GitHub/MotorolaMobilityLLC/kernel-slsi.git
11 years agoqlcnic: Support atomic commands
Rajesh Borundia [Fri, 29 Mar 2013 05:46:38 +0000 (05:46 +0000)]
qlcnic: Support atomic commands

o VFs might get scheduled out after sending a command to a PF
  and scheduled in after receiving a response. Implement a
  worker thread to handle atomic commands.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Support VF-PF communication channel commands.
Rajesh Borundia [Fri, 29 Mar 2013 05:46:37 +0000 (05:46 +0000)]
qlcnic: Support VF-PF communication channel commands.

o Add support for commands from VF to PF.
o PF validates the commands sent by the VF before sending
  it to adapter.
o vPort is a container of resources. PF creates vPort
  for VFs and attach resources to it. vPort is
  transparent to the VF.
o Separate 83xx TX and RX hardware resource cleanup from 82xx.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: VF-PF communication channel implementation
Rajesh Borundia [Fri, 29 Mar 2013 05:46:36 +0000 (05:46 +0000)]
qlcnic: VF-PF communication channel implementation

o Adapter provides communication channel between VF and PF.
  Any control commands from the VF driver are sent to the PF driver
  through this communication channel. PF driver validates the
  commands before sending them to the adapter. Similarly PF driver
  forwards any control command responses  to the VF driver
  through this communication channel.  Adapter sends message pending
  event to VF or PF when there is an outstanding response or a command
  for VF or PF respectively. When a command or a response is sent over
  a channel VF or PF cannot send another command or a response
  until adapter sends a channel free event. Adapter allocates 1K area to
  VF and  PF each for this communication.
o Commands and  responses are encapsulated in a header. Header determines
  sequence id, number of fragments, fragment number etc.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Use shared interrupt vector for Tx and Rx
Rajesh Borundia [Fri, 29 Mar 2013 05:46:35 +0000 (05:46 +0000)]
qlcnic: Use shared interrupt vector for Tx and Rx

o VF will use shared MSI-X interrupt vector for Tx and Rx.
o When QLCNIC_INTR_SHARED flag is set Tx and Rx will
  share MSI-X interrupt vector. Tx will use a separate
  MSI-X interrupt vector from Rx otherwise.

Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: SR-IOV VF probe
Rajesh Borundia [Fri, 29 Mar 2013 05:46:34 +0000 (05:46 +0000)]
qlcnic: SR-IOV VF probe

o Add PCI device entry for VF.
o Add HW operations for VF.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoqlcnic: Support SR-IOV enable and disable
Rajesh Borundia [Fri, 29 Mar 2013 05:46:33 +0000 (05:46 +0000)]
qlcnic: Support SR-IOV enable and disable

o Add QLCNIC_SRIOV to Kconfig.
o Provide PCI sysfs hooks to enable and disable SR-IOV.
o Allow enabling only when CONFIG_QLCNIC_SRIOV is defined.
o qlcnic_sriov_pf.c has all the PF related SR-IOV
  functionality.
o qlcnic_sriov_common.c has VF functionality and SR-IOV
  functionality which is common between VF and PF.
o qlcnic_sriov.h is a common header file for SR-IOV defines.

Signed-off-by: Manish Chopra <manish.chopra@qlogic.com>
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: Rajesh Borundia <rajesh.borundia@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocirrus: cs89x0: remove two obsolete Kconfig macros
Paul Bolle [Fri, 29 Mar 2013 00:51:30 +0000 (00:51 +0000)]
cirrus: cs89x0: remove two obsolete Kconfig macros

The CONFIG_ARCH_IXDP2X01 and CONFIG_MACH_IXDP2351 Kconfig macros are
unused since the ixp23xx and ixp2000 platforms were removed in v3.5. So
remove the last code still depending on these macros. And since
CS89x0_NONISA_IRQ was only set if either of these two macros was defined
we can also remove that macro and the code depending on it.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agosh_eth: add R-Car support for real
Sergei Shtylyov [Thu, 28 Mar 2013 11:51:31 +0000 (11:51 +0000)]
sh_eth: add R-Car support for real

Commit d0418bb7123f44b23d69ac349eec7daf9103472f (net: sh_eth: Add eth support
for R8A7779 device) was a failed attempt to add support for one of members of
the R-Car SoC family.  That's for three reasons: it treated R8A7779 the  same
as SH7724 except including quite dirty hack adding ECMR_ELB  bit  to the mask
in sh_eth_set_rate() while not removing ECMR_RTM bit (despite it's reserved in
R-Car Ether), and it didn't add a new register offset array despite the closest
SH_ETH_REG_FAST_SH4 mapping differs by 0x200 to the offsets all the R-Car Ether
registers have, and also some of the registers in this old mapping don't exist
on R-Car Ether (due to this, SH7724's 'sh_eth_my_cpu_data' structure is not
adequeate for R-Car too).  Fix all these shortcomings, restoring the SH7724
related section to its pristine state...

Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agosh_eth: move data from header file to driver
Sergei Shtylyov [Thu, 28 Mar 2013 11:48:26 +0000 (11:48 +0000)]
sh_eth: move data from header file to driver

The driver's header file contains initialized register offset tables which (as
any data definitions), of course, have no business being there.  Move them  to
the driver's body, somewhat beautifying the initializers, while at it...

Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: rtnetlink: fdb dflt dump must set idx used for cb->arg[0]
John Fastabend [Fri, 29 Mar 2013 08:18:37 +0000 (08:18 +0000)]
net: rtnetlink: fdb dflt dump must set idx used for cb->arg[0]

In rtnl_fdb_dump() when the fdb_dump ndo op is not populated
we never set the idx value so that cb->arg[0] is always 0.
Resulting in a endless loop of messages.

Introduced with this commit,

commit 090096bf3db1c281ddd034573260045888a68fea
Author: Vlad Yasevich <vyasevic@redhat.com>
Date:   Wed Mar 6 15:39:42 2013 +0000

    net: generic fdb support for drivers without ndo_fdb_<op>

CC: Vlad Yasevich <vyasevic@redhat.com>
Signed-off-by: John Fastabend <john.r.fastabend@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoip_tunnel: Fix off-by-one error in forming dev name.
Pravin B Shelar [Thu, 28 Mar 2013 08:21:46 +0000 (08:21 +0000)]
ip_tunnel: Fix off-by-one error in forming dev name.

As Ben pointed out following patch fixes bug in checking device
name length limits while forming tunnel device name.

CC: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: simplify the getting percpu of flow_cache
Li RongQing [Thu, 28 Mar 2013 02:24:11 +0000 (02:24 +0000)]
net: simplify the getting percpu of flow_cache

replace per_cpu with per_cpu_ptr to save conversion between address and pointer

Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonetlink: fix the warning introduced by netlink API replacement
Hong zhi guo [Fri, 29 Mar 2013 05:09:35 +0000 (05:09 +0000)]
netlink: fix the warning introduced by netlink API replacement

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoaudit: pass int* to nlmsg_next
Alexandru Copot [Thu, 28 Mar 2013 21:31:29 +0000 (23:31 +0200)]
audit: pass int* to nlmsg_next

Commit 941912133025926307c7a65b203fa38403b1063a replaced the macros
NLMSG_NEXT with calls to nlmsg_next which produces this warning:

kernel/audit.c: In function ‘audit_receive_skb’:
kernel/audit.c:928:3: warning: passing argument 2 of ‘nlmsg_next’ makes pointer from integer without a cast
In file included from include/net/rtnetlink.h:5:0,
                 from include/net/neighbour.h:28,
                 from include/net/dst.h:17,
                 from include/net/sock.h:68,
                 from kernel/audit.c:55:
include/net/netlink.h:359:1: note: expected ‘int *’ but argument is of type ‘int’

Fix this by sending the intended pointer.

Signed-off-by: Alexandru Copot <alex.mihai.c@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobridge: remove unused variable ifm
Hong zhi guo [Thu, 28 Mar 2013 06:21:22 +0000 (06:21 +0000)]
bridge: remove unused variable ifm

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Acked-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/macb: make clk_enable atomic
Steffen Trumtrar [Wed, 27 Mar 2013 23:07:07 +0000 (23:07 +0000)]
net/macb: make clk_enable atomic

Use clk_prepare_enable/clk_disable_unprepare to be safe on SMP systems.

Signed-off-by: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Cc: Fabio Estevam <festevam@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/macb: force endian_swp_pkt_en to off
Steffen Trumtrar [Wed, 27 Mar 2013 23:07:06 +0000 (23:07 +0000)]
net/macb: force endian_swp_pkt_en to off

The core has a bit for swapping packet data endianism.
Reset default from Cadence is off. Xilinx however, who uses this core on the
Zynq SoCs, opted for on.
Force it to off. This shouldn't change the behaviour for current users of the
macb, but enables usage on Zynq devices.

Signed-off-by: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet/macb: clear tx/rx completion flags in ISR
Steffen Trumtrar [Wed, 27 Mar 2013 23:07:05 +0000 (23:07 +0000)]
net/macb: clear tx/rx completion flags in ISR

At least in the cadence IP core on the Xilinx Zynq SoC the TCOMP/RCOMP flags
are not auto-cleaned. As these flags are evaluated, they need to be cleaned.

Signed-off-by: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Cc: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net...
David S. Miller [Thu, 28 Mar 2013 18:38:54 +0000 (14:38 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net-next

Jeff Kirsher says:

====================
This series contains updates to e1000e, ixgbe and ixgbevf.

Majority of the changes are against e1000e (from Bruce Allan).
Bruce adds additional error handling on PHY register access, as
well as improve slow performance on 82579 when connected to a
10Mbit hub.  In addition, fixes LED blink logic for cathode
LED design.  Most notable is added EEE support which is enabled
by default and the added support for LTR on I217/I218.

The ixgbe and ixgbevf from Greg Rose changes the VM so that if a user
does not assign a MAC address, the MAC address is set to all zeros
instead of a random MAC address.  This ensures that we always know when
we have a random address and udev won't get upset about it.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge
David S. Miller [Thu, 28 Mar 2013 18:34:23 +0000 (14:34 -0400)]
Merge tag 'batman-adv-for-davem' of git://git.open-mesh.org/linux-merge

Included changes:
- A fix for the network coding component which has been added within the last
  pull request (so it is in linux-3.10). The problem has been spotted thanks to
  Fengguang Wu's automated daily checks on our tree.
- Implementation of the RTNL API for virtual interface creation/deletion and slave
  manipulation
- substitution of seq_printf with seq_puts when possible
- minor cleanups

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoconnector: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:54:56 +0000 (06:54 +0000)]
connector: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoscsi: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:53:15 +0000 (06:53 +0000)]
scsi: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Reviewed-by: Mike Christie <michaelc@cs.wisc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agogdm72xx: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:52:17 +0000 (06:52 +0000)]
gdm72xx: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoselinux: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:49:35 +0000 (06:49 +0000)]
selinux: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoaudit: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:49:06 +0000 (06:49 +0000)]
audit: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Acked-by: Thomas Graf <tgraf@suug.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet-next: replace obsolete NLMSG_* with type safe nlmsg_*
Hong zhi guo [Wed, 27 Mar 2013 06:47:04 +0000 (06:47 +0000)]
net-next: replace obsolete NLMSG_* with type safe nlmsg_*

Signed-off-by: Hong Zhiguo <honkiko@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoixgbevf: Adjust to handle unassigned MAC address from PF
Greg Rose [Wed, 13 Feb 2013 03:02:05 +0000 (03:02 +0000)]
ixgbevf: Adjust to handle unassigned MAC address from PF

If the administrator has not assigned a MAC address to the VF via the
PF then handle it gracefully by generating a temporary MAC address.
This ensures that we always know when we have a random address and
udev won't get upset about it.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: Stefan Assmann <sassmann@kpanic.de>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoixgbe: Don't give VFs random MAC addresses
Greg Rose [Fri, 15 Feb 2013 05:20:09 +0000 (05:20 +0000)]
ixgbe: Don't give VFs random MAC addresses

If the user has not assigned a MAC address to a VM, then don't give it a
random one. Instead, just give it zeros and let it figure out what to do
with them.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
CC: Andy Gospodarek <andy@greyhouse.net>
CC: Stefan Assmann <sassmann@kpanic.de>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
11 years agoe1000e: fix scheduling while atomic bugs
Bruce Allan [Tue, 19 Mar 2013 01:47:12 +0000 (01:47 +0000)]
e1000e: fix scheduling while atomic bugs

The previous commit ce43a2168c59bc47b5f0c1825fd5f9a2a9e3b447 (e1000e:
cleanup USLEEP_RANGE checkpatch checks) converted a number of delays and
sleeps as recommended in ./Documentation/timers/timers-howto.txt.
Unfortunately, a few of the udelay() to usleep_range() conversions are in
code paths that are in an atomic context in which usleep_range() should
not be used.  Revert those specific changes.

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>
11 years agoe1000e: increase driver version number
Bruce Allan [Wed, 6 Mar 2013 09:03:13 +0000 (09:03 +0000)]
e1000e: increase driver version number

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>
11 years agoe1000e: cleanup unused defines
Bruce Allan [Wed, 6 Mar 2013 09:03:08 +0000 (09:03 +0000)]
e1000e: cleanup unused defines

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>
11 years agoe1000e: add support for LTR on I217/I218
Bruce Allan [Wed, 6 Mar 2013 09:03:02 +0000 (09:03 +0000)]
e1000e: add support for LTR on I217/I218

Set the Latency Tolerance Reporting (LTR) values for the "PCIe-like"
GbE MAC in the Lynx Point PCH based on Rx buffer size and link speed
when link is up (which must not exceed the maximum latency supported
by the platform), otherwise specify there is no LTR requirement.
Unlike true-PCIe devices which set the LTR maximum snoop/no-snoop
latencies in the LTR Extended Capability Structure in the PCIe Extended
Capability register set, on this device LTR is set by writing the
equivalent snoop/no-snoop latencies in the LTRV register in the MAC and
set the SEND bit to send an Intel On-chip System Fabric sideband (IOSF-SB)
message to the PMC.

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>
11 years agoe1000e: enable EEE by default
Bruce Allan [Wed, 6 Mar 2013 09:02:57 +0000 (09:02 +0000)]
e1000e: enable EEE by default

Now that IEEE802.3az-2010 Energy Efficient Ethernet has been approved as
standard (September 2010) and the driver can enable and disable it via
ethtool, enable the feature by default on parts which support it.

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>
11 years agoe1000e: EEE capability advertisement not set/disabled as required
Bruce Allan [Wed, 20 Mar 2013 07:23:11 +0000 (07:23 +0000)]
e1000e: EEE capability advertisement not set/disabled as required

Devices supported by the driver which support EEE (currently 82579, I217
and I218) are advertising EEE capabilities during auto-negotiation even
when EEE has been disabled.  In addition to not acting as expected, this
also caused the EEE status reported by 'ethtool --show-eee' to be wrong
when two of these devices are connected back-to-back and EEE is disabled
on one.  In addition to fixing this issue, the ability for the user to
specify which speeds (100 or 1000 full-duplex) to advertise EEE support
has been added.

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>
11 years agoe1000e: long access timeouts when I217/I218 MAC and PHY are out of sync
Bruce Allan [Wed, 6 Mar 2013 09:02:47 +0000 (09:02 +0000)]
e1000e: long access timeouts when I217/I218 MAC and PHY are out of sync

When the MAC and PHY are in two different modes (different power levels
and interconnect speeds), it could take a long time before a PHY register
access timed out using the existing MAC-PHY interconnect configuration
coded into the driver for ICH- and PCH-based LOMs.  Introduce an I217/I218-
specific .setup_physical_interface operation which does not override the
interconnect configuration in the NVM.

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>
11 years agoe1000e: fix LED blink logic for designs with LEDs driven by cathode
Bruce Allan [Wed, 6 Mar 2013 09:02:41 +0000 (09:02 +0000)]
e1000e: fix LED blink logic for designs with LEDs driven by cathode

When the LEDs are driven by cathode, the bit logic is reversed.  Use the
LED Invert bit to invert the logic.  Cleanup use of a magic number and
change the for loop increment to reduce the number of shifts.

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>
11 years agoe1000e: slow performance between two 82579 connected via 10Mbit hub
Bruce Allan [Wed, 6 Mar 2013 09:02:36 +0000 (09:02 +0000)]
e1000e: slow performance between two 82579 connected via 10Mbit hub

Two 82579 LOMs connected via a 10Mb hub experience extraordinarily low
performance.  This is because 82579 is excessively aggressive on transmit
at 10Mb half-duplex and will not provide sufficient time for the link
partner to transmit.  When the link partner is also 82579, the result is a
lot of collisions (and corresponding re-transmits) that cause the poor
performance. To work-around this issue, significantly increase the IPG in
the MAC to allow enough gap for the link partner to transmit and reduce the
Rx latency in the analog PHY to 0 to reduce the number of collisions.

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>
11 years agoe1000e: additional error handling on PHY register accesses
Bruce Allan [Wed, 6 Mar 2013 09:02:31 +0000 (09:02 +0000)]
e1000e: additional error handling on PHY register accesses

PHY reads/writes via the MDIC register could potentially return results
from a previous PHY register access.  If that happens, the offset in the
returned results will be that of the previous access and if that is
different from the expected offset, log a debug message and error out.

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>
11 years agonet: add ETH_P_802_3_MIN
Simon Horman [Thu, 28 Mar 2013 04:38:25 +0000 (13:38 +0900)]
net: add ETH_P_802_3_MIN

Add a new constant ETH_P_802_3_MIN, the minimum ethernet type for
an 802.3 frame. Frames with a lower value in the ethernet type field
are Ethernet II.

Also update all the users of this value that David Miller and
I could find to use the new constant.

Also correct a bug in util.c. The comparison with ETH_P_802_3_MIN
should be >= not >.

As suggested by Jesse Gross.

Compile tested only.

Cc: David Miller <davem@davemloft.net>
Cc: Jesse Gross <jesse@nicira.com>
Cc: Karsten Keil <isdn@linux-pingi.de>
Cc: John W. Linville <linville@tuxdriver.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: Bart De Schuymer <bart.de.schuymer@pandora.be>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Patrick McHardy <kaber@trash.net>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: Gustavo Padovan <gustavo@padovan.org>
Cc: Johan Hedberg <johan.hedberg@gmail.com>
Cc: linux-bluetooth@vger.kernel.org
Cc: netfilter-devel@vger.kernel.org
Cc: bridge@lists.linux-foundation.org
Cc: linux-wireless@vger.kernel.org
Cc: linux1394-devel@lists.sourceforge.net
Cc: linux-media@vger.kernel.org
Cc: netdev@vger.kernel.org
Cc: dev@openvswitch.org
Acked-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Acked-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Simon Horman <horms@verge.net.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: fix compilation without CONFIG_BNX2X_SRIOV
Dmitry Kravkov [Wed, 27 Mar 2013 08:56:10 +0000 (08:56 +0000)]
bnx2x: fix compilation without CONFIG_BNX2X_SRIOV

Move mutex initialization by allocation of the mailbox it protects.

introduced in commit 1d6f3cd89 'bnx2x: Prevent VF race'

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agotokenring: delete last holdout of CONFIG_TR
Paul Bolle [Wed, 27 Mar 2013 10:52:28 +0000 (10:52 +0000)]
tokenring: delete last holdout of CONFIG_TR

Tokenring support was deleted in v3.5. One last holdout of the macro
CONFIG_TR escaped that fate. Until now.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: fix compile error of implicit declaration of skb_probe_transport_header
Ying Xue [Wed, 27 Mar 2013 16:46:06 +0000 (16:46 +0000)]
net: fix compile error of implicit declaration of skb_probe_transport_header

The commit 40893fd(net: switch to use skb_probe_transport_header())
involes a new error accidently. When NET_SKBUFF_DATA_USES_OFFSE is
not enabled, below compile error happens:

  CC      net/packet/af_packet.o
  net/packet/af_packet.c: In function ‘packet_sendmsg_spkt’:
  net/packet/af_packet.c:1516:2: error: implicit declaration of function ‘skb_probe_transport_header’ [-Werror=implicit-function-declaration]
  cc1: some warnings being treated as errors
  make[2]: *** [net/packet/af_packet.o] Error 1
  make[1]: *** [net/packet] Error 2
  make: *** [net] Error 2

As it seems skb_probe_transport_header() is not related to
NET_SKBUFF_DATA_USES_OFFSE, we should move the definition of
skb_probe_transport_header() out of scope of
NET_SKBUFF_DATA_USES_OFFSE macro.

Cc: Jason Wang <jasowang@redhat.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoyam: remove redundant null check on dev
Colin Ian King [Wed, 27 Mar 2013 18:25:04 +0000 (18:25 +0000)]
yam: remove redundant null check on dev

yam_open has a redundant null check on null,  it will
never be called with dev == NULL. Remove this redundant check.
This also cleans up a smatch warning:

drivers/net/hamradio/yam.c:869 yam_open() warn: variable
  dereferenced before check 'dev' (see line 867)

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
David S. Miller [Wed, 27 Mar 2013 17:52:49 +0000 (13:52 -0400)]
Merge git://git./linux/kernel/git/davem/net

Conflicts:
include/net/ipip.h

The changes made to ipip.h in 'net' were already included
in 'net-next' before that header was moved to another location.

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'for-davem' of git://gitorious.org/linux-can/linux-can-next
David S. Miller [Wed, 27 Mar 2013 17:14:03 +0000 (13:14 -0400)]
Merge branch 'for-davem' of git://gitorious.org/linux-can/linux-can-next

Marc Kleine-Budde says:

===================
this is a pull-request for net-next/master. It consists of three
patches by Lars-Peter Clausen to clean up the mcp251x spi-can driver,
two patches from Ludovic Desroches to bring device tree support to the
at91_can driver and a patch by me to fix a sparse warning in the
blackfin driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agophy: Elimination the forced speed reduction algorithm.
Kirill Kapranov [Wed, 27 Mar 2013 01:16:13 +0000 (01:16 +0000)]
phy: Elimination the forced speed reduction algorithm.

In case of fixed speed set up for a NIC (e.g. ethtool -s eth0 autoneg off speed
100 duplex full) with an ethernet cable plugged off, the mentioned algorithm
slows down a NIC speed, so further cable hook-up leads to nonoperable link state.

Signed-off-by: Kirill Kapranov <kapranoff@inbox.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: use the frag lru_lock to protect netns_frags.nqueues update
Jesper Dangaard Brouer [Wed, 27 Mar 2013 05:55:56 +0000 (05:55 +0000)]
net: use the frag lru_lock to protect netns_frags.nqueues update

Move the protection of netns_frags.nqueues updates under the LRU_lock,
instead of the write lock.  As they are located on the same cacheline,
and this is also needed when transitioning to use per hash bucket locking.

Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: frag, avoid several CPUs grabbing same frag queue during LRU evictor loop
Jesper Dangaard Brouer [Wed, 27 Mar 2013 05:55:25 +0000 (05:55 +0000)]
net: frag, avoid several CPUs grabbing same frag queue during LRU evictor loop

The LRU list is protected by its own lock, since commit 3ef0eb0db4
(net: frag, move LRU list maintenance outside of rwlock), and
no-longer by a read_lock.

This makes it possible, to remove the inet_frag_queue, which is about
to be "evicted", from the LRU list head.  This avoids the problem, of
several CPUs grabbing the same frag queue.

Note, cannot remove the inet_frag_lru_del() call in fq_unlink()
called by inet_frag_kill(), because inet_frag_kill() is also used in
other situations.  Thus, we use list_del_init() to allow this
double list_del to work.

Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoppp: reuse print_hex_dump_bytes
Andy Shevchenko [Wed, 27 Mar 2013 05:54:14 +0000 (05:54 +0000)]
ppp: reuse print_hex_dump_bytes

There is a native function to dump hex buffers.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: core: let's use native isxdigit instead of custom
Andy Shevchenko [Wed, 27 Mar 2013 05:54:13 +0000 (05:54 +0000)]
net: core: let's use native isxdigit instead of custom

In kernel we have fast and pretty implementation of the isxdigit() function.
Let's use it.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Cosmetic changes
Yaniv Rosner [Wed, 27 Mar 2013 01:05:19 +0000 (01:05 +0000)]
bnx2x: Cosmetic changes

Make few alignments, comment fixes and debug messages.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Support reading I2C EEPROM SFF8472
Yaniv Rosner [Wed, 27 Mar 2013 01:05:18 +0000 (01:05 +0000)]
bnx2x: Support reading I2C EEPROM SFF8472

Add full support for "ethtool -m" command.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Prevent VF race
Dmitry Kravkov [Wed, 27 Mar 2013 01:05:17 +0000 (01:05 +0000)]
bnx2x: Prevent VF race

The mail box containing the Vf-Pf messages is susceptible
to a race - it's possible for 2 flows to try and write commands,
causing one to override the other's message.
Use a mutex to synchronize the access, preventing said race.

Signed-off-by: Dmitry Kravkov <dmitry@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Fix VF outer vlan removal
Ariel Elior [Wed, 27 Mar 2013 01:05:16 +0000 (01:05 +0000)]
bnx2x: Fix VF outer vlan removal

Outer vlan removal in VF queues was made according to the VF's
multi-function mode (which is never set).
Instead, the PF's multi-function mode should be used to determine
if outer vlan removal is needed.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Fix VF statistics
Ariel Elior [Wed, 27 Mar 2013 01:05:15 +0000 (01:05 +0000)]
bnx2x: Fix VF statistics

After a VF performs load/unload its statistics become corrupt -
we now zero the statistics structures upon a VF device load.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: missing ARI should not be lethal
Ariel Elior [Wed, 27 Mar 2013 01:05:14 +0000 (01:05 +0000)]
bnx2x: missing ARI should not be lethal

If ARI forwarding flag is missing from the PCI bridge, remove SR-IOV
support instead of failing the probe process.

Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobnx2x: Fix AER semaphore release
Yuval Mintz [Tue, 26 Mar 2013 23:28:03 +0000 (23:28 +0000)]
bnx2x: Fix AER semaphore release

Commit 7fa6f34 "AER revised" erroneously inserted an error-flow
in which a semaphore is released even though the attempt to take it
has failed.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: switch to use skb_probe_transport_header()
Jason Wang [Tue, 26 Mar 2013 23:11:22 +0000 (23:11 +0000)]
net: switch to use skb_probe_transport_header()

Switch to use the new help skb_probe_transport_header() to do the l4 header
probing for untrusted sources. For packets with partial csum, the header should
already been set by skb_partial_csum_set().

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: core: introduce skb_probe_transport_header()
Jason Wang [Tue, 26 Mar 2013 23:11:21 +0000 (23:11 +0000)]
net: core: introduce skb_probe_transport_header()

Sometimes, we need probe and set the transport header for packets (e.g from
untrusted source). This patch introduces a new helper
skb_probe_transport_header() which tries to probe and set the l4 header through
skb_flow_dissect(), if not just set the transport header to the hint passed by
caller.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: core: let skb_partial_csum_set() set transport header
Jason Wang [Tue, 26 Mar 2013 23:11:20 +0000 (23:11 +0000)]
net: core: let skb_partial_csum_set() set transport header

For untrusted packets with partial checksum, we need to set the transport header
for precise packet length estimation. We can just let skb_pratial_csum_set() to
do this to avoid extra call to skb_flow_dissect() and simplify the caller.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoptp_pch: eliminate a number of sparse warnings
Sahara [Tue, 26 Mar 2013 19:07:23 +0000 (19:07 +0000)]
ptp_pch: eliminate a number of sparse warnings

This fixes a number of sparse warnings like:
  warning: incorrect type in argument 2 (different address spaces)
    expected void volatile [noderef] <asn:2>*addr
    got unsigned int *<noident>

  warning: Using plain integer as NULL pointer

Additionally this fixes a warning from checkpatch.pl like:
  WARNING: sizeof pch_param.station should be sizeof(pch_param.station)

Signed-off-by: Sahara <keun-o.park@windriver.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge tag 'iommu-fixes-v3.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 27 Mar 2013 16:25:11 +0000 (09:25 -0700)]
Merge tag 'iommu-fixes-v3.9-rc4' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU fixes from Joerg Roedel:
 "Here are some fixes which have collected since Linux v3.9-rc1.

  The most important one fixes a long-standing regressen which make
  re-hotplugged devices unusable when AMD IOMMU is used.

  The other patches fix build issues (build regression on OMAP and a
  section mismatch).  One patch just removes a duplicate header include."

* tag 'iommu-fixes-v3.9-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu:
  iommu/amd: Make sure dma_ops are set for hotplug devices
  x86, io_apic: remove duplicated include from irq_remapping.c
  iommu: OMAP: build only on OMAP2+
  amd_iommu_init: remove __init from amd_iommu_erratum_746_workaround

11 years agovfs/splice: Fix missed checks in new __kernel_write() helper
Al Viro [Wed, 27 Mar 2013 15:20:30 +0000 (15:20 +0000)]
vfs/splice: Fix missed checks in new __kernel_write() helper

Commit 06ae43f34bcc ("Don't bother with redoing rw_verify_area() from
default_file_splice_from()") lost the checks to test existence of the
write/aio_write methods.  My apologies ;-/

Eventually, we want that in fs/splice.c side of things (no point
repeating it for every buffer, after all), but for now this is the
obvious minimal fix.

Reported-by: Dave Jones <davej@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
11 years agocan: bfin_can: declare locally used functions static
Marc Kleine-Budde [Thu, 11 Oct 2012 21:20:27 +0000 (23:20 +0200)]
can: bfin_can: declare locally used functions static

This patch fixes the following sparse warning:

drivers/net/can/bfin_can.c:415:13: warning: symbol 'bfin_can_interrupt' was not declared. Should it be static?
drivers/net/can/bfin_can.c:507:19: warning: symbol 'alloc_bfin_candev' was not declared. Should it be static?

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agocan: mcp251x: Use dev_pm_ops
Lars-Peter Clausen [Tue, 12 Mar 2013 12:13:53 +0000 (13:13 +0100)]
can: mcp251x: Use dev_pm_ops

Use dev_pm_ops instead of the deprecated legacy suspend/resume callbacks.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agocan: mcp251x: Use module_spi_driver
Lars-Peter Clausen [Tue, 12 Mar 2013 12:13:52 +0000 (13:13 +0100)]
can: mcp251x: Use module_spi_driver

By using module_spi_driver we can eliminate a few lines of boilerplate code.

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agocan: mcp251x: Remove redundant spi driver bus initialization
Lars-Peter Clausen [Tue, 12 Mar 2013 12:13:51 +0000 (13:13 +0100)]
can: mcp251x: Remove redundant spi driver bus initialization

In ancient times it was necessary to manually initialize the bus field of an
spi_driver to spi_bus_type. These days this is done in spi_driver_register() so
we can drop the manual assignment.

The patch was generated using the following coccinelle semantic patch:
// <smpl>
@@
identifier _driver;
@@
struct spi_driver _driver = {
.driver = {
- .bus = &spi_bus_type,
},
};
// </smpl>

Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agocan: Kconfig: CAN_AT91 depends on ARM
Ludovic Desroches [Mon, 11 Mar 2013 17:26:04 +0000 (18:26 +0100)]
can: Kconfig: CAN_AT91 depends on ARM

SAMA5D3 devices also embed CAN feature. Moreover if we want to produce a single
kernel image it is not useful to be too restrictive.

Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agocan: at91_can: add dt support
Ludovic Desroches [Mon, 11 Mar 2013 17:26:03 +0000 (18:26 +0100)]
can: at91_can: add dt support

Add device tree support.

Signed-off-by: Ludovic Desroches <ludovic.desroches@atmel.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
11 years agobatman-adv: use seq_puts instead of seq_printf when the format is constant
Antonio Quartulli [Thu, 21 Mar 2013 08:23:29 +0000 (09:23 +0100)]
batman-adv: use seq_puts instead of seq_printf when the format is constant

As reported by checkpatch, seq_puts has to be preferred with
respect to seq_printf when the format is a constant string
(no va_args)

Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
11 years agobatman-adv: update Makefile copyright years
Antonio Quartulli [Wed, 13 Mar 2013 22:02:28 +0000 (23:02 +0100)]
batman-adv: update Makefile copyright years

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Start new development cycle
Simon Wunderlich [Sun, 3 Mar 2013 17:30:29 +0000 (18:30 +0100)]
batman-adv: Start new development cycle

Signed-off-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: free an hard-interface before adding it
Antonio Quartulli [Sat, 16 Feb 2013 13:42:39 +0000 (14:42 +0100)]
batman-adv: free an hard-interface before adding it

When adding a new hard interface (e.g. wlan0) to a soft interface (e.g. bat0)
and the former is already enslaved in another virtual interface (e.g. a software
bridge) batman-adv has to free it first and then continue with the adding
mechanism.

In this way the behaviour becomes consistent with what "ip link set master"
does. At the moment batman-adv enslaves the hard interface without checking for
the master device, possibly causing strange behaviours which are never wanted by
the users.

Reported-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
11 years agobatman-adv: Allow to modify slaves of soft-interfaces through rntl_link
Sven Eckelmann [Mon, 11 Feb 2013 09:10:27 +0000 (17:10 +0800)]
batman-adv: Allow to modify slaves of soft-interfaces through rntl_link

The sysfs configuration interface of batman-adv to add/remove slaves of an
soft-iface is not deadlock free and doesn't follow the currently common way to
modify slaves of an interface.

An additional configuration interface though rtnl_link is introduced which
provides easy device adding/removing with tools like "ip":
$ ip link set dev eth0 master bat0
$ ip link set dev eth0 nomaster

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Allow to use rntl_link for device creation/deletion
Sven Eckelmann [Mon, 11 Feb 2013 09:10:26 +0000 (17:10 +0800)]
batman-adv: Allow to use rntl_link for device creation/deletion

The sysfs configuration interface of batman-adv to add/remove soft-interfaces
is not deadlock free and doesn't follow the currently common way to create new
virtual interfaces.

An additional interface though rtnl_link is introduced which provides easy device
creation/deletion with tools like "ip":

$ ip link add dev bat0 type batadv
$ ip link del dev bat0

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: rename batadv_softif_destroy to reflect sysfs use case
Marek Lindner [Mon, 11 Feb 2013 09:10:25 +0000 (17:10 +0800)]
batman-adv: rename batadv_softif_destroy to reflect sysfs use case

Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
CC: Sven Eckelmann <sven@narfation.org>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Don't always delete softif when last slave was removed
Sven Eckelmann [Mon, 11 Feb 2013 09:10:24 +0000 (17:10 +0800)]
batman-adv: Don't always delete softif when last slave was removed

batman-adv has an unusual way to manage softinterfaces. These will be created
automatically when a user writes to the batman-adv/mesh_iface file in sysfs and
removed when no slave device exists anymore.

This behaviour cannot be changed without breaking compatibility with existing
code. Instead other interfaces should be able to slightly reduce this behaviour
and provide a more common reaction to a removal of a slave interface.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Move deinitialization of soft-interface to destructor
Sven Eckelmann [Mon, 11 Feb 2013 09:10:23 +0000 (17:10 +0800)]
batman-adv: Move deinitialization of soft-interface to destructor

The deinitialization of the soft-interface created in ndo_init/constructor
should be done in the destructor and not directly before calling
unregister_netdevice

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Move soft-interface initialization to ndo_init
Sven Eckelmann [Mon, 11 Feb 2013 09:10:22 +0000 (17:10 +0800)]
batman-adv: Move soft-interface initialization to ndo_init

The initialization of an net_device object should be done in the
init/constructor function and not from the outside after the register_netdevice
was done to avoid race conditions.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Acked-by: Antonio Quartulli <ordex@autistici.org>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agobatman-adv: Fix endianness errors for network coding
Martin Hundebøll [Thu, 14 Mar 2013 20:30:21 +0000 (21:30 +0100)]
batman-adv: Fix endianness errors for network coding

Add a htonl() in network_coding.c when reading the sequence number
from received ogm_packet, to avoid wrong byte ordering when comparing
with a host value. This bug was introduced in
3ed7ada3f0bbcd058567bc0a8f9729a73eba7db6 ("batman-adv: network coding -
detect coding nodes and remove these after timeout").

Change the type of coded_packet->coded_len from uint16 to __be16 to
avoid wrong assumptions about endianness in later uses. Introduced in
c3289f3650d34b60296000a629c99f2488f7c3dd ("batman-adv: network coding -
code and transmit packets if possible").

Reported-by: Fengguang Wu <fengguang.wu@intel.com>
Signed-off-by: Martin Hundebøll <martin@hundeboll.net>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Antonio Quartulli <ordex@autistici.org>
11 years agoiommu/amd: Make sure dma_ops are set for hotplug devices
Joerg Roedel [Tue, 26 Mar 2013 21:48:23 +0000 (22:48 +0100)]
iommu/amd: Make sure dma_ops are set for hotplug devices

There is a bug introduced with commit 27c2127 that causes
devices which are hot unplugged and then hot-replugged to
not have per-device dma_ops set. This causes these devices
to not function correctly. Fixed with this patch.

Cc: stable@vger.kernel.org
Reported-by: Andreas Degert <andreas.degert@googlemail.com>
Signed-off-by: Joerg Roedel <joro@8bytes.org>
11 years ago6lowpan: use IEEE802154_ADDR_LEN instead of a magic number
Tony Cheneau [Tue, 26 Mar 2013 18:09:25 +0000 (18:09 +0000)]
6lowpan: use IEEE802154_ADDR_LEN instead of a magic number

Signed-off-by: Tony Cheneau <tony.cheneau@amnesiak.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years ago6lowpan: fix a small formatting issue
Tony Cheneau [Tue, 26 Mar 2013 18:09:24 +0000 (18:09 +0000)]
6lowpan: fix a small formatting issue

This formatting issue was introduced with commit
d4ac32365dcbfd341a87eae444c26679f889249a

Signed-off-by: Tony Cheneau <tony.cheneau@amnesiak.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoieee802154/at86rf230: Fix register names for RX_AACK_ON and TX_ARET_ON
stefan@datenfreihafen.org [Tue, 26 Mar 2013 12:41:31 +0000 (12:41 +0000)]
ieee802154/at86rf230: Fix register names for RX_AACK_ON and TX_ARET_ON

The register names have been wrong since the beginning but it only showed up now
as they are actualy used for the upcoming auto ACK support.

Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoieee802154/at86rf230: Implement hardware address filter callback.
stefan@datenfreihafen.org [Tue, 26 Mar 2013 12:41:30 +0000 (12:41 +0000)]
ieee802154/at86rf230: Implement hardware address filter callback.

Implement the filter function to update short address, pan id and ieee
address on change. Allowing for hardware address filtering needed for
auto ACK.

Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoieee802154/dgram: Pass source address in dgram_recvmsg
Stephen Röttger [Tue, 26 Mar 2013 12:41:29 +0000 (12:41 +0000)]
ieee802154/dgram: Pass source address in dgram_recvmsg

This patch lets dgram_recvmsg fill in the sockaddr struct in
msg->msg_name with the source address of the packet.
This is used by the userland functions recvmsg and recvfrom to get the
senders address.

[Stefan: Changed from old zigbee legacy tree to mainline]

Signed-off-by: Stephen Röttger <stephen.roettger@zero-entropy.de>
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoVXLAN: Fix sparse warnings.
Pravin B Shelar [Tue, 26 Mar 2013 08:29:30 +0000 (08:29 +0000)]
VXLAN: Fix sparse warnings.

Fixes following warning:-
drivers/net/vxlan.c:471:35: warning: symbol 'dev' shadows an earlier one
drivers/net/vxlan.c:433:26: originally declared here
drivers/net/vxlan.c:794:34: warning: symbol 'vxlan' shadows an earlier one
drivers/net/vxlan.c:757:26: originally declared here

CC: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Wed, 27 Mar 2013 00:42:55 +0000 (17:42 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull vfs fixes from Al Viro:
 "stable fodder; assorted deadlock fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  vt: synchronize_rcu() under spinlock is not nice...
  Nest rename_lock inside vfsmount_lock
  Don't bother with redoing rw_verify_area() from default_file_splice_from()

11 years agovt: synchronize_rcu() under spinlock is not nice...
Al Viro [Wed, 27 Mar 2013 00:30:17 +0000 (20:30 -0400)]
vt: synchronize_rcu() under spinlock is not nice...

vcs_poll_data_free() calls unregister_vt_notifier(), which calls
atomic_notifier_chain_unregister(), which calls synchronize_rcu().
Do it *after* we'd dropped ->f_lock.

Cc: stable@vger.kernel.org (all kernels since 2.6.37)
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agoNest rename_lock inside vfsmount_lock
Al Viro [Tue, 26 Mar 2013 22:25:57 +0000 (18:25 -0400)]
Nest rename_lock inside vfsmount_lock

... lest we get livelocks between path_is_under() and d_path() and friends.

The thing is, wrt fairness lglocks are more similar to rwsems than to rwlocks;
it is possible to have thread B spin on attempt to take lock shared while thread
A is already holding it shared, if B is on lower-numbered CPU than A and there's
a thread C spinning on attempt to take the same lock exclusive.

As the result, we need consistent ordering between vfsmount_lock (lglock) and
rename_lock (seq_lock), even though everything that takes both is going to take
vfsmount_lock only shared.

Spotted-by: Brad Spengler <spender@grsecurity.net>
Cc: stable@vger.kernel.org
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
11 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Tue, 26 Mar 2013 21:24:29 +0000 (14:24 -0700)]
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) Always increment IPV4 ID field in encapsulated GSO packets, even
    when DF is set.  Regression fix from Pravin B Shelar.

 2) Fix per-net subsystem initialization in netfilter conntrack,
    otherwise we may access dynamically allocated memory before it is
    actually allocated.  From Gao Feng.

 3) Fix DMA buffer lengths in iwl3945 driver, from Stanislaw Gruszka.

 4) Fix race between submission of sync vs async commands in mwifiex
    driver, from Amitkumar Karwar.

 5) Add missing cancel of command timer in mwifiex driver, from Bing
    Zhao.

 6) Missing SKB free in rtlwifi USB driver, from Jussi Kivilinna.

 7) Thermal layer tries to use a genetlink multicast string that is
    longer than the 16 character limit.  Fix it and add a BUG check to
    prevent this kind of thing from happening in the future.

 From Masatake YAMATO.

 8) Fix many bugs in the handling of the teardown of L2TP connections,
    UDP encapsulation instances, and sockets.  From Tom Parkin.

 9) Missing socket release in IRDA, from Kees Cook.

10) Fix fec driver modular build, from Fabio Estevam.

11) Erroneous use of kfree() instead of free_netdev() in lantiq_etop,
    from Wei Yongjun.

12) Fix bugs in handling of queue numbers and steering rules in mlx4
    driver, from Moshe Lazer, Hadar Hen Zion, and Or Gerlitz.

13) Some FOO_DIAG_MAX constants were defined off by one, fix from Andrey
    Vagin.

14) TCP segmentation deferral is unintentionally done too strongly,
    breaking ACK clocking.  Fix from Eric Dumazet.

15) net_enable_timestamp() can legitimately be invoked from software
    interrupts, and in a way that is safe, so remove the WARN_ON().
    Also from Eric Dumazet.

16) Fix use after free in VLANs, from Cong Wang.

17) Fix TCP slow start retransmit storms after SACK reneging, from
    Yuchung Cheng.

18) Unix socket release should mark a socket dead before NULL'ing out
    sock->sk, otherwise we can race.  Fix from Paul Moore.

19) IPV6 addrconf code can try to free static memory, from Hong Zhiguo.

20) Fix register mis-programming, NULL pointer derefs, and wrong PHC
    clock frequency in IGB driver.  From Lior LevyAlex Williamson, Jiri
    Benc, and Jeff Kirsher.

21) skb->ip_summed logic in pch_gbe driver is reversed, breaking packet
    forwarding.  Fix from Veaceslav Falico.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (65 commits)
  ipv4: Fix ip-header identification for gso packets.
  bonding: remove already created master sysfs link on failure
  af_unix: dont send SCM_CREDENTIAL when dest socket is NULL
  pch_gbe: fix ip_summed checksum reporting on rx
  igb: fix PHC stopping on max freq
  igb: make sensor info static
  igb: SR-IOV init reordering
  igb: Fix null pointer dereference
  igb: fix i350 anti spoofing config
  ixgbevf: don't release the soft entries
  ipv6: fix bad free of addrconf_init_net
  unix: fix a race condition in unix_release()
  tcp: undo spurious timeout after SACK reneging
  bnx2x: fix assignment of signed expression to unsigned variable
  bridge: fix crash when set mac address of br interface
  8021q: fix a potential use-after-free
  net: remove a WARN_ON() in net_enable_timestamp()
  tcp: preserve ACK clocking in TSO
  net: fix *_DIAG_MAX constants
  net/mlx4_core: Disallow releasing VF QPs which have steering rules
  ...

11 years agoMerge tag 'nfs-for-3.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Tue, 26 Mar 2013 21:23:45 +0000 (14:23 -0700)]
Merge tag 'nfs-for-3.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client bugfixes from Trond Myklebust:
 - Fix an NFSv4 idmapper regression
 - Fix an Oops in the pNFS blocks client
 - Fix up various issues with pNFS layoutcommit
 - Ensure correct read ordering of variables in
   rpc_wake_up_task_queue_locked

* tag 'nfs-for-3.9-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  SUNRPC: Add barriers to ensure read ordering in rpc_wake_up_task_queue_locked
  NFSv4.1: Add a helper pnfs_commit_and_return_layout
  NFSv4.1: Always clear the NFS_INO_LAYOUTCOMMIT in layoutreturn
  NFSv4.1: Fix a race in pNFS layoutcommit
  pnfs-block: removing DM device maybe cause oops when call dev_remove
  NFSv4: Fix the string length returned by the idmapper

11 years agoipv4: Fix ip-header identification for gso packets.
Pravin B Shelar [Sun, 24 Mar 2013 17:36:29 +0000 (17:36 +0000)]
ipv4: Fix ip-header identification for gso packets.

ip-header id needs to be incremented even if IP_DF flag is set.
This behaviour was changed in commit 490ab08127cebc25e3a26
(IP_GRE: Fix IP-Identification).

Following patch fixes it so that identification is always
incremented.

Reported-by: Cong Wang <amwang@redhat.com>
Signed-off-by: Pravin B Shelar <pshelar@nicira.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agobonding: remove already created master sysfs link on failure
Veaceslav Falico [Tue, 26 Mar 2013 16:43:28 +0000 (17:43 +0100)]
bonding: remove already created master sysfs link on failure

If slave sysfs symlink failes to be created - we end up without removing
the master sysfs symlink. Remove it in case of failure.

Signed-off-by: Veaceslav Falico <vfalico@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMAINTAINERS: Update qlge maintainers list
Shahed Shaikh [Tue, 26 Mar 2013 05:34:55 +0000 (05:34 +0000)]
MAINTAINERS: Update qlge maintainers list

Add myself to qlge maintainers list.

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agonet: fec: TX Buffer incorrectly initialized
Jim Baxter [Tue, 26 Mar 2013 05:25:07 +0000 (05:25 +0000)]
net: fec: TX Buffer incorrectly initialized

The TX Buffer in fec_enet_alloc_buffers was being initialized
with the receive register define BD_ENET_RX_INT instead of
the transmit register define BD_ENET_TX_INT

Signed-off-by: Jim Baxter <jim_baxter@mentor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agoMerge branch 'stmmac'
David S. Miller [Tue, 26 Mar 2013 16:54:13 +0000 (12:54 -0400)]
Merge branch 'stmmac'

Giuseppe CAVALLARO says:

====================
These patches enhance the driver adding the PTP support and the initial code
for RGMII/SGMII/TBI/RTBI modes.
Also this patches review the driver removing some Koption for selecting between
chain and ring modes. REally useful to validate the driver also at build time.
Before adding PTP, the extended descriptor support has been added because it
is mandatory to save HW timestamp in new dedicated descriptors. Also in this
case no Koption added.

Concerning the PTP, I have hacked/reviewed and tested many
part of these patches also verifying the back compatibility on
several HW and chips.

Concerning the SGMII/RGMII we have already discussed about the support
in the net.dev Mailing list with Byungho where these patchs were partially
analysed.
So I have only ported them against the latest net-next (and on
top of PTP). I have added some missing things: e.g. some parts of the
ethtool for ANE. As we clarified with Byungho, we will add further
enhancements on top of these patches if needed.

I have also built all against ARM/SH/X68 platforms and no issues on
ST-Boxes.

Thx goes to Rayagond that wrote and tested the PTP and to Byungho for SGMII.

V2: This Version 2 has the fixes discussed in the ML, for example:
    o completely remove the Koption... all the decisions are made at probe time
    o review the PTP patches and better organize them just in two patches
    o added all the fixes provided by Richard on PTP and CLK driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agostmmac: update the Doc and Version (PTP+SGMII)
Giuseppe CAVALLARO [Tue, 26 Mar 2013 04:43:12 +0000 (04:43 +0000)]
stmmac: update the Doc and Version (PTP+SGMII)

This patch updates the stmmac.txt file adding information related to the PTP
and SGMII/RGMII supports.

Also the patch updates the driver version to: March_2013.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agostmmac: add the support for PTP hw clock driver
Rayagond Kokatanur [Tue, 26 Mar 2013 04:43:11 +0000 (04:43 +0000)]
stmmac: add the support for PTP hw clock driver

This patch implements PHC (ptp hardware clock) driver for stmmac
driver to support 1588 PTP.

V2: added support for FINE method, reduced loop delay and review spinlock.

Signed-off-by: Rayagond Kokatanur <rayagond@vayavyalabs.com>
Hacked-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agostmmac: add IEEE PTPv1 and PTPv2 support.
Rayagond Kokatanur [Tue, 26 Mar 2013 04:43:10 +0000 (04:43 +0000)]
stmmac: add IEEE PTPv1 and PTPv2 support.

This patch enhances the stmmac driver to support IEEE 1588-2002
PTP (Precision Time Protocol) version 1 and IEEE 1588-2008 PPT
version 2.

Precision Time Protocol(PTP),which enables precise synchronization
of clocks in measurement and control systems implemented with
technologies such as network communication,local computing,
& distributed objects.

Both PTPv1 and PTPv2 is selected at run-time using the HW capability
register.

The PTPv1 TimeStamp support can be used on chips that have the normal
descriptor structures and PTPv2 TimeStamp support can be used on chips
that have the Extended descriptors(DES4-5-6-7). All such sanity checks
are done and verified by using HW capability register.

V2: in this version the ethtool support has been included in this patch;
Koptions have been completely removed (previously added to select
PTP and PTPv2). PTPv1 and PTPv2 is now added in a single patch instead of
two patches.
get_timestamp() and get_systemtime() L/H have been combined into single APIs.

Signed-off-by: Rayagond Kokatanur <rayagond@vayavyalabs.com>
Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>