macvlan: remove one synchronize_rcu() call

When one macvlan device is dismantled, we can avoid one
synchronize_rcu() call done after deletion from hash list, since caller
will perform a synchronize_net() call after its ndo_stop() call.

Add a new netdev->dismantle field to signal this dismantle intent.

Reduces RTNL hold time.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Patrick McHardy <kaber@trash.net>
CC: Ben Greear <greearb@candelatech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

networking: NET_CLS_ROUTE4 depends on INET

IP_ROUTE_CLASSID depends on INET and NET_CLS_ROUTE4 selects
IP_ROUTE_CLASSID, but when INET is not enabled, this kconfig warning
is produced, so fix it by making NET_CLS_ROUTE4 depend on INET.

warning: (NET_CLS_ROUTE4) selects IP_ROUTE_CLASSID which has unmet direct dependencies (NET && INET)

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

irda: Fix error propagation in ircomm_lmp_connect_response()

The variable 'ret' is set but unused, and this pointed out that
errors from irlmp_connect_response() are not propagated to the
caller.

Note that this is currently academic since irlmp_connect_response()
always returns 0. :-)

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

irda: Kill set but unused variable 'bytes' in irlan_check_command_param()

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

irda: Kill set but unused variable 'clen' in ircomm_connect_indication()

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

rxrpc: Fix set but unused variable 'usage' in rxrpc_get_transport()

This is identical to the case I fixed in rxrpc_get_peer()

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

be2net: Kill set but unused variable 'req' in lancer_fw_download()

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

irda: Kill set but unused vars 'saddr' and 'daddr' in irlan_provider_connect_indication()

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

atl1c: atl1c_resume() is only used when CONFIG_PM_SLEEP is defined.

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

rxrpc: Fix set but unused variable 'usage' in rxrpc_get_peer().

I backed off from trying to just eliminate this variable, since
transforming atomic_inc_return() into atomic_inc() takes away
the memory barriers.

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

rxrpc: Kill set but unused variable 'local' in rxrpc_UDP_error_handler()

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

rxrpc: Kill set but unused variable 'sp' in rxrpc_process_connection()

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

rxrpc: Kill set but unused variable 'sp' in rxrpc_rotate_tx_window()

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

pkt_sched: Kill set but unused variable 'protocol' in tc_classify()

I checked the history and this has been like this since the
beginning of time.

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

isdn: capi: Use pr_debug() instead of ifdefs.

I was investigating some warnings that spew because of the
_DEBUG_FOO ifdef'ery in here.

Instead of adding more ifdefs to fix that warning, let's use
pr_debug() and get rid of these CPP checks altogether.

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

tg3: Update version to 3.119

This patch updates the tg3 version to 3.119.

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

tg3: Apply rx_discards fix to 5719/5720

Commit 4d95847381228639844c7197deb8b2211274ef22, entitled
"tg3: Workaround rx_discards stat bug", was intended to be applied to
the 5717, 5718, 5719_A0, and 5720 A0 chip revisions.  The implementation
missed the latter two when applying the fix in a critical area.  This
patch fixes the problem.

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

tg3: Remove excessive parenthesis

This patch removes some excessive parenthesizing.

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

tg3: Consolidate all netdev feature assignments

This patch consolidates all the netdev feature bit assignments to one
location.

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

tg3: Move TSO_CAPABLE assignment

This patch moves the code that asserts the TSO_CAPABLE flag closer to
where the TSO capabilities flags are set.  There isn't a good enough
reason for the code to be separated.

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

tg3: Fix stats for 5704 and later devices

Commit 4d95847381228639844c7197deb8b2211274ef22, entitled
"tg3: Workaround rx_discards stat bug" modified the hardware statistics
data structure.  The modification shifted the statistics so that the
labels no longer corresponded to the counter values.  This patch fixes
the problem by utilizing reserved space for the new counters.

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

tg3: Fix TSO loopback test

Commit bb158d696489244f79fd4c3abd47968a06b48c79, entitled
"tg3: Add TSO loopback test", mistakenly inverted the checksum field
test from the receive BD.  This patch corrects the problem.

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

tg3: Consolidate autoneg advertisement setup code

Autonegotiation setup has gotten a little more complicated since the tg3
driver was created.  This patch consolidates autoneg setup into one
routine and modifies the call sites accordingly.

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

tg3: Fix EEE interoperability workaround

Commit 21a00ab270f95d32e502d92f166dd75c518d3c5f, entitled
"tg3: Fix EEE interoperability issue", added an EEE interoperability
fix.  We found that the fix doesn't work if applied too early though.
This patch delays the fix until right before allowing LPI assertion.

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

tg3: Fix IPv6 TCP problems for 5719

Commit 4d163b75e979833979cc401ae433cb1d7743d57e, entitled
"tg3: Fix 5719 A0 tx completion bug" turned off TSO to fix a hardware
bug.  In doing so, it accidentally turned off all IPv6 TCP checksum
offloading too.  This patch fixes the problem by reenabling the hardware
bit that control both features.  The TSO capability is still not exposed
to the kernel.

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

tg3: Fix 57765 B0 data corruption

The PCIe max FTS limit is too aggressive on these chips.  This patch
loosens the limit a little to eliminate data corruption issues.

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

tg3: Cleanup transmit error path

This patch consolidates the skb cleanup code into a function named
tg3_skb_error_unmap().  The modification addresses a long-standing bug
where pci_unmap_single() was incorrectly being called instead of
pci_unmap_page() in tigon3_dma_hwbug_workaround().

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

tg3: Set tx bug flags for more devices

It has been recently discovered that all tg3 devices have a 4Gb boundary
DMA problem, and that all 5755 and newer devices can't handle fragments
less than or equal to 8 bytes in size.  This patch adjusts the flags and
removes tg3_start_xmit().  tg3_start_xmit_dma_bug() has been renamed to
tg3_start_xmit().

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

9p: Kill unused variable 'ret' in trans_rdma.c:parse_opts()

Probably just cut and pasted from the other parse_opts() implementations
in the 9p sources.

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

9p: Kill set but not used variable 'ret' in parse_opts()

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

ipconfig wait for carrier

v3 -> v4: fix return boolean false instead of 0 for ic_is_init_dev

Currently the ip auto configuration has a hardcoded delay of 1 second.
When (ethernet) link takes longer to come up (e.g. more than 3 seconds),
nfs root may not be found.

Remove the hardcoded delay, and wait for carrier on at least one network
device.

Signed-off-by: Micha Nelissen <micha@neli.hopto.org>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

SCTP: fix race between sctp_bind_addr_free() and sctp_bind_addr_conflict()

During the sctp_close() call, we do not use rcu primitives to
destroy the address list attached to the endpoint.  At the same
time, we do the removal of addresses from this list before
attempting to remove the socket from the port hash

As a result, it is possible for another process to find the socket
in the port hash that is in the process of being closed.  It then
proceeds to traverse the address list to find the conflict, only
to have that address list suddenly disappear without rcu() critical
section.

Fix issue by closing address list removal inside RCU critical
section.

Race can result in a kernel crash with general protection fault or
kernel NULL pointer dereference:

kernel: general protection fault: 0000 [#1] SMP
kernel: RIP: 0010:[<ffffffffa02f3dde>]  [<ffffffffa02f3dde>] sctp_bind_addr_conflict+0x64/0x82 [sctp]
kernel: Call Trace:
kernel:  [<ffffffffa02f415f>] ? sctp_get_port_local+0x17b/0x2a3 [sctp]
kernel:  [<ffffffffa02f3d45>] ? sctp_bind_addr_match+0x33/0x68 [sctp]
kernel:  [<ffffffffa02f4416>] ? sctp_do_bind+0xd3/0x141 [sctp]
kernel:  [<ffffffffa02f5030>] ? sctp_bindx_add+0x4d/0x8e [sctp]
kernel:  [<ffffffffa02f5183>] ? sctp_setsockopt_bindx+0x112/0x4a4 [sctp]
kernel:  [<ffffffff81089e82>] ? generic_file_aio_write+0x7f/0x9b
kernel:  [<ffffffffa02f763e>] ? sctp_setsockopt+0x14f/0xfee [sctp]
kernel:  [<ffffffff810c11fb>] ? do_sync_write+0xab/0xeb
kernel:  [<ffffffff810e82ab>] ? fsnotify+0x239/0x282
kernel:  [<ffffffff810c2462>] ? alloc_file+0x18/0xb1
kernel:  [<ffffffff8134a0b1>] ? compat_sys_setsockopt+0x1a5/0x1d9
kernel:  [<ffffffff8134aaf1>] ? compat_sys_socketcall+0x143/0x1a4
kernel:  [<ffffffff810467dc>] ? sysenter_dispatch+0x7/0x32

Signed-off-by: Jacek Luczak <luczak.jacek@gmail.com>
Acked-by: Vlad Yasevich <vladislav.yasevich@hp.com>
CC: Eric Dumazet <eric.dumazet@gmail.com>
Reviewed-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: reduce per device ICMP mib sizes

ipv6 has per device ICMP SNMP counters, taking too much space because
they use percpu storage.

needed size per device is :
(512+4)*sizeof(long)*number_of_possible_cpus*2

On a 32bit kernel, 16 possible cpus, this wastes more than 64kbytes of
memory per ipv6 enabled network device, taken in vmalloc pool.

Since ICMP messages are rare, just use shared counters (atomic_long_t)

Per network space ICMP counters are still using percpu memory, we might
also convert them to shared counters in a future patch.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Denys Fedoryshchenko <denys@visp.net.lb>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ping: fix the coding style

The characters in a line should be no more than 80.

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

net: ping: make local functions static

As these functions are only used in this file.

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

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

ipv4: Pass explicit destination address to rt_bind_peer().

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

ipv4: Pass explicit destination address to rt_get_peer().

This will next trickle down to rt_bind_peer().

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

ipv4: Make caller provide flowi4 key to inet_csk_route_req().

This way the caller can get at the fully resolved fl4->{daddr,saddr}
etc.

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

ipv4: Kill RT_CACHE_DEBUG

It's way past it's usefulness.  And this gets rid of a bunch
of stray ->rt_{dst,src} references.

Even the comment documenting the macro was inaccurate (stated
default was 1 when it's 0).

If reintroduced, it should be done properly, with dynamic debug
facilities.

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

be2net: Enable SR-IOV for Lancer

Enable SR-IOV for Lancer

Signed-off-by: Mammatha Edhala <mammatha.edhala@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bnx2x: add support for retrieving dcb peer configuration

This patch adds support to the bnx2x for retrieving dcb peer (remote)
configuration from the embedded DCBX stack.

Signed-off-by: Shmulik Ravid <shmulikr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

qeth: use ndo_set_features callback for initial setup and recovery

This patch uses the ndo_set_features callback during normal device
startup or recovery to turn on hardware RX checksum. Patch was done
with much help from Michal Miroslaw, thx!!!

Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Reviewed-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

wl12xx: fix compilation error when CONFIG_PM is not set

There was a compilation error when PM is not enabled:

  CC [M]  drivers/net/wireless/wl12xx/main.o
drivers/net/wireless/wl12xx/main.c:3653: error: unknown field 'suspend' specified in initializer
drivers/net/wireless/wl12xx/main.c:3653: warning: initialization from incompatible pointer type
drivers/net/wireless/wl12xx/main.c:3654: error: unknown field 'resume' specified in initializer
drivers/net/wireless/wl12xx/main.c:3654: warning: initialization from incompatible pointer type

Fix this by adding #ifdef's in the appropriate places.

Cc: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: add skb_dst_force() in sock_queue_err_skb()

Commit 7fee226ad239 (add a noref bit on skb dst) forgot to use
skb_dst_force() on packets queued in sk_error_queue

This triggers following warning, for applications using IP_CMSG_PKTINFO
receiving one error status

------------[ cut here ]------------
WARNING: at include/linux/skbuff.h:457 ip_cmsg_recv_pktinfo+0xa6/0xb0()
Hardware name: 2669UYD
Modules linked in: isofs vboxnetadp vboxnetflt nfsd ebtable_nat ebtables
lib80211_crypt_ccmp uinput xcbc hdaps tp_smapi thinkpad_ec radeonfb fb_ddc
radeon ttm drm_kms_helper drm ipw2200 intel_agp intel_gtt libipw i2c_algo_bit
i2c_i801 agpgart rng_core cfbfillrect cfbcopyarea cfbimgblt video raid10 raid1
raid0 linear md_mod vboxdrv
Pid: 4697, comm: miredo Not tainted 2.6.39-rc6-00569-g5895198-dirty #22
Call Trace:
 [<c17746b6>] ? printk+0x1d/0x1f
 [<c1058302>] warn_slowpath_common+0x72/0xa0
 [<c15bbca6>] ? ip_cmsg_recv_pktinfo+0xa6/0xb0
 [<c15bbca6>] ? ip_cmsg_recv_pktinfo+0xa6/0xb0
 [<c1058350>] warn_slowpath_null+0x20/0x30
 [<c15bbca6>] ip_cmsg_recv_pktinfo+0xa6/0xb0
 [<c15bbdd7>] ip_cmsg_recv+0x127/0x260
 [<c154f82d>] ? skb_dequeue+0x4d/0x70
 [<c1555523>] ? skb_copy_datagram_iovec+0x53/0x300
 [<c178e834>] ? sub_preempt_count+0x24/0x50
 [<c15bdd2d>] ip_recv_error+0x23d/0x270
 [<c15de554>] udp_recvmsg+0x264/0x2b0
 [<c15ea659>] inet_recvmsg+0xd9/0x130
 [<c1547752>] sock_recvmsg+0xf2/0x120
 [<c11179cb>] ? might_fault+0x4b/0xa0
 [<c15546bc>] ? verify_iovec+0x4c/0xc0
 [<c1547660>] ? sock_recvmsg_nosec+0x100/0x100
 [<c1548294>] __sys_recvmsg+0x114/0x1e0
 [<c1093895>] ? __lock_acquire+0x365/0x780
 [<c1148b66>] ? fget_light+0xa6/0x3e0
 [<c1148b7f>] ? fget_light+0xbf/0x3e0
 [<c1148aee>] ? fget_light+0x2e/0x3e0
 [<c1549f29>] sys_recvmsg+0x39/0x60

Close bug https://bugzilla.kernel.org/show_bug.cgi?id=34622

Reported-by: Witold Baryluk <baryluk@smp.if.uj.edu.pl>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sfc: Don't use enums as a bitmask.

This fixes:

drivers/net/sfc/mcdi_mac.c: In function ‘efx_mcdi_set_mac’:
drivers/net/sfc/mcdi_mac.c:36:2: warning: case value ‘3’ not in enumerated type ‘enum efx_fc_type’

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

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

Conflicts:
drivers/net/vmxnet3/vmxnet3_ethtool.c
net/core/dev.c

ipv4: Don't use enums as bitmasks in ip_fragment.c

Noticed by Joe Perches.

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

net: ethtool: fix IPV6 checksum feature name string

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

net: Change netdev_fix_features messages loglevel

Cool, how about we make 'Features changed' debug as well?
This way userspace can't fill up the log just by tweaking tun features
with an ioctl.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: recvmmsg: Strip MSG_WAITFORONE when calling recvmsg

recvmmsg fails on a raw socket with EINVAL. The reason for this is
packet_recvmsg checks the incoming flags:

        err = -EINVAL;
        if (flags & ~(MSG_PEEK|MSG_DONTWAIT|MSG_TRUNC|MSG_CMSG_COMPAT|MSG_ERRQUEUE))
                goto out;

This patch strips out MSG_WAITFORONE when calling recvmmsg which
fixes the issue.

Signed-off-by: Anton Blanchard <anton@samba.org>
Cc: stable@kernel.org [2.6.34+]
Signed-off-by: David S. Miller <davem@davemloft.net>

xen: netback: use __CONST_RING_SIZE not __RING_SIZE

The later causes warnings with gcc 4.5+. __CONST_RING_SIZE was introduced in
667c78afaec0 to fix this but as netback wasn't upstream at the time it did not
benefit, hence:

  CC      drivers/net/xen-netback/netback.o
drivers/net/xen-netback/netback.c:110:37: warning: variably modified 'grant_copy_op' at file scope [enabled by default]
drivers/net/xen-netback/netback.c:111:30: warning: variably modified 'meta' at file scope [enabled by default]
drivers/net/xen-netback/netback.c: In function 'xen_netbk_rx_action':
drivers/net/xen-netback/netback.c:584:6: warning: variable 'irq' set but not used [-Wunused-but-set-variable]

Thanks to Witold Baryluk for pointing this out.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Witold Baryluk <baryluk@smp.if.uj.edu.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

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

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

net: ping: fix build failure

If CONFIG_PROC_SYSCTL=n the building process fails:

    ping.c:(.text+0x52af3): undefined reference to `inet_get_ping_group_range_net'

Moved inet_get_ping_group_range_net() to ping.c.

Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Vasiliy Kulikov <segoon@openwall.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: use hlist_del_rcu() in dev_change_name()

Using plain hlist_del() in dev_change_name() is wrong since a
concurrent reader can crash trying to dereference LIST_POISON1.

Bug introduced in commit 72c9528bab94 (net: Introduce
dev_get_by_name_rcu())

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

bluetooth: Fix warnings in l2cap_core.c

net/bluetooth/l2cap_core.c: In function ‘l2cap_recv_frame’:
net/bluetooth/l2cap_core.c:3758:15: warning: ‘sk’ may be used uninitialized in this function
net/bluetooth/l2cap_core.c:3758:15: note: ‘sk’ was declared here
net/bluetooth/l2cap_core.c:3791:15: warning: ‘sk’ may be used uninitialized in this function
net/bluetooth/l2cap_core.c:3791:15: note: ‘sk’ was declared here

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

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

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

Conflicts:
drivers/net/wireless/iwlwifi/iwl-agn-tx.c
net/mac80211/sta_info.h

sfc: Use netif_device_{detach,attach}() around reset and self-test

We need to keep the TX queues stopped throughout a reset, without
triggering the TX watchdog and regardless of the link state.  The
proper way to do this is to use netif_device_{detach,attach}() just as
we do around suspend/resume, rather than the current bodge of faking
link-down.

Since we also need to do this during an offline self-test and we
perform a reset during that, add these function calls outside of
efx_reset_down() and efx_reset_up().

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

net: Change netdev_fix_features messages loglevel

Those reduced to DEBUG can possibly be triggered by unprivileged processes
and are nothing exceptional. Illegal checksum combinations can only be
caused by driver bug, so promote those messages to WARN.

Since GSO without SG will now only cause DEBUG message from
netdev_fix_features(), remove the workaround from register_netdevice().

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

vmxnet3: Fix inconsistent LRO state after initialization

During initialization of vmxnet3, the state of LRO
gets out of sync with netdev->features.

This leads to very poor TCP performance in a IP forwarding
setup and is hitting many VMware users.

Simplified call sequence:
1. vmxnet3_declare_features() initializes "adapter->lro" to true.

2. The kernel automatically disables LRO if IP forwarding is enabled,
so vmxnet3_set_flags() gets called. This also updates netdev->features.

3. Now vmxnet3_setup_driver_shared() is called. "adapter->lro" is still
set to true and LRO gets enabled again, even though
netdev->features shows it's disabled.

Fix it by updating "adapter->lro", too.

The private vmxnet3 adapter flags are scheduled for removal
in net-next, see commit a0d2730c9571aeba793cb5d3009094ee1d8fda35
"net: vmxnet3: convert to hw_features".

Patch applies to 2.6.37 / 2.6.38 and 2.6.39-rc6.

Please CC: comments.

Signed-off-by: Thomas Jarosch <thomas.jarosch@intra2net.com>
Acked-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sfc: Fix oops in register dump after mapping change

Commit 747df2258b1b9a2e25929ef496262c339c380009 ('sfc: Always map MCDI
shared memory as uncacheable') introduced a separate mapping for the
MCDI shared memory (MC_TREG_SMEM).  This means we can no longer easily
include it in the register dump.  Since it is not particularly useful
in debugging, substitute a recognisable dummy value.

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

net: convert to new cpumask API

We plan to remove cpu_xx() old api later. Thus this patch
convert it.

This patch has no functional change.

Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: more compliant RFC 3168 support

Commit 6623e3b24a5e (ipv4: IP defragmentation must be ECN aware) was an
attempt to not lose "Congestion Experienced" (CE) indications when
performing datagram defragmentation.

Stefanos Harhalakis raised the point that RFC 3168 requirements were not
completely met by this commit.

In particular, we MUST detect invalid combinations and eventually drop
illegal frames.

Reported-by: Stefanos Harhalakis <v13@v13.gr>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: FW download for Lancer

Added implementation of FW download feature for Lancer.

Signed-off-by: Shripad Nunjundarao <shripad.nunjundarao@emulex.com>
Signed-off-by: Sevin Xavier <selvin.xavier@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Stats for Lancer

Added Lancer stats implementation.

Signed-off-by: Selvin Xavier <selvin.xavier@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

be2net: Support for version 1 of stats for BE3

Added support to get version 1 of the stats for BE3.
Use old stats command for BE2.

Signed-off-by: Ajit Khaparde <ajit.khaparde@emulex.com>
Signed-off-by: Selvin Xavier <selvin.xavier@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

mwifiex: use ieee80211_amsdu_to_8023s routine

mwifiex was using its own implementation of converting 802.11n
AMSDU to 802.3s.  This patch removes mwifiex specific
implementation and uses existing ieee80211_amsdu_to_8023s
routine.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

cfg80211: make stripping of 802.11 header optional from AMSDU

Currently the devices that have already stripped IEEE 802.11
header from the AMSDU SKB can not use ieee80211_amsdu_to_8023s
routine. This patch enhances ieee80211_amsdu_to_8023s() API by
changing mandatory removing of IEEE 802.11 header from AMSDU
to optional.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

nl80211: Move peer link state definition to nl80211

These definitions need to be exposed now that we can set the peer link
states via NL80211_ATTR_STA_PLINK_STATE.  They were already being
(opaquely) reported by NL80211_STA_INFO_PLINK_STATE.

Signed-off-by: Javier Cardona <javier@cozybit.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: make npending frames check as bool

we are not doing anything by tracking the number of pending frames.
bail out when we first find a pending frame in any one of the 10 queues.

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: make sure main_rssi is positive

some times the rssi control descriptor for the main antenna may be
negative like that of alternate antenna, hence before  incrementing
packet counts/rssi of main/alternate antenna make sure both main_rssi
and alt_rssi are positive only. this avoids wrong selection of antenna
due to diversity

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: configure fast_div_bias based on diversity group

configure fast diversity bias based on the antenna diversity group and
based on main/alt LNA configurations. also configure main antenna and
alternate antenna to gain-table 0 for diversity group 2(AR9485)

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k: Implement an API to swap main/ALT LNA's

for the diversity group 2(AR9485) we swap the LNA's of main/ALT antenna
based on alternate antenna's rssi average in comparision with main
antenna's rssi, while for AR9285(antenna diversity group 0)we still
follow the older method of looking at the packet count in alternate
antenna

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: define antenna diversity group

AR9285 belongs to diversity group 0 and AR9485 belongs to diversity
group 2. Based on the diversity group we configure certain antenna
diversity paramaters such as lna1_lna2_delta and fast diversity
bias values. For AR9485 we have some gain table parameter which
selects the gain table 0/1 for main and alternate antenna

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: define modules to get/set Antenna diversity paramaters

these are the two important modules that will be called by the antenna
diversity algorithm module in the rx. this will continuosly configure
the hardware based on the current diversity status obtained
from the algorithm

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: config diversity based on eeprom contents

* enable LNA-diversity, fast-diversity for AR9485 based  on
  the value read from EEPROM content
* if antenna diversity/combining is supported, set LNA1 for the main
  antenna and LNA2 for the alternate antenna based on the new diversity
  algorithm

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Cc: Luis Rodriguez <Luis.Rodriguez@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: define registers/macros to support Antenna diversity

define few registers and macros to configure/enable Antenna diversity
parameters in AR9485

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: enable Antenna diversity for AR9485

read antenna diversity and combining information from the EEPROM.
Enable antenna diversity/combining feature only when both LNA
diversity and fast diversity are supported

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Tested-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: make antenna diversity modules chip specific

this is necessary to support Antenna diversity and combining in new chip
sets such as AR9485, previously Antenna diversity support is available
only in AR9285

Cc: Gabriel Tseng <Gabriel.Tseng@Atheros.com>
Cc: Senthilkumar Balasubramanian <Senthilkumar.Balasubramanian@Atheros.com>
Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

net/rfkill/core.c: Avoid leaving freed data in a list

The list_for_each_entry loop can fail, in which case the list element is
not removed from the list rfkill_fds.  Since this list is not accessed by
the loop, the addition of &data->list into the list is just moved after the
loop.

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

// <smpl>
@@
expression E,E1,E2;
identifier l;
@@

*list_add(&E->l,E1);
... when != E1
    when != list_del(&E->l)
    when != list_del_init(&E->l)
    when != E = E2
*kfree(E);// </smpl>

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: sparse RCU annotations

This adds sparse RCU annotations to most of
mac80211, only the mesh code remains to be
done.

Due the the previous patches, the annotations
are pretty simple. The only thing that this
actually changes is removing the RCU usage of
key->sta in debugfs since this pointer isn't
actually an RCU-managed pointer (it only has
a single assignment done before the key even
goes live). As that is otherwise harmless, I
decided to make it part of this patch.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

mac80211: fix TX a-MPDU locking

During my quest to make mac80211 not have any RCU
warnings from sparse, I came across the a-MPDU code
again and it wasn't quite clear why it isn't racy.
So instead of assigning the tid_tx array with just
the spinlock held in ieee80211_start_tx_ba_session
use a separate temporary array protected only by
the spinlock and protect all assignments to the
"live" array by both the spinlock and the mutex so
that other code is easily verified to be correct.

Due to pointer assignment atomicity I don't think
this is a real issue, but I'm not sure, especially
on Alpha the current code might be problematic.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

cfg80211: advertise possible interface combinations

Add the ability to advertise interface combinations in nl80211.
This allows the driver to indicate what the combinations are
that it supports. "Combinations" of just a single interface are
implicit, as previously. Note that cfg80211 will enforce that
the restrictions are met, but not for all drivers yet (once all
drivers are updated, we can remove the flag and enforce for all).

When no combinations are actually supported, an empty list will
be exported so that userspace can know if the kernel exported
this info or not (although it isn't clear to me what tools using
the info should do if the kernel didn't export it).

Since some interface types are purely virtual/software and don't
fit the restrictions, those are exposed in a new list of pure SW
types, not subject to restrictions. This mainly exists to handle
AP-VLAN and monitor interfaces in mac80211.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlwifi: Silence DEBUG_STRICT_USER_COPY_CHECKS=y warning

Enabling DEBUG_STRICT_USER_COPY_CHECKS causes the following
warning:

In file included from arch/x86/include/asm/uaccess.h:573,
                 from include/net/checksum.h:25,
                 from include/linux/skbuff.h:28,
                 from drivers/net/wireless/iwlwifi/iwl-agn-rs.c:28:
In function 'copy_from_user',
    inlined from 'rs_sta_dbgfs_scale_table_write' at
    drivers/net/wireless/iwlwifi/iwl-agn-rs.c:3099:
arch/x86/include/asm/uaccess_64.h:65:
warning: call to 'copy_from_user_overflow' declared with
attribute warning: copy_from_user() buffer size is not provably
correct

presumably due to buf_size being signed causing GCC to fail to
see that buf_size can't become negative.

Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

iwlegacy: Silence DEBUG_STRICT_USER_COPY_CHECKS=y warning

Enabling DEBUG_STRICT_USER_COPY_CHECKS causes the following
warning:

In file included from arch/x86/include/asm/uaccess.h:573,
                 from include/net/checksum.h:25,
                 from include/linux/skbuff.h:28,
                 from drivers/net/wireless/iwlegacy/iwl-4965-rs.c:28:
In function 'copy_from_user',
    inlined from 'iwl4965_rs_sta_dbgfs_scale_table_write' at
    drivers/net/wireless/iwlegacy/iwl-4965-rs.c:2616:
arch/x86/include/asm/uaccess_64.h:65:
warning: call to 'copy_from_user_overflow' declared with
attribute warning: copy_from_user() buffer size is not provably
correct

presumably due to buf_size being signed causing GCC to fail to
see that buf_size can't become negative.

Cc: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

bnx2x: Allow ethtool to enable/disable loopback.

This patch updates the bnx2x_set_features() to handle loopback mode.
When enabled; it sets internal-MAC loopback.

Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv4: Trivial rt->rt_src conversions in net/ipv4/route.c

At these points we have a fully filled in value via the IP
header the form of ip_hdr(skb)->saddr

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

sfc: Fix TX queue numbering when separate_tx_channels=1

This option appears to have been broken by commit
8313aca38b3937947fffebca6e34bac8e24300c8 ('sfc: Allocate each channel
separately, along with its RX and TX queues').

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

net: ping: dont call udp_ioctl()

udp_ioctl() really handles UDP and UDPLite protocols.

1) It can increment UDP_MIB_INERRORS in case first_packet_length() finds
a frame with bad checksum.

2) It has a dependency on sizeof(struct udphdr), not applicable to
ICMP/PING

If ping sockets need to handle SIOCINQ/SIOCOUTQ ioctl, this should be
done differently.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Vasiliy Kulikov <segoon@openwall.com>
Acked-by: Vasiliy Kulikov <segoon@openwall.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netfilter: nf_ct_sip: fix SDP parsing in TCP SIP messages for some Cisco phones

Some Cisco phones do not place the Content-Length field at the end of the
SIP message. This is valid, due to a misunderstanding of the specification
the parser expects the SDP body to start directly after the Content-Length
field. Fix the parser to scan for \r\n\r\n to locate the beginning of the
SDP body.

Reported-by: Teresa Kang <teresa_kang@gemtek.com.tw>
Signed-off-by: Patrick McHardy <kaber@trash.net>

netfilter: nf_ct_sip: validate Content-Length in TCP SIP messages

Verify that the message length of a single SIP message, which is calculated
based on the Content-Length field contained in the SIP message, does not
exceed the packet boundaries.

Signed-off-by: Patrick McHardy <kaber@trash.net>

Merge branch 'pablo/nf-2.6-updates' of git://1984.lsi.us.es/net-2.6

net: drivers: kill two unused macro definitions

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

caif: remove unesesarry exports

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

caif: Bugfix debugfs directory name must be unique.

Race condition caused debugfs_create_dir() to fail due to duplicate
name. Use atomic counter to create unique directory name.

net_ratelimit() is introduced to limit debug printouts.

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

caif: Handle dev_queue_xmit errors.

Do proper handling of dev_queue_xmit errors in order to
avoid double free of skb and leaks in error conditions.
In cfctrl pending requests are removed when CAIF Link layer goes down.

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

caif: prepare support for namespaces

Use struct net to reference CAIF configuration object instead of static variables.
Refactor functions caif_connect_client, caif_disconnect_client and squach
files cfcnfg.c and caif_config_utils.

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

caif: Protected in-flight packets using dev or sock refcont.

CAIF Socket Layer and ip-interface registers reference counters
in CAIF service layer. The functions sock_hold, sock_put and
dev_hold, dev_put are used by CAIF Stack to protect from freeing
memory while packets are in-flight.

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