openwrt/staging/blogic.git
11 years agoMerge branch 'for-davem' of git://git.kernel.org/pub/scm/linux/kernel/git/linville...
David S. Miller [Wed, 28 Nov 2012 22:49:16 +0000 (17:49 -0500)]
Merge branch 'for-davem' of git://git./linux/kernel/git/linville/wireless-next

John W. Linville says:

====================
This pull request is intended for the 3.8 stream.  It is a bit large
-- I guess Thanksgiving got me off track!  At least the code got to
spend some time in linux-next... :-)

This includes the usual batch of pulls for Bluetooth, NFC, and mac80211
as well as iwlwifi.  Also here is an ath6kl pull, and a new driver
in the rtlwifi family.  The brcmfmac, brcmsmac, ath9k, and mwl8k get
their usual levels of attention, and a handful of other updates tag
along as well.

For more detail on the pulls, please see below...

On Bluetooth, Gustavo says:

"Another set of patches for integration in wireless-next. There are two big set
of changes in it: Andrei Emeltchenko and Mat Martineau added more patches
towards a full Bluetooth High Speed support and Johan Hedberg improve the
single mode support for Bluetooth dongles. Apart from that we have small fixes
and improvements."

...and:

"A few patches to 3.8. The majority of the work here is from Andrei on the High
Speed support. Other than that Johan added support for setting LE advertising
data. The rest are fixes and clean ups and small improvements like support for
a new broadcom hardware."

On mac80211, Johannes says:

"This is for mac80211, for -next (3.8). Plenty of changes, as you can see
below. Some fixes for previous changes like the export.h include, the
beacon listener fix from Ben Greear, etc. Overall, no exciting new
features, though hwsim does gain channel context support for people to
try it out and look at."

...and...:

"This one contains the mac80211-next material. Apart from a few small new
features and cleanups I have two fixes for the channel context code. The
RX_END timestamp support will probably be reworked again as Simon Barber
noted the calculations weren't really valid, but the discussions there
are still going on and it's better than what we had before."

...and:

"Please pull (see below) to get the following changes:
 * a fix & a debug aid in IBSS from Antonio,
 * mesh cleanups from Marco,
 * a few bugfixes for some of my previous patches from Arend and myself,
 * and the big initial VHT support patchset"

And on iwlwifi, Johannes says:

"In addition to the previous four patches that I'm not resending,
we have a number of cleanups, message reduction, firmware error
handling improvements (yes yes... we need to fix them instead)
and various other small things all over."

...and:

"In his quest to try to understand the current iwlwifi problems (like
stuck queues etc.) Emmanuel has first cleaned up the PCIe code, I'm
including his changes in this pull request. Other than that I only have
a small cleanup from Sachin Kamat to remove a duplicate include and a
bugfix to turn off MFP if software crypto is enabled, but this isn't
really interesting as MFP isn't supported right now anyway."

On NFC, Samuel says:

"With this one we have:

- A few HCI improvements in preparation for an upcoming HCI chipset support.
- A pn544 code cleanup after the old driver was removed.
- An LLCP improvement for notifying user space when one peer stops ACKing I
  frames."

On ath6kl, Kalle says:

"Major changes this time are firmware recover support to gracefully
handle if firmware crashes, support for changing regulatory domain and
support for new ar6004 hardware revision 1.4. Otherwise there are just
smaller fixes or cleanups from different people."

Thats about it... :-)  Please let me know if there are problems!
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
11 years agocxgb3: Restore dependency on INET
Ben Hutchings [Wed, 28 Nov 2012 20:03:03 +0000 (20:03 +0000)]
cxgb3: Restore dependency on INET

Commit ff33c0e1885cda44dd14c79f70df4706f83582a0 ('net: Remove bogus
dependencies on INET') wrongly removed this dependency.  cxgb3 uses
the arp_send() function defined in net/ipv4/arp.c.

Reported-by: kbuild test robot <fengguang.wu@intel.com>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoip6tnl/sit: drop packet if ECN present with not-ECT
Nicolas Dichtel [Tue, 27 Nov 2012 03:07:11 +0000 (03:07 +0000)]
ip6tnl/sit: drop packet if ECN present with not-ECT

This patch reports the change made by Stephen Hemminger in ipip and gre[6] in
commit eccc1bb8d4b4 (tunnel: drop packet if ECN present with not-ECT).

Goal is to handle RFC6040, Section 4.2:

Default Tunnel Egress Behaviour.
 o If the inner ECN field is Not-ECT, the decapsulator MUST NOT
      propagate any other ECN codepoint onwards.  This is because the
      inner Not-ECT marking is set by transports that rely on dropped
      packets as an indication of congestion and would not understand or
      respond to any other ECN codepoint [RFC4774].  Specifically:

      *  If the inner ECN field is Not-ECT and the outer ECN field is
         CE, the decapsulator MUST drop the packet.

      *  If the inner ECN field is Not-ECT and the outer ECN field is
         Not-ECT, ECT(0), or ECT(1), the decapsulator MUST forward the
         outgoing packet with the ECN field cleared to Not-ECT.

The patch takes benefits from common function added in net/inet_ecn.h.

Like it was done for Xin4 tunnels, it adds logging to allow detecting broken
systems that set ECN bits incorrectly when tunneling (or an intermediate
router might be changing the header). Errors are also tracked via
rx_frame_error.

CC: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobe2net: fix INTx ISR for interrupt behaviour on BE2
Sathya Perla [Tue, 27 Nov 2012 19:50:02 +0000 (19:50 +0000)]
be2net: fix INTx ISR for interrupt behaviour on BE2

On BE2 chip, an interrupt may be raised even when EQ is in un-armed state.
As a result be_intx()::events_get() and be_poll:events_get() can race and
notify an EQ wrongly.

Fix this by counting events only in be_poll(). Commit 0b545a629 fixes
the same issue in the MSI-x path.

But, on Lancer, INTx can be de-asserted only by notifying num evts. This
is not an issue as the above BE2 behavior doesn't exist/has never been
seen on Lancer.

Signed-off-by: Sathya Perla <sathya.perla@emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoewrk3: remove outdated comment
Paul Bolle [Mon, 26 Nov 2012 23:48:34 +0000 (23:48 +0000)]
ewrk3: remove outdated comment

Remove an outdated comment, that should have been removed in the
patch named "MODULE_PARM conversions" from early 2005.

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoewrk3: silence GCC warning
Paul Bolle [Mon, 26 Nov 2012 23:47:17 +0000 (23:47 +0000)]
ewrk3: silence GCC warning

Building ewrk3.o triggers this GCC warning:
    drivers/net/ethernet/dec/ewrk3.c: In function '__check_irq':
    drivers/net/ethernet/dec/ewrk3.c:1915:1: warning: return from incompatible pointer type [enabled by default]

This can be trivially fixed by changing the 'irq' parameter from int to
byte (which is an alias for unsigned char for module parameters).

Signed-off-by: Paul Bolle <pebolle@tiscali.nl>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMerge branch 'for-davem' of git://gitorious.org/linux-can/linux-can-next
David S. Miller [Wed, 28 Nov 2012 16:30:57 +0000 (11:30 -0500)]
Merge branch 'for-davem' of git://gitorious.org/linux-can/linux-can-next

Marc Kleine-Budde says:

====================
this is pull request is for net-next. Contains a patch by Andreas
Larsson, which enables the sja1000 of driver to work under sparc.
AnilKumar Ch contributed a patch to improve the c_can support under
omap, Olivier Sobrie's patch brings support for the CAN/USB dongles
from Kvaser. In a bunch of patches by me missing MODULE_ALIAS and/or
MODULE_DEVICE_TABLE entries were added to the CAN drivers.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobonding: in balance-rr mode, set curr_active_slave only if it is up
Michal Kubeček [Thu, 22 Nov 2012 02:48:39 +0000 (02:48 +0000)]
bonding: in balance-rr mode, set curr_active_slave only if it is up

If all slaves of a balance-rr bond with ARP monitor are enslaved
with down link state, bond keeps down state even after slaves
go up.

This is caused by bond_enslave() setting curr_active_slave to
first slave not taking into account its link state. As
bond_loadbalance_arp_mon() uses curr_active_slave to identify
whether slave's down->up transition should update bond's link
state, bond stays down even if slaves are up (until first slave
goes from up to down at least once).

Before commit f31c7937 "bonding: start slaves with link down for
ARP monitor", this was masked by slaves always starting in UP
state with ARP monitor (and MII monitor not relying on
curr_active_slave being NULL if there is no slave up).

Signed-off-by: Michal Kubecek <mkubecek@suse.cz>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agopkt_sched: QFQ Plus: fair-queueing service at DRR cost
Paolo Valente [Fri, 23 Nov 2012 11:03:19 +0000 (11:03 +0000)]
pkt_sched: QFQ Plus: fair-queueing service at DRR cost

This patch turns QFQ into QFQ+, a variant of QFQ that provides the
following two benefits: 1) QFQ+ is faster than QFQ, 2) differently
from QFQ, QFQ+ correctly schedules also non-leaves classes in a
hierarchical setting. A detailed description of QFQ+, plus a
performance comparison with DRR and QFQ, can be found in [1].

[1] P. Valente, "Reducing the Execution Time of Fair-Queueing Schedulers"
http://algo.ing.unimo.it/people/paolo/agg-sched/agg-sched.pdf

Signed-off-by: Paolo Valente <paolo.valente@unimore.it>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: don't enable remote wakeup directly
Steve Glendinning [Wed, 28 Nov 2012 05:59:50 +0000 (05:59 +0000)]
smsc75xx: don't enable remote wakeup directly

As pointed out by Bjorn Mork, the generic "usb" driver sets this
for us so no need to directly set it in this driver.

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Cc: Bjorn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: add support for USB dynamic autosuspend
Steve Glendinning [Wed, 28 Nov 2012 05:59:49 +0000 (05:59 +0000)]
smsc75xx: add support for USB dynamic autosuspend

This patch adds support for USB dynamic autosuspend to the
smsc75xx driver.  This saves virtually no power in the USB
device but enables power savings in upstream hosts and
the host CPU.

Note currently Linux doesn't automatically enable this
functionality by default for devices so to test this:

 echo auto > /sys/bus/usb/devices/2-1.2/power/control

where 2-1.2 is the USB bus address of the LAN7500.

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: fix error handling in suspend failure case
Steve Glendinning [Wed, 28 Nov 2012 05:59:48 +0000 (05:59 +0000)]
smsc75xx: fix error handling in suspend failure case

This patch ensures that if we fail to suspend the LAN7500 device
we call usbnet_resume before returning failure, instead of
leaving the usbnet driver in an unusable state.

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: support PHY wakeup source
Steve Glendinning [Wed, 28 Nov 2012 05:59:47 +0000 (05:59 +0000)]
smsc75xx: support PHY wakeup source

This patch enables LAN7500 family devices to wake from suspend
on either link up or link down events.

It also adds _nopm versions of mdio access functions, so we can
safely call them from suspend and resume functions

Updated patch to add newlines to printk messages

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: refactor entering suspend modes
Steve Glendinning [Wed, 28 Nov 2012 05:59:46 +0000 (05:59 +0000)]
smsc75xx: refactor entering suspend modes

This patch splits out the logic for entering suspend modes
to separate functions, to reduce the complexity of the
smsc75xx_suspend function.

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc75xx: check return code from smsc75xx_reset
Steve Glendinning [Wed, 28 Nov 2012 05:59:45 +0000 (05:59 +0000)]
smsc75xx: check return code from smsc75xx_reset

This patch adds a missing check and error message if smsc75xx_reset
fails.

Signed-off-by: Steve Glendinning <steve.glendinning@shawell.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: fix coding style issues in qlcnic_minidump.c
Sony Chacko [Wed, 28 Nov 2012 04:34:31 +0000 (04:34 +0000)]
qlcnic: fix coding style issues in qlcnic_minidump.c

Fix coding style violations in qlcnic_minidump.c

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: create file qlcnic_minidump.c for dump utility
Sony Chacko [Wed, 28 Nov 2012 04:34:30 +0000 (04:34 +0000)]
qlcnic: create file qlcnic_minidump.c for dump utility

Physical refactoring of 82xx adapter register dump utility.

Move register dump routines to new file qlcnic_minidump.c
Existing register dump routines has coding style issues, the code
is moved to the new file without fixing the style issues.

There is a seperate patch to fix the style issues in qlcnic_minidump.c

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: fix style issues in qlcnic_sysfs.c file
Sony Chacko [Wed, 28 Nov 2012 04:34:29 +0000 (04:34 +0000)]
qlcnic: fix style issues in qlcnic_sysfs.c file

Fix coding style issues in qlcnic_sysfs.c file

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: create file qlcnic_sysfs.c for sysfs routines
Sony Chacko [Wed, 28 Nov 2012 04:34:28 +0000 (04:34 +0000)]
qlcnic: create file qlcnic_sysfs.c for sysfs routines

Physical refactoring of 82xx adapter sysfs routines.

Move sysfs routines to new file qlcnic_sysfs.c
Existing sysfs routines has coding style issues, this code is
moved to the new file without fixing the style issues.

There is a seperate patch to fix the style issues in qlcnic_sysfs.c

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: fix coding style issues in qlcnic_io.c
Sony Chacko [Wed, 28 Nov 2012 04:34:27 +0000 (04:34 +0000)]
qlcnic: fix coding style issues in qlcnic_io.c

Fix coding style issues in qlcnic_io.c

Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: create file qlcnic_io.c for datapath routines
Sony Chacko [Wed, 28 Nov 2012 04:34:26 +0000 (04:34 +0000)]
qlcnic: create file qlcnic_io.c for datapath routines

Physical refactoring of 82xx adapter data path routines.

Move data path code to new file qlcnic_io.c
Existing data path code has coding stye issues, the code is
moved to the new file without fixing the style issues.

There is a seperate patch to fix the style issues in qlcnic_io.c

Signed-off-by: Sony Chacko <sony.chacko@qlogic.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 [Wed, 28 Nov 2012 16:05:00 +0000 (11:05 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/jkirsher/net-next

Jeff Kirsher says:

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

Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Change duplex setting in EEE function
Yaniv Rosner [Tue, 27 Nov 2012 03:46:36 +0000 (03:46 +0000)]
bnx2x: Change duplex setting in EEE function

This is not a real problem, since the EEE is supported for devices where the
actual_phy_selection is zero, such that the req_duplex of params will match
the one of the phy struct.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Add warning message in case of non-10G SFP module
Yaniv Rosner [Tue, 27 Nov 2012 03:46:35 +0000 (03:46 +0000)]
bnx2x: Add warning message in case of non-10G SFP module

The string was split to several lines since it reached over 180 chars, which
seems too much.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Cosmetic changes
Yaniv Rosner [Tue, 27 Nov 2012 03:46:34 +0000 (03:46 +0000)]
bnx2x: Cosmetic changes

This patch makes some cosmetic changes to the code:
1. Code alignment.
2. Merge read-modify-write into a single function (read_or_write /
read_and_write).
3. Merge several write registers into a for-loop write using a static array.
4. Remove empty lines.
5. Fix comments.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Always take PHY lock
Yaniv Rosner [Tue, 27 Nov 2012 03:46:33 +0000 (03:46 +0000)]
bnx2x: Always take PHY lock

Taking PHY lock is not required on some older designs, but we are removing this
complication and always taking it since it is always required on newer designs
and does not worth the code complication on the older boards.

Taking PHY lock was initially required only on specific boards which had their
MDC/MDIO bus crossed, but since this lock is now always required, for example,
when NCSI is present, the PHY lock will always be taken.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Add support for BCM84834
Yaniv Rosner [Tue, 27 Nov 2012 03:46:32 +0000 (03:46 +0000)]
bnx2x: Add support for BCM84834

Add support for the 10G-baseT PHY - BCM84834, which is the quad-port version of
the dual-port BCM84833.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Fix SFP+ current leakage
Yaniv Rosner [Tue, 27 Nov 2012 03:46:31 +0000 (03:46 +0000)]
bnx2x: Fix SFP+ current leakage

Per measurements, the SFP+ suffered from small current leakage in two cases:
 - When no module was plugged and TX laser was disabled. The fix was to enable
   it, and when module is plugged in, check if it needs to be disabled.
 - When over-current event occurs due to invalid SFP+ module, the HW basically
   shuts down the current for this module, but the SW needs to complete this
   by issuing a power down via a GPIO.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Change MDIO clock settings
Yaniv Rosner [Tue, 27 Nov 2012 03:46:30 +0000 (03:46 +0000)]
bnx2x: Change MDIO clock settings

When drivers works on top of an old bootcode, it is theoretically subjected to
MDC/MDIO failures since the MDIO clock is set in the beginning of each sequence,
rather than per CL45 command. On rare cases an old bootcodes may change that in
the middle, so to address that, the MDIO clock is set for each CL45 access.
In addition, setting the MDIO clock is now done per EMAC base, and
not per port number, since a specific port can potentially use both EMACs for
different PHY accesses.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Add support for 20G-KR2
Yaniv Rosner [Tue, 27 Nov 2012 03:46:29 +0000 (03:46 +0000)]
bnx2x: Add support for 20G-KR2

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobnx2x: Activate LFA
Yaniv Rosner [Tue, 27 Nov 2012 03:46:28 +0000 (03:46 +0000)]
bnx2x: Activate LFA

In case Link Flap Avoidance feature is supported by the MCP, bnx2x will enable
it, and will pass the appropriate parameter when load request is sent to
the MCP.

Signed-off-by: Yaniv Rosner <yanivr@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.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/linville/wirel...
John W. Linville [Wed, 28 Nov 2012 15:56:03 +0000 (10:56 -0500)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next into for-davem

Conflicts:
drivers/net/wireless/iwlwifi/pcie/tx.c

12 years agoixgbe: bump version number
Don Skidmore [Fri, 9 Nov 2012 05:03:53 +0000 (05:03 +0000)]
ixgbe: bump version number

Move the version string to better reflect the driver functionality with
that of the out of tree driver.  Also since we no longer need the MAJ,
MIN, BUILD defines remove them to clean up the code.

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: Make the bridge mode setting sticky
Greg Rose [Thu, 8 Nov 2012 02:41:35 +0000 (02:41 +0000)]
ixgbe: Make the bridge mode setting sticky

The internal bridge mode setting needs to be sticky so that it can be
configured correctly after a device reset.  This change is required now
that the driver supports setting the bridge mode to VEB or VEPA.

Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Sibai Li <Sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Fix incorrect disabling of Tx hang check in case of PFC
Parikh, Neerav [Thu, 27 Sep 2012 12:02:22 +0000 (12:02 +0000)]
ixgbe: Fix incorrect disabling of Tx hang check in case of PFC

The XOFF received statistic registers are per priority based and not per
traffic class. The ixgbe driver was incorrectly considering them to be for
each traffic class; and then disabling the "Tx hang" check for the queues
that belonged to the particular traffic class that had received PFC frames.

The above logic worked fine in scenario where the user priority and traffic
class number matched e.g. priority 0 is mapped to traffic class 0 and so on.
But, when multiple user priorities are mapped to a single traffic class or
when user priorities and traffic class numbers do not line up; the ixgbe
driver may disable the "Tx hang" check for queues belonging to a traffic
class that did not receive PFC frames and keep the "Tx hang" check enabled
for the queues that did receive the PFC frames.

This patch corrects the above in the code by considering the statistics
on a per priority basis; then getting the traffic class the user priority
belongs to and disabling the "Tx hang" check for queues that belong
to that traffic class.

Signed-off-by: Neerav Parikh <Neerav.Parikh@intel.com>
Acked-by: John Fastabend <john.r.fastabend@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Marcus Dennis <marcusx.e.dennis@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoixgbe: Drop RLPML configuration from x540 RXDCTL register configuration
Alexander Duyck [Sat, 8 Sep 2012 02:13:38 +0000 (02:13 +0000)]
ixgbe: Drop RLPML configuration from x540 RXDCTL register configuration

Since we are doing a page based receive there is no point in setting a maximum
packet length on the x540 RXDCTL register.  As such we can drop the code from
the driver entirely.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Tested-by: Marcus Dennis <marcusx.e.dennis@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigb: Use a 32-bit mask when calculating the flow control watermarks
Matthew Vick [Thu, 8 Nov 2012 04:03:58 +0000 (04:03 +0000)]
igb: Use a 32-bit mask when calculating the flow control watermarks

For some devices, the result of the flow control high watermark gets
truncated when programming it into the registers because of the mask used.
Switch the mask to 32-bit to prevent this from happening.

Signed-off-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 agoigbvf: update version number
Mitch A Williams [Sat, 17 Nov 2012 07:10:16 +0000 (07:10 +0000)]
igbvf: update version number

Update version number.

Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agoigbvf: work around i350 erratum
Mitch A Williams [Sat, 17 Nov 2012 07:10:16 +0000 (07:10 +0000)]
igbvf: work around i350 erratum

On i350 VF devices, VLAN tags will be byte-swapped in the receive
descriptor only when received packets are looped back from other
VFs. Check for this condition and swab the tag if needed.

Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
12 years agocan: mpc5xxx_can: add MODULE_DEVICE_TABLE
Marc Kleine-Budde [Fri, 12 Oct 2012 08:19:27 +0000 (10:19 +0200)]
can: mpc5xxx_can: add MODULE_DEVICE_TABLE

This patch adds a MODULE_DEVICE_TABLE for the of bindings, so that the module
can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: c_can_platform: add MODULE_DEVICE_TABLE
Marc Kleine-Budde [Fri, 12 Oct 2012 07:48:06 +0000 (09:48 +0200)]
can: c_can_platform: add MODULE_DEVICE_TABLE

This patch adds a MODULE_DEVICE_TABLE for the of and platform bindings, so that
the module can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: at91_can: add MODULE_DEVICE_TABLE
Marc Kleine-Budde [Fri, 12 Oct 2012 07:44:51 +0000 (09:44 +0200)]
can: at91_can: add MODULE_DEVICE_TABLE

This patch adds a MODULE_DEVICE_TABLE for the platform bindings, so that the
module can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: flexcan: add MODULE_DEVICE_TABLE
Marc Kleine-Budde [Thu, 4 Oct 2012 08:55:35 +0000 (10:55 +0200)]
can: flexcan: add MODULE_DEVICE_TABLE

This patch adds a MODULE_DEVICE_TABLE for the of and platform bindings, so that
the module can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: cc770_platform: add MODULE_ALIAS and MODULE_DEVICE_TABLE
Marc Kleine-Budde [Fri, 12 Oct 2012 07:51:19 +0000 (09:51 +0200)]
can: cc770_platform: add MODULE_ALIAS and MODULE_DEVICE_TABLE

This patch adds a MODULE_ALIAS for the platform bindings and a
MODULE_DEVICE_TABLE for of bindings, so that the module can be loaded
automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: sja1000_platform: add MODULE_ALIAS
Marc Kleine-Budde [Fri, 12 Oct 2012 07:45:14 +0000 (09:45 +0200)]
can: sja1000_platform: add MODULE_ALIAS

This patch adds a MODULE_ALIAS for the platform bindings, so that the module
can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: ti_hecc: add MODULE_ALIAS
Marc Kleine-Budde [Mon, 29 Oct 2012 13:18:04 +0000 (14:18 +0100)]
can: ti_hecc: add MODULE_ALIAS

This patch adds a MODULE_ALIAS for the platform bindings, so that
the module can be loaded automatically udev.

Tested-by: Jan Lübbe <jlu@pengutronix.de>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: bfin_can: add MODULE_ALIAS
Marc Kleine-Budde [Fri, 12 Oct 2012 07:45:14 +0000 (09:45 +0200)]
can: bfin_can: add MODULE_ALIAS

This patch adds a MODULE_ALIAS for the platform bindings, so that the module
can be loaded automatically by udev.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: sja1000: Make sja1000_of_platform selectable and compilable on SPARC
Andreas Larsson [Thu, 4 Oct 2012 13:59:10 +0000 (15:59 +0200)]
can: sja1000: Make sja1000_of_platform selectable and compilable on SPARC

Signed-off-by: Andreas Larsson <andreas@gaisler.com>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: kvaser_usb: Add support for Kvaser CAN/USB devices
Olivier Sobrie [Wed, 21 Nov 2012 07:11:13 +0000 (08:11 +0100)]
can: kvaser_usb: Add support for Kvaser CAN/USB devices

This driver provides support for several Kvaser CAN/USB devices.
Such kind of devices supports up to three CAN network interfaces.

It has been tested with a Kvaser USB Leaf Light (one network interface)
connected to a pch_can interface.
The firmware version of the Kvaser device was 2.5.205.

List of Kvaser devices supported by the driver:
  - Kvaser Leaf Light
  - Kvaser Leaf Professional HS
  - Kvaser Leaf SemiPro HS
  - Kvaser Leaf Professional LS
  - Kvaser Leaf Professional SWC
  - Kvaser Leaf Professional LIN
  - Kvaser Leaf SemiPro LS
  - Kvaser Leaf SemiPro SWC
  - Kvaser Memorator II HS/HS
  - Kvaser USBcan Professional HS/HS
  - Kvaser Leaf Light GI
  - Kvaser Leaf Professional HS (OBD-II connector)
  - Kvaser Memorator Professional HS/LS
  - Kvaser Leaf Light "China"
  - Kvaser BlackBird SemiPro
  - Kvaser USBcan R

Signed-off-by: Daniel Berglund <db@kvaser.com>
Signed-off-by: Olivier Sobrie <olivier@sobrie.be>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agocan: c_can: Add d_can raminit support
AnilKumar Ch [Wed, 21 Nov 2012 05:44:10 +0000 (11:14 +0530)]
can: c_can: Add d_can raminit support

Add D_CAN raminit support to C_CAN driver to enable D_CAN RAM,
which holds all the message objects during transmission or
receiving of data. This initialization/de-initialization should
be done in synchronous with D_CAN clock.

In case of AM335X-EVM (current user of D_CAN driver) message RAM is
controlled through control module register for both instances. So
control module register details is required to initialization or
de-initialization of message RAM according to instance number.

Control module memory resource is obtained from D_CAN dt node and
instance number obtained from device tree aliases node.

This patch was tested on AM335x-EVM along with pinctrl data addition
patch, d_can dt aliases addition and control module data addition.
pinctrl data addition is not added to am335x-evm.dts (only supports
CPLD profile#0) because d_can1 is supported under CPLD profile#1.

Signed-off-by: AnilKumar Ch <anilkumar@ti.com>
[mkl: fix instance for non DT in probe, cleaned up raminit]
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
12 years agoip6mr: Add sizeof verification to MRT6_ASSERT and MT6_PIM
Joe Perches [Sun, 25 Nov 2012 18:26:34 +0000 (18:26 +0000)]
ip6mr: Add sizeof verification to MRT6_ASSERT and MT6_PIM

Verify the length of the user-space arguments.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosockopt: Change getsockopt() of SO_BINDTODEVICE to return an interface name
Brian Haley [Mon, 26 Nov 2012 05:21:08 +0000 (05:21 +0000)]
sockopt: Change getsockopt() of SO_BINDTODEVICE to return an interface name

Instead of having the getsockopt() of SO_BINDTODEVICE return an index, which
will then require another call like if_indextoname() to get the actual interface
name, have it return the name directly.

This also matches the existing man page description on socket(7) which mentions
the argument being an interface name.

If the value has not been set, zero is returned and optlen will be set to zero
to indicate there is no interface name present.

Added a seqlock to protect this code path, and dev_ifname(), from someone
changing the device name via dev_change_name().

v2: Added seqlock protection while copying device name.

v3: Fixed word wrap in patch.

Signed-off-by: Brian Haley <brian.haley@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agocpts: add missing kconfig dependency
Richard Cochran [Mon, 26 Nov 2012 02:07:18 +0000 (02:07 +0000)]
cpts: add missing kconfig dependency

The Common Platform Time Sync function of the CPSW does not depend the
CPSW configuration option as it should. This patch fixes the issue by
adding the dependency.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoptp: reduce stack usage when measuring the system time offset
Richard Cochran [Mon, 26 Nov 2012 01:44:35 +0000 (01:44 +0000)]
ptp: reduce stack usage when measuring the system time offset

This patch removes the large buffer from the stack of the system
offset ioctl and replaces it with a kmalloced buffer.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoptp: reduce stack usage when reading external time stamps
Richard Cochran [Mon, 26 Nov 2012 01:44:34 +0000 (01:44 +0000)]
ptp: reduce stack usage when reading external time stamps

This patch removes the large buffer from the stack of the read file
operation and replaces it with a kmalloced buffer.

Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: update the driver version to Nov_2012
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:46 +0000 (23:10 +0000)]
stmmac: update the driver version to Nov_2012

Many new feauture have been introduced in the driver:
ethtool coalesce options, Rx HW watchdog... so this patch updates the
driver's version.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: update the doc with new IRQ mitigation
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:45 +0000 (23:10 +0000)]
stmmac: update the doc with new IRQ mitigation

This patch updates the stmmac.txt adding some information
about the new rx/tx mitigation schema adopted in the driver.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: get/set coalesce parameters via ethtool
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:44 +0000 (23:10 +0000)]
stmmac: get/set coalesce parameters via ethtool

This patch is to get/set the tx/rx coalesce parameters
via ethtool interface.

Tests have been done on several platform with different GMAC chips w/o and w/
RX watchdog feature.

V2: reject coalesce settings that are not supported.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Reviewed-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: add Rx watchdog support to mitigate the DMA irqs
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:43 +0000 (23:10 +0000)]
stmmac: add Rx watchdog support to mitigate the DMA irqs

GMAC devices newer than databook 3.40 has an embedded timer
that can be used for mitigating the number of interrupts.
So this patch adds this optimizations.

At any rate, the Rx watchdog can be disable (on bugged HW) by
passing from the platform the riwt_off field.

In this implementation the rx timer stored in the Reg9 is fixed
to the max value. This will be tuned by using ethtool.

V2: added a platform parameter to force to disable the rx-watchdog
for example on new core where it is bugged.

V3: do not disable NAPI when Rx watchdog is used.

V4: a new extra statistic field has been added to show the early
receive status in the interrupt handler.
This patch also adds an extra check to avoid to call
napi_schedule when the DMA_INTR_ENA_RIE bit is disabled in the
Interrupt Mask register.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: add the initial tx coalesce schema
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:42 +0000 (23:10 +0000)]
stmmac: add the initial tx coalesce schema

This patch adds a new schema used for mitigating the
number of transmit interrupts.
It is based on a SW timer and a threshold value.
The timer is used to periodically call the stmmac_tx_clean
function; the threshold is used for setting the IC (Interrupt
on Completion bit). The ISR will then invoke the poll method.
Also the patch improves some ethtool stat fields.

V2: review the logic to manage the IC bit in the TDESC
that was bugged because it didn't take care about the
fragments. Also fix the tx_count_frames that has not to be
limited to TX DMA ring. Thanks to Ben Hutchings.

V3: removed the spin_lock irqsave/restore as D. Miller suggested.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agostmmac: remove dead code for STMMAC_TIMER support
Giuseppe CAVALLARO [Sun, 25 Nov 2012 23:10:41 +0000 (23:10 +0000)]
stmmac: remove dead code for STMMAC_TIMER support

The TIMER option is not longer supported and this
code can be considered dead for this driver in
the new kernel series.
In fact, It was not updated at all and never used.

Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agotun: put correct method name in a debug message.
Rami Rosen [Sun, 25 Nov 2012 22:07:41 +0000 (22:07 +0000)]
tun: put correct method name in a debug message.

This patch puts the correct method name, tun_do_read, in a debug message.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agovtun: fix typos.
Rami Rosen [Sun, 25 Nov 2012 22:07:40 +0000 (22:07 +0000)]
vtun: fix typos.

This patch fixes four typos in drivers/net/vtun.c.

Signed-off-by: Rami Rosen <ramirose@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoatm: br2684: Fix excessive queue bloat
David Woodhouse [Sun, 25 Nov 2012 12:06:52 +0000 (12:06 +0000)]
atm: br2684: Fix excessive queue bloat

There's really no excuse for an additional wmem_default of buffering
between the netdev queue and the ATM device. Two packets (one in-flight,
and one ready to send) ought to be fine. It's not as if it should take
long to get another from the netdev queue when we need it.

If necessary we can make the queue space configurable later, but I don't
think it's likely to be necessary.

cf. commit 9d02daf754238adac48fa075ee79e7edd3d79ed3 (pppoatm: Fix
excessive queue bloat) which did something very similar for PPPoATM.

Note that there is a tremendously unlikely race condition which may
result in qspace temporarily going negative. If a CPU running the
br2684_pop() function goes off into the weeds for a long period of time
after incrementing qspace to 1, but before calling netdev_wake_queue()...
and another CPU ends up calling br2684_start_xmit() and *stopping* the
queue again before the first CPU comes back, the netdev queue could
end up being woken when qspace has already reached zero.

An alternative approach to coping with this race would be to check in
br2684_start_xmit() for qspace==0 and return NETDEV_TX_BUSY, but just
using '> 0' and '< 1' for comparison instead of '== 0' and '!= 0' is
simpler. It just warranted a mention of *why* we do it that way...

Move the call to atmvcc->send() to happen *after* the accounting and
potentially stopping the netdev queue, in br2684_xmit_vcc(). This matters
if the ->send() call suffers an immediate failure, because it'll call
br2684_pop() with the offending skb before returning. We want that to
happen *after* we've done the initial accounting for the packet in
question. Also make it return an appropriate success/failure indication
while we're at it.

Tested by running 'ping -l 1000 bottomless.aaisp.net.uk' from within my
network, with only a single PPPoE-over-BR2684 link running. And after
setting txqueuelen on the nas0 interface to something low (5, in fact).
Before the patch, we'd see about 15 packets being queued and a resulting
latency of ~56ms being reached. After the patch, we see only about 8,
which is fairly much what we expect. And a max latency of ~36ms. On this
OpenWRT box, wmem_default is 163840.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Reviewed-by: Krzysztof Mazur <krzysiek@podlesie.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agodsa: Hide core config options; make drivers select what they need
Ben Hutchings [Mon, 26 Nov 2012 09:16:10 +0000 (09:16 +0000)]
dsa: Hide core config options; make drivers select what they need

Commit 82167cb8c6b2f8166d5c7532e5ef4b5e0cc46a72 ('net: dsa/slave: Fix
compilation warnings') fixed one possible invalid configuration
(NET_DSA enabled with no trailer formats) but added others: drivers
can select NET_DSA without its dependencies being met.

It's not very useful to make either the DSA core or the tagging
formats manually selectable without a driver to use them, so:

1. Define a hidden HAVE_NET_DSA option and move the dependencies of
   NET_DSA to that.  While we're at it, drop the deprecated
   EXPERIMENTAL dependency.
2. Make NET_DSA and the drivers dependent on HAVE_NET_DSA.
3. Hide the tagging format options again.
4. Make drivers select both NET_DSA and the appropriate tagging format
   option.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agobcma: add more package IDs
Hauke Mehrtens [Sat, 24 Nov 2012 18:34:17 +0000 (19:34 +0100)]
bcma: add more package IDs

Signed-off-by: Hauke Mehrtens <hauke@hauke-m.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agop54: improve TSF timestamp precision
Christian Lamparter [Sat, 24 Nov 2012 13:23:25 +0000 (14:23 +0100)]
p54: improve TSF timestamp precision

The LMAC API states that the TSF clock value of
every rx'ed frame is a "usec accurate timestamp
of the hardware clock at the end of frame
(before OFDM SIFS EOF padding)".

Signed-off-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoiwlwifi: initialize rx_status
Johannes Berg [Fri, 23 Nov 2012 22:46:19 +0000 (23:46 +0100)]
iwlwifi: initialize rx_status

The vendor radiotap patch added a few fields to
struct ieee80211_rx_status that need to be zero,
initialize the struct instead of using whatever
was left on the stack.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Tested-by: Kalle Valo <kvalo@adurom.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoiwlegacy: initialize rx_status
Johannes Berg [Thu, 22 Nov 2012 11:58:16 +0000 (12:58 +0100)]
iwlegacy: initialize rx_status

The vendor radiotap patch added a few fields to
struct ieee80211_rx_status that need to be zero,
initialize the struct instead of using whatever
was left on the stack.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agobrcmsmac: Add __printf verification to logging prototypes
Joe Perches [Wed, 21 Nov 2012 18:17:34 +0000 (10:17 -0800)]
brcmsmac: Add __printf verification to logging prototypes

Adding __printf helps spot format and argument mismatches.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
12 years agoMerge branch 'for-john' of git://git.kernel.org/pub/scm/linux/kernel/git/jberg/mac802...
John W. Linville [Mon, 26 Nov 2012 19:46:41 +0000 (14:46 -0500)]
Merge branch 'for-john' of git://git./linux/kernel/git/jberg/mac80211-next

12 years agocfg80211: fix some tracing output issues
Johannes Berg [Fri, 16 Nov 2012 16:22:28 +0000 (17:22 +0100)]
cfg80211: fix some tracing output issues

In some cases, e.g. probe_status, there were spaces
missing so the trace output was confusing. Also make
it more like mac80211 when printing netdevs/wiphys
to make reading a combined log easier.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211_hwsim: remove printing scan config
Johannes Berg [Tue, 20 Nov 2012 16:34:10 +0000 (17:34 +0100)]
mac80211_hwsim: remove printing scan config

The frequencies will be printed when actually
doing the scan, and the IEs can be captured
on the hwsim0 monitor.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211_hwsim: advertise VHT support
Johannes Berg [Mon, 12 Nov 2012 10:51:50 +0000 (11:51 +0100)]
mac80211_hwsim: advertise VHT support

If the number of channels is > 1, which means that
hwsim will use mac80211 channel contexts, it can
also advertise VHT support.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agowireless: add definitions for VHT MCS support
Johannes Berg [Mon, 12 Nov 2012 10:44:18 +0000 (11:44 +0100)]
wireless: add definitions for VHT MCS support

Add definitions for the VHT MCS support values that
are used to indicate, for each number of streams
(1 through 8) which MCSes are supported.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: support VHT rates in TX info
Johannes Berg [Fri, 9 Nov 2012 17:38:32 +0000 (18:38 +0100)]
mac80211: support VHT rates in TX info

To achieve this, limit the number of retries to
31 (instead of 255) and use the three bits that
are then free for VHT flags.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: support drivers reporting VHT RX
Johannes Berg [Fri, 9 Nov 2012 14:07:02 +0000 (15:07 +0100)]
mac80211: support drivers reporting VHT RX

Add support to mac80211 for having drivers report
received VHT MCS information.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agonl80211/cfg80211: add VHT MCS support
Johannes Berg [Fri, 9 Nov 2012 13:56:41 +0000 (14:56 +0100)]
nl80211/cfg80211: add VHT MCS support

Add support for reporting and calculating VHT MCSes.

Note that I'm not completely sure that the bitrate
calculations are correct, nor that they can't be
simplified.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: convert to channel definition struct
Johannes Berg [Fri, 9 Nov 2012 10:39:59 +0000 (11:39 +0100)]
mac80211: convert to channel definition struct

Convert mac80211 (and where necessary, some drivers a
little bit) to the new channel definition struct.

This will allow extending mac80211 for VHT, which is
currently restricted to channel contexts since there
are no drivers using that which makes it easier. As
I also don't care about VHT for drivers not using the
channel context API, I won't convert the previous API
to VHT support.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agonl80211/cfg80211: support VHT channel configuration
Johannes Berg [Thu, 8 Nov 2012 22:14:50 +0000 (23:14 +0100)]
nl80211/cfg80211: support VHT channel configuration

Change nl80211 to support specifying a VHT (or HT)
using the control channel frequency (as before) and
new attributes for the channel width and first and
second center frequency. The old channel type is of
course still supported for HT.

Also change the cfg80211 channel definition struct
to support these by adding the relevant fields to
it (and removing the _type field.)

This also adds new helper functions:
 - cfg80211_chandef_create to create a channel def
   struct given the control channel and channel type,
 - cfg80211_chandef_identical to check if two channel
   definitions are identical
 - cfg80211_chandef_compatible to check if the given
   channel definitions are compatible, and return the
   wider of the two

This isn't entirely complete, but that doesn't matter
until we have a driver using it. In particular, it's
missing
 - regulatory checks on the usable bandwidth (if that
   even makes sense)
 - regulatory TX power (database can't deal with it)
 - a proper channel compatibility calculation for the
   new channel types

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agocfg80211: pass a channel definition struct
Johannes Berg [Thu, 8 Nov 2012 20:25:48 +0000 (21:25 +0100)]
cfg80211: pass a channel definition struct

Instead of passing a channel pointer and channel type
to all functions and driver methods, pass a new channel
definition struct. Right now, this struct contains just
the control channel and channel type, but for VHT this
will change.

Also, add a small inline cfg80211_get_chandef_type() so
that drivers don't need to use the _type field of the
new structure all the time, which will change.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agonl80211: add documentation for channel type
Johannes Berg [Thu, 8 Nov 2012 18:20:56 +0000 (19:20 +0100)]
nl80211: add documentation for channel type

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agocfg80211: remove remain-on-channel channel type
Johannes Berg [Thu, 8 Nov 2012 17:31:02 +0000 (18:31 +0100)]
cfg80211: remove remain-on-channel channel type

As mwifiex (and mac80211 in the software case) are the
only drivers actually implementing remain-on-channel
with channel type, userspace can't be relying on it.
This is the case, as it's used only for P2P operations
right now.

Rather than adding a flag to tell userspace whether or
not it can actually rely on it, simplify all the code
by removing the ability to use different channel types.
Leave only the validation of the attribute, so that if
we extend it again later (with the needed capability
flag), it can't break userspace sending invalid data.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: fix managed mode channel flags handling
Johannes Berg [Mon, 26 Nov 2012 10:57:41 +0000 (11:57 +0100)]
mac80211: fix managed mode channel flags handling

If ieee80211_prep_channel() decides that HT should be
disabled (because the HT IEs from the AP were invalid)
it will set the IEEE80211_STA_DISABLE_HT to not send
HT capabilities to the AP when associating. If this
happens during authentication, the flag will be lost
and we send HT frames, even if the channel config was
set up for non-HT. This can lead to issues.

Fix this by always resetting the ifmgd flags to zero
when the channel context is released so that the flag
resetting in ieee80211_mgd_assoc() isn't necessary.

To make the code a bit easier move the call to release
the channel in ieee80211_set_disassoc() to the end of
the function together with the flag resetting (which
needs to be at the end to avoid timers setting flags.)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: remove mesh config macros from mesh_plink.c
Marco Porsch [Thu, 22 Nov 2012 02:40:32 +0000 (18:40 -0800)]
mac80211: remove mesh config macros from mesh_plink.c

Use shortcut pointer instead where it is appropriate.

Signed-off-by: Marco Porsch <marco.porsch@etit.tu-chemnitz.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: refactor ieee80211_set_qos_hdr
Marco Porsch [Thu, 22 Nov 2012 02:40:31 +0000 (18:40 -0800)]
mac80211: refactor ieee80211_set_qos_hdr

Return early if not a QoS Data frame.
Give proper documentation.

Signed-off-by: Marco Porsch <marco.porsch@etit.tu-chemnitz.de>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: move Mesh Capability field definition to ieee80211.h
Marco Porsch [Thu, 22 Nov 2012 02:40:30 +0000 (18:40 -0800)]
mac80211: move Mesh Capability field definition to ieee80211.h

Signed-off-by: Marco Porsch <marco.porsch@etit.tu-chemnitz.de>
[prefix with IEEE80211_]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: in ADHOC print debug message for every Auth message
Antonio Quartulli [Sun, 25 Nov 2012 22:24:27 +0000 (23:24 +0100)]
mac80211: in ADHOC print debug message for every Auth message

The debug message has to be printed also for an Auth message with
auth_sequence != 1. This helps understanding whether the two Auth
messages are exchanged correctly or not.

Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agomac80211: in ADHOC don't update last_rx if sta is not authorized
Antonio Quartulli [Sun, 25 Nov 2012 22:13:42 +0000 (23:13 +0100)]
mac80211: in ADHOC don't update last_rx if sta is not authorized

It does not make sense to keep a station alive if it is not authorised
at all. If IBSS/RSN is used it could also be the case that something
went wrong during the keys exchange and the stations ended up in a not
recoverable state.

By not updating last_rx we are giving the station a chance to be
deleted and to start the key exchange once again from scratch.

Signed-off-by: Antonio Quartulli <antonio@open-mesh.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agocfg80211: change function signature of cfg80211_get_p2p_attr()
Arend van Spriel [Sun, 25 Nov 2012 18:13:28 +0000 (19:13 +0100)]
cfg80211: change function signature of cfg80211_get_p2p_attr()

The function cfg80211_get_p2p_attr() can fail and returns
a negative error code. However, the return type is unsigned
int. The largest positive number is determined by desired_len
variable in the function, which is u16. So changing the return
type to int to allow easy error checking. Also change the type
for the attribute to enum for improved type checking.

Signed-off-by: Arend van Spriel <arend@broadcom.com>
[fix indentation, don't use u8 attr variable]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
12 years agoipv4/ipmr and ipv6/ip6mr: Convert int mroute_do_<foo> to bool
Joe Perches [Sun, 25 Nov 2012 09:35:30 +0000 (09:35 +0000)]
ipv4/ipmr and ipv6/ip6mr: Convert int mroute_do_<foo> to bool

Save a few bytes per table by convert mroute_do_assert and
mroute_do_pim from int to bool.

Remove !! as the compiler does that when assigning int to bool.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoipv4: ipmr: various fixes and cleanups
Eric Dumazet [Sun, 25 Nov 2012 06:41:45 +0000 (06:41 +0000)]
ipv4: ipmr: various fixes and cleanups

1) ip_mroute_setsockopt() & ip_mroute_getsockopt() should not
   access/set raw_sk(sk)->ipmr_table before making sure the socket
   is a raw socket, and protocol is IGMP

2) MRT_INIT should return -EINVAL if optlen != sizeof(int), not
   -ENOPROTOOPT

3) MRT_ASSERT & MRT_PIM should validate optlen

4) " (v) ? 1 : 0 " can be written as " !!v "

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoMAINTAINERS: Add Q: patchwork entries for net/ and drivers/net/
Joe Perches [Sat, 24 Nov 2012 11:18:25 +0000 (11:18 +0000)]
MAINTAINERS: Add Q: patchwork entries for net/ and drivers/net/

Add the netdev patchwork entries for networking drivers.
Change the W: entry to Q:for networking.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agosmsc: Add logging message newlines
Joe Perches [Sat, 24 Nov 2012 01:27:49 +0000 (01:27 +0000)]
smsc: Add logging message newlines

Avoid any possible message logging interleaving by adding
missing newlines.

Align arguments.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agoqlcnic: fix sparse check endian warnings
Shahed Shaikh [Fri, 23 Nov 2012 23:56:52 +0000 (23:56 +0000)]
qlcnic: fix sparse check endian warnings

Signed-off-by: Shahed Shaikh <shahed.shaikh@qlogic.com>
Signed-off-by: Sony Chacko <sony.chacko@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: dsa/slave: Fix compilation warnings
viresh kumar [Fri, 23 Nov 2012 19:53:54 +0000 (19:53 +0000)]
net: dsa/slave: Fix compilation warnings

Currently when none of CONFIG_NET_DSA_TAG_DSA, CONFIG_NET_DSA_TAG_EDSA and
CONFIG_NET_DSA_TAG_TRAILER is defined, we get following compilation warnings:

net/dsa/slave.c:51:12: warning: 'dsa_slave_init' defined but not used [-Wunused-function]
net/dsa/slave.c:60:12: warning: 'dsa_slave_open' defined but not used [-Wunused-function]
net/dsa/slave.c:98:12: warning: 'dsa_slave_close' defined but not used [-Wunused-function]
net/dsa/slave.c:116:13: warning: 'dsa_slave_change_rx_flags' defined but not used [-Wunused-function]
net/dsa/slave.c:127:13: warning: 'dsa_slave_set_rx_mode' defined but not used [-Wunused-function]
net/dsa/slave.c:136:12: warning: 'dsa_slave_set_mac_address' defined but not used [-Wunused-function]
net/dsa/slave.c:164:12: warning: 'dsa_slave_ioctl' defined but not used [-Wunused-function]

Earlier approach to fix this was discussed here:

lkml.org/lkml/2012/10/29/549

This is another approach to fix it. This is done by some changes in config
options, which make more sense than the earlier approach. As, atleast one
tagging option must always be selected for using net/dsa/ infrastructure, this
patch selects NET_DSA from tagging configs instead of having it as an selectable
config.

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years agonet: sched: enable CAN Identifier to be build into kernel
Marc Kleine-Budde [Fri, 23 Nov 2012 00:44:57 +0000 (00:44 +0000)]
net: sched: enable CAN Identifier to be build into kernel

This patch makes it possible to build the CAN Identifier into the kernel, even
if the CAN support is build as a module.

Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp: re-enable interrupts after tx timeout
David Woodhouse [Sat, 24 Nov 2012 12:11:21 +0000 (12:11 +0000)]
8139cp: re-enable interrupts after tx timeout

Recovery doesn't work too well if we leave interrupts disabled...

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-by: Francois Romieu <romieu@fr.zoreil.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp: enable bql
David Woodhouse [Thu, 22 Nov 2012 03:16:58 +0000 (03:16 +0000)]
8139cp: enable bql

This adds support for byte queue limits on RTL8139C+

Tested on real hardware.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Acked-By: Dave Täht <dave.taht@bufferbloat.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
12 years ago8139cp: set ring address after enabling C+ mode
David Woodhouse [Wed, 21 Nov 2012 10:27:19 +0000 (10:27 +0000)]
8139cp: set ring address after enabling C+ mode

This fixes (for me) a regression introduced by commit b01af457 ("8139cp:
set ring address before enabling receiver"). That commit configured the
descriptor ring addresses earlier in the initialisation sequence, in
order to avoid the possibility of triggering stray DMA before the
correct address had been set up.

Unfortunately, it seems that the hardware will scribble garbage into the
TxRingAddr registers when we enable "plus mode" Tx in the CpCmd
register. Observed on a Traverse Geos router board.

To deal with this, while not reintroducing the problem which led to the
original commit, we augment cp_start_hw() to write to the CpCmd register
*first*, then set the descriptor ring addresses, and then finally to
enable Rx and Tx in the original 8139 Cmd register. The datasheet
actually indicates that we should enable Tx/Rx in the Cmd register
*before* configuring the descriptor addresses, but that would appear to
re-introduce the problem that the offending commit b01af457 was trying
to solve. And this variant appears to work fine on real hardware.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Cc: stable@kernel.org [3.5+]
Signed-off-by: David S. Miller <davem@davemloft.net>