier <rolandd@cisco.com>
Cc: netdev@vger.kernel.org
Cc: "David S. Miller" <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

econet: kill unnecessary spin_lock_init()

The spinlock aun_queue_lock is initialized statically. It is unnecessary
to initialize by spin_lock_init() at module load time.

This is detected by the semantic patch.

// <smpl>
@def@
declarer name DEFINE_SPINLOCK;
identifier spinlock;
@@

DEFINE_SPINLOCK(spinlock);

@@
identifier def.spinlock;
@@

- spin_lock_init(&spinlock);
// </smpl>

Signed-off-by: Akinobu Mita <akinobu.mita@gmail.com>
Cc: Julia Lawall <julia@diku.dk>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>

slip: Use net_device_stats from struct net_device

Use net_device->stats for stats instead of private variable copies in
struct slip. Use ndo_get_stat64 so the additions can be performed on a private
destination buffer.

Cc: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: fix wrong return from bnx2x_trylock_hw_lock

bnx2x_trylock_hw_lock() returns a bool :

true if succeeded to acquire the lock.
false in case of error.

-EINVAL is not an acceptable value, since its promoted to true.

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

gro: __napi_gro_receive() optimizations

compare_ether_header() can have a special implementation on 64 bit
arches if CONFIG_HAVE_EFFICIENT_UNALIGNED_ACCESS is defined.

__napi_gro_receive() and vlan_gro_common() can avoid a conditional
branch to perform device match.

On x86_64, __napi_gro_receive() has now 38 instructions instead of 53

As gcc-4.4.3 still choose to not inline it, add inline keyword to this
performance critical function.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: Make time functionality depend upon RTC_HCTOSYS_DEVICE

Based upon a report by Stephen Rothwell.

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

qlcnic: PCI ID addition

Added PCI ID to board info.

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 driver load issue in FW hang

If there is a FW hang when the driver loads, it can not determine the FW operational
mode. Fix it by checking the FW state first before issuing any FW commands to
determine its capabilities and thereby detecting driver operational mode.

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

qlcnic: Fix missing error codes

In the original code, the error codes returned from some of the functions
are not caught and sent up the caller chain. Fixed it here.

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>

net: use scnprintf() to avoid potential buffer overflow

strlcpy() returns the total length of the string they tried to create, so
we should not use its return value without any check. scnprintf() returns
the number of characters written into @buf not including the trailing '\0',
so use it instead here.

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

net/sctp: Use pr_fmt and pr_<level>

Change SCTP_DEBUG_PRINTK and SCTP_DEBUG_PRINTK_IPADDR to
use do { print } while (0) guards.
Add SCTP_DEBUG_PRINTK_CONT to fix errors in log when
lines were continued.
Add #define pr_fmt(fmt) KBUILD_MODNAME ": " fmt
Add a missing newline in "Failed bind hash alloc"

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

ipvs: switch to GFP_KERNEL allocations

Switch from GFP_ATOMIC allocations to GFP_KERNEL ones in
ip_vs_add_service() and ip_vs_new_dest(), as we hold a mutex and are
allowed to sleep in this context.

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

IPVS: convert __ip_vs_securetcp_lock to a spinlock

Also rename __ip_vs_securetcp_lock to ip_vs_securetcp_lock.

Spinlock conversion was suggested by Eric Dumazet.

Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

IPVS: convert __ip_vs_sched_lock to a spinlock

Also rename __ip_vs_sched_lock to ip_vs_sched_lock.

Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

IPVS: ICMPv6 checksum calculation

Cc: Xiaoyu Du <tingsrain@gmail.com>
Signed-off-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

bna: Fixed build break for allyesconfig

This is the patch to fix the build break caused by multiple
definitions of symbols between Brocade's FC/FCOE driver(BFA)
and 10G Networking Driver(BNA).

Changes are:

1. locally used functions are made static

2. unused functions are removed

3. using unique namespaces for the function names that must be
globally visible

Signed-off-by: Debashis Dutt <ddutt@brocade.com>
Signed-off-by: Rasesh Mody <rmody@brocade.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r6040: Free irq line on error path

Free irq line on error path.

Signed-off-by: Denis Kirjanov <dkirjanov@kernel.org>
Acked-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

s2io: remove lro parameter

Remove "lro" parameter of s2io driver.

Signed-off-by: WANG Cong <amwang@redhat.com>
Cc: jon.mason@exar.com
Acked-by: Jon Mason <jon.mason@exar.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: stats for packets received due to internal switching in ASIC.

Counters for packets received due to internal switching are already available.
This change will start displaying them in ethtool -S

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: fix to dynamically generate MAC Address for VFs

The BE ASIC/firmware doesnot reserve and assign MAC address for VFs.
This results in the VF interfaces being created with MAC Address 0.
The code change proposed takes the MAC address of PF to generate a seed.
MAC Address for VFs are assigned incrementally starting from the seed.
These addresses are programmed in the ASIC by the PF and the VF driver
queries for the MAC address during its probe.

Signed-off-by: Ajit Khaparde <ajitk@serverengines.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: avoid ethtool on non running interface

If bridge port is offline, don't call ethtool to query speed.

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

qlge: Fix a deadlock when the interface is going down

Currently qlge can deadlock when the interface is going
down, and the mpi_port_cfg_work() is executing on another
processor. It happens because unregister_netdev() holds
the rtnl lock, and the mpi_port_cfg_work() also request
this lock.

Since unregiter_netdev() may wait mpi_port_cfg_work(), who
also request the holding lock, it can cause an deadlock,
displaying the following error:

 "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
 rmmod         D 00000080c6c1d190     0  3993   2081 0x00008080
 Call Trace:
 [c000000975f56ee0] [c0000000000152a0] .__switch_to+0x100/0x1d0
 [c000000975f56f70] [c0000000005781b4] .schedule+0x3a4/0x8c0
 [c000000975f570c0] [c000000000578e8c] .schedule_timeout+0x24c/0x350
 [c000000975f571e0] [c000000000578a88] .wait_for_common+0x198/0x210
 [c000000975f572c0] [c0000000000abbb4] .__cancel_work_timer+0x2c4/0x2e0
 [c000000975f57400] [d0000000078e7a20] .ql_adapter_down+0x80/0x260 [qlge]
 [c000000975f574b0] [d0000000078e7d80] .qlge_close+0x70/0x130 [qlge]
 [c000000975f57540] [c000000000497ef8] .__dev_close+0x98/0xf0
 [c000000975f575d0] [c000000000497f74] .dev_close+0x24/0x60
 [c000000975f57650] [c000000000498080] .rollback_registered_many+0xd0/0x2b0
 [c000000975f576f0] [c000000000498338] .rollback_registered+0x38/0x50
 [c000000975f57780] [c0000000004983d8] .unregister_netdevice_queue+0x88/0xe0
 [c000000975f57810] [c000000000498574] .unregister_netdev+0x24/0x40
 [c000000975f57890] [d0000000078f6f38] .qlge_remove+0x3c/0x78 [qlge]
 [c000000975f57920] [c0000000002d9298] .pci_device_remove+0x48/0x90
 [c000000975f579a0] [c000000000372850] .__device_release_driver+0xa0/0x130
 [c000000975f57a30] [c000000000372a08] .driver_detach+0x128/0x150
 [c000000975f57ad0] [c000000000371134] .bus_remove_driver+0xc4/0x1a0
 [c000000975f57b70] [c00000000037357c] .driver_unregister+0x8c/0xd0
 [c000000975f57c00] [c0000000002d968c] .pci_unregister_driver+0x5c/0x110
 [c000000975f57ca0] [d0000000078f6ee4] .qlge_exit+0x1c/0x34 [qlge]

Signed-off-by: Breno Leitao <leitao@linux.vnet.ibm.com>
Signed-off-by: Ron Mercer <ron.mercer@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: fix locking comment

The carrier check is not called from work queue in current code.

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

net/netfilter/ipvs: Eliminate memory leak

__ip_vs_service_get and __ip_vs_svc_fwm_get increment a reference count, so
that reference count should be decremented before leaving the function in an
error case.

A simplified version of the semantic match that finds this problem is:
(http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
local idexpression x;
expression E;
identifier f1;
iterator I;
@@

x = __ip_vs_service_get(...);
<... when != x
     when != true (x == NULL || ...)
     when != if (...) { <+...x...+> }
     when != I (...) { <+...x...+> }
(
 x == NULL
|
 x == E
|
 x->f1
)
...>
* return ...;
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net/irda: Eliminate memory leak

dev_alloc_skb allocates some memory, so that memory should be freed before
leaving the function in an error case.

Corrected some typos in a nearby comment as well.

A simplified version of the semantic match that finds this problem is:
(http://coccinelle.lip6.fr/)

// <smpl>
@r exists@
local idexpression x;
expression E;
identifier f1;
iterator I;
@@

x = dev_alloc_skb(...);
<... when != x
     when != true (x == NULL || ...)
     when != if (...) { <+...x...+> }
     when != I (...) { <+...x...+> }
(
 x == NULL
|
 x == E
|
 x->f1
)
...>
* return ...;
// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David S. Miller <davem@davemloft.net>

typhoon: fix a race in typhoon_do_get_stats

Its important to store 'final' values in counters,
not using them as temporary variables,
or this might break some SNMP applications.

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

plip: remove superflous return

This return isn't reachable and it obscures the goto on the line before.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

stmmac: make ioaddr 'void __iomem *' rather than unsigned long

This avoids unnecessary casting and adds the ioaddr in the
private structure.
This patch also removes many warning when compile the driver.

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

stmmac: remove dead option in the driver's Kconfig

This patch removes the CPU_SUBTYPE_ST40 dependency in the
driver's Kconfig.
In fact, this option has been removed in the commit:
 f96691872439ab2071171d4531c4a95b5d493ae5
as reported by Christian Dietrich.

Note that the driver remains tested on STM platforms, only.

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

stmmac: fix_mac_speed is called during 10/100<->1000 speed changes

This patch modifies the stmmac_adjust_link() function so the
fix_mac_speed() is called not only when link speed is changing
between 10 and 100 Mbps (as required in RMII mode) but also
for 1000 Mbps.

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

bnx2x: small fix in stats handling

Use a private variable to fold rx_dropped value, instead of shared
destination buffer, as it might break SNMP applications.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Eilon Greenstein <eilong@broadcom.com>
CC: Dmitry Kravkov <dmitry@broadcom.com>
Acked-By: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

drivers/net: Remove address use from assignments of function pointers

"foo = &function" is more commonly written "foo = function"

Done with coccinelle script:

// <smpl>
@r@
identifier f;
@@

f(...) { ... }

@@
identifier r.f;
@@

- &f
+ f
// </smpl>

drivers/net/tehuti.c used a function and struct with the
same name, the function was renamed.

Compile tested x86 only.

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

qlcnic: change reg name

Change QLCNIC_CRB_DEV_REF_COUNT to QLCNIC_CRB_DRV_ACTIVE to match
document.

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

qlcnic: fix fw recovery for PF

Privilege function should wait for npar state to be operational
before creating context.

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

qlcnic: support port vlan id

On NIC Partition capable adapter, Administrator can configure to
tag packet with particular vlan id. Packet will be tagged and strip with
that vlan id. Also if 'Tagging' flag is disable, other packet will be drop.

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

qlcnic: eswitch config fixes

o remove validation before deleting vlan id
o Add missing 'break' while deleting vlan id.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: Amit Kumar Salecha <amit.salecha@qlogic.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/libertas/if_sdio.c

mlx4_en: Fix build warning in mlx4_en_create_rx_ring.

drivers/net/mlx4/en_rx.c: In function ‘mlx4_en_create_rx_ring’:
drivers/net/mlx4/en_rx.c:305: warning: label ‘err_map’ defined but not used

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

mlx4_en: updated driver version

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Moving to work with GRO

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: UDP RSS support

Adding capability for RSS for UDP traffic, hashing is done based on
IP addresses and UDP port number.
The support depends on HW/FW capabilities.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: reconfiguring mac address.

When Mac address is removed from one port of the CX2 device, the other device
should reconfigure its Mac.
This fixes an issue with failover, when both ports have the same Mac address,
and one of the ports, is closed, the second one stops receiving traffic.
(bugzilla #1965 at bugs.openfabrics.org)

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: get/set ringsize uses actual ring size

In case that the rings were not fully filled, the report in Ethtool
should be according to the actual size that was filled.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Fixing report in Ethtool get_settings

The report now based on query from FW, giving the correct tranciever type
and link speed.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Added self diagnostics test implementation

The selftest includes 5 features:
1. Interrupt test: Executing commands and receiving command completion
   on all our interrupt vectors.
2. Link test: Verifying we are connected to valid link partner.
3. Speed test: Check that we negotiated link speed correctly.
4. Registers test: Activate HW health check command.
5. Loopback test: Send a packet on loopback interface and catch it on RX side.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Validate port up prior to transmitting

We might try to use resources which were not activated.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Reporting link state with KERN_INFO

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Fixed MAX_TX_RINGS definition

Defined as sum of default number of TX rings and number of
rings used for priority flow control mode.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: performing CLOSE_PORT at the end of tear-down process

All port's resources should be closed before we give the
CLOSE_PORT command to the firmware.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Setting dev->perm_addr field

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Setting actual RX ring size

When configuring HW resources, the RX ring size that is passed should be the actual
size that depends on number of buffers that we succeeded to allocate.
The mask for ring size should also be configured accordingly

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>

mlx4_en: Fixed incorrect unmapping on RX flow.

When allocating new fragments to replace the ones that would be passed to the stack,
The fragments that should be replaced, are the ones that were already used.

Signed-off-by: Yevgeny Petrilin <yevgenyp@mellanox.co.il>
Signed-off-by: David S. Miller <davem@davemloft.net>