platform/kernel/linux-exynos.git
12 years agoigb: Update get cable length function for i210/i211
Carolyn Wyborny [Wed, 10 Oct 2012 04:42:59 +0000 (04:42 +0000)]
igb: Update get cable length function for i210/i211

There was a problem in the initial implementation of the get cable length
function for i210 and it did not work properly.  This patch fixes that
problem for i210/i211 devices.

Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agomaintainers: update with official intel support link, new maintainer
Jesse Brandeburg [Tue, 9 Oct 2012 02:52:14 +0000 (02:52 +0000)]
maintainers: update with official intel support link, new maintainer

Add an official link which is designed to guide the user to the appropriate
support resource (be it community, OEM, Intel phone, Intel email, etc)

Add the current e1000 maintainer to the list of Intel maintainers.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoe1000e: Minimum packet size must be 17 bytes
Tushar Dave [Fri, 14 Sep 2012 02:21:37 +0000 (02:21 +0000)]
e1000e: Minimum packet size must be 17 bytes

This is a HW requirement. Although a buffer as short as 1 byte is allowed,
the total length of packet before, padding and CRC insertion, must be at
least 17 bytes.  So pad all small packets manually up to 17 bytes before
delivering them to HW.

Signed-off-by: Tushar Dave <tushar.n.dave@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net...
David S. Miller [Tue, 23 Oct 2012 06:51:00 +0000 (02:51 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net-next

Jeff Kirsher says:

====================
This series contains updates to ixgbe only.  Only change to this series
is I dropped the "ixgbe: Add support for pipeline reset" due to
change requested by Martin Josefsson.

Alexander Duyck (7):
  ixgbe: Add support for IPv6 and UDP to ixgbe_get_headlen
  ixgbe: Add support for tracking the default user priority to SR-IOV
  ixgbe: Add support for GET_QUEUES message to get DCB configuration
  ixgbe: Enable support for VF API version 1.1 in the PF.
  ixgbevf: Add VF DCB + SR-IOV support
  ixgbe: Drop unnecessary addition from ixgbe_set_rx_buffer_len
  ixgbe: Fix possible memory leak in ixgbe_set_ringparam

Don Skidmore (1):
  ixgbe: Add function ixgbe_reset_pipeline_82599

Emil Tantilov (1):
  ixgbe: add WOL support for new subdevice id

Jacob Keller (1):
  ixgbe: (PTP) refactor init, cyclecounter and reset

Tushar Dave (1):
  ixgbe: Correcting small packet padding

Wei Yongjun (1):
  ixgbe: using is_zero_ether_addr() to simplify the code
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_mbim: Device Service Stream support
Bjørn Mork [Mon, 22 Oct 2012 10:56:40 +0000 (10:56 +0000)]
net: cdc_mbim: Device Service Stream support

MBIM devices can support up to 256 generic streams called
Device Service Streams (DSS). The MBIM spec says

   The format of the Device Service Stream payload depends
   on the device service (as identified by the corresponding
   UUID) that is used when opening the data stream.

Example use cases are serial AT command interfaces and NMEA
data streams. We cannot make any assumptions about these
device services.

Adding support for Device Service Stream by extending
the MBIM session to VLAN mapping scheme, allocating
VLAN IDs 256 to 511 for DSS, using the DSS SessionID
as the lower 8bit of the VLAN ID.

Using a netdev for DSS keeps the device framing intact and
allows userspace to do whatever it want with the streams.
For example, exporting an AT command interface using DSS
session #0 to a PTY for use with a terminal application like
minicom:

  vconfig add wwan0 256
  ip link set dev wwan0 up
  ip link set dev wwan0.256 up
  socat INTERFACE:wwan0.256,type=2 PTY:,echo=0,link=/tmp/modem

Device configuration must be done using MBIM control commands
over the /dev/cdc-wdmx device. The userspace management
application should coordinate host VLAN configuration and the
device MBIM configuration using the device capabilities to
find out if it needs to set up PTY mappings etc.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: map MBIM IPS SessionID to VLAN ID
Bjørn Mork [Mon, 22 Oct 2012 10:56:39 +0000 (10:56 +0000)]
net: cdc_ncm: map MBIM IPS SessionID to VLAN ID

MBIM devices can support up to 256 independent IP Streams.
The main network device will only handle SessionID 0. Mapping
SessionIDs 1 to 255 to VLANs using the SessionID as VLAN ID
allow userspace to use these streams with traditional tools
like vconfig.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: do not bind to NCM compatible MBIM devices
Bjørn Mork [Mon, 22 Oct 2012 10:56:38 +0000 (10:56 +0000)]
net: cdc_ncm: do not bind to NCM compatible MBIM devices

The MBIM specification allows a MBIM device to disguise
itself as NCM for backwards compatibility, using additional
altsettings with different subclass (control) or protocol
(data):

C:* #Ifs= 2 Cfg#= 1 Atr=80 MxPwr=100mA
I:  If#= 0 Alt= 0 #EPs= 1 Cls=02(comm.) Sub=0d Prot=00 Driver=cdc_mbim
E:  Ad=81(I) Atr=03(Int.) MxPS=  64 Ivl=7ms
I:* If#= 0 Alt= 1 #EPs= 1 Cls=02(comm.) Sub=0e Prot=00 Driver=cdc_mbim
E:  Ad=81(I) Atr=03(Int.) MxPS=  64 Ivl=7ms
I:  If#= 1 Alt= 0 #EPs= 0 Cls=0a(data ) Sub=00 Prot=01 Driver=cdc_mbim
I:  If#= 1 Alt= 1 #EPs= 2 Cls=0a(data ) Sub=00 Prot=01 Driver=cdc_mbim
E:  Ad=81(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 2 #EPs= 2 Cls=0a(data ) Sub=00 Prot=02 Driver=cdc_mbim
E:  Ad=81(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms

If the MBIM driver is enabled then that should have priority
for devices providing such a NCM 1.0 backward compatibility
mode.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Greg Suarez <gsuarez@smithmicro.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_mbim: build the MBIM driver
Bjørn Mork [Mon, 22 Oct 2012 10:56:37 +0000 (10:56 +0000)]
net: cdc_mbim: build the MBIM driver

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_mbim: adding MBIM driver
Greg Suarez [Mon, 22 Oct 2012 10:56:36 +0000 (10:56 +0000)]
net: cdc_mbim: adding MBIM driver

The CDC Mobile Broadband Interface Model (MBIM) specification
extends CDC NCM by
 - removing the redundant ethernet header from the point-to-point
   USB channel
 - adding support for multiple IP (v4 and/or v6) sessions multiplexed
   on the same USB channel
 - adding a MBIM control channel encapsulated in CDC
 - adding Device Service Streams (DSS), which are non IP generic data
   streams multiplexed on the same USB channel as the IP sessions

MBIM devices are managed using the dedicated control channel, and no
data will flow on the data channel until a control session has been
established.  This driver has no knowledge of MBIM control messages.
It just exports the control channel to a /dev/cdc-wdmX character
device for userspace management applications. Such an application is
therefore required to use this driver.

This patch implements basic MBIM support, reusing the NCM and WDM driver
APIs, currently limited to IP sessions with SessionID 0. DSS and
multiplexed IP sessions are not yet supported.

Signed-off-by: Greg Suarez <gsuarez@smithmicro.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: export shared symbols and definitions
Bjørn Mork [Mon, 22 Oct 2012 10:56:35 +0000 (10:56 +0000)]
net: cdc_ncm: export shared symbols and definitions

Move symbols and definitons which can be shared with a
MBIM driver in a new header.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: refactoring for tx multiplexing
Bjørn Mork [Mon, 22 Oct 2012 10:56:34 +0000 (10:56 +0000)]
net: cdc_ncm: refactoring for tx multiplexing

Adding multiplexed NDP support to cdc_ncm_fill_tx_frame, allowing
transmissions of multiple independent sessions within the same NTB.

Refactoring the code quite a bit to avoid having to store copies
of multiple NDPs being prepared for tx. The old code would still
reserve enough room for a maximum sized NDP in the skb so we might
as well keep them in the skb while they are being prepared.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: splitting rx_fixup for code reuse
Bjørn Mork [Mon, 22 Oct 2012 10:56:33 +0000 (10:56 +0000)]
net: cdc_ncm: splitting rx_fixup for code reuse

Verifying and handling received MBIM and NCM frames will need
to be different in three areas:
 - verifying the NDP signature
 - checking valid datagram length
 - datagram header manipulation

This makes it inconvenient to share rx_fixup in whole.  But
some verification parts are common.  Split these out in separate
functions.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: process chained NDPs
Bjørn Mork [Mon, 22 Oct 2012 10:56:32 +0000 (10:56 +0000)]
net: cdc_ncm: process chained NDPs

The NCM 1.0 spefication makes provisions for linking more than
one NDP into a single NTB. This is important for MBIM support,
where these NDPs might be of different types.

Following the chain of NDPs is also correct for NCM, and will
not change anything in the common case where there is only
one NDP

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: refactor bind preparing for MBIM support
Greg Suarez [Mon, 22 Oct 2012 10:56:31 +0000 (10:56 +0000)]
net: cdc_ncm: refactor bind preparing for MBIM support

NCM and MBIM can share most of the bind function.  Split
out the shareable part and add MBIM functional descriptor
parsing.

Signed-off-by: Greg Suarez <gsuarez@smithmicro.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: adding MBIM support to ncm_setup
Greg Suarez [Mon, 22 Oct 2012 10:56:30 +0000 (10:56 +0000)]
net: cdc_ncm: adding MBIM support to ncm_setup

MBIM and NCM are very similar, so we can reuse most of the
setup and bind logic in cdc_ncm for CDC MBIM devices.  Handle
a few minor differences in ncm_setup.

Signed-off-by: Greg Suarez <gsuarez@smithmicro.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoUSB: cdc: add MBIM constants and structures
Greg Suarez [Mon, 22 Oct 2012 10:56:29 +0000 (10:56 +0000)]
USB: cdc: add MBIM constants and structures

Based on revision 1.0 of "Universal Serial Bus Communications
Class Subclass Specification for Mobile Broadband Interface
Model" available from www.usb.org

Signed-off-by: Greg Suarez <gsuarez@smithmicro.com>
[bmork: added DSS defines]
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: cdc_ncm: workaround NTB input size firmware bug
Bjørn Mork [Mon, 22 Oct 2012 10:56:28 +0000 (10:56 +0000)]
net: cdc_ncm: workaround NTB input size firmware bug

Some devices do not support the 8 byte variants of the NTB input
size control messages despite announcing such support in their
NCM or MBIM functional descriptor.

According to the NCM specification, all devices must support the
4 byte variant regardless of whether or not the flag is set:

  If bit D5 is set in the bmNetworkCapabilities field of
  function’s NCM Functional Descriptor, the host may
  set wLength either to 4 or to 8. If wLength is 4, the
  function shall assume that wNtbInMaxDatagrams is to be
  set to zero. If wLength is 8, then the function shall
  use the provided value as the limit. The function shall
  return an error response (a STALL PID) if wLength is set
  to any other value.

We do not set wNtbInMaxDatagrams in any case, so we can just as
well unconditionally use the 4 byte variant without losing any
functionality.  This works around the known firmware bug, and
simplifies the code considerably.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/at91_ether: add pdata flag for reverse Eth addr
Joachim Eastwood [Mon, 22 Oct 2012 08:45:34 +0000 (08:45 +0000)]
net/at91_ether: add pdata flag for reverse Eth addr

This will allow us to remove the last mach include from at91_ether
and also make it easier to share address setup with macb.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/at91_ether: select MACB in Kconfig
Joachim Eastwood [Mon, 22 Oct 2012 08:45:33 +0000 (08:45 +0000)]
net/at91_ether: select MACB in Kconfig

Now that HAVE_NET_MACB is gone let's just select MACB to
satisfy the dependecies in at91_ether.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/cadence: get rid of HAVE_NET_MACB
Joachim Eastwood [Mon, 22 Oct 2012 08:45:32 +0000 (08:45 +0000)]
net/cadence: get rid of HAVE_NET_MACB

macb is a platform driver and there is nothing that prevents
this driver from being built on non-ARM/AVR32 platforms.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet/macb: fix truncate warnings
Joachim Eastwood [Mon, 22 Oct 2012 08:45:31 +0000 (08:45 +0000)]
net/macb: fix truncate warnings

When building macb on x86_64 the following warnings show up:
  drivers/net/ethernet/cadence/macb.c: In function macb_interrupt:
  drivers/net/ethernet/cadence/macb.c:556:4: warning: large integer implicitly truncated to unsigned type [-Woverflow]
  drivers/net/ethernet/cadence/macb.c: In function macb_reset_hw:
  drivers/net/ethernet/cadence/macb.c:792:2: warning: large integer implicitly truncated to unsigned type [-Woverflow]
  drivers/net/ethernet/cadence/macb.c:793:2: warning: large integer implicitly truncated to unsigned type [-Woverflow]
  drivers/net/ethernet/cadence/macb.c:796:2: warning: large integer implicitly truncated to unsigned type [-Woverflow]

Use -1 insted of ~0UL, as done in other places in the driver,
to silence these warnings.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv6: add support of equal cost multipath (ECMP)
Nicolas Dichtel [Mon, 22 Oct 2012 03:42:09 +0000 (03:42 +0000)]
ipv6: add support of equal cost multipath (ECMP)

Each nexthop is added like a single route in the routing table. All routes
that have the same metric/weight and destination but not the same gateway
are considering as ECMP routes. They are linked together, through a list called
rt6i_siblings.

ECMP routes can be added in one shot, with RTA_MULTIPATH attribute or one after
the other (in both case, the flag NLM_F_EXCL should not be set).

The patch is based on a previous work from
Luc Saillard <luc.saillard@6wind.com>.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoixgbe: Fix possible memory leak in ixgbe_set_ringparam
Alexander Duyck [Wed, 12 Sep 2012 07:09:51 +0000 (07:09 +0000)]
ixgbe: Fix possible memory leak in ixgbe_set_ringparam

We were not correctly freeing the temporary rings on error in
ixgbe_set_ring_param.  In order to correct this I am unwinding a number of
changes that were made in order to get things back to the original working
form with modification for the current ring layouts.

This approach has multiple advantages including a smaller memory footprint,
and the fact that the interface is stopped while we are allocating the rings
meaning that there is less potential for some sort of memory corruption on the
ring.

The only disadvantage I see with this approach is that on a Rx allocation
failure we will report an error and only update the Tx rings.  However the
adapter should be fully functional in this state and the likelihood of such
an error is very low.  In addition it is not unreasonable to expect the
user to need to recheck the ring configuration should they experience an
error setting the ring sizes.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Add function ixgbe_reset_pipeline_82599
Don Skidmore [Tue, 11 Sep 2012 06:58:19 +0000 (06:58 +0000)]
ixgbe: Add function ixgbe_reset_pipeline_82599

This patch adds a function that forces a full pipeline reset.  This
function will be used in following patches to completely reset the PHY
during resets.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Drop unnecessary addition from ixgbe_set_rx_buffer_len
Alexander Duyck [Sat, 8 Sep 2012 03:17:01 +0000 (03:17 +0000)]
ixgbe: Drop unnecessary addition from ixgbe_set_rx_buffer_len

We still had some code floating around from the old single buffer receive
path.  As a result we were adding VLAN_HLEN to max_frame although the
resultant value was never used.  Since that is the case we can drop this from
the function.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Correcting small packet padding
Tushar Dave [Fri, 14 Sep 2012 04:24:49 +0000 (04:24 +0000)]
ixgbe: Correcting small packet padding

Driver pad skb up to 17 bytes because of the HW requirement. However, that code
implementation mess up the skb tail pointer after padding. This patch sets
skb->tail correctly.

Signed-off-by: Tushar Dave <tushar.n.dave@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: using is_zero_ether_addr() to simplify the code
Wei Yongjun [Sun, 26 Aug 2012 16:59:37 +0000 (16:59 +0000)]
ixgbe: using is_zero_ether_addr() to simplify the code

Using is_zero_ether_addr() to simplify the code.

Signed-off-by: Wei Yongjun <yongjun_wei@trendmicro.com.cn>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: (PTP) refactor init, cyclecounter and reset
Jacob Keller [Sat, 25 Aug 2012 03:54:19 +0000 (03:54 +0000)]
ixgbe: (PTP) refactor init, cyclecounter and reset

This patch modifies when and where PTP registers and data are set. Previously
a work-around was used inside cyclecounter_start in order to reset some of the
time registers. This patch creates a new ixgbe_ptp_reset specifically for this
purpose. The cyclecounter configuration has trimmed down to only modify what
is necessary. Due to hardware conditions after probe and before open, PTP init
has now moved into the ixgbe_open call. This allows the ptp device name in the
sysfs to be the ethernet device name instead of the MAC address.

The cyclecounter check flag is renamed to PTP_ENABLED and is used to prevent
PTP init from happening when PTP has not been enabled.

CC: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: add WOL support for new subdevice id
Emil Tantilov [Thu, 16 Aug 2012 08:13:07 +0000 (08:13 +0000)]
ixgbe: add WOL support for new subdevice id

This patch adds a subdevice id for new 82599 device. The define is needed
to allow enabling WOL support.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbevf: Add VF DCB + SR-IOV support
Alexander Duyck [Fri, 20 Jul 2012 08:10:03 +0000 (08:10 +0000)]
ixgbevf: Add VF DCB + SR-IOV support

This change adds support for DCB and SR-IOV from the VF.  With this change
in place the VF will correctly use a traffic class other than 0 in the case
that the PF is configured with the default user priority belonging to a
traffic class other than 0.

Cc: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Enable support for VF API version 1.1 in the PF.
Alexander Duyck [Fri, 20 Jul 2012 08:09:37 +0000 (08:09 +0000)]
ixgbe: Enable support for VF API version 1.1 in the PF.

This change switches on the last few bits for us enabling version 1.1 VF
support in the PF.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Robert Garrett <RobertX.Garrett@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Add support for GET_QUEUES message to get DCB configuration
Alexander Duyck [Fri, 20 Jul 2012 08:09:32 +0000 (08:09 +0000)]
ixgbe: Add support for GET_QUEUES message to get DCB configuration

This patch addresses several issues in regards to the combination of DCB
and SR-IOV. Specifically it allows us to send information to the VF on
which queues it should be using.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Add support for tracking the default user priority to SR-IOV
Alexander Duyck [Tue, 2 Oct 2012 00:17:03 +0000 (00:17 +0000)]
ixgbe: Add support for tracking the default user priority to SR-IOV

It is necessary to track the default user priority in the PF so that we can
force it upon the VFs.  The motivation behind this is to keep the VFs from
getting access to user priorities meant for things like storage.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Add support for IPv6 and UDP to ixgbe_get_headlen
Alexander Duyck [Thu, 24 May 2012 08:26:29 +0000 (08:26 +0000)]
ixgbe: Add support for IPv6 and UDP to ixgbe_get_headlen

This change adds support for IPv6 and UDP to ixgbe_get_headlen. The
advantage to this is that we can now handle ipv4/UDP, ipv6/TCP, and
ipv6/UDP with a single memcpy instead of having to do them in multiple
pskb_may_pull calls.

A quick bit of testing shows that we increase throughput for a single
session of netperf from 8800Mpbs to about 9300Mpbs in the case of ipv6/TCP.
As such overall ipv6 performance should improve with this change.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Stephen Ko <stephen.s.ko@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoipv4: 16 slots in initial fib_info hash table
Eric Dumazet [Sun, 21 Oct 2012 20:12:09 +0000 (20:12 +0000)]
ipv4: 16 slots in initial fib_info hash table

A small host typically needs ~10 fib_info structures, so create initial
hash table with 16 slots instead of only one. This removes potential
false sharing and reallocs/rehashes (1->2->4->8->16)

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: speedup SIOCINQ ioctl
Eric Dumazet [Sun, 21 Oct 2012 20:06:56 +0000 (20:06 +0000)]
tcp: speedup SIOCINQ ioctl

SIOCINQ can use the lock_sock_fast() version to avoid double acquisition
of socket lock.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: RFC 5961 5.2 Blind Data Injection Attack Mitigation
Eric Dumazet [Sun, 21 Oct 2012 19:57:11 +0000 (19:57 +0000)]
tcp: RFC 5961 5.2 Blind Data Injection Attack Mitigation

RFC 5961 5.2 [Blind Data Injection Attack].[Mitigation]

  All TCP stacks MAY implement the following mitigation.  TCP stacks
  that implement this mitigation MUST add an additional input check to
  any incoming segment.  The ACK value is considered acceptable only if
  it is in the range of ((SND.UNA - MAX.SND.WND) <= SEG.ACK <=
  SND.NXT).  All incoming segments whose ACK value doesn't satisfy the
  above condition MUST be discarded and an ACK sent back.

Move tcp_send_challenge_ack() before tcp_ack() to avoid a forward
declaration.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Jerry Chu <hkchu@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agopkt_sched: use ns_to_ktime() helper
Eric Dumazet [Sat, 20 Oct 2012 00:40:51 +0000 (00:40 +0000)]
pkt_sched: use ns_to_ktime() helper

ns_to_ktime() seems better than ktime_set() + ktime_add_ns()

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Update driver version
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:04:53 +0000 (06:04 +0000)]
be2net: Update driver version

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix skyhawk VF PCI Device ID
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:04:40 +0000 (06:04 +0000)]
be2net: Fix skyhawk VF PCI Device ID

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix FW flashing on Skyhawk-R
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:04:16 +0000 (06:04 +0000)]
be2net: Fix FW flashing on Skyhawk-R

FW flash layout on Skyhawk-R is different from BE3-R.
Hence the code needs to be fixed to flash FW on Skyhawk-R.
Also cleaning up code in BE3-R flashing function.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Enabling Wake-on-LAN is not supported in S5 state
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:04:00 +0000 (06:04 +0000)]
be2net: Enabling Wake-on-LAN is not supported in S5 state

be_shutdown is enabling wake-on-lan by calling be_setup_wol.
Emulex adapter do not support wake-on-lan in S5 state.

Signed-off-by: Vasundhara Volam <vasundhara.volam@emulex.com>
Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix VF driver load on newer Lancer FW
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:03:49 +0000 (06:03 +0000)]
be2net: Fix VF driver load on newer Lancer FW

PF driver should enable VF so that VF goes to ready state in
new Lancer FW.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix unnecessary delay in PCI EEH
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:03:37 +0000 (06:03 +0000)]
be2net: Fix unnecessary delay in PCI EEH

During PCI EEH, driver waits for all functions in the card.
Wait is needed only once per card. Fix is to wait only for the
first PCI function.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix issues in error recovery due to wrong queue state
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:03:25 +0000 (06:03 +0000)]
be2net: Fix issues in error recovery due to wrong queue state

During recovery from a FW error, destroy queue operation may fail.
Queue should be marked as destroyed so that recovery code can recreate
the queue. Also fix queue created state not getting checked at one instance.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix ethtool get_settings output for VF
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:03:04 +0000 (06:03 +0000)]
be2net: Fix ethtool get_settings output for VF

Return default values for fields for which VFs dont have privilege to get the
required information from FW.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix error messages while driver load for VFs
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:02:52 +0000 (06:02 +0000)]
be2net: Fix error messages while driver load for VFs

VF does not have privileges to execute many commands. When VFs try
to execute those commands there are unnecessary error messages.
Fix this by executing only those commands for which VF has privilege.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix configuring VLAN for VF for Lancer
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:02:40 +0000 (06:02 +0000)]
be2net: Fix configuring VLAN for VF for Lancer

Allow adding VLANs for Lancer VF.
VLAN ID 0 should not be added to list of VLANs sent to FW.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Wait till resources are available for VF in error recovery
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:02:27 +0000 (06:02 +0000)]
be2net: Wait till resources are available for VF in error recovery

After FW error, driver should wait for NO_RESOURCE error to disappear before
proceeding with recovery.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix change MAC operation for VF for Lancer
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:02:13 +0000 (06:02 +0000)]
be2net: Fix change MAC operation for VF for Lancer

For changing MAC of VF from PF, delete MAC operation needs to be done before
assigning new MAC. Also in ndo_set_mac_address operation avoid delete MAC if
it has been already deleted by PF.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix setting QoS for VF for Lancer
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:01:53 +0000 (06:01 +0000)]
be2net: Fix setting QoS for VF for Lancer

Use Lancer specific command to set QoS for VF.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: Fix driver load failure for different FW configs in Lancer
Padmanabh Ratnakar [Sat, 20 Oct 2012 06:01:41 +0000 (06:01 +0000)]
be2net: Fix driver load failure for different FW configs in Lancer

Driver assumes FW resource counts and capabilities while creating queues and
using functionality like RSS. This causes driver load to fail in FW configs
where resources and capabilities are reduced. Fix this by querying FW
configuration during probe and using resources and capabilities accordingly.

Signed-off-by: Padmanabh Ratnakar <padmanabh.ratnakar@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet:dev: remove double indentical assignment in dev_change_net_namespace().
Rami Rosen [Fri, 19 Oct 2012 01:09:30 +0000 (01:09 +0000)]
net:dev: remove double indentical assignment in dev_change_net_namespace().

This patch removes double assignment of err to -EINVAL in dev_change_net_namespace().

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Acked-by: Serge E. Hallyn <serge.hallyn@ubuntu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosockopt: Make SO_BINDTODEVICE readable
Pavel Emelyanov [Thu, 18 Oct 2012 23:55:56 +0000 (23:55 +0000)]
sockopt: Make SO_BINDTODEVICE readable

The SO_BINDTODEVICE option is the only SOL_SOCKET one that can be set, but
cannot be get via sockopt API. The only way we can find the device id a
socket is bound to is via sock-diag interface. But the diag works only on
hashed sockets, while the opt in question can be set for yet unhashed one.

That said, in order to know what device a socket is bound to (we do want
to know this in checkpoint-restore project) I propose to make this option
getsockopt-able and report the respective device index.

Another solution to the problem might be to teach the sock-diag reporting
info on unhashed sockets. Should I go this way instead?

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agounix: Remove unused field from unix_sock
Pavel Emelyanov [Thu, 18 Oct 2012 20:28:28 +0000 (20:28 +0000)]
unix: Remove unused field from unix_sock

The struct sock *other one seem to be unused. Grep and make do not object.

Signed-off-by: Pavel Emelyanov <xemul@parallels.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agopktgen: Use ipv6_addr_any
Joe Perches [Thu, 18 Oct 2012 17:55:31 +0000 (17:55 +0000)]
pktgen: Use ipv6_addr_any

Use the standard test for a non-zero ipv6 address.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agohyperv: Remove unnecessary comments in rndis_filter_receive_data()
Haiyang Zhang [Thu, 18 Oct 2012 11:00:52 +0000 (11:00 +0000)]
hyperv: Remove unnecessary comments in rndis_filter_receive_data()

Checked with Windows networking team, there is only one RNDIS message
in each netvsc packet.

Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/net...
David S. Miller [Sat, 20 Oct 2012 02:19:23 +0000 (22:19 -0400)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net-next

Jeff Kirsher says:

====================
This series contains updates to ixgbe and igb.

Alexander Duyck (13):
  ixgbe: Initialize q_vector cpu and affinity masks correctly
  ixgbe: Enable jumbo frames support w/ SR-IOV
  ixgbe: Move message handling routines into their own functions
  ixgbe: Add mailbox API version negotiation support to ixgbe PF
  igb: Split Rx timestamping into two separate functions
  igb: Do not use header split, instead receive all frames into a
    single buffer
  igb: Combine post-processing of skb into a single function
  igb: Map entire page and sync half instead of mapping and unmapping
    half pages
  igb: Move rx_buffer related code in Rx cleanup path into separate
    function
  igb: Lock buffer size at 2K even on systems with larger pages
  igb: Combine q_vector and ring allocation into a single function
  igb: Move the calls to set the Tx and Rx queues into igb_open
  igb: Split igb_update_dca into separate Tx and Rx functions

Tushar Dave (1):
  igb: Correcting and improving small packet check and padding
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'at91'
David S. Miller [Fri, 19 Oct 2012 17:47:00 +0000 (13:47 -0400)]
Merge branch 'at91'

Joachim Eastwood says;

====================
This patch series prepares the old at91_ether driver for code sharing
with the macb driver. The hardware is similar except for DMA TX/RX, so
its not quite clear if it is practical to support both in one
driver. But stuff like MDIO and statistics should be possible to
share.

Patch 1 adds some register defines and bits that is only found on
RM9200.

Patch 2-4 uses the register defines and access functions from the macb
header. These can be squashed if it cause too much churn.

Patch 5 merges the private at91_ether struct with the private macb
struct. This makes it easier to later share code with the macb. The
private macb struct becomes quite large, but most at91_ether specific
members are removed in later patches.

Patch 8 make macb compile when we select at91_ether. Is this approach
okey?

Patch 9 makes use of MDIO code from macb. This rips out the private
phy handling code in at91_ether. One thing that is lost is the
interrupt support for phy. But this should easy to add to macb which
will then benefit both drivers.

Patch 10 makes use of the macb_set_rx_mode from macb.

Patch 11-12 makes at91_ether share the rx dma struct members from
macb. Patch also moves the rx buffer allocation into netdev open and
dealloc into netdev close.

Last patch remove the now unused rm9200 emac header from include/mach.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoARM: AT91: remove old RM9200 EMAC register definitions
Joachim Eastwood [Thu, 18 Oct 2012 11:01:19 +0000 (11:01 +0000)]
ARM: AT91: remove old RM9200 EMAC register definitions

This file is unused after at91_ether was converted to use macb.h

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: convert to devm_* functions
Joachim Eastwood [Thu, 18 Oct 2012 11:01:18 +0000 (11:01 +0000)]
net/at91_ether: convert to devm_* functions

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: clean up rx buffer handling
Joachim Eastwood [Thu, 18 Oct 2012 11:01:17 +0000 (11:01 +0000)]
net/at91_ether: clean up rx buffer handling

This patch does two things:
* Use macb struct members and remove at91_ether ones
* Alloc DMA buffers on netdev start and dealloc on stop

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use macb dma description struct
Joachim Eastwood [Thu, 18 Oct 2012 11:01:16 +0000 (11:01 +0000)]
net/at91_ether: use macb dma description struct

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: share macb_set_rx_mode with macb
Joachim Eastwood [Thu, 18 Oct 2012 11:01:15 +0000 (11:01 +0000)]
net/at91_ether: share macb_set_rx_mode with macb

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use ethtool and mdio from macb
Joachim Eastwood [Thu, 18 Oct 2012 11:01:14 +0000 (11:01 +0000)]
net/at91_ether: use ethtool and mdio from macb

This rips out the at91_ether phy handling and ethtool stuff
and replace it with equivalent stuff from macb.

The only thing lost is the phy irq support from at91_ether,
but this can be added to macb and then benefit all users.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: compile macb for exported functions
Joachim Eastwood [Thu, 18 Oct 2012 11:01:13 +0000 (11:01 +0000)]
net/at91_ether: compile macb for exported functions

Comile macb as well as at91_ether to access exported functions.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/macb: export some symbols for at91_ether
Joachim Eastwood [Thu, 18 Oct 2012 11:01:12 +0000 (11:01 +0000)]
net/macb: export some symbols for at91_ether

Export some symbols to start sharing code between
macb and at91_ether drivers.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use pclk member instead of ether_clk
Joachim Eastwood [Thu, 18 Oct 2012 11:01:11 +0000 (11:01 +0000)]
net/at91_ether: use pclk member instead of ether_clk

Remove old at91_priv member and use pclk member from macb.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether/macb: absorb at91_private in to macb private struct
Joachim Eastwood [Thu, 18 Oct 2012 11:01:10 +0000 (11:01 +0000)]
net/at91_ether/macb: absorb at91_private in to macb private struct

This will make it easier to share code between the drivers and
eventually merge them into one driver.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use macb defs for rx dma buffers
Joachim Eastwood [Thu, 18 Oct 2012 11:01:09 +0000 (11:01 +0000)]
net/at91_ether: use macb defs for rx dma buffers

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use macb access functions
Joachim Eastwood [Thu, 18 Oct 2012 11:01:08 +0000 (11:01 +0000)]
net/at91_ether: use macb access functions

Use macb read/write funtions and remove the old at91_ether ones.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/at91_ether: use macb register definitions
Joachim Eastwood [Thu, 18 Oct 2012 11:01:07 +0000 (11:01 +0000)]
net/at91_ether: use macb register definitions

Use register and bits definitions from the macb header. This makes it
possible to have one header file for this hardware.

Process was scripted and the resulting object file has the same checksum.

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agonet/macb: add AT91RM9200 specific registers and bits to header
Joachim Eastwood [Thu, 18 Oct 2012 11:01:06 +0000 (11:01 +0000)]
net/macb: add AT91RM9200 specific registers and bits to header

Signed-off-by: Joachim Eastwood <manabian@gmail.com>
12 years agoigb: Split igb_update_dca into separate Tx and Rx functions
Alexander Duyck [Tue, 25 Sep 2012 00:31:27 +0000 (00:31 +0000)]
igb: Split igb_update_dca into separate Tx and Rx functions

This change makes it so that igb_update_dca is broken into two halves, one
for Rx and one for Tx.  The advantage to this is primarily readability.

In addition I am enabling relaxed ordering for reads from hardware since
this is supported on all of the igb parts.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Move the calls to set the Tx and Rx queues into igb_open
Alexander Duyck [Tue, 25 Sep 2012 00:31:22 +0000 (00:31 +0000)]
igb: Move the calls to set the Tx and Rx queues into igb_open

This change helps to address locking issues seen with
netif_set_real_num_tx_queues and netif_set_real_num_rx_queues when used in
the igb_set_interrupt_capability function.  To resolve these locking issues
I have moved the two function calls into __igb_open so that they can be
called while the RTNL lock is held.

An added advantage to this is that the number of queues is not updated
until the last possible moment so if there are any issues in allocating
MSI-X interrupts or resources for the rings we have time to change the
values prior to updating the netdev.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Combine q_vector and ring allocation into a single function
Alexander Duyck [Tue, 25 Sep 2012 00:31:17 +0000 (00:31 +0000)]
igb: Combine q_vector and ring allocation into a single function

This change combines the the allocation of q_vectors and rings into a single
function.  The advantage of this is that we are guaranteed we will avoid
overlap in the L1 cache sets.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Lock buffer size at 2K even on systems with larger pages
Alexander Duyck [Tue, 25 Sep 2012 00:31:12 +0000 (00:31 +0000)]
igb: Lock buffer size at 2K even on systems with larger pages

This change locks us in at 2K buffers even on a system that supports larger
frames.  The reason for this change is to make better use of pages and to
reduce the overall truesize of frames generated by igb.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Move rx_buffer related code in Rx cleanup path into separate function
Alexander Duyck [Tue, 25 Sep 2012 00:31:07 +0000 (00:31 +0000)]
igb: Move rx_buffer related code in Rx cleanup path into separate function

In order to try and isolate things a bit further I am moving the code
related to retrieving data from the rx_buffer_info structure into a
separate function.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Map entire page and sync half instead of mapping and unmapping half pages
Alexander Duyck [Tue, 25 Sep 2012 00:31:02 +0000 (00:31 +0000)]
igb: Map entire page and sync half instead of mapping and unmapping half pages

This change makes it so that we map the entire page and just sync half of
it for the device at a time.  The advantage to this approach is that we can
avoid the locking on map/unmap seen in many IOMMU implementations.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Combine post-processing of skb into a single function
Alexander Duyck [Tue, 25 Sep 2012 00:30:57 +0000 (00:30 +0000)]
igb: Combine post-processing of skb into a single function

This change is meant to just clean-up a number of function calls that were
made at the end of the Rx clean-up path by combining them into a single
function call.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Do not use header split, instead receive all frames into a single buffer
Alexander Duyck [Tue, 25 Sep 2012 00:30:52 +0000 (00:30 +0000)]
igb: Do not use header split, instead receive all frames into a single buffer

This change makes it so that we no longer use header split.  The idea is to
reduce partial cache line writes by hardware when handling frames larger
then header size.  We can compensate for the extra overhead of having to
memcpy the header buffer by avoiding the cache misses seen by leaving an
full skb allocated and sitting on the ring.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Split Rx timestamping into two separate functions
Alexander Duyck [Tue, 25 Sep 2012 05:14:55 +0000 (05:14 +0000)]
igb: Split Rx timestamping into two separate functions

In order to support page based receive we will need to split up the two
different types of timestamping into two separate functions.  The first one
will handle legacy timestamps with the value in the register, and the new
one will handle timestamps in the Rx buffer itself.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Acked-by: Matthew Vick <matthew.vick@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Correcting and improving small packet check and padding
Tushar Dave [Fri, 14 Sep 2012 03:43:43 +0000 (03:43 +0000)]
igb: Correcting and improving small packet check and padding

Current implementation mess up the tail pointer. This patch sets skb->tail
correctly.
Also, the small packet check and padding is optimized by using unlikely and
calling skb_pad directly.

Signed-off-by: Tushar Dave <tushar.n.dave@intel.com>
Tested-by: Jeff Pieper <jeffrey.e.pieper@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Add mailbox API version negotiation support to ixgbe PF
Alexander Duyck [Fri, 20 Jul 2012 08:09:22 +0000 (08:09 +0000)]
ixgbe: Add mailbox API version negotiation support to ixgbe PF

This change allows us to add a mailbox versioning API.  This will allow us
to determine the features supported by the VFs from the PF.  For example we
will be implementing a version 1.1 API for the VF that will indicate that
it can support us enabling Jumbo frames as the VF will support buffer
chaining.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Robert Garrett <RobertX.Garrett@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Move message handling routines into their own functions
Alexander Duyck [Fri, 20 Jul 2012 08:09:17 +0000 (08:09 +0000)]
ixgbe: Move message handling routines into their own functions

Instead of trying to maintain one large monolithic function that handles
most of the different messages from the VF it makes sense to break the
message handling function up so that we can just go through one switch
statement and call the correct routine for a given message.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Enable jumbo frames support w/ SR-IOV
Alexander Duyck [Wed, 15 Aug 2012 02:10:43 +0000 (02:10 +0000)]
ixgbe: Enable jumbo frames support w/ SR-IOV

This change makes it so that we can have limited support for jumbo frames
when SR-IOV is enabled.  In order to accomplish this it is necessary to
disable all VFs when the PF has jumbo frames enabled.  If the VFs then
request the same maximum frame size as the PF they will be re-enabled.  A
follow on patch will add a means of identifying when a VF can support
spanning buffers and does not need to be worried about the actual supported
max frame size.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Tested-by: Robert Garrett <robertx.e.garrett@intel.com>
Tested-by: Sibai Li <Sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Initialize q_vector cpu and affinity masks correctly
Alexander Duyck [Fri, 27 Jul 2012 23:49:30 +0000 (23:49 +0000)]
ixgbe: Initialize q_vector cpu and affinity masks correctly

When enabling DCB the rings belonging to a q_vector on CPU 0 were not
reinitializing their DCA registers.  Upon closer inspection the issue was
that the q_vector CPU variable was left at 0 resulting in the driver not
updating the DCA registers.

In order to guarantee the DCA registers will be updated I am adding a
couple line change so that we initialize the CPU variable to -1 which will
force a DCA update the first time an interrupt fires on that q_vector.

In addition we were setting the CPU affinity hint to all CPUs when we were
not specifying a CPU.  Instead we should leave it as all zeros to avoid any
possible confusion about the fact that we shouldn't be giving a hint.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Ross Brattain <ross.b.brattain@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoMerge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge
David S. Miller [Thu, 18 Oct 2012 19:36:59 +0000 (15:36 -0400)]
Merge tag 'batman-adv-fix-for-davem' of git://git.open-mesh.org/linux-merge

Included fixes:
- Fix broadcast packet CRC calculation which can lead to ~80% broadcast packet
  loss
- Fix a race condition in duplicate broadcast packet check

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotcp: fix FIONREAD/SIOCINQ
Eric Dumazet [Thu, 18 Oct 2012 09:14:12 +0000 (09:14 +0000)]
tcp: fix FIONREAD/SIOCINQ

tcp_ioctl() tries to take into account if tcp socket received a FIN
to report correct number bytes in receive queue.

But its flaky because if the application ate the last skb,
we return 1 instead of 0.

Correct way to detect that FIN was received is to test SOCK_DONE.

Reported-by: Elliot Hughes <enh@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Cc: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: qmi_wwan: adding more ZTE devices
Bjørn Mork [Thu, 18 Oct 2012 05:11:29 +0000 (05:11 +0000)]
net: qmi_wwan: adding more ZTE devices

Analyzed a few Windows driver description files, supporting
this long list of devices:

%ztewwan.DeviceDesc0002%    = ztewwan.ndi, USB\VID_19D2&PID_0002&MI_01
%ztewwan.DeviceDesc0012%    = ztewwan.ndi, USB\VID_19D2&PID_0012&MI_01
%ztewwan.DeviceDesc0017%    = ztewwan.ndi, USB\VID_19D2&PID_0017&MI_03
%ztewwan.DeviceDesc0021%    = ztewwan.ndi, USB\VID_19D2&PID_0021&MI_04
%ztewwan.DeviceDesc0025%    = ztewwan.ndi, USB\VID_19D2&PID_0025&MI_01
%ztewwan.DeviceDesc0031%    = ztewwan.ndi, USB\VID_19D2&PID_0031&MI_04
%ztewwan.DeviceDesc0042%    = ztewwan.ndi, USB\VID_19D2&PID_0042&MI_04
%ztewwan.DeviceDesc0049%    = ztewwan.ndi, USB\VID_19D2&PID_0049&MI_05
%ztewwan.DeviceDesc0052%    = ztewwan.ndi, USB\VID_19D2&PID_0052&MI_04
%ztewwan.DeviceDesc0055%    = ztewwan.ndi, USB\VID_19D2&PID_0055&MI_01
%ztewwan.DeviceDesc0058%    = ztewwan.ndi, USB\VID_19D2&PID_0058&MI_04
%ztewwan.DeviceDesc0063%    = ztewwan.ndi, USB\VID_19D2&PID_0063&MI_04
%ztewwan.DeviceDesc2002%    = ztewwan.ndi, USB\VID_19D2&PID_2002&MI_04
%ztewwan.DeviceDesc0104%    = ztewwan.ndi, USB\VID_19D2&PID_0104&MI_04
%ztewwan.DeviceDesc0113%    = ztewwan.ndi, USB\VID_19D2&PID_0113&MI_05
%ztewwan.DeviceDesc0118%    = ztewwan.ndi, USB\VID_19D2&PID_0118&MI_05
%ztewwan.DeviceDesc0121%    = ztewwan.ndi, USB\VID_19D2&PID_0121&MI_05
%ztewwan.DeviceDesc0123%    = ztewwan.ndi, USB\VID_19D2&PID_0123&MI_04
%ztewwan.DeviceDesc0124%    = ztewwan.ndi, USB\VID_19D2&PID_0124&MI_05
%ztewwan.DeviceDesc0125%    = ztewwan.ndi, USB\VID_19D2&PID_0125&MI_06
%ztewwan.DeviceDesc0126%    = ztewwan.ndi, USB\VID_19D2&PID_0126&MI_05
%ztewwan.DeviceDesc1008%    = ztewwan.ndi, USB\VID_19D2&PID_1008&MI_04
%ztewwan.DeviceDesc1010%    = ztewwan.ndi, USB\VID_19D2&PID_1010&MI_04
%ztewwan.DeviceDesc1012%    = ztewwan.ndi, USB\VID_19D2&PID_1012&MI_04
%ztewwan.DeviceDesc1402%    = ztewwan.ndi, USB\VID_19D2&PID_1402&MI_02
%ztewwan.DeviceDesc0157%    = ztewwan.ndi, USB\VID_19D2&PID_0157&MI_05
%ztewwan.DeviceDesc0158%    = ztewwan.ndi, USB\VID_19D2&PID_0158&MI_03
%ztewwan.DeviceDesc1401%    = ztewwan.ndi, USB\VID_19D2&PID_1401&MI_02
%ztewwan.DeviceDesc0130%    = ztewwan.ndi, USB\VID_19D2&PID_0130&MI_01
%ztewwan.DeviceDesc0133%    = ztewwan.ndi, USB\VID_19D2&PID_0133&MI_03
%ztewwan.DeviceDesc0176%    = ztewwan.ndi, USB\VID_19D2&PID_0176&MI_03
%ztewwan.DeviceDesc0178%    = ztewwan.ndi, USB\VID_19D2&PID_0178&MI_03
%ztewwan.DeviceDesc0168%    = ztewwan.ndi, USB\VID_19D2&PID_0168&MI_04
;EuFi890
%ztewwan.DeviceDesc0191%    = ztewwan.ndi, USB\VID_19D2&PID_0191&MI_04
;AL621
%ztewwan.DeviceDesc0167%    = ztewwan.ndi, USB\VID_19D2&PID_0167&MI_04
;MF821
%ztewwan.DeviceDesc0199%    = ztewwan.ndi, USB\VID_19D2&PID_0199&MI_01
%ztewwan.DeviceDesc0200%    = ztewwan.ndi, USB\VID_19D2&PID_0200&MI_01
%ztewwan.DeviceDesc0257%    = ztewwan.ndi, USB\VID_19D2&PID_0257&MI_03
;MF821V
%ztewwan.DeviceDesc1018%    = ztewwan.ndi, USB\VID_19D2&PID_1018&MI_03
;MF91
%ztewwan.DeviceDesc1426%    = ztewwan.ndi, USB\VID_19D2&PID_1426&MI_02
;0141
%ztewwan.DeviceDesc1247%    = ztewwan.ndi, USB\VID_19D2&PID_1247&MI_04
%ztewwan.DeviceDesc1425%    = ztewwan.ndi, USB\VID_19D2&PID_1425&MI_02
%ztewwan.DeviceDesc1424%    = ztewwan.ndi, USB\VID_19D2&PID_1424&MI_02
%ztewwan.DeviceDesc1252%    = ztewwan.ndi, USB\VID_19D2&PID_1252&MI_04
%ztewwan.DeviceDesc1254%    = ztewwan.ndi, USB\VID_19D2&PID_1254&MI_04
%ztewwan.DeviceDesc1255A%   = ztewwan.ndi, USB\VID_19D2&PID_1255&MI_03
%ztewwan.DeviceDesc1255B%   = ztewwan.ndi, USB\VID_19D2&PID_1255&MI_04
%ztewwan.DeviceDesc1256%    = ztewwan.ndi, USB\VID_19D2&PID_1256&MI_04
%ztewwan.DeviceDesc1245%    = ztewwanCombB.ndi, USB\VID_19D2&PID_1245&MI_04
%ztewwan.DeviceDesc1021%    = ztewwan.ndi, USB\VID_19D2&PID_1021&MI_02

Adding the ones we were missing.

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonetlink: use kfree_rcu() in netlink_release()
Eric Dumazet [Thu, 18 Oct 2012 03:21:55 +0000 (03:21 +0000)]
netlink: use kfree_rcu() in netlink_release()

On some suspend/resume operations involving wimax device, we have
noticed some intermittent memory corruptions in netlink code.

Stéphane Marchesin tracked this corruption in netlink_update_listeners()
and suggested a patch.

It appears netlink_release() should use kfree_rcu() instead of kfree()
for the listeners structure as it may be used by other cpus using RCU
protection.

netlink_release() must set to NULL the listeners pointer when
it is about to be freed.

Also have to protect netlink_update_listeners() and
netlink_has_listeners() if listeners is NULL.

Add a nl_deref_protected() lockdep helper to properly document which
locks protects us.

Reported-by: Jonathan Kliegman <kliegs@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Stéphane Marchesin <marcheu@google.com>
Cc: Sam Leffler <sleffler@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv4: Fix flushing of cached routing informations
Steffen Klassert [Wed, 17 Oct 2012 21:17:44 +0000 (21:17 +0000)]
ipv4: Fix flushing of cached routing informations

Currently we can not flush cached pmtu/redirect informations via
the ipv4_sysctl_rtcache_flush sysctl. We need to check the rt_genid
of the old route and reset the nh exeption if the old route is
expired when we bind a new route to a nh exeption.

Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agovlan: allow to change type when no vlan device is hooked on netdev
Jiri Pirko [Wed, 17 Oct 2012 01:37:36 +0000 (01:37 +0000)]
vlan: allow to change type when no vlan device is hooked on netdev

vlan_info might be present but still no vlan devices might be there.
That is in case of vlan0 automatically added.

So in that case, allow to change netdev type.

Reported-by: Jon Stanley <jstanley@rmrf.net>
Signed-off-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agophy: add AT803x driver
Matus Ujhelyi [Sun, 14 Oct 2012 19:07:16 +0000 (19:07 +0000)]
phy: add AT803x driver

This driver add support for wake over lan on AT803x phys.

Signed-off-by: Matus Ujhelyi <ujhelyi.m@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agousb/ipheth: Add iPhone 5 support
Jay Purohit [Sun, 14 Oct 2012 07:07:21 +0000 (07:07 +0000)]
usb/ipheth: Add iPhone 5 support

I noticed that the iPhone ethernet driver did not support
iPhone 5. I quickly added support to it in my kernel, here's
a patch.

Signed-off-by: Jay Purohit <jspurohit@velocitylimitless.com>
Acked-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
Signed-off-by: Jan Ceuleers <jan.ceuleers@computer.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'master' of git://1984.lsi.us.es/nf
David S. Miller [Thu, 18 Oct 2012 19:17:09 +0000 (15:17 -0400)]
Merge branch 'master' of git://1984.lsi.us.es/nf

Pablo Neira Ayuso Says:

====================
The following patchset contains Netfilter/IPVS updates for your
net tree, they are:

* Fix incorrect hooks for SNAT and DNAT (bug introduced in recent IPv6
  NAT changes), from Elison Niven.

* Fix xt_TEE (got broken with recent rt_gateway semantic change),
  from Eric Dumazet.

* Fix custom conntrack timeout policy attachment for IPv6, from myself.

* Always initialize ip_vs_timeout_user in case that TCP or UDP protocols
  is disabled, from Arnd Bergmann.

Note that I had to pull from your tree to obtain:

(c92b96553a80c1 ipv4: Add FLOWI_FLAG_KNOWN_NH)

which was required for the xt_TEE fix.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobatman-adv: Fix potential broadcast BLA-duplicate-check race condition
Linus Lüssing [Wed, 17 Oct 2012 12:53:05 +0000 (14:53 +0200)]
batman-adv: Fix potential broadcast BLA-duplicate-check race condition

Threads in the bottom half of batadv_bla_check_bcast_duplist() might
otherwise for instance overwrite variables which other threads might
be using/reading at the same time in the top half, potentially
leading to messing up the bcast_duplist, possibly resulting in false
bridge loop avoidance duplicate check decisions.

Signed-off-by: Linus Lüssing <linus.luessing@web.de>
Acked-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
12 years agobatman-adv: Fix broadcast packet CRC calculation
Linus Lüssing [Wed, 17 Oct 2012 12:53:04 +0000 (14:53 +0200)]
batman-adv: Fix broadcast packet CRC calculation

So far the crc16 checksum for a batman-adv broadcast data packet, received
on a batman-adv hard interface, was calculated over zero bytes of its
content leading to many incoming broadcast data packets wrongly being
dropped (60-80% packet loss).

This patch fixes this issue by calculating the crc16 over the actual,
complete broadcast payload.

The issue is a regression introduced by
("batman-adv: add broadcast duplicate check").

Signed-off-by: Linus Lüssing <linus.luessing@web.de>
Acked-by: Simon Wunderlich <siwu@hrz.tu-chemnitz.de>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
12 years agoapparmor: fix apparmor OOPS in audit_log_untrustedstring+0x1c/0x40
John Johansen [Wed, 17 Oct 2012 20:29:33 +0000 (13:29 -0700)]
apparmor: fix apparmor OOPS in audit_log_untrustedstring+0x1c/0x40

The capability defines have moved causing the auto generated names
of capabilities that apparmor uses in logging to be incorrect.

Fix the autogenerated table source to uapi/linux/capability.h

Reported-by: YanHong <clouds.yan@gmail.com>
Reported-by: Krzysztof Kolasa <kkolasa@winsoft.pl>
Analyzed-by: Al Viro <viro@ZenIV.linux.org.uk>
Signed-off-by: John Johansen <john.johansen@canonical.com>
Acked-by: David Howells <dhowells@redhat.com>
Acked-by: James Morris <james.l.morris@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
12 years agoMerge tag 'uapi-fixes-20121017' of git://git.infradead.org/users/dhowells/linux-headers
Linus Torvalds [Wed, 17 Oct 2012 20:40:22 +0000 (13:40 -0700)]
Merge tag 'uapi-fixes-20121017' of git://git.infradead.org/users/dhowells/linux-headers

Pull misc UAPI fixes from David Howells:
 "They do a number of things:

  (1) Import a patch from Catalin Marinas to extend the generic-y in
      Kbuild facility to uapi directories.

  (2) Make arch/tile's ucontext.h file use (1) and remove the header-y
      line from the kernel internal side of things.

  (3) Remove some now-empty conditional bits from include/linux/Kbuild.
      The contents got moved to the UAPI side of things along with new
      conditionals.

  (4) Deal with now-empty files:

     (a) Empty Kbuild files under include/ get removed.

     (b) Empty Kbuild files under arch/ get comments to hold them as
         they are likely to end up with generic-y or genhdr-y lines.
         Deleting them appears to work if we want to go that route.

     (c) Put a comment into uapi/asm-generic/kvm_para.h to prevent the
         patch program from deleting that, and made the arches with
         empty kvm_para.h uapi files use that instead of having their
         own files.

     (d) Put comments into four other empty uapi/ headers to prevent the
         patch program from deleting them.

  A question: Is this the right way to deal with the now-empty Kbuild
  files?

  The ones under include/ are unlikely to be used - even for generated
  files, I think - so getting rid of them is probably okay.  Once all
  the bits are in, we can probably remove all the Kbuild files under
  include/ that aren't also under include/uapi/.

  The ones under arch/ are more of an issue because of the potential for
  generic-y and genhdr-y."

* tag 'uapi-fixes-20121017' of git://git.infradead.org/users/dhowells/linux-headers:
  UAPI: Make arch/sparc/include/uapi/asm/sigcontext.h non-empty
  UAPI: Make arch/sh/include/uapi/asm/hw_breakpoint.h non-empty
  UAPI: Make arch/mn10300/include/uapi/asm/setup.h non-empty
  UAPI: Put a comment into uapi/asm-generic/kvm_para.h and use it from arches
  UAPI: The tile arch uses the generic ucontext.h file
  UAPI: Place comments in empty arch Kbuilds to make them non-empty
  UAPI: Remove empty non-UAPI Kbuild files
  UAPI: Remove empty conditionals from include/linux/Kbuild
  UAPI: Make uapi/linux/irqnr.h non-empty
  uapi: Allow automatic generation of uapi/asm/ header files