platform/kernel/linux-3.10.git
13 years agoe1000e: remove redundant reverse dependency on CRC32
Bruce Allan [Thu, 26 May 2011 05:24:38 +0000 (05:24 +0000)]
e1000e: remove redundant reverse dependency on CRC32

Commit 5d03078a6804bf4c7f943c5b68bef80468c0717f added a redundant 'select
CRC32'; remove it.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: Clear host wakeup bit on 82577/8 without touching PHY page 800
Bruce Allan [Fri, 13 May 2011 07:20:14 +0000 (07:20 +0000)]
e1000e: Clear host wakeup bit on 82577/8 without touching PHY page 800

The Host Wakeup Active bit in the PHY Port General Configuration register
(page 769 register 17) must be cleared after every PHY reset to prevent an
unexpected wake signal from the PHY. Originally, this was accomplished by
simply reading the PHY Wakeup Control register on page 800 which clears the
Host Wakeup Active bit as a side-effect. Unfortunately, a hardware bug on
the 82577 and 82578 PHY can cause unexpected behavior when registers on
page 800 are accessed while in gigabit mode.

This patch changes the remaining instances when the Host Wakeup Active bit
needs to be cleared while possibly in gigabit mode by accessing the Port
General Configuration register directly instead of accessing any register
on page 800.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: access multiple PHY registers on same page at the same time
Bruce Allan [Fri, 13 May 2011 07:20:09 +0000 (07:20 +0000)]
e1000e: access multiple PHY registers on same page at the same time

Doing a PHY page select can take a long time, relatively speaking. This
can cause a significant delay when updating a number of PHY registers on
the same page by unnecessarily setting the page for each PHY access. For
example when going to Sx, all the PHY wakeup registers (WUC, RAR[], MTA[],
SHRAR[], IP4AT[], IP6AT[], etc.) on 82577/8/9 need to be updated which
takes a long time which can cause issues when suspending.

This patch introduces new PHY ops function pointers to allow callers to
set the page directly and do any number of PHY accesses on that page.
This feature is currently only implemented for 82577, 82578 and 82579
PHYs for both the normally addressed registers as well as the special-
case addressing of the PHY wakeup registers on page 800. For the latter
registers, the existing function for accessing the wakeup registers has
been divided up into three- 1) enable access to the wakeup register page,
2) perform the register access and 3) disable access to the wakeup register
page. The two functions that enable/disable access to the wakeup register
page are necessarily available to the caller so that the caller can restore
the value of the Port Control (a.k.a. Wakeup Enable) register after the
wakeup register accesses are done.

All instances of writing to multiple PHY registers on the same page are
updated to use this new method and to acquire any PHY locking mechanism
before setting the page and performing the register accesses, and release
the locking mechanism afterward.

Some affiliated magic number cleanup is done as well.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: do not schedule the Tx queue until ready
Bruce Allan [Fri, 13 May 2011 07:20:03 +0000 (07:20 +0000)]
e1000e: do not schedule the Tx queue until ready

Start the Tx queue when the interface is brought up in e1000e_up() but do
not schedule the queue until link is up as detected in the watchdog task
which sets netif_carrier_on.

Also flush the descriptors and clean the Tx and Rx rings before resetting
the hardware when bringing the interface down otherwise there is a small
window where the watchdog task can be triggered with netif_carrier_off
and the Tx ring not yet empty which causes an additional and unnecessary
reset.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: log when swflag is cleared unexpectedly on ICH/PCH devices
Bruce Allan [Fri, 13 May 2011 07:19:53 +0000 (07:19 +0000)]
e1000e: log when swflag is cleared unexpectedly on ICH/PCH devices

Since EXTCNF_CTRL.SWFLAG (used in the ownership arbitration of shared
resources, e.g. the PHY shared between the s/w, f/w, and h/w clients)
can be cleared by any of those clients, log a debug message when
software attempts to clear it and it is already cleared unexpectedly.
And since the swflag is cleared by a hardware reset, the driver does
not need to do that, but the mutex acquired when the bit is set must
still be cleared.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: 82579 intermittently disabled during S0->Sx
Bruce Allan [Fri, 13 May 2011 07:19:48 +0000 (07:19 +0000)]
e1000e: 82579 intermittently disabled during S0->Sx

When repeatedly cycling Sx->S0 states with the network cable unplugged,
the 82579 PHY may not initialize as expected and may require a full power
cycle to recover functionality to the device.  Workaround this by testing
access of the PHY registers after resuming; if that returns unexpected
results toggle the LANPHYPC signal to power cycle the PHY.

This is implemented in the new function e1000_resume_workarounds_pchlan()
which calls another new function, e1000_toggle_lanphypc_value_ich8lan(),
which has been created to reduce code duplication (same functionality
required by a previous workaround).  Also, e1000e_disable_gig_wol_ich8lan
is now e1000_suspend_workarounds_ich8lan to better reflect what it does.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agoe1000e: disable far-end loopback mode on ESB2
Bruce Allan [Fri, 13 May 2011 07:19:42 +0000 (07:19 +0000)]
e1000e: disable far-end loopback mode on ESB2

The ESB2 LAN includes a debug feature that enables far-end loopback (FELB)
of the SerDes/Kumeran interface.  This feature is activated when receiving
a sequence of symbols that includes a reserved codeword.  On a perfect
link, FELB would never be activated.  In the presence of bit errors, there
is a very small, but non-zero, probability of FELB being activated.

If the FELB is activated, the SerDes link becomes non-functional and must
be reset.  It could also corrupt the switching tables in the switch since
the ESB2 is transmitting packets with a different source MAC address.

This patch disables the FELB feature.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
13 years agobonding: delete unused arp_mon_pt
Peter Pan(潘卫平) [Wed, 8 Jun 2011 21:19:05 +0000 (21:19 +0000)]
bonding: delete unused arp_mon_pt

Now all received packets are handled by bond_handle_frame,
and arp_mon_pt isn't used any more.

Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Weiping Pan <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding: delete unused ad_timer
Peter Pan(潘卫平) [Wed, 8 Jun 2011 21:19:04 +0000 (21:19 +0000)]
bonding: delete unused ad_timer

Now we use agg_select_timer and ad_work.

Reviewed-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Weiping Pan <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding:delete agg_select_mode from ad_bond_info
Peter Pan(潘卫平) [Wed, 8 Jun 2011 21:19:03 +0000 (21:19 +0000)]
bonding:delete agg_select_mode from ad_bond_info

bond_params->ad_select and ad_bond_info->agg_select_mode have the same
meaning, they are duplicate and need extra synchronization.

__get_agg_selection_mode() get ad_select from bond_params directly.

Signed-off-by: Weiping Pan <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding:delete lacp_fast from ad_bond_info
Peter Pan(潘卫平) [Wed, 8 Jun 2011 21:19:02 +0000 (21:19 +0000)]
bonding:delete lacp_fast from ad_bond_info

These is also a bug, that if you modify lacp_rate via sysfs,
and add new slaves in bonding, new slaves won't use the latest lacp_rate,
since ad_bond_info->lacp_fast is initialized only once,
in bond_3ad_initialize().

Since both struct bond_params and ad_bond_info have lacp_fast,
they are duplicate and need extra synchronization.

bond_3ad_bind_slave() can use bond_params->lacp_fast to initialize port.
So we can just remove lacp_fast from struct ad_bond_info.

Signed-off-by: Weiping Pan <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding: make 802.3ad use latest lacp_rate
Peter Pan(潘卫平) [Wed, 8 Jun 2011 21:19:01 +0000 (21:19 +0000)]
bonding: make 802.3ad use latest lacp_rate

There is bug that when you modify lacp_rate via sysfs,
802.3ad won't use the new value of lacp_rate to transmit packets.
This is because port->actor_oper_port_state isn't changed.

Signed-off-by: Weiping Pan <panweiping3@gmail.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: Bug Fix: Fix hardware transmit queue indexing in enic_poll_controller
Vasanthy Kolluri [Thu, 9 Jun 2011 10:37:07 +0000 (10:37 +0000)]
enic: Bug Fix: Fix hardware transmit queue indexing in enic_poll_controller

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: Get/Set interrupt resource index for transmit and receive queues
Vasanthy Kolluri [Thu, 9 Jun 2011 10:37:02 +0000 (10:37 +0000)]
enic: Get/Set interrupt resource index for transmit and receive queues

Instead of deriving the index of a transmit/receive interrupt resource
from the transmit/receive queue index, always save and retrieve it
using an additional variable.

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: Log device configuration in detail during driver load
Vasanthy Kolluri [Thu, 9 Jun 2011 10:36:57 +0000 (10:36 +0000)]
enic: Log device configuration in detail during driver load

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: Pass 802.1p bits for packets tagged with vlan zero
Vasanthy Kolluri [Thu, 9 Jun 2011 10:36:52 +0000 (10:36 +0000)]
enic: Pass 802.1p bits for packets tagged with vlan zero

enic driver currently passes 802.1p bits to the upper layers for packets
tagged with non-zero vlan ids only. This patch extends such behaviour to
zero vlan tagged packets also.

The patch is dependant on the following kernel patches:

1) vlan_dev: VLAN 0 should be treated as "no vlan tag" (802.1p packet)
   - net-next-2.6 git commit: ad1afb00393915a51c21b1ae8704562bf036855f
   - Available 2.6.36 and later
2) vlan: Centralize handling of hardware acceleration.
  - net-next-2.6 git commit: 3701e51382a026cba10c60b03efabe534fba4ca4
  - Available 2.6.37 and later

Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: Danny Guo <dannguo@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'batman-adv/next' of git://git.open-mesh.org/ecsv/linux-merge
David S. Miller [Thu, 9 Jun 2011 21:56:13 +0000 (14:56 -0700)]
Merge branch 'batman-adv/next' of git://git.open-mesh.org/ecsv/linux-merge

13 years agobatman-adv: use NO_FLAGS define instead of hard-coding 0
Marek Lindner [Thu, 9 Jun 2011 15:13:09 +0000 (17:13 +0200)]
batman-adv: use NO_FLAGS define instead of hard-coding 0

The definition NO_FLAGS was introduced to make the code more
readable and shall be used to initialize flag fields.

Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agobatman-adv: Use enums for related constants
Sven Eckelmann [Sat, 4 Jun 2011 09:26:00 +0000 (11:26 +0200)]
batman-adv: Use enums for related constants

CodingStyle "Chapter 12: Macros, Enums and RTL" recommends to use enums
for several related constants. Internal states can be used without
defining the actual value, but all values which are visible to the
outside must be defined as before. Normal values are assigned as usual
and flags are defined by shifts of a bit.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agobatman-adv: Rewrite debugfs kobj_to_* helpers as functions
Sven Eckelmann [Sun, 5 Jun 2011 08:20:19 +0000 (10:20 +0200)]
batman-adv: Rewrite debugfs kobj_to_* helpers as functions

CodingStyle "Chapter 12: Macros, Enums and RTL" highly recommends to use
functions instead of macros were possible. This ensures type safety and
prevents shadowing of other variables.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agobatman-adv: Fix signedness problem in parse_gw_bandwidth
Sven Eckelmann [Sat, 4 Jun 2011 12:00:49 +0000 (14:00 +0200)]
batman-adv: Fix signedness problem in parse_gw_bandwidth

strict_strtoul as used in parse_gw_bandwidth is defined for unsigned
long and strict_strtol should be used instead for long.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agobatman-adv: Don't return value in void function
Sven Eckelmann [Sat, 4 Jun 2011 10:40:37 +0000 (12:40 +0200)]
batman-adv: Don't return value in void function

gw_node_delete is defined with "void" as return type, but still tries to
return a value. The called function gw_node_delete is also return as
void and thus doesn't provide a value for us.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agobatman-adv: accept delayed rebroadcasts to avoid bogus routing under heavy load
Daniele Furlan [Mon, 6 Jun 2011 22:45:55 +0000 (00:45 +0200)]
batman-adv: accept delayed rebroadcasts to avoid bogus routing under heavy load

When a link is saturated (re)broadcasts of OGMs are delayed. Under heavy
load this delay may exceed the orig interval which leads to OGMs being
dropped (the code would only accept an OGM rebroadcast if it arrived
before the next OGM was broadcasted). With this patch batman-adv will
also accept delayed OGMs in order to avoid a bogus influence on the
routing metric.

Signed-off-by: Daniele Furlan <daniele.furlan@gmail.com>
Signed-off-by: Sven Eckelmann <sven@narfation.org>
13 years agonet: emaclite: Add missing linux/interrupt.h header
Michal Simek [Thu, 9 Jun 2011 08:29:13 +0000 (01:29 -0700)]
net: emaclite: Add missing linux/interrupt.h header

Log:
drivers/net/xilinx_emaclite.c: In function 'xemaclite_open':
drivers/net/xilinx_emaclite.c:961: error: implicit declaration of function 'request_irq'
drivers/net/xilinx_emaclite.c: In function 'xemaclite_close':
drivers/net/xilinx_emaclite.c:995: error: implicit declaration of function 'free_irq'
make[2]: *** [drivers/net/xilinx_emaclite.o] Error 1
make[1]: *** [drivers/net] Error 2
make: *** [drivers] Error 2

Signed-off-by: Michal Simek <monstr@monstr.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotun: do not put self in waitq if doing a nonblock read
Amos Kong [Thu, 9 Jun 2011 07:27:10 +0000 (00:27 -0700)]
tun: do not put self in waitq if doing a nonblock read

Perf shows a relatively high rate (about 8%) race in
spin_lock_irqsave() when doing netperf between external host and
guest. It's mainly becuase the lock contention between the
tun_do_read() and tun_xmit_skb(), so this patch do not put self into
waitqueue to reduce this kind of race. After this patch, it drops to
4%.

Signed-off-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Amos Kong <akong@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotun: dont force inline of functions
stephen hemminger [Wed, 8 Jun 2011 14:33:08 +0000 (14:33 +0000)]
tun: dont force inline of functions

Current standard practice is to not mark most functions as inline
and  let compiler decide instead.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotun: reserves space for network in skb
stephen hemminger [Wed, 8 Jun 2011 14:33:07 +0000 (14:33 +0000)]
tun: reserves space for network in skb

The tun driver allocates skb's to hold data from user and then passes
the data into the network stack as received data. Most network devices
allocate the receive skb with routines like dev_alloc_skb() that reserves
additional space for use by network protocol stack but tun does not.

Because of the lack of padding, when the packet is passed through bridge
netfilter a new skb has to be allocated.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agocnic, bnx2: Check iSCSI support early in bnx2_init_one()
Michael Chan [Wed, 8 Jun 2011 19:29:36 +0000 (19:29 +0000)]
cnic, bnx2: Check iSCSI support early in bnx2_init_one()

Based on earlier patch from Neil Horman <nhorman@tuxdriver.com>

If iSCSI is not supported on a bnx2 device, bnx2_cnic_probe() will
return NULL and the cnic device will not be visible to bnx2i.  This
will prevent bnx2i from registering and then unregistering during
cnic_start() and cause the warning message:

bnx2 0003:01:00.1: eth1: Failed waiting for ULP up call to complete

Signed-off-by: Michael Chan <mchan@broadcom.com>
Cc: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agocnic: Improve NETDEV_UP event handling
Michael Chan [Wed, 8 Jun 2011 19:29:35 +0000 (19:29 +0000)]
cnic: Improve NETDEV_UP event handling

During NETDEV_UP, we use symbol_get() to get the net driver's cnic
probe function.  This sometimes doesn't work if NETDEV_UP happens
right after NETDEV_REGISTER and the net driver is still running module
init code.  As a result, the cnic device may not be discovered.  We
fix this by probing on all NETDEV events if the device's netif_running
state is up.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agocnic: Randomize initial TCP port for iSCSI connections
Eddie Wai [Wed, 8 Jun 2011 19:29:34 +0000 (19:29 +0000)]
cnic: Randomize initial TCP port for iSCSI connections

This reduces the likelihood of port re-use when re-loading the driver.

Signed-off-by: Eddie Wai <eddie.wai@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agocnic: Fix race conditions with firmware
Michael Chan [Wed, 8 Jun 2011 19:29:33 +0000 (19:29 +0000)]
cnic: Fix race conditions with firmware

During iSCSI connection terminations, if the target is also terminating
at about the same time, the firmware may not complete the driver's
request to close or reset the connection.  This is fixed by handling
other events (instead of the expected completion event) as an indication
that the driver's request has been rejected.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agocnic: Fix interrupt logic
Michael Chan [Wed, 8 Jun 2011 19:29:32 +0000 (19:29 +0000)]
cnic: Fix interrupt logic

We need to keep looping until cnic_get_kcqes() returns 0.  cnic_get_kcqes()
returns a maximum of 64 entries.  If there are more entries in the queue
and we don't loop back, the remaining entries may not be serviced for a
long time.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoinetpeer: lower false sharing effect
Eric Dumazet [Thu, 9 Jun 2011 06:31:27 +0000 (23:31 -0700)]
inetpeer: lower false sharing effect

Profiles show false sharing in addr_compare() because refcnt/dtime
changes dirty the first inet_peer cache line, where are lying the keys
used at lookup time. If many cpus are calling inet_getpeer() and
inet_putpeer(), or need frag ids, addr_compare() is in 2nd position in
"perf top".

Before patch, my udpflood bench (16 threads) on my 2x4x2 machine :

             5784.00  9.7% csum_partial_copy_generic [kernel]
             3356.00  5.6% addr_compare              [kernel]
             2638.00  4.4% fib_table_lookup          [kernel]
             2625.00  4.4% ip_fragment               [kernel]
             1934.00  3.2% neigh_lookup              [kernel]
             1617.00  2.7% udp_sendmsg               [kernel]
             1608.00  2.7% __ip_route_output_key     [kernel]
             1480.00  2.5% __ip_append_data          [kernel]
             1396.00  2.3% kfree                     [kernel]
             1195.00  2.0% kmem_cache_free           [kernel]
             1157.00  1.9% inet_getpeer              [kernel]
             1121.00  1.9% neigh_resolve_output      [kernel]
             1012.00  1.7% dev_queue_xmit            [kernel]
# time ./udpflood.sh

real 0m44.511s
user 0m20.020s
sys 11m22.780s

# time ./udpflood.sh

real 0m44.099s
user 0m20.140s
sys 11m15.870s

After patch, no more addr_compare() in profiles :

             4171.00 10.7% csum_partial_copy_generic   [kernel]
             1787.00  4.6% fib_table_lookup            [kernel]
             1756.00  4.5% ip_fragment                 [kernel]
             1234.00  3.2% udp_sendmsg                 [kernel]
             1191.00  3.0% neigh_lookup                [kernel]
             1118.00  2.9% __ip_append_data            [kernel]
             1022.00  2.6% kfree                       [kernel]
              993.00  2.5% __ip_route_output_key       [kernel]
              841.00  2.2% neigh_resolve_output        [kernel]
              816.00  2.1% kmem_cache_free             [kernel]
              658.00  1.7% ia32_sysenter_target        [kernel]
              632.00  1.6% kmem_cache_alloc_node       [kernel]

# time ./udpflood.sh

real 0m41.587s
user 0m19.190s
sys 10m36.370s

# time ./udpflood.sh

real 0m41.486s
user 0m19.290s
sys 10m33.650s

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoniu: support 64 bit stats interface
stephen hemminger [Wed, 8 Jun 2011 14:54:04 +0000 (14:54 +0000)]
niu: support 64 bit stats interface

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agomyricom: update to 64 bit stats
stephen hemminger [Wed, 8 Jun 2011 14:54:03 +0000 (14:54 +0000)]
myricom: update to 64 bit stats

Driver was already keeping 64 bit counters, just not using the new interface.

Ps: IMHO drivers should not be duplicating network device
stats into ethtool stats. It is useless duplication.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: update to support 64 bit stats
stephen hemminger [Wed, 8 Jun 2011 14:54:02 +0000 (14:54 +0000)]
enic: update to support 64 bit stats

The device driver already uses 64 bit statistics, it just
doesn't use the 64 bit interface.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: convert to 64 bit statistics
stephen hemminger [Wed, 8 Jun 2011 14:54:01 +0000 (14:54 +0000)]
netxen: convert to 64 bit statistics

Change to 64 bit statistics interface, driver was already maintaining 64 bit
value.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Acked-by: Amit Kumar Salecha <amit.salecha@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoveth: convert to 64 bit statistics
stephen hemminger [Wed, 8 Jun 2011 14:53:59 +0000 (14:53 +0000)]
veth: convert to 64 bit statistics

Not much change, device was already keeping per cpu statistics.
Use recent 64 statistics interface.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agovmxnet3: convert to 64 bit stats interface
stephen hemminger [Wed, 8 Jun 2011 14:53:57 +0000 (14:53 +0000)]
vmxnet3: convert to 64 bit stats interface

Convert vmxnet3 driver to 64 bit statistics interface.
This driver was already counting packet per queue in a 64 bit value so not
a huge change. Eliminate unused old net_device_stats structure.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Scott J. Goldman <scottjg@vmware.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoinetpeer: remove unused list
Eric Dumazet [Wed, 8 Jun 2011 13:35:34 +0000 (13:35 +0000)]
inetpeer: remove unused list

Andi Kleen and Tim Chen reported huge contention on inetpeer
unused_peers.lock, on memcached workload on a 40 core machine, with
disabled route cache.

It appears we constantly flip peers refcnt between 0 and 1 values, and
we must insert/remove peers from unused_peers.list, holding a contended
spinlock.

Remove this list completely and perform a garbage collection on-the-fly,
at lookup time, using the expired nodes we met during the tree
traversal.

This removes a lot of code, makes locking more standard, and obsoletes
two sysctls (inet_peer_gc_mintime and inet_peer_gc_maxtime). This also
removes two pointers in inet_peer structure.

There is still a false sharing effect because refcnt is in first cache
line of object [were the links and keys used by lookups are located], we
might move it at the end of inet_peer structure to let this first cache
line mostly read by cpus.

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Andi Kleen <andi@firstfloor.org>
CC: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotcp: RFC2988bis + taking RTT sample from 3WHS for the passive open side
Jerry Chu [Wed, 8 Jun 2011 11:08:38 +0000 (11:08 +0000)]
tcp: RFC2988bis + taking RTT sample from 3WHS for the passive open side

This patch lowers the default initRTO from 3secs to 1sec per
RFC2988bis. It falls back to 3secs if the SYN or SYN-ACK packet
has been retransmitted, AND the TCP timestamp option is not on.

It also adds support to take RTT sample during 3WHS on the passive
open side, just like its active open counterpart, and uses it, if
valid, to seed the initRTO for the data transmission phase.

The patch also resets ssthresh to its initial default at the
beginning of the data transmission phase, and reduces cwnd to 1 if
there has been MORE THAN ONE retransmission during 3WHS per RFC5681.

Signed-off-by: H.K. Jerry Chu <hkchu@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoipv6: generate link local address for GRE tunnel
stephen hemminger [Wed, 8 Jun 2011 10:44:30 +0000 (10:44 +0000)]
ipv6: generate link local address for GRE tunnel

Use same logic as SIT tunnel to handle link local address
for GRE tunnel. OSPFv3 requires link-local address to function.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoiph: use default get_stats
stephen hemminger [Wed, 8 Jun 2011 08:09:13 +0000 (08:09 +0000)]
iph: use default get_stats

This driver keeps stats in net_device stats therefore it
does not need to define it's own get_stats hook.

Also, use standard format for net_device_ops (without &).

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agov2 ethtool: remove support for ETHTOOL_GRXNTUPLE
Alexander Duyck [Wed, 8 Jun 2011 12:35:08 +0000 (12:35 +0000)]
v2 ethtool: remove support for ETHTOOL_GRXNTUPLE

This change is meant to remove all support for displaying an ntuple as
strings via ETHTOOL_GRXNTUPLE.  The reason for this change is due to the
fact that multiple issues have been found including:
 - Multiple buffer overruns for strings being displayed.
 - Incorrect filters displayed, cleared filters with ring of -2 are displayed
 - Setting get_rx_ntuple displays no rules if defined.
 - Endianess wrong on displayed values.
 - Hard limit of 1024 filters makes display functionality extremely limited

The only driver that had supported this interface was ixgbe.  Since it no
longer uses the interface and due to the issues mentioned above I am
submitting this patch to remove it.

v2:
Updated based on comments from Ben Hutchings
 - Left ETH_SS_NTUPLE_FILTERS in code but commented on it being deprecated
 - Removed ethtool_rx_ntuple_list and ethtool_rx_ntuple_flow_spec_container
 - Left ETHTOOL_GRXNTUPLE but commented it as deprecated

Also cleaned up set_rx_ntuple since there is no flow spec container to
maintain we can drop all the code for the alloc and free of it and just
return ops->set_rx_ntuple().
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Wed, 8 Jun 2011 17:44:21 +0000 (13:44 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next-2.6 into for-davem

13 years agonet: add needed interrupt.h
Stephen Rothwell [Wed, 8 Jun 2011 05:49:33 +0000 (15:49 +1000)]
net: add needed interrupt.h

Fixes these errors after the removal of interrupt.h from netdevice.h:

drivers/net/ll_temac_main.c: In function 'temac_open':
drivers/net/ll_temac_main.c:859:2: error: implicit declaration of function 'request_irq'
drivers/net/ll_temac_main.c:870:2: error: implicit declaration of function 'free_irq'
drivers/net/ll_temac_main.c: In function 'temac_poll_controller':
drivers/net/ll_temac_main.c:903:2: error: implicit declaration of function 'disable_irq'
drivers/net/ll_temac_main.c:909:2: error: implicit declaration of function 'enable_irq'

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoath9k: make use of a helper to get paprd scale factor
Mohammed Shafi Shajakhan [Mon, 6 Jun 2011 17:32:33 +0000 (23:02 +0530)]
ath9k: make use of a helper to get paprd scale factor

Signed-off-by: Mohammed Shafi Shajakhan <mshajakhan@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: disable transmission buffer aggregation for AMSDU packets
Yogesh Ashok Powar [Mon, 6 Jun 2011 09:24:17 +0000 (14:54 +0530)]
mwifiex: disable transmission buffer aggregation for AMSDU packets

Padding per MSDU will affect the length of next packet and hence
the exact length of next packet is uncertain here.

Also, aggregation of transmission buffer, while downloading the
data to the card, wont gain much on the AMSDU packets as the AMSDU
packets utilizes the transmission buffer space to the maximum
(adapter->tx_buf_size).

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: adding check for enough space before padding
Yogesh Ashok Powar [Mon, 6 Jun 2011 09:23:02 +0000 (14:53 +0530)]
mwifiex: adding check for enough space before padding

All MSDUs, except the last one in an AMSDU, should end up at 4
bytes boundary. There is need to check if enough skb_tailroom
space exists before padding the skb.

Also re-arranging code for better readablity.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: Remove redundant skb_queue_empty checks
Yogesh Ashok Powar [Mon, 6 Jun 2011 09:20:58 +0000 (14:50 +0530)]
mwifiex: Remove redundant skb_queue_empty checks

The check of skb list empty before calling skb_peek and skb_dequeue is
redundant. These functions returns NULL if the list is empty.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwifiex: remove list traversal in mwifiex_num_pkts_in_txq
Yogesh Ashok Powar [Mon, 6 Jun 2011 09:19:32 +0000 (14:49 +0530)]
mwifiex: remove list traversal in mwifiex_num_pkts_in_txq

Instead of counting the number of packets in txq
for particular RA list before AMSDU creation,
maintain a counter which will keep track of the
same.

This will reduce some MIPS while generating AMSDU
traffic as we only have to check the counter instead
of traversing through skb list.

Signed-off-by: Yogesh Ashok Powar <yogeshp@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlwifi: remove unecessary if statement
Greg Dietsche [Mon, 6 Jun 2011 00:30:11 +0000 (19:30 -0500)]
iwlwifi: remove unecessary if statement

the code always returns ret regardless, so if(ret) check is unecessary.

Signed-off-by: Greg Dietsche <Gregory.Dietsche@cuw.edu>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: refactor iwlagn_mac_channel_switch
Stanislaw Gruszka [Thu, 2 Jun 2011 14:46:20 +0000 (16:46 +0200)]
iwlagn: refactor iwlagn_mac_channel_switch

Use less indentions and remove uneeded irq-save flags.

Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agowl12xx: Stop BA session event from device
Shahar Levi [Sun, 22 May 2011 13:10:22 +0000 (16:10 +0300)]
wl12xx: Stop BA session event from device

Adding new event that close RX BA session in case of periodic BT activity
limiting WLAN activity.

Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: Stop BA session event from device
Shahar Levi [Sun, 22 May 2011 13:10:21 +0000 (16:10 +0300)]
mac80211: Stop BA session event from device

Some devices support BT/WLAN co-existence algorigthms.
In order not to harm the system performance and user experience, the device
requests not to allow any RX BA session and tear down existing RX BA sessions
based on system constraints such as periodic BT activity that needs to limit
WLAN activity (eg.SCO or A2DP).
In such cases, the intention is to limit the duration of the RX PPDU and
therefore prevent the peer device to use A-MPDU aggregation.

Adding ieee80211_stop_rx_ba_session() callback
that can be used by the driver to stop existing BA sessions.

Signed-off-by: Shahar Levi <shahar_levi@ti.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
John W. Linville [Tue, 7 Jun 2011 18:07:11 +0000 (14:07 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-2.6

13 years agoMerge branch 'can/at91-sam9x5' of git://git.pengutronix.de/git/mkl/linux-2.6
David S. Miller [Tue, 7 Jun 2011 08:03:56 +0000 (01:03 -0700)]
Merge branch 'can/at91-sam9x5' of git://git.pengutronix.de/git/mkl/linux-2.6

13 years agonet: remove interrupt.h inclusion from netdevice.h
Alexey Dobriyan [Mon, 6 Jun 2011 10:43:46 +0000 (10:43 +0000)]
net: remove interrupt.h inclusion from netdevice.h

* remove interrupt.g inclusion from netdevice.h -- not needed
* fixup fallout, add interrupt.h and hardirq.h back where needed.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobe2net: Fix Rx pause counter for lancer
Selvin Xavier [Mon, 6 Jun 2011 02:27:13 +0000 (02:27 +0000)]
be2net: Fix Rx pause counter for lancer

Fixed Rx pause counter for Lancer. Swapping hi and lo words.

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>
13 years agoqlge: remove unecessary if statement
Greg Dietsche [Sun, 5 Jun 2011 14:44:13 +0000 (14:44 +0000)]
qlge: remove unecessary if statement

the code always returns 'status' regardless, so if(status) check is unecessary.

Signed-off-by: Greg Dietsche <Gregory.Dietsche@cuw.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoenic: Add support for MTU change via port profile on a dynamic vnic
Roopa Prabhu [Fri, 3 Jun 2011 14:35:17 +0000 (14:35 +0000)]
enic: Add support for MTU change via port profile on a dynamic vnic

enic driver gets MTU change notifications for MTU changes in the
port profile associated to a dynamic vnic. This patch adds support
in enic driver to set new MTU on the dynamic vnic and dynamically
adjust its buffers with new MTU size in response to such notifications.

Signed-off-by: Roopa Prabhu <roprabhu@cisco.com>
Signed-off-by: Vasanthy Kolluri <vkolluri@cisco.com>
Signed-off-by: David Wang <dwang2@cisco.com>
Signed-off-by: Christian Benvenuti <benve@cisco.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agomacvlan: add VLAN filters to lowerdev
John Fastabend [Mon, 6 Jun 2011 04:27:16 +0000 (04:27 +0000)]
macvlan: add VLAN filters to lowerdev

Stacking VLANs on top of the macvlan device does not
work if the lowerdev device is using vlan filters set
by NETIF_F_HW_VLAN_FILTER. Add ndo ops to pass vlan
calls to lowerdev.

Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosctp: Guard IPV6 specific code properly.
David S. Miller [Mon, 6 Jun 2011 20:05:55 +0000 (13:05 -0700)]
sctp: Guard IPV6 specific code properly.

Outside of net/sctp/ipv6.c, IPV6 specific code needs to
be ifdef guarded.

This fixes build failures with IPV6 disabled.

Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoiwl4965: set tx power after rxon_assoc
Stanislaw Gruszka [Mon, 6 Jun 2011 13:11:30 +0000 (15:11 +0200)]
iwl4965: set tx power after rxon_assoc

Setting tx power can be deferred during scan or changing channel.
If after that correct tx power settings will not be sent to device,
we can observe transmission problems and timeouts. Force to send
tx power settings also after partial rxon change, to assure device
always be configured with up-to-date settings.

Resolves:
https://bugzilla.kernel.org/show_bug.cgi?id=36492

Cc: stable@kernel.org # 2.6.39+
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: fix rmmod crash
Stanislaw Gruszka [Sat, 4 Jun 2011 14:48:54 +0000 (16:48 +0200)]
rt2x00: fix rmmod crash

Avoid queue and run autowakeup_work when device is not present anymore.
That prevent rmmod and device remove crash introduced by:

commit 1c0bcf89d85cc97a0d9ce4cd909351a81fa4fdde
Author: Ivo van Doorn <ivdoorn@gmail.com>
Date:   Sat Apr 30 17:18:18 2011 +0200

    rt2x00: Add autowake support for USB hardware

Signed-off-by: Stanislaw Gruszka <stf_xl@wp.pl>
Acked-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: use cts-to-self protection on 5000 adapters series
Stanislaw Gruszka [Thu, 26 May 2011 15:14:22 +0000 (17:14 +0200)]
iwlagn: use cts-to-self protection on 5000 adapters series

This patch fixes 802.11n stability and performance regression we have
since 2.6.35. It boost performance on my 5GHz N-only network from about
5MB/s to 8MB/s. Similar percentage boost can be observed on 2.4 GHz.

These are test results of 5x downloading of approximately 700MB iso
image:

vanilla: 5.27 5.22 4.94 4.47 5.31 ; avr 5.0420 std 0.35110
patched: 8.07 7.95 8.06 7.99 7.96 ; avr 8.0060 std 0.055946

This was achieved with NetworkManager configured to do not perform
periodical scans, by configuring constant BSSID. With periodical scans,
after some time, performance downgrade to unpatched driver level, like
in example below:

patched: 7.40 7.61 4.28 4.37 4.80 avr 5.6920 std 1.6683

However patch still make better here, since similar test on unpatched
driver make link disconnects with below messages after some time:

wlan1: authenticate with 00:23:69:35:d1:3f (try 1)
wlan1: authenticate with 00:23:69:35:d1:3f (try 2)
wlan1: authenticate with 00:23:69:35:d1:3f (try 3)
wlan1: authentication with 00:23:69:35:d1:3f timed out

On 2.6.35 kernel patch helps against connection hangs with messages:

iwlagn 0000:20:00.0: queue 10 stuck 3 time. Fw reload.
iwlagn 0000:20:00.0: On demand firmware reload
iwlagn 0000:20:00.0: Stopping AGG while state not ON or starting

Cc: stable@kernel.org # 2.6.35+
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Acked-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoRevert "mac80211: Skip tailroom reservation for full HW-crypto devices"
John W. Linville [Mon, 6 Jun 2011 18:35:27 +0000 (14:35 -0400)]
Revert "mac80211: Skip tailroom reservation for full HW-crypto devices"

This reverts commit aac6af5534fade2b18682a0b9efad1a6c04c34c6.

Conflicts:

net/mac80211/key.c

That commit has a race that causes a warning, as documented in the thread
here:

http://marc.info/?l=linux-wireless&m=130717684914101&w=2

Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agonet/can: allow CAN_AT91 on AT91SAM9X5
Uwe Kleine-König [Tue, 26 Apr 2011 13:05:59 +0000 (15:05 +0200)]
net/can: allow CAN_AT91 on AT91SAM9X5

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: add support for the AT91SAM9X5 SOCs
Marc Kleine-Budde [Sat, 16 Apr 2011 22:08:45 +0000 (00:08 +0200)]
can: at91_can: add support for the AT91SAM9X5 SOCs

The AT91SAM9X5 SOCs have a similar CAN core, but they only have 8 compared
to 16 mailboxes on the AT91SAM9263 SOC. Another difference is that the bits
defining the state of the CAN core are cleared on read, thus the driver
has to derive the state by looking at the error counters.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: register mb0 sysfs entry only on at91sam9263
Marc Kleine-Budde [Tue, 31 May 2011 22:20:17 +0000 (00:20 +0200)]
can: at91_can: register mb0 sysfs entry only on at91sam9263

This patch prepares the driver for the at91sam9X5 processors,
which don't have the mb0 bug.
(See commit 3a5655a5b545e9647c3437473ee3d815fe1b9050 for more details.)

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: add id_table and convert prime mailbox constats to functions
Marc Kleine-Budde [Tue, 3 May 2011 15:41:09 +0000 (17:41 +0200)]
can: at91_can: add id_table and convert prime mailbox constats to functions

This is the second of two patches converting the at91_can driver from a
compile time mailbox setup to a dynamic one.

This patch first adds a id_table to the platform driver. Depending on the
driver_data the constants for the mailbox setup is selected. Then all
remaining prime mailbox constants are converted to functions, using the
run time selected mailbox constants.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: convert derived mailbox constants into functions
Marc Kleine-Budde [Tue, 3 May 2011 15:31:40 +0000 (17:31 +0200)]
can: at91_can: convert derived mailbox constants into functions

This is the first of two patches converting the at91_can driver from a
compile time mailbox setup to a dynamic one.

This patch converts all derived mailbox constants to functions.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: rename AT91_MB_RX_MASK to AT91_IRQ_MB_RX
Marc Kleine-Budde [Tue, 3 May 2011 14:37:16 +0000 (16:37 +0200)]
can: at91_can: rename AT91_MB_RX_MASK to AT91_IRQ_MB_RX

... and use it for AT91_NEXT_MB_MASK,
AT91_IRQ_MB_RX and AT91_IRQ_MB_RX, too.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: directly define AT91_MB_RX_LAST
Marc Kleine-Budde [Sat, 30 Apr 2011 18:46:12 +0000 (20:46 +0200)]
can: at91_can: directly define AT91_MB_RX_LAST

...instead of deriving it from AT91_MB_RX_FIRST and AT91_MB_RX_NUM.
This removes a level of computation, when switching the driver from
compile time constants to runtime values.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: let get_tx_* functions return unsigned int
Marc Kleine-Budde [Tue, 3 May 2011 15:47:55 +0000 (17:47 +0200)]
can: at91_can: let get_tx_* functions return unsigned int

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: don't copy data to rx'ed RTR frames
Marc Kleine-Budde [Thu, 21 Oct 2010 16:39:26 +0000 (18:39 +0200)]
can: at91_can: don't copy data to rx'ed RTR frames

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Acked-by: Wolfgang Grandegger <wg@grandegger.com>
13 years agocan: at91_can: fix comment about priv->tx_next
Marc Kleine-Budde [Sat, 16 Apr 2011 11:25:15 +0000 (13:25 +0200)]
can: at91_can: fix comment about priv->tx_next

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agocan: at91_can: don't align struct definitions
Marc Kleine-Budde [Sun, 30 Jan 2011 21:14:49 +0000 (22:14 +0100)]
can: at91_can: don't align struct definitions

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
13 years agoinclude/net: Remove unnecessary semicolons
Joe Perches [Fri, 3 Jun 2011 11:51:21 +0000 (11:51 +0000)]
include/net: Remove unnecessary semicolons

Semicolons are not necessary after switch/while/for/if braces
so remove them.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agodrivers/net: Remove unnecessary semicolons
Joe Perches [Fri, 3 Jun 2011 11:51:20 +0000 (11:51 +0000)]
drivers/net: Remove unnecessary semicolons

Semicolons are not necessary after switch/while/for/if braces
so remove them.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: Remove unnecessary semicolons
Joe Perches [Fri, 3 Jun 2011 11:51:19 +0000 (11:51 +0000)]
net: Remove unnecessary semicolons

Semicolons are not necessary after switch/while/for/if braces
so remove them.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoaf-packet: Use existing netdev reference for bound sockets.
Ben Greear [Wed, 1 Jun 2011 07:18:53 +0000 (07:18 +0000)]
af-packet: Use existing netdev reference for bound sockets.

This saves a network device lookup on each packet transmitted,
for sockets that are bound to a network device.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoaf-packet: Hold reference to bound network devices.
Ben Greear [Wed, 1 Jun 2011 07:18:52 +0000 (07:18 +0000)]
af-packet: Hold reference to bound network devices.

Old code was probably safe, but with this change we
can actually use the netdev object, not just compare
the pointer values.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
David S. Miller [Sat, 4 Jun 2011 20:38:31 +0000 (13:38 -0700)]
Merge branch 'master' of /linux/kernel/git/davem/net-2.6

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Sat, 4 Jun 2011 14:19:51 +0000 (23:19 +0900)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6:
  [SCSI] Fix oops caused by queue refcounting failure

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 4 Jun 2011 14:16:00 +0000 (23:16 +0900)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (40 commits)
  tg3: Fix tg3_skb_error_unmap()
  net: tracepoint of net_dev_xmit sees freed skb and causes panic
  drivers/net/can/flexcan.c: add missing clk_put
  net: dm9000: Get the chip in a known good state before enabling interrupts
  drivers/net/davinci_emac.c: add missing clk_put
  af-packet: Add flag to distinguish VID 0 from no-vlan.
  caif: Fix race when conditionally taking rtnl lock
  usbnet/cdc_ncm: add missing .reset_resume hook
  vlan: fix typo in vlan_dev_hard_start_xmit()
  net/ipv4: Check for mistakenly passed in non-IPv4 address
  iwl4965: correctly validate temperature value
  bluetooth l2cap: fix locking in l2cap_global_chan_by_psm
  ath9k: fix two more bugs in tx power
  cfg80211: don't drop p2p probe responses
  Revert "net: fix section mismatches"
  drivers/net/usb/catc.c: Fix potential deadlock in catc_ctrl_run()
  sctp: stop pending timers and purge queues when peer restart asoc
  drivers/net: ks8842 Fix crash on received packet when in PIO mode.
  ip_options_compile: properly handle unaligned pointer
  iwlagn: fix incorrect PCI subsystem id for 6150 devices
  ...

13 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-block
Linus Torvalds [Fri, 3 Jun 2011 23:11:26 +0000 (08:11 +0900)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-block

* 'for-linus' of git://git.kernel.dk/linux-block:
  block: Use hlist_entry() for io_context.cic_list.first
  cfq-iosched: Remove bogus check in queue_fail path
  xen/blkback: potential null dereference in error handling
  xen/blkback: don't call vbd_size() if bd_disk is NULL
  block: blkdev_get() should access ->bd_disk only after success
  CFQ: Fix typo and remove unnecessary semicolon
  block: remove unwanted semicolons
  Revert "block: Remove extra discard_alignment from hd_struct."
  nbd: adjust 'max_part' according to part_shift
  nbd: limit module parameters to a sane value
  nbd: pass MSG_* flags to kernel_recvmsg()
  block: improve the bio_add_page() and bio_add_pc_page() descriptions

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Fri, 3 Jun 2011 23:04:10 +0000 (08:04 +0900)]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin:
  Blackfin: strncpy: fix handling of zero lengths

13 years agoMerge branch 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux...
Linus Torvalds [Fri, 3 Jun 2011 23:03:16 +0000 (08:03 +0900)]
Merge branch 'stable' of git://git./linux/kernel/git/cmetcalf/linux-tile

* 'stable' of git://git.kernel.org/pub/scm/linux/kernel/git/cmetcalf/linux-tile:
  asm-generic/unistd.h: support sendmmsg syscall
  tile: enable CONFIG_BUGVERBOSE

13 years agoMerge branch 'linux-next' of git://git.infradead.org/ubifs-2.6
Linus Torvalds [Fri, 3 Jun 2011 22:59:32 +0000 (07:59 +0900)]
Merge branch 'linux-next' of git://git.infradead.org/ubifs-2.6

* 'linux-next' of git://git.infradead.org/ubifs-2.6:
  UBIFS: fix-up free space earlier
  UBIFS: intialize LPT earlier
  UBIFS: assert no fixup when writing a node
  UBIFS: fix clean znode counter corruption in error cases
  UBIFS: fix memory leak on error path
  UBIFS: fix shrinker object count reports
  UBIFS: fix recovery broken by the previous recovery fix
  UBIFS: amend ubifs_recover_leb interface
  UBIFS: introduce a "grouped" journal head flag
  UBIFS: supress false error messages

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
Linus Torvalds [Fri, 3 Jun 2011 22:58:48 +0000 (07:58 +0900)]
Merge branch 'for-linus' of git://git./linux/kernel/git/rostedt/linux-2.6-ktest

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-2.6-ktest:
  ktest: Ignore unset values of the minconfig in config_bisect
  ktest: Fix result of rebooting the kernel
  ktest: Fix off-by-one in config bisect result

13 years agoMerge branch 'rmobile-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 3 Jun 2011 22:53:23 +0000 (07:53 +0900)]
Merge branch 'rmobile-fixes-for-linus' of git://git./linux/kernel/git/lethal/sh-2.6

* 'rmobile-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  ARM: mach-shmobile: add DMAC clock definitions on SH7372
  ARM: arch-shmobile: support SDHI card detection on mackerel, using a GPIO
  sh_mobile_meram: MERAM platform data for LCDC

13 years agoMerge branch 'sh-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 3 Jun 2011 22:04:25 +0000 (07:04 +0900)]
Merge branch 'sh-fixes-for-linus' of git://git./linux/kernel/git/lethal/sh-2.6

* 'sh-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6:
  dmaengine: shdma: fix a regression: initialise DMA channels for memcpy
  dmaengine: shdma: Fix up fallout from runtime PM changes.
  Revert "clocksource: sh_cmt: Runtime PM support"
  Revert "clocksource: sh_tmu: Runtime PM support"
  sh: Fix up asm-generic/ptrace.h fallout.
  sh64: Move from P1SEG to CAC_ADDR for consistent sync.
  sh64: asm/pgtable.h needs asm/mmu.h
  sh: asm/tlb.h needs linux/swap.h
  sh: mark DMA slave ID 0 as invalid
  sh: Update shmin to reflect PIO dependency.
  sh: arch/sh/kernel/process_32.c needs linux/prefetch.h.
  sh: add MMCIF runtime PM support on ecovec
  sh: switch ap325rxa to dynamically manage the platform camera

13 years agoRevert "ASoC: Update cx20442 for TTY API change"
Linus Torvalds [Fri, 3 Jun 2011 22:00:50 +0000 (07:00 +0900)]
Revert "ASoC: Update cx20442 for TTY API change"

This reverts commit ed0bd2333cffc3d856db9beb829543c1dfc00982.

Since we reverted the TTY API change, we should revert the ASoC update
to it too.

Cc: Mark Brown <broonie@opensource.wolfsonmicro.com>
Cc: Liam Girdwood <lrg@ti.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoRevert "tty: make receive_buf() return the amout of bytes received"
Linus Torvalds [Fri, 3 Jun 2011 21:33:24 +0000 (06:33 +0900)]
Revert "tty: make receive_buf() return the amout of bytes received"

This reverts commit b1c43f82c5aa265442f82dba31ce985ebb7aa71c.

It was broken in so many ways, and results in random odd pty issues.

It re-introduced the buggy schedule_work() in flush_to_ldisc() that can
cause endless work-loops (see commit a5660b41af6a: "tty: fix endless
work loop when the buffer fills up").

It also used an "unsigned int" return value fo the ->receive_buf()
function, but then made multiple functions return a negative error code,
and didn't actually check for the error in the caller.

And it didn't actually work at all.  BenH bisected down odd tty behavior
to it:
  "It looks like the patch is causing some major malfunctions of the X
   server for me, possibly related to PTYs.  For example, cat'ing a
   large file in a gnome terminal hangs the kernel for -minutes- in a
   loop of what looks like flush_to_ldisc/workqueue code, (some ftrace
   data in the quoted bits further down).

   ...

   Some more data: It -looks- like what happens is that the
   flush_to_ldisc work queue entry constantly re-queues itself (because
   the PTY is full ?) and the workqueue thread will basically loop
   forver calling it without ever scheduling, thus starving the consumer
   process that could have emptied the PTY."

which is pretty much exactly the problem we fixed in a5660b41af6a.

Milton Miller pointed out the 'unsigned int' issue.

Reported-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Reported-by: Milton Miller <miltonm@bga.com>
Cc: Stefan Bigler <stefan.bigler@keymile.com>
Cc: Toby Gray <toby.gray@realvnc.com>
Cc: Felipe Balbi <balbi@ti.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoiwlagn: call commit_rxon function directly
Wey-Yi Guy [Fri, 3 Jun 2011 14:54:16 +0000 (07:54 -0700)]
iwlagn: call commit_rxon function directly

No need to go though multiple levels of indirect call to send RXON command.
Call it directly

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: direct call to post_scan function
Wey-Yi Guy [Fri, 3 Jun 2011 14:54:15 +0000 (07:54 -0700)]
iwlagn: direct call to post_scan function

After driver split, no need to use function "ops" for post_scan.

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: move all post scan functions in one place
Wey-Yi Guy [Fri, 3 Jun 2011 14:54:14 +0000 (07:54 -0700)]
iwlagn: move all post scan functions in one place

Both tx power and power save are being done after scan complete, move into
post_scan function

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: set smps mode after assoc for 1000 device
Wey-Yi Guy [Fri, 3 Jun 2011 14:54:13 +0000 (07:54 -0700)]
iwlagn: set smps mode after assoc for 1000 device

For some timing reason, 1000 device having problem to kick-in to aggregation
without sending rxon assoc command. This is a W/A until find the real reason

Signed-off-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoiwlagn: save the latest smps mode
Wey-Yi Guy [Fri, 3 Jun 2011 14:54:12 +0000 (07:54 -0700)]
iwlagn: save the latest smps mode

When change smps mode due to bt coex, save it

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