pandora-kernel.git
13 years agobonding: Fix useless test: int > INT_MAX
Nicolas de Pesloüan [Fri, 28 Aug 2009 13:18:34 +0000 (13:18 +0000)]
bonding: Fix useless test: int > INT_MAX

max_bonds is of type int and cannot be greater than INT_MAX.

Signed-off-by: Nicolas de Pesloüan <nicolas.2p.debian@free.fr>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding: use compare_ether_addr
Stephen Hemminger [Fri, 28 Aug 2009 12:05:13 +0000 (12:05 +0000)]
bonding: use compare_ether_addr

Bonding can use compare_ether_addr() in bond_release.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobonding: propogate vlan_features to bonding master
Jay Vosburgh [Fri, 28 Aug 2009 12:05:12 +0000 (12:05 +0000)]
bonding: propogate vlan_features to bonding master

Propogate the vlan_features of the slave devices to the bonding
master device, using the same logic as for regular features.

Tested by Or Gerlitz <ogerlitz@voltaire.com>, who also removed
the debug logic from the original test patch.

Signed-off-by: Or Gerlitz <ogerlitz@voltaire.com>
Signed-off-by: Jay Vosburgh <fubar@us.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMAINTAINERS: update information for sfc network driver
Joe Perches [Wed, 26 Aug 2009 08:47:47 +0000 (08:47 +0000)]
MAINTAINERS: update information for sfc network driver

Based upon a patch by Ben Hutchings.

Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosfc: Improve reliability of RX queue flushing
Ben Hutchings [Wed, 26 Aug 2009 08:18:13 +0000 (08:18 +0000)]
sfc: Improve reliability of RX queue flushing

Reconfiguring the port requires us to flush all DMA queues.  In
repeated testing we have found that RX flushes would sometimes fail
because the RX DMA engine was not properly isolated from the MACs.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosfc: Work around XMAC bug causing packet loss with some peers
Ben Hutchings [Wed, 26 Aug 2009 08:17:59 +0000 (08:17 +0000)]
sfc: Work around XMAC bug causing packet loss with some peers

Received frames must be re-clocked by the local XGXS to the 156.25 MHz
(DDR) clock of the XGMII.  If the remote clock is slightly faster this
can reduce a minimum IPG of 64 bit-times (1 cycle) to 32 bit-times
(half a cycle).  If the XMAC detects that a frame has reached the
maximum RX frame length in the same cycle that it receives one of
these reduced IPGs, it may miss the IPG, causing two valid frames to
be treated as a single invalid frame (over-length with bad CRC).

We work around this by increasing the maximum RX frame length so that
peers with matched MTU will not provoke this bug.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosfc: QT2025C: Use hard reset only
Ben Hutchings [Wed, 26 Aug 2009 08:17:19 +0000 (08:17 +0000)]
sfc: QT2025C: Use hard reset only

At probe time, falcon_reset_hw() performs a hard reset of the PHY
along with Falcon.  There is no need to perform a soft reset later,
and any access to standard MDIO registers before the PHY firmware has
booted can interrupt the boot process, making the port unusable.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agomdio: Check MDIO_STAT1_FAULT in mdio45_links_ok()
Ben Hutchings [Wed, 26 Aug 2009 08:16:58 +0000 (08:16 +0000)]
mdio: Check MDIO_STAT1_FAULT in mdio45_links_ok()

Some PHYs will report that the link is up even though there is a fault
condition.  Therefore, check the fault flag too.  We must also read
STAT2 to reset this flag.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosfc: Do not reinitialise XAUI serdes before it has completed reset
Ben Hutchings [Wed, 26 Aug 2009 08:16:46 +0000 (08:16 +0000)]
sfc: Do not reinitialise XAUI serdes before it has completed reset

falcon_reset_xaui() waits for XGXS reset to complete, but the XAUI
serdes reset may take longer.  It needs to check both reset active
bits.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agosfc: Fix ordering of device registration and initial netif_carrier_off()
Ben Hutchings [Wed, 26 Aug 2009 08:16:27 +0000 (08:16 +0000)]
sfc: Fix ordering of device registration and initial netif_carrier_off()

We must call netif_carrier_off() after the device is registered, not
before, to set the operational state and user-space IFF_RUNNING flag
correctly.

Since we don't want observers to see an intermediate state, open-code
register_netdev() and add efx_update_name() and netif_carrier_off()
into the locked region.

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoqeth: display "undefined" value of sysfs-attribute "layer2"
Ursula Braun [Wed, 26 Aug 2009 02:01:10 +0000 (02:01 +0000)]
qeth: display "undefined" value of sysfs-attribute "layer2"

If nothing has been written into the qeth sysfs-attribute layer2,
its value is "-1" meaning "not yet defined". But the value is
displayed as "1" meaning "layer2 selected". The patch changes the
reading of this "-1"-value to "-1" to make clear the layer2-attribute
has not yet been defined.

Signed-off-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoqeth: display service_level info only when valid.
Klaus-Dieter Wacker [Wed, 26 Aug 2009 02:01:09 +0000 (02:01 +0000)]
qeth: display service_level info only when valid.

qeth displayed an entry in /proc/service_level even when no valid
MCL-string was available (the MCL info is blank). The change is to
create an entry in /proc/service_level only when MCL-string is
non-zero.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoqeth: Cleanup for cast-type determination.
Klaus-Dieter Wacker [Wed, 26 Aug 2009 02:01:08 +0000 (02:01 +0000)]
qeth: Cleanup for cast-type determination.

Clear separation of cast-type determination (send path) for layer-2
resp. layer-3. Allowing to have inline functions for qeth layer-
discipline.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoqeth: scheduling while atomic during ifconfig online sequence
Frank Blaschka [Wed, 26 Aug 2009 02:01:07 +0000 (02:01 +0000)]
qeth: scheduling while atomic during ifconfig online sequence

In case the IP address list contains entries (not removed when the device was set
offline) this entries should be registered next time the device is brought online.
In the past this was done implicitly with the device open call but since we wait
in the set IPv4 IPA and the device open common code holds various locks this
does not work any longer.

Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoqeth: HiperSockets SIGA retry support on CC=2.
Klaus-Dieter Wacker [Wed, 26 Aug 2009 02:01:06 +0000 (02:01 +0000)]
qeth: HiperSockets SIGA retry support on CC=2.

Qeth HiperSockets support now retries sending of packets when the
IBM System z signals a temporary resource shortage (e.g. target
buffer full). The packet is enqueued into the device queue.
After 3 times of unsuccessful send the packet is dropped.

Signed-off-by: Klaus-Dieter Wacker <kdwacker@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonet: Temporarily backout SKB sources tracer.
David S. Miller [Wed, 26 Aug 2009 23:32:37 +0000 (16:32 -0700)]
net: Temporarily backout SKB sources tracer.

Steven Rostedt has suggested that Neil work with the tracing
folks, trying to use TRACE_EVENT as the mechanism for
implementation.  And if that doesn't workout we can investigate
other solutions such as that one which was tried here.

This reverts the following 2 commits:

5a165657bef7c47e5ff4cd138f7758ef6278e87b
("net: skb ftracer - Add config option to enable new ftracer (v3)")

9ec04da7489d2c9ae01ea6e9b5fa313ccf3d35fb
("net: skb ftracer - Add actual ftrace code to kernel (v3)")

Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Update version to 3.100
Matt Carlson [Tue, 25 Aug 2009 10:11:55 +0000 (10:11 +0000)]
tg3: Update version to 3.100

This patch updates the tg3 version to 3.100.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobroadcom: Make the 57780 IEEE compliant
Matt Carlson [Tue, 25 Aug 2009 10:11:26 +0000 (10:11 +0000)]
broadcom: Make the 57780 IEEE compliant

This brings the 57780's phy into IEEE compliance by suppressing the
common mode oscillation.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobroadcom: Add AC131 phy support
Matt Carlson [Tue, 25 Aug 2009 10:10:58 +0000 (10:10 +0000)]
broadcom: Add AC131 phy support

This patch adds support for the AC131 fast ethernet transceiver.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobroadcom: Add BCM50610M support
Matt Carlson [Tue, 25 Aug 2009 10:10:30 +0000 (10:10 +0000)]
broadcom: Add BCM50610M support

This patch adds support for the BCM50610M phy ID.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Convert code to use PHY_IS_FET
Matt Carlson [Tue, 25 Aug 2009 10:10:03 +0000 (10:10 +0000)]
tg3: Convert code to use PHY_IS_FET

This patch converts the code to use the PHY_IS_FET flag rather than the
ASIC revision to decide whether or not to use FET paths.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Create MII_TG3_FET namespace
Matt Carlson [Tue, 25 Aug 2009 10:09:36 +0000 (10:09 +0000)]
tg3: Create MII_TG3_FET namespace

Broadcom's phys come in two distinctly different register layouts.  For
the lack of an official term to distinguish between the two formats, we
can loosely categorize them by their fast ethernet or gigabit ethernet
transceiver description.  This patch creates the (driver-internal) Fast
Ethernet Transceiver (FET) namespace and converts the 5906 EPHY
definitions over.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Tune 5785 clock switching
Matt Carlson [Tue, 25 Aug 2009 10:09:07 +0000 (10:09 +0000)]
tg3: Tune 5785 clock switching

This patch tunes the timeouts the CPMU uses to decide when to switch
from the clocks output by the PHY to internal clock sources.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Add 57788, remove 57720
Matt Carlson [Tue, 25 Aug 2009 10:08:42 +0000 (10:08 +0000)]
tg3: Add 57788, remove 57720

This patch adds support for the 57788 and removes support for the 57720
which was never released.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Preserve PCIe MPS setting for new devs
Matt Carlson [Tue, 25 Aug 2009 10:08:16 +0000 (10:08 +0000)]
tg3: Preserve PCIe MPS setting for new devs

Most older tg3 devices only supported a PCIe maximum payload size of
128 bytes.  More recent devices bump this limit up to 256 bytes
though.  This patch modifies the code so that the MPS limit is only
enforced on those devices that only allow the 128 byte setting.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Fix TSO test against wrong flags var
Matt Carlson [Tue, 25 Aug 2009 10:07:54 +0000 (10:07 +0000)]
tg3: Fix TSO test against wrong flags var

Julia Lawall discovered that the TG3_FLG2_TSO_CAPABLE flag was being
compared against the wrong flags device member.  This patch implements
the fix.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Prevent tx BD corruption
Matt Carlson [Tue, 25 Aug 2009 10:07:27 +0000 (10:07 +0000)]
tg3: Prevent tx BD corruption

This patch prevents a tx BD corruption bug by preventing the device from
powering down the PLL from L1 if the link speed is 10Mbps or 100Mbps.

The same bits are also used to prevent a system hang during chip reset
resulting from a complicated set of events that ultimately leads to
PCIe block register corruption.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agotg3: Fix 57780 asic rev PCIe link receiver errors
Matt Carlson [Tue, 25 Aug 2009 10:06:01 +0000 (10:06 +0000)]
tg3: Fix 57780 asic rev PCIe link receiver errors

This patch fixes some PCIe link receiver errors by decreasing the internal
electrical idle timeout.

Signed-off-by: Matt Carlson <mcarlson@broadcom.com>
Reviewed-by: Michael Chan <mchan@broadcom.com>
Reviewed-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoe1000: Remove unused function e1000_mta_set.
Graham, David [Tue, 25 Aug 2009 04:43:17 +0000 (04:43 +0000)]
e1000: Remove unused function e1000_mta_set.

Remove function e1000_mta_set, as it is no longer called

Signed-off-by: Dave Graham <david.graham@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoigb/ixgbe: add IPV6_CSUM support to vlan_features
Alexander Duyck [Tue, 25 Aug 2009 04:47:50 +0000 (04:47 +0000)]
igb/ixgbe: add IPV6_CSUM support to vlan_features

We were already exporting TSO6 to the vlan, but we weren't exporting the
checksum support for IPV6 which was causing warning messages to be
displayed when doing IPv6 TSO over a vlan.

Signed-off-by: Alexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoixgbe: cleanup functions that should have been defined static
Don Skidmore [Tue, 25 Aug 2009 04:47:32 +0000 (04:47 +0000)]
ixgbe: cleanup functions that should have been defined static

We have some ~40 functions that were being called out with 'make
namespacecheck'.  This patch changes these functions to be static.

Signed-off-by: Don Skidmore <donald.c.skidmore@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoixgbe: Fix isues while reporting 8259x backplane link capabilities
Mallikarjuna R Chilakala [Tue, 25 Aug 2009 04:47:11 +0000 (04:47 +0000)]
ixgbe: Fix isues while reporting 8259x backplane link capabilities

Fix ethtool get_settings logic to report 10G & 1G advertised and
supported link modes in all 8259x 10G backplane connection types
except for 82598EB BX network connection type.

Signed-off-by: Mallikarjuna R Chilakala <mallikarjuna.chilakala@intel.com>
Acked-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: bridged mode optimizations
Narender Kumar [Mon, 24 Aug 2009 19:23:28 +0000 (19:23 +0000)]
netxen: bridged mode optimizations

When the interface is put in bridged mode, destination mac
addresses are unknown to firmware. So packets take a slow
path (lower priority) in firmware reducing performance.

Firmware can cache limited number of remote unicast mac
addresses for certain interval, if "dynamic mac learning"
mode is enabled.

Driver needs to enable this "mac learning" mode in firmware.
Currently this is done through net device class sysfs entry,
possibly this can also be done upon netlink notifications to
from bridge.

Signed-off-by: Narender Kumar <narender.kumar@qlogic.com>
Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: remove netxen_nic_niu.c
Dhananjay Phadke [Mon, 24 Aug 2009 19:23:27 +0000 (19:23 +0000)]
netxen: remove netxen_nic_niu.c

Consolidate all MAC/PHY access functions into netxen_nic_hw.c

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: implement generic pcie semaphore functions
Dhananjay Phadke [Mon, 24 Aug 2009 19:23:26 +0000 (19:23 +0000)]
netxen: implement generic pcie semaphore functions

Implement common function for locking/unlocking 8 hardware
semaphores used for serializing access to shared resouces
on a NIC board by different PCI functions.

As by definition, callers of these semaphore API can be
put to sleep till the semaphore is locked.

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: remove unused code
Dhananjay Phadke [Mon, 24 Aug 2009 19:23:25 +0000 (19:23 +0000)]
netxen: remove unused code

Remove code for phy access on unreleased NX2031 based quad-gig board.

NX3031 based production quad-gig boards do not require direct phy
access by driver.

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/lowpan/lowpan
David S. Miller [Mon, 24 Aug 2009 02:19:30 +0000 (19:19 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/lowpan/lowpan

13 years agor6040: bump to version 0.25 and 20 August 2009 release date
Florian Fainelli [Thu, 20 Aug 2009 01:26:23 +0000 (01:26 +0000)]
r6040: bump to version 0.25 and 20 August 2009 release date

Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agor6040: move down second PHY detection to r6040_init_one
Mark Kelly [Thu, 20 Aug 2009 01:26:20 +0000 (01:26 +0000)]
r6040: move down second PHY detection to r6040_init_one

Instead of fully initializing an unusable second r6040
device, perform the PHY detection earlier and bail out
in r6040_init_one when we could not read the PHY identifier.

Signed-off-by: Mark Kelly <mark@bifferos.com>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agor6040: remove unused pioaddr variable
Mark Kelly [Thu, 20 Aug 2009 01:26:12 +0000 (01:26 +0000)]
r6040: remove unused pioaddr variable

This patch removes the unused pioaddr variable.

Signed-off-by: Mark Kelly <mark@bifferos.com>
Signed-off-by: Florian Fainelli <florian@openwrt.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoRDS: Add a debug message suggesting to load transport modules
Andy Grover [Fri, 21 Aug 2009 12:28:35 +0000 (12:28 +0000)]
RDS: Add a debug message suggesting to load transport modules

Now that RDS transports are no longer compiled-in to RDS core,
there is now the possibility that they will not be loaded. This
adds a helpful suggestion when rds_bind() fails to find a transport.

Signed-off-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoRDS: Track transports via an array, not a list
Andy Grover [Fri, 21 Aug 2009 12:28:34 +0000 (12:28 +0000)]
RDS: Track transports via an array, not a list

Now that transports can be loaded in arbitrary order,
it is important for rds_trans_get_preferred() to look
for them in a particular order, instead of walking the list
until it finds a transport that works for a given address.
Now, each transport registers for a specific transport slot,
and these are ordered so that preferred transports come first,
and then if they are not loaded, other transports are queried.

Signed-off-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoRDS: Modularize RDMA and TCP transports
Andy Grover [Fri, 21 Aug 2009 12:28:33 +0000 (12:28 +0000)]
RDS: Modularize RDMA and TCP transports

Enable the building of transports as modules.

Also, improve consistency of Kconfig messages in relation to other
protocols, and move build dependency on IB from the RDS core code
to the rds_rdma module.

Signed-off-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoRDS: Export symbols from core RDS
Andy Grover [Fri, 21 Aug 2009 12:28:32 +0000 (12:28 +0000)]
RDS: Export symbols from core RDS

Now that rdma and tcp transports will be modularized,
we need to export a number of functions so they can call them.

Signed-off-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoRDS: Add TCP transport to RDS
Andy Grover [Fri, 21 Aug 2009 12:28:31 +0000 (12:28 +0000)]
RDS: Add TCP transport to RDS

This code allows RDS to be tunneled over a TCP connection.

RDMA operations are disabled when using TCP transport,
but this frees RDS from the IB/RDMA stack dependency, and allows
it to be used with standard Ethernet adapters, or in a VM.

Signed-off-by: Andy Grover <andy.grover@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: remove netxen_nic_phan_reg.h
Dhananjay Phadke [Sun, 23 Aug 2009 08:35:13 +0000 (08:35 +0000)]
netxen: remove netxen_nic_phan_reg.h

Consolidate register definitions in netxen_nic_hdr.h

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: implement pci driver shutdown
Amit Kumar Salecha [Sun, 23 Aug 2009 08:35:12 +0000 (08:35 +0000)]
netxen: implement pci driver shutdown

Implement pci driver shutdown functionality, this helps
quiescing all PCI transaction before chipset is reset.

Amit Kumar Salecha <amit@qlogic.com>
Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: refactor tx dma mapping code
Dhananjay Phadke [Sun, 23 Aug 2009 08:35:11 +0000 (08:35 +0000)]
netxen: refactor tx dma mapping code

Move all tx skb mapping code into netxen_map_tx_skb().

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: fix firmware reset logic
Dhananjay Phadke [Sun, 23 Aug 2009 08:35:10 +0000 (08:35 +0000)]
netxen: fix firmware reset logic

If netxen_need_fw_reset() return 0 [ implies firmware is up
and running], still go through dma mask check, etc.

Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agonetxen: ethtool statistics and control for LRO
Narender Kumar [Sun, 23 Aug 2009 08:35:09 +0000 (08:35 +0000)]
netxen: ethtool statistics and control for LRO

Add ethtool -K knob to control LRO in firmware.
LRO path is completely separated from GRO, LRO packets
are still fed with netif_receive_skb().

Also fix ethtool statistics to include LRO packets.
Also use correct message type while configuring interrupt coalescing.

Signed-off-by: Narender Kumar <narender.kumar@qlogic.com>
Signed-off-by: Dhananjay Phadke <dhananjay@netxen.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Update version to 2.0.2.
Michael Chan [Fri, 21 Aug 2009 16:20:50 +0000 (16:20 +0000)]
bnx2: Update version to 2.0.2.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Use const on flash_table structure.
Michael Chan [Fri, 21 Aug 2009 16:20:49 +0000 (16:20 +0000)]
bnx2: Use const on flash_table structure.

The structure, once initialized, never changes.

Signed-off-by: Benjamin Li <benli@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Refine coalescing parameters.
Michael Chan [Fri, 21 Aug 2009 16:20:48 +0000 (16:20 +0000)]
bnx2: Refine coalescing parameters.

- Set the USE_INT_PARAM bit so the rx-frames-irq and tx-frames-irq will take
  effect on 5709.
- Increase the default rx-frames to reduce interrupt count.
- Decrease the default rx-frames-irq and tx-frames-irq to catch more events
  during NAPI poll.

All these will reduce interrupts without affecting latency.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Report FTQ discard counter.
Michael Chan [Fri, 21 Aug 2009 16:20:47 +0000 (16:20 +0000)]
bnx2: Report FTQ discard counter.

Report this counter to ethtool -S and include it in netstat.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Apply BROKEN_STATS workaround to 5706 and 5708.
Michael Chan [Fri, 21 Aug 2009 16:20:46 +0000 (16:20 +0000)]
bnx2: Apply BROKEN_STATS workaround to 5706 and 5708.

Add flag to expand the workaround to both chips.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Close device if MTU change or ring size change fails.
Michael Chan [Fri, 21 Aug 2009 16:20:45 +0000 (16:20 +0000)]
bnx2: Close device if MTU change or ring size change fails.

When unable to allocate memory for new MTU or new ring size, we need
to close the device to prevent it from crashing.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Check if_running() before touching chip registers.
Michael Chan [Fri, 21 Aug 2009 16:20:44 +0000 (16:20 +0000)]
bnx2: Check if_running() before touching chip registers.

Add this check to bnx2_netif_stop() and bnx2_vlan_rx_register() to
prevent bus lockups on some systems when the chip is in low power state.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agobnx2: Zero out status block before chip reset.
Michael Chan [Fri, 21 Aug 2009 16:20:43 +0000 (16:20 +0000)]
bnx2: Zero out status block before chip reset.

In case IRQs are shared, we will not mistakenly start processing
the ring based on old status block indices.

Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
13 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git/linville/wirel...
David S. Miller [Fri, 21 Aug 2009 08:56:56 +0000 (01:56 -0700)]
Merge branch 'master' of git://git./linux/kernel/git/linville/wireless-next-2.6

13 years agoMAINTAINERS: add information for mwl8k wireless driver
Lennert Buytenhek [Tue, 18 Aug 2009 03:13:48 +0000 (05:13 +0200)]
MAINTAINERS: add information for mwl8k wireless driver

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: update copyright and version number
Lennert Buytenhek [Tue, 18 Aug 2009 02:34:26 +0000 (04:34 +0200)]
mwl8k: update copyright and version number

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove mwl8k_queue_work()
Lennert Buytenhek [Tue, 18 Aug 2009 02:08:01 +0000 (04:08 +0200)]
mwl8k: remove mwl8k_queue_work()

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_configure_filter() workqueue use
Lennert Buytenhek [Tue, 18 Aug 2009 02:06:20 +0000 (04:06 +0200)]
mwl8k: get rid of mwl8k_configure_filter() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_get_stats() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:26:27 +0000 (07:26 +0200)]
mwl8k: get rid of mwl8k_get_stats() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_conf_tx() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:25:59 +0000 (07:25 +0200)]
mwl8k: get rid of mwl8k_conf_tx() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_set_rts_threshold() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:24:15 +0000 (07:24 +0200)]
mwl8k: get rid of mwl8k_set_rts_threshold() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_bss_info_changed() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:21:46 +0000 (07:21 +0200)]
mwl8k: get rid of mwl8k_bss_info_changed() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_config() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:19:37 +0000 (07:19 +0200)]
mwl8k: get rid of mwl8k_config() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_stop() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:15:49 +0000 (07:15 +0200)]
mwl8k: get rid of mwl8k_stop() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: get rid of mwl8k_start() workqueue use
Lennert Buytenhek [Fri, 17 Jul 2009 05:11:37 +0000 (07:11 +0200)]
mwl8k: get rid of mwl8k_start() workqueue use

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: fix firmware command serialisation
Lennert Buytenhek [Tue, 18 Aug 2009 01:18:01 +0000 (03:18 +0200)]
mwl8k: fix firmware command serialisation

The current mwl8k_priv->fw_lock spinlock doesn't actually protect
against multiple commands being submitted at once, as it is not kept
held over the entire firmware command submission.  And since waiting
for command completion sleeps, we can't use a spinlock anyway.

To fix mwl8k firmware command serialisation properly, we have the
following requirements:
- Some commands require that the packet transmit path is idle when
  the command is issued.  (For simplicity, we'll just quiesce the
  transmit path for every command.)
- There are certain sequences of commands that need to be issued to
  the hardware sequentially, with no other intervening commands.

This leads to an implementation of a "firmware lock" as a mutex that
can be taken recursively, and which is taken by both the low-level
command submission function (mwl8k_post_cmd) as well as any users of
that function that require issuing of an atomic sequence of commands,
and quiesces the transmit path whenever it's taken.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: mwl8k_queue_work() cleanup
Lennert Buytenhek [Thu, 16 Jul 2009 23:48:41 +0000 (01:48 +0200)]
mwl8k: mwl8k_queue_work() cleanup

Delete most of the mwl8k_work_struct fields and options, since most
of them are unused or never changed from their defaults:
- We always use priv->config_wq, so delete the wqueue argument from
  mwl8k_queue_work().
- MWL8K_WQ_SPIN and MWL8K_WQ_POST_REQUEST are never used, as all
  callers sleep for request completion, so sleep unconditionally.
- MWL8K_WQ_FREE_WORKSTRUCT is never used.
- MWL8K_WQ_TX_WAIT_EMPTY is always set, so assume it unconditionally.
- timeout_ms/txwait_attempts/tx_timeout_ms are never changed from
  their defaults, so just hardcode these in the workqueue worker.
- step is never used.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: mwl8k_txq_xmit() rework
Lennert Buytenhek [Fri, 17 Jul 2009 03:21:04 +0000 (05:21 +0200)]
mwl8k: mwl8k_txq_xmit() rework

Various mwl8k_txq_xmit changes:
- Extract the QoS field before adding the DMA header.
- Only write to tx->status once, and only after all the other
  descriptor fields have been set.
- Do all tx state manipulation under the tx spinlock.
- Remove the priv->inconfig check, as all transmit queues will
  be frozen during config cycles, so we won't ever be asked to
  transmit if a config cycle is running.
- Remove some more dead code.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: implement idle mode
Lennert Buytenhek [Mon, 17 Aug 2009 21:59:40 +0000 (23:59 +0200)]
mwl8k: implement idle mode

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: no need to hold ->tx_lock while setting the hardware interrupt mask
Lennert Buytenhek [Tue, 18 Aug 2009 02:19:00 +0000 (04:19 +0200)]
mwl8k: no need to hold ->tx_lock while setting the hardware interrupt mask

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: don't hardcode the number of transmit queues
Lennert Buytenhek [Tue, 18 Aug 2009 02:15:22 +0000 (04:15 +0200)]
mwl8k: don't hardcode the number of transmit queues

Use MWL8K_TX_QUEUES instead of a hardcoded "4" in a couple of places.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: don't touch 'command done' interrupt during firmware load
Lennert Buytenhek [Thu, 16 Jul 2009 10:44:45 +0000 (12:44 +0200)]
mwl8k: don't touch 'command done' interrupt during firmware load

Since firmware load commands don't generate 'command done' interrupts
like normal commands do, polling for command done interrupts just
unnecessarily slows down the firmware load process.  Removing this bit
of code speeds up loading a typical firmware image from 840 msec to
180 msec.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: dma header manipulations can't fail
Lennert Buytenhek [Thu, 16 Jul 2009 09:07:09 +0000 (11:07 +0200)]
mwl8k: dma header manipulations can't fail

Adding and removing the DMA header that the mwl8k hardware requires
on tx and provides on rx can never fail, since we are guaranteed to
have enough headroom on the tx path to expand the packet, and we only
ever shrink the packet on the rx path.  (And on both paths we are
guaranteed to be the only user of the skb we are handling.)

This allows removing all of the skb clone handling in the tx and tx
reclaim paths, and eliminates error checks in both the tx and rx
paths, simplifying the code a bit more.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: ->add_interface() is not called for monitor interfaces
Lennert Buytenhek [Thu, 16 Jul 2009 12:15:44 +0000 (14:15 +0200)]
mwl8k: ->add_interface() is not called for monitor interfaces

There is no need to check for NL80211_IFTYPE_MONITOR in
->add_interface(), as this function is never called for monitor
interfaces.  Also, there is no need to advertise this bit in our
wiphy's ->interface_modes.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: fix mwl8k_configure_filter() parameter lifetime issue
Lennert Buytenhek [Tue, 18 Aug 2009 01:55:42 +0000 (03:55 +0200)]
mwl8k: fix mwl8k_configure_filter() parameter lifetime issue

mwl8k_configure_filter() passes pointers to total_flags and the
multicast address list to a workqueue function, while there is no
guarantee that those pointers will still be valid by the time the
workqueue function runs.

Solve this by passing total_flags by value, and by passing an
already built multicast address setup command packet to the workqueue
function so that we don't have to look at the multicast address list
itself outside of mwl8k_configure_filter().

Also, since ->configure_filter() can sleep now, wait synchronously
for the worker to finish.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove MWL8K_*_SLOTTIME defines
Lennert Buytenhek [Thu, 16 Jul 2009 14:06:53 +0000 (16:06 +0200)]
mwl8k: remove MWL8K_*_SLOTTIME defines

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove MWL8K_WMM_* defines
Lennert Buytenhek [Thu, 16 Jul 2009 10:34:02 +0000 (12:34 +0200)]
mwl8k: remove MWL8K_WMM_* defines

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove MWL8K_RADIO_*_PREAMBLE defines
Lennert Buytenhek [Thu, 16 Jul 2009 10:26:57 +0000 (12:26 +0200)]
mwl8k: remove MWL8K_RADIO_*_PREAMBLE defines

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove MWL8K_RADIO_* defines
Lennert Buytenhek [Thu, 16 Jul 2009 10:14:58 +0000 (12:14 +0200)]
mwl8k: remove MWL8K_RADIO_* defines

Instead of passing a flag bitmask to mwl8k_cmd_802_11_radio_control,
pass the 'enable' and 'force' arguments as separate parameters, and
introduce wrappers for the common cases of enabling and disabling
without forcing.

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: various coding style cleanups
Lennert Buytenhek [Thu, 16 Jul 2009 12:00:45 +0000 (14:00 +0200)]
mwl8k: various coding style cleanups

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: sort firmware command list by opcode, and trim unused commands
Lennert Buytenhek [Thu, 16 Jul 2009 09:50:36 +0000 (11:50 +0200)]
mwl8k: sort firmware command list by opcode, and trim unused commands

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: s/IEEE80211_ADDR_LEN/ETH_ALEN/g
Lennert Buytenhek [Thu, 16 Jul 2009 07:54:27 +0000 (09:54 +0200)]
mwl8k: s/IEEE80211_ADDR_LEN/ETH_ALEN/g

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomwl8k: remove various unused struct members and defines
Lennert Buytenhek [Thu, 16 Jul 2009 11:49:55 +0000 (13:49 +0200)]
mwl8k: remove various unused struct members and defines

Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Nuke struct ath9k_node_stats
Vasanthakumar Thiagarajan [Wed, 19 Aug 2009 10:53:40 +0000 (16:23 +0530)]
ath9k: Nuke struct ath9k_node_stats

Other than ns_avgbrssi (Average beacon rssi) no data of
ath9k_node_stats is being used anywhere. Nuke this structure
and move its only useful member to ar5416Anistate. Also cleanup
this redundant data in ath_softc.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoath9k: Fix bug in retrieving average beacon rssi
Vasanthakumar Thiagarajan [Thu, 20 Aug 2009 08:11:14 +0000 (13:41 +0530)]
ath9k: Fix bug in retrieving average beacon rssi

Currently the beacon rssi that LPF gives is divided and rounded
up by ATH_RSSI_EP_MULTIPLIER twice. This will leave the incorrect rssi
in ANI. Having correct rssi in ANI fixes the connection stability at
< 30dB rssi range. This patch removes the unncessary computation of average
rssi over already valid average rssi. Also removes the redundant macros to
find average rssi.

Signed-off-by: Vasanthakumar Thiagarajan <vasanth@atheros.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: add hardware tkip
gregor kowski [Wed, 19 Aug 2009 20:35:45 +0000 (22:35 +0200)]
b43: add hardware tkip

This add hardware tkip for b43.

Signed-off-by: Gregor Kowski <gregor.kowski@gmail.com>
Acked-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agomac80211: fix register_hw error path
Johannes Berg [Wed, 19 Aug 2009 17:45:50 +0000 (19:45 +0200)]
mac80211: fix register_hw error path

"cfg80211: fix alignment problem in scan request"
introduced a bug into the error path, because now
we allocate the entire scan request and not just
the channel list (the channel list is allocated
together with the scan request) -- on errors we
thus also need to free the entire scan request.

Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: LP-PHY: Implement spec updates and remove resolved FIXMEs
Gábor Stefanik [Tue, 18 Aug 2009 17:18:13 +0000 (19:18 +0200)]
b43: LP-PHY: Implement spec updates and remove resolved FIXMEs

Larry has started re-checking all current routines against a new
version of the Broadcom MIPS driver. This patch implements the first
round of changes he documented on the specs wiki.

Also remove a few FIXMEs regarding missing initial values for variables
with dynamic initial values where reading the values has been implemented.

Signed-off-by: Gábor Stefanik <netrolller.3d@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: allow cfg80211_connect_result with bssid == NULL
Zhu Yi [Wed, 19 Aug 2009 08:08:22 +0000 (16:08 +0800)]
cfg80211: allow cfg80211_connect_result with bssid == NULL

In case of connection failure, the bssid info is not a must have.

Signed-off-by: Zhu Yi <yi.zhu@intel.com>
Acked-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoorinoco: remove spare whitespace
David Kilroy [Tue, 18 Aug 2009 23:44:43 +0000 (00:44 +0100)]
orinoco: remove spare whitespace

Signed-off-by: David Kilroy <kilroyd@googlemail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agocfg80211: fix leaks of wdev->conn->ie
David Kilroy [Tue, 18 Aug 2009 23:43:31 +0000 (00:43 +0100)]
cfg80211: fix leaks of wdev->conn->ie

This only occurs in the following error situations:
 - driver calls connect_result with failure
 - error scheduling authentication on connect
 - error initiating scan (to get BSSID and channel) on
   connect
 - userspace calls disconnect while in the SCANNING or
   SCAN_AGAIN states

Signed-off-by: David Kilroy <kilroyd@googlemail.com>
Reviewed-by: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agoar9170: refactor configure_filter
Christian Lamparter [Wed, 19 Aug 2009 10:43:47 +0000 (12:43 +0200)]
ar9170: refactor configure_filter

Thanks to "mac80211: allow configure_filter callback to sleep",
we no longer have to defer the work to the workqueue.

Signed-off-by: Christian Lamparter <chunkeey@web.de>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agob43: LP-PHY: Two small spec updates
Gábor Stefanik [Tue, 18 Aug 2009 20:08:31 +0000 (22:08 +0200)]
b43: LP-PHY: Two small spec updates

The specs are beginning to support rev3 LP-PHYs - implement one of
the changes needed for rev3 support.
Also, in the new MIPS driver, the "Japan TX filter" was renamed to
"analog TX filter init" - however, calling it "init" is confusing,
so name it "set analog filter", with a comment for easier future
identification.

Signed-off-by: Gábor Stefanik <netrolller.3d@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: bss_info_changed() callback is allowed to sleep
Ivo van Doorn [Tue, 18 Aug 2009 18:33:12 +0000 (20:33 +0200)]
rt2x00: bss_info_changed() callback is allowed to sleep

The bss_info_changed() callback function no longer needs
to be atomic. Remove the scheduled work structure and
call into the driver directly.

Additionaly this makes the DRIVER_REQUIRE_SCHEDULED
flag redundant so it can be removed.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>
13 years agort2x00: configure_filter() callback is allowed to sleep
Ivo van Doorn [Tue, 18 Aug 2009 17:54:23 +0000 (19:54 +0200)]
rt2x00: configure_filter() callback is allowed to sleep

The configure_filter() callback function no longer needs
to be atomic. Remove the scheduled work structure and
call into the driver configure_filter() directly.

Signed-off-by: Ivo van Doorn <IvDoorn@gmail.com>
Signed-off-by: John W. Linville <linville@tuxdriver.com>