ixgbe: ixgbe_down needs to stop dev_watchdog

There is a small race between when the tx queues are stopped
and when netif_carrier_off() is called in ixgbe_down.  If the
dev_watchdog() timer fires during this time it is possible for
a false tx timeout to occur.

This patch moves the netif_carrier_off() so that it is called before
the tx queues are stopped preventing the dev_watchdog timer from
detecting false tx timeouts.  The race is seen occosionally when
FCoE or DCB settings are being configured or changed.

Testing note, running ifconfig up/down will not reproduce this
issue because dev_open/dev_close call dev_deactivate() and then
dev_activate().

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Acked-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

igb: add support for reporting 5GT/s during probe on PCIe Gen2

This change corrects the fact that we were not reporting Gen2 link speeds
when we were in fact connected at Gen2 rates.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: fix bug when EITR=0 causing no writebacks

writebacks can be held indefinitely by hardware if EITR=0, when
combined with TXDCTL.WTHRESH=8.  When EITR=0, WTHRESH should be
set back to zero.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: enable extremely low latency

82598/82599 can support EITR == 0, which allows for the
absolutely lowest latency setting in the hardware.  This disables
writeback batching and anything else that relies upon a delayed
interrupt. This patch enables the feature of "override" when a
user sets rx-usecs to zero, the driver will respect that setting
over using RSC, and automatically disable RSC.  If rx-usecs is
used to set the EITR value to 0, then the driver should disable
LRO (aka RSC) internally until EITR is set to non-zero again.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

igbvf: double increment nr_frags

There is no need to increment nr_frags because skb_fill_page_desc increments
it.

Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

igb: double increment nr_frags

There is no need to increment nr_frags because skb_fill_page_desc increments
it.

Signed-off-by: Koki Sanagi <sanagi.koki@jp.fujitsu.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: fix a lockdep rcu warning in __sk_dst_set()

__sk_dst_set() might be called while no state can be integrated in a
rcu_dereference_check() condition.

So use rcu_dereference_raw() to shutup lockdep warnings (if
CONFIG_PROVE_RCU is set)

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rps: inet_rps_save_rxhash() argument is not const

const qualifier on sock argument is misleading, since we can modify rxhash.

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

TCP: avoid to send keepalive probes if receiving data

RFC 1122 says the following:
...
  Keep-alive packets MUST only be sent when no data or
  acknowledgement packets have been received for the
  connection within an interval.
...

The acknowledgement packet is reseting the keepalive
timer but the data packet isn't. This patch fixes it by
checking the timestamp of the last received data packet
too when the keepalive timer expires.

Signed-off-by: Flavio Leitner <fleitner@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: multicast router list manipulation

I prefer that the hlist be only accessed through the hlist macro
objects. Explicit twiddling of links (especially with RCU) exposes
the code to future bugs.

Compile tested only.

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

bridge: use is_multicast_ether_addr

Use existing inline function.

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

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

Conflicts:
drivers/net/e100.c
drivers/net/e1000e/netdev.c

net: suppress RCU lockdep false positive in twsk_net()

Calls to twsk_net() are in some cases protected by reference counting
as an alternative to RCU protection.  Cases covered by reference counts
include __inet_twsk_kill(), inet_twsk_free(), inet_twdr_do_twkill_work(),
inet_twdr_twcal_tick(), and tcp_timewait_state_process().  RCU is used
by inet_twsk_purge().  Locking is used by established_get_first()
and established_get_next().  Finally, __inet_twsk_hashdance() is an
initialization case.

It appears to be non-trivial to locate the appropriate locks and
reference counts from within twsk_net(), so used rcu_dereference_raw().

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cxgb3: Wait longer for control packets on initialization

In some Power7 platforms, when using VIOS (Virtual I/O Server), we
need to wait longer for control packets to finish transfer during
initialization.
Without this change, initialization may fail prematurely.

Signed-off-by: Wen Xiong <wenxiong@us.ibm.com>
Signed-off-by: Andre Detsch <adetsch@br.ibm.com>
Acked-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

e1000e: enable/disable ASPM L0s and L1 and ERT according to hardware errata

Prompted by a previous patch submitted by Matthew Garret <mjg@redhat.com>,
further digging into errata documentation reveals the current enabling or
disabling of ASPM L0s and L1 states for certain parts supported by this
driver are incorrect.  82571 and 82572 should always disable L1.  For
standard frames, 82573/82574/82583 can enable L1 but L0s must be disabled,
and for jumbo frames 82573/82574 must disable L1.  This allows for some
parts to enable L1 in certain configurations leading to better power
savings.

Also according to the same errata, Early Receive (ERT) should be disabled
on 82573 when using jumbo frames.

Cc: Matthew Garret <mjg@redhat.com>
Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ixgbe: Power down PHY during driver resets

The PHY laser is still on during driver init.  It's allowing
garbage to hit our FIFO, which eventually can cause the entire
device to die.  Power down the laser while setting up the device,
and re-enable the laser before getting link.

Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge: Fix build of ipv6 multicast code.

Based upon a report from Stephen Rothwell:

--------------------
net/bridge/br_multicast.c: In function 'br_ip6_multicast_alloc_query':
net/bridge/br_multicast.c:469: error: implicit declaration of function 'csum_ipv6_magic'

Introduced by commit 08b202b6726459626c73ecfa08fcdc8c3efc76c2 ("bridge
br_multicast: IPv6 MLD support") from the net tree.

csum_ipv6_magic is declared in net/ip6_checksum.h ...
--------------------

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

r8169: more broken register writes workaround

78f1cd02457252e1ffbc6caa44a17424a45286b8 ("fix broken register writes")
does not work for Al Viro's r8169 (XID 18000000).

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

r8169: failure to enable mwi should not be fatal

Few (6) network drivers enable mwi explicitly. Fewer worry about a
failure.

It is not a fix but it should avoid some annoyance like
http://bugzilla.kernel.org/show_bug.cgi?id=15454

Signed-off-by: Francois Romieu <romieu@fr.zoreil.com>
Cc: Conrad Kostecki <conikost@gmx.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge br_multicast: Ensure to initialize BR_INPUT_SKB_CB(skb)->mrouters_only.

Even with commit 32dec5dd0233ebffa9cae25ce7ba6daeb7df4467 ("bridge
br_multicast: Don't refer to BR_INPUT_SKB_CB(skb)->mrouters_only
without IGMP snooping."), BR_INPUT_SKB_CB(skb)->mrouters_only is
not appropriately initialized if IGMP/MLD snooping support is
compiled and disabled, so we can see garbage.

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

bridge br_multicast: Ensure to initialize BR_INPUT_SKB_CB(skb)->mrouters_only.

Even with commit 32dec5dd0233ebffa9cae25ce7ba6daeb7df4467 ("bridge
br_multicast: Don't refer to BR_INPUT_SKB_CB(skb)->mrouters_only
without IGMP snooping."), BR_INPUT_SKB_CB(skb)->mrouters_only is
not appropriately initialized if IGMP snooping support is
compiled and disabled, so we can see garbage.

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

ieee802154: Fix oops during ieee802154_sock_ioctl

Trying to run izlisten (from lowpan-tools tests) on a device that does not
exists I got the oops below. The problem is that we are using get_dev_by_name
without checking if we really get a device back. We don't in this case and
writing to dev->type generates this oops.

[Oops code removed by Dmitry Eremin-Solenikov]

If possible this patch should be applied to the current -rc fixes branch.

Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: use sk_sleep()

Commit aa395145 (net: sk_sleep() helper) missed three files in the
conversion.

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

pppoe: use pppoe_pernet instead of directly net_generic

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

phonet: use phonet_pernet instead of directly net_generic

As in for example pppoe introduce phonet_pernet and use it instead of calling
net_generic directly.

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

tg3: Fix INTx fallback when MSI fails

tg3: Fix INTx fallback when MSI fails

MSI setup changes the value of irq_vec in struct tg3 *tp.
This attribute must be taken into account and restored before
we try to do a new request_irq for INTx fallback.

In powerpc, the original code was leading to an EINVAL return within
request_irq, because the driver was trying to use the disabled MSI
virtual irq number instead of tp->pdev->irq.

Signed-off-by: Andre Detsch <adetsch@br.ibm.com>
Acked-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ipv6: Fix inet6_csk_bind_conflict()

Commit fda48a0d7a84 (tcp: bind() fix when many ports are bound)
introduced a bug on IPV6 part.
We should not call ipv6_addr_any(inet6_rcv_saddr(sk2)) but
ipv6_addr_any(inet6_rcv_saddr(sk)) because sk2 can be IPV4, while sk is
IPV6.

Reported-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Tested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

netns: rename unregister_pernet_subsys parameter

Stay consistent with other functions and with comment also and name
pernet_operations parameter properly.

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

rps: optimize rps_get_cpu()

optimize rps_get_cpu().

don't initialize ports when we can get the ports. one memory access
for ports than two.

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

e100: Fix the TX workqueue race

Nothing stops the workqueue being left to run in parallel with close or a
few other operations. This causes double unmaps and the like.

See kerneloops.org #1041230 for an example

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

sky2: add support for receive hashing

Sky2 hardware supports hardware receive hash calculation.
Now that Receive Packet Steering is available, add support
to enable it.

This version does not depend on CONFIG_RPS. Also set_flags rejects
all values except RXHASH, so driver won't have to change next time
somebody adds a new one.

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

Merge branch 'net-next-2.6_20100423a/br/br_multicast_v3' of git://git.linux-ipv6.org/gitroot/yoshfuji/linux-2.6-next

IPv6: Complete IPV6_DONTFRAG support

Finally add support to detect a local IPV6_DONTFRAG event
and return the relevant data to the user if they've enabled
IPV6_RECVPATHMTU on the socket.  The next recvmsg() will
return no data, but have an IPV6_PATHMTU as ancillary data.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

IPv6: Add dontfrag argument to relevant functions

Add dontfrag argument to relevant functions for
IPV6_DONTFRAG support, as well as allowing the value
to be passed-in via ancillary cmsg data.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

IPv6: data structure changes for new socket options

Add underlying data structure changes and basic setsockopt()
and getsockopt() support for IPV6_RECVPATHMTU, IPV6_PATHMTU,
and IPV6_DONTFRAG.  IPV6_PATHMTU is actually fully functional
at this point.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

l2tp_eth: fix memory allocation

Since .size is set properly in "struct pernet_operations l2tp_eth_net_ops",
allocating space for "struct l2tp_eth_net" by hand is not correct, even causes
memory leakage.

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

l2tp: fix memory allocation

Since .size is set properly in "struct pernet_operations l2tp_net_ops",
allocating space for "struct l2tp_net" by hand is not correct, even causes
memory leakage.

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

gianfar: Fix potential oops during OF address translation

gianfar driver may pass NULL pointer to the of_translate_address(),
which may lead to a kernel oops. Fix this by using of_iomap(), which
is also much simpler and shorter.

Signed-off-by: Anton Vorontsov <avorontsov@mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

fsl_pq_mdio: Fix kernel oops during OF address translation

Old P1020RDB device trees were not specifing tbipa address for
MDIO nodes, which is now causing this kernel oops:

 ...
 eth2: TX BD ring size for Q[6]: 256
 eth2: TX BD ring size for Q[7]: 256
 Unable to handle kernel paging request for data at address 0x00000000
 Faulting instruction address: 0xc0015504
 Oops: Kernel access of bad area, sig: 11 [#1]
 ...
 NIP [c0015504] memcpy+0x3c/0x9c
 LR [c000a9f8] __of_translate_address+0xfc/0x21c
 Call Trace:
 [df839e00] [c000a94c] __of_translate_address+0x50/0x21c (unreliable)
 [df839e50] [c01a33e8] get_gfar_tbipa+0xb0/0xe0
 ...

The old device trees are buggy, though having a dead ethernet is
better than a dead kernel, so fix the issue by using of_iomap().

Also, a somewhat similar issue exist in the probe() routine, though
there the oops is only a possibility. Nonetheless, fix it too.

Signed-off-by: Anton Vorontsov <avorontsov@mvista.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Merge branch 'master' into for-davem

Conflicts:
drivers/net/wireless/ath/ath9k/phy.c
drivers/net/wireless/iwlwifi/iwl-6000.c
drivers/net/wireless/iwlwifi/iwl-debugfs.c

bnx2x: add support for receive hashing

Add support to bnx2x to extract Toeplitz hash out of the receive
descriptor for use in skb->rxhash.

Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

bridge br_multicast: IPv6 MLD support.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

bridge br_multicast: Make functions less ipv4 dependent.

Introduce struct br_ip{} to store ip address and protocol
and make functions more generic so that we can support
both IPv4 and IPv6 with less pain.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

ipv6 mcast: Introduce include/net/mld.h for MLD definitions.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>

tcp: bind() fix when many ports are bound

Port autoselection done by kernel only works when number of bound
sockets is under a threshold (typically 30000).

When this threshold is over, we must check if there is a conflict before
exiting first loop in inet_csk_get_port()

Change inet_csk_bind_conflict() to forbid two reuse-enabled sockets to
bind on same (address,port) tuple (with a non ANY address)

Same change for inet6_csk_bind_conflict()

Reported-by: Gaspar Chilingarov <gasparch@gmail.com>
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Acked-by: Evgeniy Polyakov <zbr@ioremap.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

sky2: size status ring based on Tx/Rx ring

Sky2 status ring must be big enough to handle worst case number
of status messages. It was being oversized (to handle dual port cards),
and excessive number of tx ring entries were allowed. This patch reduces
the footprint and makes sure the value is enough.

Later patch to add RSS increases the number of possible Rx status elements.

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

remove DCB_PROTO_VERSION as we don't do netlink versioning

remove DCB_PROTO_VERSION as we don't do netlink versioning

Signed-off-by: Scott Feldman <scofeldm@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: Update X25 interface documentation

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

X25: Use identifiers for hdlc x25 device to x25 interface

Change magic numbers to identifiers for X25 interface.

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Acked-by: Krzysztof Halasa <khc@pm.waw.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: Use identifiers for cyclades device to x25 interface

Change magic numbers to identifiers for X25 interface.

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

X25: Use identifiers for lapbether device to x25 interface

Change magic numbers to identifiers for X25 interface.

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

X25: Use identifiers for x25 async device to x25 interface

Change magic numbers to identifiers for X25 interface.

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

X25: Use identifiers for isdn device to x25 interface

Change magic numbers to identifiers for X25 interface.
also minor check patch formatting.

Signed-off-by: Andrew Hendry <andrew.hendry@gmail.com>
Acked-by: Karsten Keil <isdn@linux-pingi.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25: Add if_x25.h and x25 to device identifiers

V2 Feedback from John Hughes.
- Add header for userspace implementations such as xot/xoe to use
- Use explicit values for interface stability
- No changes to driver patches

V1
- Use identifiers instead of magic numbers for X25 layer 3 to device interface.
- Also fixed checkpatch notes on updated code.

[ Add new user header to include/linux/Kbuild  -DaveM ]

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

dst: rcu check refinement

__sk_dst_get() might be called from softirq, with socket lock held.

[  159.026180] include/net/sock.h:1200 invoked rcu_dereference_check()
without protection!
[  159.026261]
[  159.026261] other info that might help us debug this:
[  159.026263]
[  159.026425]
[  159.026426] rcu_scheduler_active = 1, debug_locks = 0
[  159.026552] 2 locks held by swapper/0:
[  159.026609]  #0:  (&icsk->icsk_retransmit_timer){+.-...}, at:
[<ffffffff8104fc15>] run_timer_softirq+0x105/0x350
[  159.026839]  #1:  (slock-AF_INET){+.-...}, at: [<ffffffff81392b8f>]
tcp_write_timer+0x2f/0x1e0
[  159.027063]
[  159.027064] stack backtrace:
[  159.027172] Pid: 0, comm: swapper Not tainted
2.6.34-rc5-03707-gde498c8-dirty #36
[  159.027252] Call Trace:
[  159.027306]  <IRQ>  [<ffffffff810718ef>] lockdep_rcu_dereference
+0xaf/0xc0
[  159.027411]  [<ffffffff8138e4f7>] tcp_current_mss+0xa7/0xb0
[  159.027537]  [<ffffffff8138fa49>] tcp_write_wakeup+0x89/0x190
[  159.027600]  [<ffffffff81391936>] tcp_send_probe0+0x16/0x100
[  159.027726]  [<ffffffff81392cd9>] tcp_write_timer+0x179/0x1e0
[  159.027790]  [<ffffffff8104fca1>] run_timer_softirq+0x191/0x350
[  159.027980]  [<ffffffff810477ed>] __do_softirq+0xcd/0x200

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

net: Socket filter ancilliary data access for skb->dev->type

Add an SKF_AD_HATYPE field to the packet ancilliary data area, giving
access to skb->dev->type, as reported in the sll_hatype field.

When capturing packets on a PF_PACKET/SOCK_RAW socket bound to all
interfaces, there doesn't appear to be a way for the filter program to
actually find out the underlying hardware type the packet was captured
on. This patch adds such ability.

This patch also handles the case where skb->dev can be NULL, such as on
netlink sockets.

Signed-off-by: Paul Evans <leonerd@leonerd.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: fix outsegs stat for TSO segments

Account for TSO segments of an skb in TCP_MIB_OUTSEGS counter.  Without
doing this, the counter can be off by orders of magnitude from the
actual number of segments sent.

Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rdma: potential ERR_PTR dereference

In the original code, the "goto out" calls "rdma_destroy_id(cm_id);"
That isn't needed here and would cause problems because "cm_id" is an
ERR_PTR.  The new code just returns directly.

Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

rtnetlink: potential ERR_PTR dereference

In the original code, if rtnl_create_link() returned an ERR_PTR then that
would get passed to rtnl_configure_link() which dereferences it.

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

niu: Add skb->rxhash support.

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

qlcnic: update version 5.0.2

Update version to indicate IDC(fw recovery) changes.

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

qlcnic: protect resource access

We do netif_device_attach, even if resource allocation fails.
Driver callbacks can be called, if device is attached.
All these callbacks need to be protected by ADAPTER_UP_MAGIC check.

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

qlcnic: fix rcv buffer leak

Rcv producer value should be read in spin-lock.

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

qlcnic: fix pci semaphore checks

Driver should not go ahead with fw recovery if fails to acquire
semaphore.

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

qlcnic: define macro for driver state

Defining macro to set and clear driver state.

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

qlcnic: fix fw initialization responsibility

Now any pci-func can start fw, whoever sees the reset ack first.
Before this, pci-func which sets the RESET state has the responsibility
to start fw.

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

qlcnic: fix defines as per IDC document

Different class of drivers co-exist for CNA device,
there is some minimal interaction that will be required amongst
the drivers for performing some device level operations.

All the driver should follow inter driver coexistence document.

Fixing polling interval and spelling mistake.

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

qlcnic: additional driver statistics

Added additional driver statistics to track errors in rcv/tx path.

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

IPv6: Generic TTL Security Mechanism (final version)

This patch adds IPv6 support for RFC5082 Generalized TTL Security Mechanism.

Not to users of mapped address; the IPV6 and IPV4 socket options are seperate.
The server does have to deal with both IPv4 and IPv6 socket options
and the client has to handle the different for each family.

On client:
int ttl = 255;
getaddrinfo(argv[1], argv[2], &hint, &result);

for (rp = result; rp != NULL; rp = rp->ai_next) {
s = socket(rp->ai_family, rp->ai_socktype, rp->ai_protocol);
if (s < 0) continue;

if (rp->ai_family == AF_INET) {
setsockopt(s, IPPROTO_IP, IP_TTL, &ttl, sizeof(ttl));
} else if (rp->ai_family == AF_INET6) {
setsockopt(s, IPPROTO_IPV6,  IPV6_UNICAST_HOPS,
&ttl, sizeof(ttl)))
}

if (connect(s, rp->ai_addr, rp->ai_addrlen) == 0) {
   ...

On server:
int minttl = 255 - maxhops;

getaddrinfo(NULL, port, &hints, &result);
for (rp = result; rp != NULL; rp = rp->ai_next) {
s = socket(rp->ai_family, rp->ai_socktype, rp->ai_protocol);
if (s < 0) continue;

if (rp->ai_family == AF_INET6)
setsockopt(s, IPPROTO_IPV6,  IPV6_MINHOPCOUNT,
&minttl, sizeof(minttl));
setsockopt(s, IPPROTO_IP, IP_MINTTL, &minttl, sizeof(minttl));

if (bind(s, rp->ai_addr, rp->ai_addrlen) == 0)
break
...

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

net: Orphan and de-dst skbs earlier in xmit path.

This way GSO packets don't get handled differently.

With help from Eric Dumazet.

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

rps: immediate send IPI in process_backlog()

If some skb are queued to our backlog, we are delaying IPI sending at
the end of net_rx_action(), increasing latencies. This defeats the
queueing, since we want to quickly dispatch packets to the pool of
worker cpus, then eventually deeply process our packets.

It's better to send IPI before processing our packets in upper layers,
from process_backlog().

Change the _and_disable_irq suffix to _and_enable_irq(), since we enable
local irq in net_rps_action(), sorry for the confusion.

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

cxgb4: Make unnecessarily global functions static

Also put t4_write_indirect() inside "#if 0" to avoid a "defined but not
used" compile warning.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cxgb4: Use ntohs() on __be16 value instead of htons()

Use the correct direction of byte-swapping function to fix a mistake
shown by sparse endianness checking -- c.fl0id is __be16.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Acked-by: Dimitris Michailidis <dm@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: ipv6 bind to device issue

The issue raises when having 2 NICs both assigned the same
IPv6 global address.

If a sender binds to a particular NIC (SO_BINDTODEVICE),
the outgoing traffic is being sent via the first found.
The bonded device is thus not taken into an account during the
routing.

From the ip6_route_output function:

If the binding address is multicast, linklocal or loopback,
the RT6_LOOKUP_F_IFACE bit is set, but not for global address.

So binding global address will neglect SO_BINDTODEVICE-binded device,
because the fib6_rule_lookup function path won't check for the
flowi::oif field and take first route that fits.

Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Signed-off-by: Scott Otto <scott.otto@alcatel-lucent.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ethernet: print protocol in host byte order

Eric's recent patch added __force, but this
place would seem to require actually doing
a byte order conversion so the printk is
consistent across architectures.

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

net: Introduce skb_orphan_try()

At this point, skb->destructor is not the original one (stored in
DEV_GSO_CB(skb)->destructor)

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

ipv6: allow to send packet after receiving ICMPv6 Too Big message with MTU field less than IPV6_MIN_MTU

According to RFC2460, PMTU is set to the IPv6 Minimum Link
MTU (1280) and a fragment header should always be included
after a node receiving Too Big message reporting PMTU is
less than the IPv6 Minimum Link MTU.

After receiving a ICMPv6 Too Big message reporting PMTU is
less than the IPv6 Minimum Link MTU, sctp *can't* send any
data/control chunk that total length including IPv6 head
and IPv6 extend head is less than IPV6_MIN_MTU(1280 bytes).

The failure occured in p6_fragment(), about reason
see following(take SHUTDOWN chunk for example):
sctp_packet_transmit (SHUTDOWN chunk, len=16 byte)
|------sctp_v6_xmit (local_df=0)
   |------ip6_xmit
       |------ip6_output (dst_allfrag is ture)
           |------ip6_fragment

In ip6_fragment(), for local_df=0, drops the the packet
and returns EMSGSIZE.

The patch fixes it with adding check length of skb->len.
In this case, Ipv6 not to fragment upper protocol data,
just only add a fragment header before it.

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

drivers/net/usb: Add new driver ipheth

Add new driver to use tethering with an iPhone device. After initial submission,
apply fixes to fit the new driver into the kernel standards.

There are still a couple of minor (almost cosmetic-level) issues, but the driver
is fully functional right now.

Signed-off-by: L. Alberto Giménez <agimenez@sysvalve.es>
Signed-off-by: Diego Giagio <diego@giagio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ehea: fix possible DLPAR/mem deadlock

Force serialization of userspace-triggered DLPAR/mem operations

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ehea: error handling improvement

Reset a port's resources only if they're actually in an error state

Signed-off-by: Thomas Klein <tklein@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

cxgb3: fix linkup issue

I encountered an issue that not to link up on cxgb3 fabric.
I bisected and found that this regression was introduced by
0f07c4ee8c800923ae7918c231532a9256233eed.

Correct to pass phy_addr to cphy_init() at t3_xaui_direct_phy_prep().

Signed-off-by: Hiroshi Shimamoto <h-shimamoto@ct.jp.nec.com>
Acked-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

ks8842: Add platform data for setting mac address

This patch adds platform data to the ks8842 driver.

Via the platform data a MAC address, to be used by the controller,
can be passed.

To ensure this MAC address is used, the MAC address is written
after each hardware reset.

Signed-off-by: Richard Röjfors <richard.rojfors@pelagicore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

X25 fix dead unaccepted sockets

1, An X25 program binds and listens
2, calls arrive waiting to be accepted
3, Program exits without accepting
4, Sockets time out but don't get correctly cleaned up
5, cat /proc/net/x25/socket shows the dead sockets with bad inode fields.

This line borrowed from AX25 sets the dying socket so the timers clean up later.

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

KS8851: NULL pointer dereference if list is empty

Fix NULL pointer dereference in ks8851_tx_work by checking if dequeued
list is already empty before writing the packet to TX FIFO

 Unable to handle kernel NULL pointer dereference at virtual address 00000050
 PC is at ks8851_tx_work+0xdc/0x1b0
 LR is at wait_for_common+0x148/0x164
 pc : [<c01c0df4>]    lr : [<c025a980>]    psr: 20000013
 Backtrace:
  ks8851_tx_work+0x0/0x1b0
  worker_thread+0x0/0x190
  kthread+0x0/0x90

Signed-off-by: Abraham Arce <x0066660@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: 3c574_cs fix stats.tx_bytes counter

Update the stats counter calculation in 3c574_cs, similar
to the method used in 3c589_cs. This corrects the contents
of the counter on tests using a "Megahertz 574B" card.

[linux@dominikbrodowski.net: clean up commit message]
Signed-off-by: Alexander Kurz <linux@kbdbabel.org>
Signed-off-by: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: David S. Miller <davem@davemloft.net>

xfrm6: ensure to use the same dev when building a bundle

When building a bundle, we set dst.dev and rt6.rt6i_idev.
We must ensure to set the same device for both fields.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: small cleanup of lib8390

Remove the always true #if 1. Also the unecessary re-test of ei_local->irqlock
and the unreachable printk format string.

Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
Signed-off-by: David S. Miller <davem@davemloft.net>

fasync: RCU and fine grained locking

kill_fasync() uses a central rwlock, candidate for RCU conversion, to
avoid cache line ping pongs on SMP.

fasync_remove_entry() and fasync_add_entry() can disable IRQS on a short
section instead during whole list scan.

Use a spinlock per fasync_struct to synchronize kill_fasync_rcu() and
fasync_{remove|add}_entry(). This spinlock is IRQ safe, so sock_fasync()
doesnt need its own implementation and can use fasync_helper(), to
reduce code size and complexity.

We can remove __kill_fasync() direct use in net/socket.c, and rename it
to kill_fasync_rcu().

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

can: Fix possible NULL pointer dereference in ems_usb.c

In ems_usb_probe(), a pointer is dereferenced after making sure it is NULL...

This patch replaces netdev->dev.parent with &intf->dev in dev_err() calls to
avoid this.

Signed-off-by: "Hans J. Koch" <hjk@linutronix.de>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

tcp: Mark v6 response packets as CHECKSUM_PARTIAL

Otherwise we only get the checksum right for data-less TCP responses.

Noticed by Herbert Xu.

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

radiotap parser: fix endian annotation

When I updated this from the corresponding
userspace library, an annotation error crept
in -- this variable needs to be annotated as
little endian. No effect on code generation.

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

ath9k_hw: make all AR9002 initvals use u32

Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

ath9k_hw: make two initvals consto for the AR9001 family

This makes ar5416Addac_9160 and ar5416Addac_9160 const
I guess we skipped them long ago.

Signed-off-by: Luis R. Rodriguez <lrodriguez@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>

tcp: Fix ipv6 checksumming on response packets for real.

Commit 6651ffc8e8bdd5fb4b7d1867c6cfebb4f309512c
("ipv6: Fix tcp_v6_send_response transport header setting.")
fixed one half of why ipv6 tcp response checksums were
invalid, but it's not the whole story.

If we're going to use CHECKSUM_PARTIAL for these things (which we are
since commit 2e8e18ef52e7dd1af0a3bd1f7d990a1d0b249586 "tcp: Set
CHECKSUM_UNNECESSARY in tcp_init_nondata_skb"), we can't be setting
buff->csum as we always have been here in tcp_v6_send_response.  We
need to leave it at zero.

Kill that line and checksums are good again.

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

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

Conflicts:
drivers/net/wireless/iwlwifi/iwl-6000.c
net/core/dev.c

net: Fix an RCU warning in dev_pick_tx()

Fix the following RCU warning in dev_pick_tx():

===================================================
[ INFO: suspicious rcu_dereference_check() usage. ]
---------------------------------------------------
net/core/dev.c:1993 invoked rcu_dereference_check() without protection!

other info that might help us debug this:

rcu_scheduler_active = 1, debug_locks = 0
2 locks held by swapper/0:
 #0:  (&idev->mc_ifc_timer){+.-...}, at: [<ffffffff81039e65>] run_timer_softirq+0x17b/0x278
 #1:  (rcu_read_lock_bh){.+....}, at: [<ffffffff812ea3eb>] dev_queue_xmit+0x14e/0x4dc

stack backtrace:
Pid: 0, comm: swapper Not tainted 2.6.34-rc5-cachefs #4
Call Trace:
 <IRQ>  [<ffffffff810516c4>] lockdep_rcu_dereference+0xaa/0xb2
 [<ffffffff812ea4f6>] dev_queue_xmit+0x259/0x4dc
 [<ffffffff812ea3eb>] ? dev_queue_xmit+0x14e/0x4dc
 [<ffffffff81052324>] ? trace_hardirqs_on+0xd/0xf
 [<ffffffff81035362>] ? local_bh_enable_ip+0xbc/0xc1
 [<ffffffff812f0954>] neigh_resolve_output+0x24b/0x27c
 [<ffffffff8134f673>] ip6_output_finish+0x7c/0xb4
 [<ffffffff81350c34>] ip6_output2+0x256/0x261
 [<ffffffff81052324>] ? trace_hardirqs_on+0xd/0xf
 [<ffffffff813517fb>] ip6_output+0xbbc/0xbcb
 [<ffffffff8135bc5d>] ? fib6_force_start_gc+0x2b/0x2d
 [<ffffffff81368acb>] mld_sendpack+0x273/0x39d
 [<ffffffff81368858>] ? mld_sendpack+0x0/0x39d
 [<ffffffff81052099>] ? mark_held_locks+0x52/0x70
 [<ffffffff813692fc>] mld_ifc_timer_expire+0x24f/0x288
 [<ffffffff81039ed6>] run_timer_softirq+0x1ec/0x278
 [<ffffffff81039e65>] ? run_timer_softirq+0x17b/0x278
 [<ffffffff813690ad>] ? mld_ifc_timer_expire+0x0/0x288
 [<ffffffff81035531>] ? __do_softirq+0x69/0x140
 [<ffffffff8103556a>] __do_softirq+0xa2/0x140
 [<ffffffff81002e0c>] call_softirq+0x1c/0x28
 [<ffffffff81004b54>] do_softirq+0x38/0x80
 [<ffffffff81034f06>] irq_exit+0x45/0x47
 [<ffffffff810177c3>] smp_apic_timer_interrupt+0x88/0x96
 [<ffffffff810028d3>] apic_timer_interrupt+0x13/0x20
 <EOI>  [<ffffffff810488dd>] ? __atomic_notifier_call_chain+0x0/0x86
 [<ffffffff810096bf>] ? mwait_idle+0x6e/0x78
 [<ffffffff810096b6>] ? mwait_idle+0x65/0x78
 [<ffffffff810011cb>] cpu_idle+0x4d/0x83
 [<ffffffff81380b05>] rest_init+0xb9/0xc0
 [<ffffffff81380a4c>] ? rest_init+0x0/0xc0
 [<ffffffff8168dcf0>] start_kernel+0x392/0x39d
 [<ffffffff8168d2a3>] x86_64_start_reservations+0xb3/0xb7
 [<ffffffff8168d38b>] x86_64_start_kernel+0xe4/0xeb

An rcu_dereference() should be an rcu_dereference_bh().

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

Merge branch 'master' of /home/davem/src/GIT/linux-2.6/

ipv6: Fix tcp_v6_send_response transport header setting.

My recent patch to remove the open-coded checksum sequence in
tcp_v6_send_response broke it as we did not set the transport
header pointer on the new packet.

Actually, there is code there trying to set the transport
header properly, but it sets it for the wrong skb ('skb'
instead of 'buff').

This bug was introduced by commit
a8fdf2b331b38d61fb5f11f3aec4a4f9fb2dedcb ("ipv6: Fix
tcp_v6_send_response(): it didn't set skb transport header")

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Remove two unnecessary exports (skbuff).

There is no need to export skb_under_panic() and skb_over_panic() in
skbuff.c, since these methods are used only in skbuff.c ; this patch
removes these two exports. It also marks these functions as 'static'
and removeS the extern declarations of them from
include/linux/skbuff.h

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

net: Fix various endianness glitches

Sparse can help us find endianness bugs, but we need to make some
cleanups to be able to more easily spot real bugs.

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