OSDN Git Service

sagit-ice-cold/kernel_xiaomi_msm8998.git
9 years agoiwlwifi: allow to define the stuck queue timer per queue
Emmanuel Grumbach [Mon, 12 Jan 2015 12:38:29 +0000 (14:38 +0200)]
iwlwifi: allow to define the stuck queue timer per queue

Different queue can have different behavior. While it can be
unacceptable for a certain queue to be stuck for 2 seconds
(e.g. the command queue), it can happen that another queue
will stay stuck for even longer (a queue servicing a power
saving client in GO).
The op_mode can even make the timeout be a function of the
listen interval.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: enable watchdog on Tx queues for mvm
Emmanuel Grumbach [Sun, 11 Jan 2015 15:19:39 +0000 (17:19 +0200)]
iwlwifi: mvm: enable watchdog on Tx queues for mvm

This watchdog allows to monitor the transmit queues. When a
queue doesn't progress for a too long time, a timer fires
and then, debug data can be collected.
This watchdog has never been enabled on dvm controlled
devices, so don't enable it there.
In order to have it running on mvm controlled devices, we
need to fix a small issue in the transport layer: mvm
controlled devices use the shadow registers optimization.
In this case, the watchdog wasn't running at all, even if
enabled by the module parameter. Fix that on the way.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: ignore stale TDLS ch-switch responses
Arik Nemtsov [Thu, 22 Jan 2015 12:24:44 +0000 (14:24 +0200)]
iwlwifi: mvm: ignore stale TDLS ch-switch responses

During out-of-channel activities (e.g. scan) TDLS ch-switch responses from
a peer are kept in FW. These packets arrive only after the out-of-channel
activity is complete, which can be in the order of several seconds.

Since TDLS ch-sw has no dialog-token-like mechanism for distinguishing
sessions, use the GP2 time of the incoming ch-switch response to discern
validity. For this purpose record the GP2 time of an outgoing TDLS ch-sw
request and compare to the Rx time of the ch-sw response.
The methods works in practice since the GP2 time of FW-deferred Rx is
accurate and contains the real Rx timestamp.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: improve TDLS ch-sw state machine
Arik Nemtsov [Thu, 22 Jan 2015 10:19:26 +0000 (12:19 +0200)]
iwlwifi: mvm: improve TDLS ch-sw state machine

Add a response-received state and add more limits on allowed requests
in each state of the connection. Previously ch-switch requests from
other peers could interrupt an outgoing active ch-switch. Also stale
packets from the current peer could disrupt the channel switch state.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: pcie: disable the SCD_BASE_ADDR when we resume from WoWLAN
Emmanuel Grumbach [Thu, 29 Jan 2015 19:34:00 +0000 (21:34 +0200)]
iwlwifi: pcie: disable the SCD_BASE_ADDR when we resume from WoWLAN

The base address of the scheduler in the device's memory
(SRAM) comes from two different sources. The periphery
register and the alive notification from the firmware.
We have a check in iwl_pcie_tx_start that ensures that
they are the same.
When we resume from WoWLAN, the firmware may have crashed
for whatever reason. In that case, the whole device may be
reset which means that the periphery register will hold a
meaningless value. When we come to compare
trans_pcie->scd_base_addr (which really holds the value we
had when we loaded the WoWLAN firmware upon suspend) and
the current value of the register, we don't see a match
unsurprisingly.
Trick the check to avoid a loud yet harmless WARN.
Note that when the WoWLAN has crashed, we will see that
in iwl_trans_pcie_d3_resume which will let the op_mode
know. Once the op_mode is informed that the WowLAN firmware
has crashed, it can't do much besides resetting the whole
device.

CC: <stable@vger.kernel.org>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: pcie: prepare the enablement of 31 TFD queues
Emmanuel Grumbach [Sun, 25 Jan 2015 08:36:31 +0000 (10:36 +0200)]
iwlwifi: pcie: prepare the enablement of 31 TFD queues

Some devices have 31 TFD queues. Don't enable it yet since
there are still issues with it, but at least prepare the
code for it. There was a bug in the read pointer assignment,
fix that. Also, move the inline functions to iwl-scd.h which
is the right place.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: enable forcing single stream Tx decision
Eyal Shapira [Wed, 28 Jan 2015 12:44:06 +0000 (14:44 +0200)]
iwlwifi: mvm: rs: enable forcing single stream Tx decision

In certain testing scenarios we'd like to force a decision
between STBC/BFER/SISO. In the normal scenario this decision
is done by the FW. Enable this option vis debugfs.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add beamformer support
Eyal Shapira [Fri, 16 Jan 2015 20:37:04 +0000 (22:37 +0200)]
iwlwifi: mvm: add beamformer support

VHT Beamformer (BFER) will be used if the peer supports it
and there's a benefit to use it vs. STBC or SISO.
The driver now tells the FW whether BFER and/or STBC are
allowed but the FW will make the decision to use either
or stick to SISO on its own.
BFER is limited to a single remote peer. The driver takes
care of ensuring this to the FW and prioritizes with which
peer BFER will be used.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: pcie: don't dump useless data when a TFD queue hangs
Emmanuel Grumbach [Sun, 25 Jan 2015 13:54:40 +0000 (15:54 +0200)]
iwlwifi: pcie: don't dump useless data when a TFD queue hangs

Printing all the scratch data of the TFDs of that queue is
useless and stuffed the kernel log with data. Remove that.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: always use mac color zero
Luciano Coelho [Thu, 29 Jan 2015 10:48:20 +0000 (12:48 +0200)]
iwlwifi: mvm: always use mac color zero

We don't really need to use different mac colors when adding mac
contexts, because they're not used anywhere.  In fact, the firmware
doesn't accept 255 as a valid color, so we get into a SYSASSERT 0x3401
when we reach that.

Remove the color increment to use always zero and avoid reaching 255.

CC: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: remove space padding after sysassert description
Johannes Berg [Wed, 28 Jan 2015 22:43:24 +0000 (23:43 +0100)]
iwlwifi: mvm: remove space padding after sysassert description

There's really no reason to pad out the field with spaces at the
end of the line - they're practically invisible there anyway.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: don't reprobe if we fail during reconfig and fw_restart is false
Luciano Coelho [Wed, 28 Jan 2015 07:00:27 +0000 (09:00 +0200)]
iwlwifi: mvm: don't reprobe if we fail during reconfig and fw_restart is false

If we don't want to restart the firmware, don't reprobe either in case
of a failure during reconfiguration.  This allows us to debug failures
in the reconfig flow as well.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: check IWL_UCODE_TLV_API_SCD_CFG in API and not in capa
Emmanuel Grumbach [Sun, 25 Jan 2015 08:03:17 +0000 (10:03 +0200)]
iwlwifi: mvm: check IWL_UCODE_TLV_API_SCD_CFG in API and not in capa

IWL_UCODE_TLV_API_SCD_CFG is a new API and hence, check if
enabled in the correct field.

Fixes: 0294d9eece86 ("iwlwifi: mvm: let the firmware configure the scheduler")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: fix rx chains configuration in phy ctxt cmd
Liad Kaufman [Thu, 22 Jan 2015 17:01:35 +0000 (19:01 +0200)]
iwlwifi: mvm: fix rx chains configuration in phy ctxt cmd

In the PHY_CTXT command sent to the FW the TX chains were
indeed configured by the values of both FW TLVs and of NVM,
but the RX chains were left out and configured only by FW
TLV.

This causes problems in 4165 HW, where there are 1x1
antennas, and the wrong configuration denies the driver
from connecting to the AP.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: pcie: support secured boot flow for family 8000 B step
Eran Harary [Tue, 20 Jan 2015 13:37:34 +0000 (15:37 +0200)]
iwlwifi: pcie: support secured boot flow for family 8000 B step

The driver loads the 2 CPU sections, then it needs to let
the firmware know to start the authentication of the
sections. This is done by writing the relevants bits to
FH_UCODE_LOAD_STATUS.

For CPU1, the driver sets the lower 16 bits. For both CPUs,
the driver sets all the 32 bits.

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: remove stats argument from functions
Johannes Berg [Wed, 21 Jan 2015 17:05:04 +0000 (18:05 +0100)]
iwlwifi: mvm: rs: remove stats argument from functions

The stats argument is always only passed as &mvm->drv_rx_stats, so
there's no point in passing it when the mvm pointer is passed.
Remove the argument entirely.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: use a new API for enabling STBC
Eyal Shapira [Thu, 15 Jan 2015 18:22:34 +0000 (20:22 +0200)]
iwlwifi: mvm: use a new API for enabling STBC

The new API tells the FW that it's allowed to use STBC
but the FW will decide on its own whether to use STBC
or SISO (and in the future Beamformer).
Keep support for the old API which sets STBC explicitly
in the rates in the LQ table while we still support old
FW revisions.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: refactor ht/vht init
Eyal Shapira [Fri, 16 Jan 2015 09:01:51 +0000 (11:01 +0200)]
iwlwifi: mvm: rs: refactor ht/vht init

Prepare to add some more code there so refactor to
separate functions.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: really disable TDLS queues
Emmanuel Grumbach [Tue, 20 Jan 2015 15:07:10 +0000 (17:07 +0200)]
iwlwifi: mvm: really disable TDLS queues

for_each_set_bit expect the size in number of bits and not
in bytes.

Fixes: a0f6bf2a5b01 ("iwlwifi: mvm: use private TFD queues for TDLS stations")
Reviewed-by: Arik Nemtsov <arik@wizery.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: Do not consider invalid HW queues in queue mask
Ilan Peer [Sun, 18 Jan 2015 03:35:00 +0000 (22:35 -0500)]
iwlwifi: mvm: Do not consider invalid HW queues in queue mask

The iwl_mvm_mac_get_queues_mask() added vif->hw_queue[ac] to the
queue mask although it might be set to IEEE80211_INVAL_HW_QUEUE.
Fix it.

Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: BT Coex - set all the co-running values to 0
Emmanuel Grumbach [Thu, 22 Jan 2015 13:14:38 +0000 (15:14 +0200)]
iwlwifi: mvm: BT Coex - set all the co-running values to 0

With this value, we de-facto disable the feature. Since it
is not working yet, disable it completely.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: support family 8000 C step
Eran Harary [Mon, 12 Jan 2015 08:43:58 +0000 (10:43 +0200)]
iwlwifi: mvm: support family 8000 C step

C step functionality in the driver is exactly the same as
B step besides the ucode name that present as iwlwifi-8000C-xx.ucode
instead of iwlwifi-8000B-xx.ucode

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: pcie: init ref_lock
Johannes Berg [Mon, 19 Jan 2015 20:09:09 +0000 (21:09 +0100)]
iwlwifi: pcie: init ref_lock

The ref_lock that was recently added is missing initialization
which makes lockdep unhappy and is generally a bad idea.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: document switch case fall-through in iwl_mvm_send_sta_key
John W. Linville [Mon, 12 Jan 2015 21:18:11 +0000 (16:18 -0500)]
iwlwifi: mvm: document switch case fall-through in iwl_mvm_send_sta_key

Add a comment indicating that the WLAN_CIPHER_SUITE_WEP104 case falls
through to the WLAN_CIPHER_SUITE_WEP40 case in iwl_mvm_send_sta_key.
This will document that the lack of a break is intentional.

Coverity: CID 1260023

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: use STBC regardless of power save mode
Eyal Shapira [Mon, 12 Jan 2015 09:53:06 +0000 (11:53 +0200)]
iwlwifi: mvm: rs: use STBC regardless of power save mode

Tx STBC was used only when in CAM mode or if powersave is disabled.
Effectively this meant we never used STBC as these modes aren't
used on most platforms by default. Change that.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoRevert "iwlwifi: mvm: drop non VO frames when flushing"
Emmanuel Grumbach [Tue, 20 Jan 2015 07:25:19 +0000 (09:25 +0200)]
Revert "iwlwifi: mvm: drop non VO frames when flushing"

This is now implemented by mac80211 (commit below).
mac80211 will flush/drop the frames on the queues before
suspending / disconnecting.
It will then send the deauth and wait until the queues are
empty.

commit 3b24f4c65386dc0f2efb41027bc6e410ea2c0049
Author: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Date:   Wed Jan 7 15:42:39 2015 +0200

    mac80211: let flush() drop packets when possible

This reverts commit 4e6c48e0984e28d064ee8fbc292aee7b7920c507.

9 years agoiwlwifi: mvm: add support for new LTR command
Emmanuel Grumbach [Tue, 13 Jan 2015 08:16:30 +0000 (10:16 +0200)]
iwlwifi: mvm: add support for new LTR command

This new command will give finer granularity to configure
the platform.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: BT Coex - fine tune the MPLUT register
Emmanuel Grumbach [Tue, 13 Jan 2015 11:56:26 +0000 (13:56 +0200)]
iwlwifi: mvm: BT Coex - fine tune the MPLUT register

This allow to better preserve the BT performance while
WiFi is running.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoMerge remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next
Emmanuel Grumbach [Tue, 20 Jan 2015 07:23:27 +0000 (09:23 +0200)]
Merge remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Conflicts:
drivers/net/wireless/iwlwifi/iwl-fw-file.h
drivers/net/wireless/iwlwifi/mvm/scan.c

9 years agoiwlwifi: mvm: move U-APSD decision to authentication
Johannes Berg [Thu, 13 Nov 2014 19:53:45 +0000 (20:53 +0100)]
iwlwifi: mvm: move U-APSD decision to authentication

In order to change the usage of U-APSD on the fly later,
move the enabling condition into a new function that is
called when authenticated.

This allows the module parameter to become writable, it
won't take effect immediately but at least on the next
association the new value will be used.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoMerge tag 'tags/mac80211-next-for-davem-2015-01-19' into iwlwifi-next
Emmanuel Grumbach [Tue, 20 Jan 2015 07:11:39 +0000 (09:11 +0200)]
Merge tag 'tags/mac80211-next-for-davem-2015-01-19' into iwlwifi-next

Some further updates for net-next:
 * fix network-manager which was broken by the previous changes
 * fix delete-station events, which were broken by me making the
   genlmsg_end() mistake
 * fix a timer left running during suspend in some race conditions
   that would cause an annoying (but harmless) warning
 * (less important, but in the tree already) remove 80+80 MHz rate
   reporting since the spec doesn't distinguish it from 160 MHz;
   as the bitrate they're both 160 MHz bandwidth

9 years agoiwlwifi: mvm: ignore temperature updates in the RX statistics notification
Luciano Coelho [Wed, 14 Jan 2015 19:39:34 +0000 (21:39 +0200)]
iwlwifi: mvm: ignore temperature updates in the RX statistics notification

If the firmware sends spontaneous DTS notfications with the
temperature (indicated in a TLV), we can ignore the temperature we get
in the RX statistics notifications.  This prevents potentially
handling the same temperature change twice.  It also ignores
notifications with temperature equal to 0 that happens from time to
time.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: cleanup unuseful and overflowing traces
Eyal Shapira [Thu, 15 Jan 2015 16:34:14 +0000 (18:34 +0200)]
iwlwifi: mvm: rs: cleanup unuseful and overflowing traces

These aren't useful and overflowing so drop them
and also fix a minor typo.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: generate statistics debugfs code
Johannes Berg [Thu, 15 Jan 2015 11:59:26 +0000 (12:59 +0100)]
iwlwifi: mvm: generate statistics debugfs code

There's no need to duplicate the structure field name in the string,
just generate the string in the macro that's there anyway. To keep
the debugfs output the same, rename one (otherwise unused) field.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: move statistics API to new header file
Johannes Berg [Wed, 14 Jan 2015 14:54:18 +0000 (15:54 +0100)]
iwlwifi: mvm: move statistics API to new header file

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: sync statistics firmware API
Johannes Berg [Wed, 14 Jan 2015 14:16:03 +0000 (15:16 +0100)]
iwlwifi: mvm: sync statistics firmware API

The firmware API structs are split differently, synchronize
the struct splits with the current firmware definitions.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: correctly set the NMI register
Emmanuel Grumbach [Wed, 7 Jan 2015 14:44:06 +0000 (16:44 +0200)]
iwlwifi: correctly set the NMI register

When we want to trigger an NMI in the device, we need to set
bit 7 and not bit 0. However, older firmwares don't register
to the interrupt issued by bit 7. Use bit 7 first so that
the correct interrupt will be issued hoping that the firmware
will react. To be on the safe side, set bit 0 in case the
firmware didn't register to the proper interrupt.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add rxf and txf to dump data
Liad Kaufman [Mon, 15 Dec 2014 15:54:16 +0000 (17:54 +0200)]
iwlwifi: mvm: add rxf and txf to dump data

When the FW is in error status - try to read the RXF and
TXF (all of them) and add them to the dump data.

This shouldn't happen in non-error statuses, as we don't
want to stop the RXF/TXF while they are running.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add support for dumping a secondary SRAM
Ido Yariv [Wed, 10 Dec 2014 17:39:27 +0000 (12:39 -0500)]
iwlwifi: mvm: add support for dumping a secondary SRAM

Some HW modules have two SRAMs. In such cases add the secondary SRAM to
the list of dumped segments.

Signed-off-by: Ido Yariv <idox.yariv@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add debugfs file for misbehaving U-APSD AP
Johannes Berg [Fri, 9 Jan 2015 13:15:32 +0000 (14:15 +0100)]
iwlwifi: mvm: add debugfs file for misbehaving U-APSD AP

As this functionality relies on getting a firmware notification
it is difficult to test. Allow accessing the data for it from
debugfs to be able to trigger all kinds of scenarios to test.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: let the firmware configure the scheduler
Emmanuel Grumbach [Mon, 5 Jan 2015 14:52:55 +0000 (16:52 +0200)]
iwlwifi: mvm: let the firmware configure the scheduler

A new host command can be used to configure the scheduler
instead of accessing the scheduler's registers from the
driver. This is easier and less error prone since accessing
the hardware at certain moments can lead to races with the
firmware.
Prefer to use the host command whenever it is available.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: scan dwell time corrections
David Spinadel [Mon, 29 Dec 2014 13:43:48 +0000 (15:43 +0200)]
iwlwifi: mvm: scan dwell time corrections

Use only basic dwell time (10 ms for active scan and 110 for passive),
regardless of the number of the probes and the band, if it is
supported by the FW. The FW will add 3 ms for each probe sent and 10
ms for low band channels.
Add a TLV flag to indicate such support in FW.

This fix is needed to fix few bugs regarding scans that take too much time.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: remove unused TLV capability flags
Emmanuel Grumbach [Tue, 6 Jan 2015 07:48:54 +0000 (09:48 +0200)]
iwlwifi: remove unused TLV capability flags

The driver doesn't support the firmwares that don't have
these capabilities. The code that actually used these
flags has been removed already, but the flags were left
for an unclear reason. Remove them.

Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add print of he nvm version
Eran Harary [Mon, 29 Dec 2014 11:02:13 +0000 (13:02 +0200)]
iwlwifi: mvm: add print of he nvm version

Print the nvm version in the log for debugging purposes.

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: set max_out_time equal to frag_passive_dwell in fragmented scan
Haim Dreyfuss [Thu, 18 Dec 2014 10:34:01 +0000 (12:34 +0200)]
iwlwifi: mvm: set max_out_time equal to frag_passive_dwell in fragmented scan

Fragmented scan should be applied for all channels, passive and active.
When scanning on passive channels the firmware uses frag_passive_dwell
to define the maximum continuous scan time before returning to the
operating channel. On active channels max_out_time is the parameter
used by the firmware to define the maximum time allowed out of the
operating channel. Since active channels' scan should also be fragmented
set max_out_time equal to frag_passive_dwell.
In addition:
- Set max_out_time and suspend_time if the firmware doesn't support
  fragmented scan to avoid unexpected behavior.
- Adjust max_out_time for second level of scan precedence.

Signed-off-by: Haim Dreyfuss <haim.dreyfuss@intel.com>
Reviewed-by: Alexander Bondar <alexander.bondar@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: allow to disable MIMO for P2P only
Emmanuel Grumbach [Mon, 5 Jan 2015 07:48:54 +0000 (09:48 +0200)]
iwlwifi: mvm: rs: allow to disable MIMO for P2P only

This is to work around interoperability bugs with devices
that don't hanle MIMO properly.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: allow to collect debug data from non-sleepable context
Emmanuel Grumbach [Mon, 29 Dec 2014 07:42:37 +0000 (09:42 +0200)]
iwlwifi: mvm: allow to collect debug data from non-sleepable context

iwl_mvm_fw_dbg_collect allows to collect debug data from
the firmware. Most of the firmware interaction is done in
non-sleepable context. It makes little sense to force the
caller of iwl_mvm_fw_dbg_collect to sleep.
Defer the actual collection to a worker so that this
function will be able to be called from any context.

Reviewed-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: make sure state isn't in d0i3 when stopping fw monitor
Liad Kaufman [Sun, 4 Jan 2015 09:03:13 +0000 (11:03 +0200)]
iwlwifi: mvm: make sure state isn't in d0i3 when stopping fw monitor

In case platform is in d0i3 - make sure it is awake when
writing the registers to stop the monitor when collecting FW
debug data. Plus, remove unneeded mutex locking currently
done.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: Add debugfs entry to enable scan offload notification
Alexander Bondar [Tue, 9 Dec 2014 17:15:49 +0000 (19:15 +0200)]
iwlwifi: mvm: Add debugfs entry to enable scan offload notification

This option enables scan offload iteration complete notification from
firmware which includes the last iteration's status and the scanned
channels from the current iteration.

Signed-off-by: Haim Dreyfuss <haim.dreyfuss@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: add new config and PCI IDs for 4165 series
Oren Givon [Tue, 23 Dec 2014 08:39:48 +0000 (10:39 +0200)]
iwlwifi: add new config and PCI IDs for 4165 series

Add a new config for 4165 series over PCI and insert support
for two new 4165 series PCI IDs.

Signed-off-by: Oren Givon <oren.givon@intel.com>
Reviewed-by: Liad Kaufman <liad.kaufman@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: make sure state isn't in d0i3 when collecting fw dbg
Liad Kaufman [Thu, 1 Jan 2015 15:42:46 +0000 (17:42 +0200)]
iwlwifi: mvm: make sure state isn't in d0i3 when collecting fw dbg

This makes sure that we're not trying to read/write any of
the FW debug data collected during d0i3.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Reviewed-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rs: repeat initial legacy rates in LQ table
Eyal Shapira [Sun, 28 Dec 2014 20:12:38 +0000 (22:12 +0200)]
iwlwifi: mvm: rs: repeat initial legacy rates in LQ table

Repeating the legacy rates avoids degrading quickly to
lower rates due to collisions which is common when doing
TCP Tx traffic in legacy.
This slightly improves TCP Tx throughput while working
in legacy in different scenarios.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: don't indicate no BA if STA was in powersave
Eyal Shapira [Wed, 31 Dec 2014 13:22:38 +0000 (15:22 +0200)]
iwlwifi: mvm: don't indicate no BA if STA was in powersave

If Tx failed because the STA was in powersave there's no point
in sending a BAR so avoid indicating AMPDU_NO_BACK to mac80211.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: validate tid and sta_id in ba_notif
Eyal Shapira [Fri, 16 Jan 2015 09:09:30 +0000 (11:09 +0200)]
iwlwifi: mvm: validate tid and sta_id in ba_notif

These are coming from the FW and are used to access arrays.
Bad values can cause an out of bounds access so discard
such ba_notifs and warn.

CC: <stable@vger.kernel.org> [3.10+]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agomac80211: delete the assoc/auth timer upon suspend
Emmanuel Grumbach [Mon, 19 Jan 2015 14:53:06 +0000 (16:53 +0200)]
mac80211: delete the assoc/auth timer upon suspend

While suspending, we destroy the authentication /
association that might be taking place. While doing so, we
forgot to delete the timer which can be firing after
local->suspended is already set, producing the warning below.

Fix that by deleting the timer.

[66722.825487] WARNING: CPU: 2 PID: 5612 at net/mac80211/util.c:755 ieee80211_can_queue_work.isra.18+0x32/0x40 [mac80211]()
[66722.825487] queueing ieee80211 work while going to suspend
[66722.825529] CPU: 2 PID: 5612 Comm: kworker/u16:69 Tainted: G        W  O  3.16.1+ #24
[66722.825537] Workqueue: events_unbound async_run_entry_fn
[66722.825545] Call Trace:
[66722.825552]  <IRQ>  [<ffffffff817edbb2>] dump_stack+0x4d/0x66
[66722.825556]  [<ffffffff81075cad>] warn_slowpath_common+0x7d/0xa0
[66722.825572]  [<ffffffffa06b5b90>] ? ieee80211_sta_bcn_mon_timer+0x50/0x50 [mac80211]
[66722.825573]  [<ffffffff81075d1c>] warn_slowpath_fmt+0x4c/0x50
[66722.825586]  [<ffffffffa06977a2>] ieee80211_can_queue_work.isra.18+0x32/0x40 [mac80211]
[66722.825598]  [<ffffffffa06977d5>] ieee80211_queue_work+0x25/0x50 [mac80211]
[66722.825611]  [<ffffffffa06b5bac>] ieee80211_sta_timer+0x1c/0x20 [mac80211]
[66722.825614]  [<ffffffff8108655a>] call_timer_fn+0x8a/0x300

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoRevert "wireless: Support of IFLA_INFO_KIND rtnl attribute"
Johannes Berg [Mon, 19 Jan 2015 17:49:50 +0000 (18:49 +0100)]
Revert "wireless: Support of IFLA_INFO_KIND rtnl attribute"

This reverts commit ba1debdfed974f25aa598c283567878657b292ee.

Oliver reported that it breaks network-manager, for some reason with
this patch NM decides that the device isn't wireless but "generic"
(ethernet), sees no carrier (as expected with wifi) and fails to do
anything else with it.

Revert this to unbreak userspace.

Reported-by: Oliver Hartkopp <socketcan@hartkopp.net>
Tested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoiwlwifi: mvm: abort scheduled scan upon RFKILL
Emmanuel Grumbach [Sun, 18 Jan 2015 19:39:30 +0000 (21:39 +0200)]
iwlwifi: mvm: abort scheduled scan upon RFKILL

When we have an active scheduled scan, and the RFKILL
interrupt kicks in, the stack will cancel the scheduled
scan as part of the down flow. But cancelling scheduled
scan usually implies sending a command to the firwmare
which has been killed as part of the RFKILL interrupt
handling.
Because of that, we returned an error to mac80211 when
it asked to stop the scheduled scan and didn't notify the
end of the scheduled scan. Besides a fat warning, this led
to a situation in which cfg80211 would refuse any new scan
request.

To disentangle this, fake that the scheduled scan has been
stopped without sending the command to the firwmare, return
0 after having properly let cfg80211 know that the scan
has been cancelled.

This is basically the same as:
commit 9b520d84957d63348e87c0f2cbd21d86e1e8f2f2
Author: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Date:   Tue Nov 4 15:54:11 2014 +0200

    iwlwifi: mvm: abort scan upon RFKILL

    This code existed but not for all the different FW APIs
    we support.
    Fix this.

but for the scheduled scan case.

Link: http://permalink.gmane.org/gmane.linux.kernel.wireless.general/133232
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agocfg80211: fix checking nl80211_send_station() return value
Johannes Berg [Fri, 16 Jan 2015 20:05:02 +0000 (21:05 +0100)]
cfg80211: fix checking nl80211_send_station() return value

The return value from nl80211_send_station() is the length of the
skb, or a negative error, so abort sending the message only when
the return value was negative.

This fixes the ibss_rsn wpa_supplicant test case.

Reported-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: remove doubled semicolon
Johannes Berg [Fri, 16 Jan 2015 12:27:56 +0000 (13:27 +0100)]
mac80211: remove doubled semicolon

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: change bandwidth reporting to explicit field
Johannes Berg [Thu, 15 Jan 2015 15:14:02 +0000 (16:14 +0100)]
cfg80211: change bandwidth reporting to explicit field

For some reason, we made the bandwidth separate flags, which
is rather confusing - a single rate cannot have different
bandwidths at the same time.

Change this to no longer be flags but use a separate field
for the bandwidth ('bw') instead.

While at it, add support for 5 and 10 MHz rates - these are
reported as regular legacy rates with their real bitrate,
but tagged as 5/10 now to make it easier to distinguish them.

In the nl80211 API, the flags are preserved, but the code
now can also clearly only set a single one of the flags.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: remove 80+80 MHz rate reporting
Johannes Berg [Thu, 15 Jan 2015 15:05:21 +0000 (16:05 +0100)]
cfg80211: remove 80+80 MHz rate reporting

These rates are treated the same as 160 MHz in the spec, so
it makes no sense to distinguish them. As no driver uses them
yet, this is also not a problem, just remove them.

In the userspace API the field remains reserved to preserve
API and ABI.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: remove 80+80 MHz rate reporting
Johannes Berg [Thu, 15 Jan 2015 15:02:46 +0000 (16:02 +0100)]
mac80211: remove 80+80 MHz rate reporting

These rates are treated the same as 160 MHz in the spec,
so it makes no sense to distinguish them. As no driver
uses them yet, this is also not a problem, just remove
them.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: docs: remove station_info_flags
Johannes Berg [Wed, 14 Jan 2015 12:57:36 +0000 (13:57 +0100)]
cfg80211: docs: remove station_info_flags

I removed the enum earlier, but forgot to remove it from the
documentation - do that now.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: send netdetect configuration info in NL80211_CMD_GET_WOWLAN
Luciano Coelho [Fri, 9 Jan 2015 12:06:37 +0000 (14:06 +0200)]
nl80211: send netdetect configuration info in NL80211_CMD_GET_WOWLAN

Send the netdetect configuration information in the response to
NL8021_CMD_GET_WOWLAN commands.  This includes the scan interval,
SSIDs to match and frequencies to scan.

Additionally, add the NL80211_WOWLAN_TRIG_NET_DETECT with
NL80211_ATTR_WOWLAN_TRIGGERS_SUPPORTED.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: avoid reg-hints in self-managed only systems
Arik Nemtsov [Wed, 7 Jan 2015 14:47:20 +0000 (16:47 +0200)]
cfg80211: avoid reg-hints in self-managed only systems

When a system contains only self-managed regulatory devices all hints
from the regulatory core are ignored. Stop hint processing early in this
case. These systems usually don't have CRDA deployed, which results in
endless (irrelevent) logs of the form:
cfg80211: Calling CRDA to update world regulatory domain

Make sure there's at least one self-managed device before discarding a
hint, in order to prevent initial hints from disappearing on CRDA
managed systems.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: introduce sync regdom set API for self-managed
Arik Nemtsov [Wed, 7 Jan 2015 14:47:19 +0000 (16:47 +0200)]
cfg80211: introduce sync regdom set API for self-managed

A self-managed device will sometimes need to set its regdomain synchronously.
Notably it should be set before usermode has a chance to query it. Expose
a new API to accomplish this which requires the RTNL.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Ilan Peer <ilan.peer@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: don't defer scans in case of radar detection
Eliad Peller [Wed, 7 Jan 2015 15:50:11 +0000 (17:50 +0200)]
mac80211: don't defer scans in case of radar detection

Radar detection can last indefinite time. There is no
point in deferring a scan request in this case - simply
return -EBUSY.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: consider only relevant vifs for radar_required calculation
Eliad Peller [Wed, 7 Jan 2015 15:50:10 +0000 (17:50 +0200)]
mac80211: consider only relevant vifs for radar_required calculation

ctx->conf.radar_enabled should reflect whether radar
detection is enabled for the channel context.

When calculating it, make it consider only the vifs
that have this context assigned (instead of all the
vifs).

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: remove local->radar_detect_enabled
Eliad Peller [Wed, 7 Jan 2015 15:50:09 +0000 (17:50 +0200)]
mac80211: remove local->radar_detect_enabled

local->radar_detect_enabled should tell whether
radar_detect is enabled on any interface belonging
to local.

However, it's not getting updated correctly
in many cases (actually, when testing with hwsim
it's never been set, even when the dfs master
is beaconing).

Instead of handling all the corner cases
(e.g. channel switch), simply check whether
radar detection is enabled only when needed,
instead of caching the result.

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: add TDLS supported channels correctly
Arik Nemtsov [Wed, 7 Jan 2015 14:45:07 +0000 (16:45 +0200)]
mac80211: add TDLS supported channels correctly

The function adding the supported channels IE during a TDLS connection had
several issues:
1. If the entire subband is usable, the function exitted the loop without
   adding it
2. The function only checked chandef_usable, ignoring flags like RADAR
   which would prevent TDLS off-channel communcation.
3. HT20 was explicitly required in the chandef, while not a requirement
   for TDLS off-channel.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: let flush() drop packets when possible
Emmanuel Grumbach [Wed, 7 Jan 2015 13:42:39 +0000 (15:42 +0200)]
mac80211: let flush() drop packets when possible

When roaming / suspending, it makes no sense to wait until
the transmit queues of the device are empty. In extreme
condition they can be starved (VO saturating the air), but
even in regular cases, it is pointless to delay the roaming
because the low level driver is trying to send packets to
an AP which is far away. We'd rather drop these packets and
let TCP retransmit if needed. This will allow to speed up
the roaming.

For suspend, the explanation is even more trivial.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoiwlwifi: mvm: set the tx cmd tid for BAR frame correctly
Eyal Shapira [Wed, 31 Dec 2014 16:34:56 +0000 (18:34 +0200)]
iwlwifi: mvm: set the tx cmd tid for BAR frame correctly

BAR tx cmd tid was set to non qos (8). This is wrong as BAR
should be sent with the tid of the BA session.
This led to a corruption in the firmware. The visible
effect of this from the driver side is the BA notification
that comes back after the BAR. It was botched and led to the
WARNING below.

------------[ cut here ]------------
WARNING: CPU: 2 PID: 17707 at /home/tester/workspace_hostap/iwlwifi/drivers/net/wireless/iwlwifi/mvm/tx.c:976 iwl_mvm_rx_ba_notif+0x4ba/0x4d0 [iwlmvm]()
Q 4500, tid 8, flow 65535
Modules linked in: iwlmvm(O) mac80211(O) iwlwifi(O) cfg80211(O) compat(O) netconsole configfs ctr ccm arc4 autofs4 microcode bnep rfcomm snd_hda_codec_hdmi snd_hda_codec_idt snd_hda_codec_generic snd_hda_intel joydev snd_hda_codec uvcvideo videobuf2_core snd_hwdep videodev snd_pcm videobuf2_vmalloc videobuf2_memops i915 snd_seq_midi snd_rawmidi snd_seq_midi_event snd_seq snd_timer snd_seq_device drm_kms_helper dell_wmi dell_laptop drm btusb bluetooth snd psmouse i2c_algo_bit sparse_keymap wmi soundcore 6lowpan_iphc dcdbas serio_raw video lpc_ich ppdev mac_hid parport_pc nfsd nfs_acl auth_rpcgss nfs fscache binfmt_misc lockd sunrpc lp parport msdos sdhci_pci sdhci mmc_core ahci libahci e1000e ptp pps_core [last unloaded: compat]
CPU: 2 PID: 17707 Comm: irq/46-iwlwifi Tainted: G        W  O 3.14.17-patched #4
Hardware name: Dell Inc. Latitude E6430/0CPWYR, BIOS A09 12/13/2012
 00000000 00000000 ebd49d6c c1616221 f985dbdc ebd49d9c c1044e44 f9861df4
 ebd49dc8 0000452b f985dbdc 000003d0 f98395da f98395da ebd49f10 eaf3d8a4
 0000ffff ebd49db4 c1044f03 00000009 ebd49dac f9861df4 ebd49dc8 ebd49e64
Call Trace:
 [<c1616221>] dump_stack+0x41/0x52
 [<c1044e44>] warn_slowpath_common+0x84/0xa0
 [<f98395da>] ? iwl_mvm_rx_ba_notif+0x4ba/0x4d0 [iwlmvm]
 [<f98395da>] ? iwl_mvm_rx_ba_notif+0x4ba/0x4d0 [iwlmvm]
 [<c1044f03>] warn_slowpath_fmt+0x33/0x40
 [<f98395da>] iwl_mvm_rx_ba_notif+0x4ba/0x4d0 [iwlmvm]
 [<c10e3952>] ? ring_buffer_unlock_commit+0xa2/0xd0
 [<c10e9767>] ? trace_buffer_unlock_commit+0x37/0x50
 [<f98568a3>] ? iwl_tm_mvm_send_rx+0x53/0x90 [iwlmvm]
 [<f98327a8>] iwl_mvm_rx_dispatch+0x108/0x130 [iwlmvm]
 [<f9eac7e7>] iwl_pcie_irq_handler+0xf17/0x15b0 [iwlwifi]
 [<c10994c1>] irq_thread_fn+0x21/0x50
 [<c109926c>] irq_thread+0xec/0x110
 [<c10994a0>] ? irq_thread_dtor+0xb0/0xb0
 [<c10993f0>] ? irq_finalize_oneshot.part.34+0xc0/0xc0
 [<c1099180>] ? wake_threads_waitq+0x40/0x40
 [<c1062fdb>] kthread+0x9b/0xb0
 [<c1627137>] ret_from_kernel_thread+0x1b/0x28
 [<c1062f40>] ? flush_kthread_worker+0x90/0x90
---[ end trace 5e0f67374816db17 ]---

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: drop non VO frames when flushing
Emmanuel Grumbach [Mon, 12 Jan 2015 14:34:26 +0000 (16:34 +0200)]
iwlwifi: mvm: drop non VO frames when flushing

This change has already been implemented in iwldvm:

commit a260e7b3f0307878b99d57ed1406cf2d497923b8
Author: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Date:   Sun Oct 5 09:11:14 2014 +0300

    iwlwifi: dvm: drop non VO frames when flushing

Since I added the flush() callback implementation in mvm,
we got reports that the queues are stuck while roaming
or suspending.
This commit above helped much for iwldvm, implement the
same behavior for iwlmvm.

CC: <stable@vger.kernel.org> [3.16+]
Fixes: c5b0e7c0565a ("iwlwifi: mvm: implement mac80211's flush callback")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: fix EBS on single scan
David Spinadel [Mon, 1 Dec 2014 21:30:07 +0000 (23:30 +0200)]
iwlwifi: mvm: fix EBS on single scan

EBS error detection isn't supported by all FWs, so turn it on
only if the FW advertises such support.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agorfkill: document rfkill module parameters
Andrew Clausen [Fri, 9 Jan 2015 20:24:55 +0000 (20:24 +0000)]
rfkill: document rfkill module parameters

Document the rfkill module parameters default_state and
master_switch_mode.

Signed-off-by: Andrew Clausen <andrew.p.clausen@gmail.com>
[rewrite commit message]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: fix handling TIM IE when stations disconnect
Johannes Berg [Fri, 9 Jan 2015 10:40:39 +0000 (11:40 +0100)]
mac80211: fix handling TIM IE when stations disconnect

When a station disconnects with frames still pending, we clear
the TIM bit, but too late - it's only cleared when the station
is already removed from the driver, and thus the driver can get
confused (and hwsim will loudly complain.)

Fix this by clearing the TIM bit earlier, when the station has
been unlinked but not removed from the driver yet. To do this,
refactor the TIM recalculation to in that case ignore traffic
and simply assume no pending traffic - this is correct for the
disconnected station even though the frames haven't been freed
yet at that point.

This patch isn't needed for current drivers though as they don't
check the station argument to the set_tim() operation and thus
don't really run into the possible confusion.

Reported-by: Jouni Malinen <j@w1.fi>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211_hwsim: fix PS debugfs file locking
Johannes Berg [Fri, 9 Jan 2015 10:29:50 +0000 (11:29 +0100)]
mac80211_hwsim: fix PS debugfs file locking

The functions called within the iterators must be called with
tasklets disabled, so use atomic iteration like the rest of
the code and disable tasklets around the whole operation.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: provide per-TID RX/TX MSDU counters
Johannes Berg [Fri, 21 Nov 2014 13:26:31 +0000 (14:26 +0100)]
mac80211: provide per-TID RX/TX MSDU counters

Implement the new counters cfg80211 can now advertise to userspace.
The TX code is in the sequence number handler, which is a bit odd,
but that place already knows the TID and frame type, so it was
easiest and least impact there.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: support per-TID station statistics
Johannes Berg [Fri, 19 Dec 2014 11:34:00 +0000 (12:34 +0100)]
nl80211: support per-TID station statistics

The base for the current statistics is pretty mixed up, support
exporting RX/TX statistics for MSDUs per TID. This (currently)
covers received MSDUs, transmitted MSDUs and retries/failures
thereof.

Doing it per TID for MSDUs makes more sense than say only per AC
because it's symmetric - we could export per-AC statistics for all
frames (which AC we used for transmission can be determined also
for management frames) but per TID is better and usually data
frames are really the ones we care about. Also, on RX we can't
determine the AC - but we do know the TID for any QoS MPDU we
received.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: clarify packet statistics descriptions
Johannes Berg [Fri, 21 Nov 2014 11:40:05 +0000 (12:40 +0100)]
nl80211: clarify packet statistics descriptions

The current statistics we keep aren't very clear, some are on
MPDUs and some on MSDUs/MMPDUs. Clarify the descriptions based
on the counters mac80211 keeps.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add nl80211 beacon-only statistics
Johannes Berg [Mon, 17 Nov 2014 13:12:22 +0000 (14:12 +0100)]
cfg80211: add nl80211 beacon-only statistics

Add these two values:
 * BEACON_RX: number of beacons received from this peer
 * BEACON_SIGNAL_AVG: signal strength average for beacons only

These can then be used for Android Lollipop's statistics request.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: remove enum station_info_flags
Johannes Berg [Mon, 17 Nov 2014 13:08:11 +0000 (14:08 +0100)]
cfg80211: remove enum station_info_flags

This is really just duplicating the list of information that's
already available in the nl80211 attribute, so remove the list.
Two small changes are needed:
 * remove STATION_INFO_ASSOC_REQ_IES complete, but the length
   (assoc_req_ies_len) can be used instead
 * add NL80211_STA_INFO_RX_DROP_MISC which exists internally
   but not in nl80211 yet

This gets rid of the duplicate maintenance of the two lists.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: allow drivers to provide most station statistics
Johannes Berg [Mon, 17 Nov 2014 10:35:23 +0000 (11:35 +0100)]
mac80211: allow drivers to provide most station statistics

In many cases, drivers can filter things like beacons that will
skew statistics reported by mac80211. To get correct statistics
in these cases, call drivers to obtain statistics and let them
override all values, filling values from mac80211 if the driver
didn't provide them. Not all of them make sense for the driver
to fill, so some are still always done by mac80211.

Note that this doesn't currently allow a driver to say "I know
this value is wrong, don't report it at all", or to sum it up
with a mac80211 value (as could be useful for "dropped misc"),
that can be added if it turns out to be needed.

This also gets rid of the get_rssi() method as is can now be
implemented using sta_statistics().

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: send statistics with delete station event
Johannes Berg [Fri, 14 Nov 2014 19:32:57 +0000 (20:32 +0100)]
mac80211: send statistics with delete station event

Use the new cfg80211_del_sta_sinfo() function to send the
statistics about the deleted station with the delete event.
This lets userspace see how much traffic etc. the deleted
station used.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow including station info in delete event
Johannes Berg [Fri, 14 Nov 2014 16:14:00 +0000 (17:14 +0100)]
cfg80211: allow including station info in delete event

When a station is removed, its statistics may be interesting to
userspace, for example for further aggregation of statistics of
all stations that ever connected to an AP.

Introduce a new cfg80211_del_sta_sinfo() function (and make the
cfg80211_del_sta() a static inline calling it) to allow passing
a struct station_info along with this, and send the data in the
nl80211 event message.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add scan time to survey data
Johannes Berg [Fri, 14 Nov 2014 15:44:11 +0000 (16:44 +0100)]
cfg80211: add scan time to survey data

Add the time spent scanning to the survey data so it can be
reported by drivers that collect such information.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: allow survey data to return global data
Johannes Berg [Fri, 14 Nov 2014 15:43:50 +0000 (16:43 +0100)]
cfg80211: allow survey data to return global data

Not all devices are able to report survey data (particularly
time spent for various operations) per channel. As all these
statistics already exist in survey data, allow such devices
to report them (if userspace requested it)

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: remove "channel" from survey names
Johannes Berg [Fri, 14 Nov 2014 15:35:34 +0000 (16:35 +0100)]
cfg80211: remove "channel" from survey names

All of the survey data is (currently) per channel anyway,
so having the word "channel" in the name does nothing. In
the next patch I'll introduce global data to the survey,
where the word "channel" is actually confusing.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: Re-fix accounting of the tailroom-needed counter
Ido Yariv [Tue, 6 Jan 2015 13:39:02 +0000 (08:39 -0500)]
mac80211: Re-fix accounting of the tailroom-needed counter

When hw acceleration is enabled, the GENERATE_IV or PUT_IV_SPACE flags
only require headroom space. Therefore, the tailroom-needed counter can
safely be decremented for most drivers.

The older incarnation of this patch (ca34e3b5) assumed that the above
holds true for all drivers. As reported by Christopher Chavez and
researched by Christian Lamparter and Larry Finger, this isn't a valid
assumption for p54 and cw1200.

Drivers that still require tailroom for ICV/MIC even when HW encryption
is enabled can use IEEE80211_KEY_FLAG_RESERVE_TAILROOM to indicate it.

Signed-off-by: Ido Yariv <idox.yariv@intel.com>
Cc: Christopher Chavez <chrischavez@gmx.us>
Cc: Christian Lamparter <chunkeey@googlemail.com>
Cc: Larry Finger <Larry.Finger@lwfinger.net>
Cc: Solomon Peachy <pizza@shaftnet.org>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoMerge branch 'mac80211' into mac80211-next
Johannes Berg [Wed, 7 Jan 2015 13:39:02 +0000 (14:39 +0100)]
Merge branch 'mac80211' into mac80211-next

Merge mac80211.git to get some changes that would otherwise
cause conflicts with new changes coming here.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agomac80211: skip disabled channels in VHT check
Arik Nemtsov [Thu, 1 Jan 2015 11:43:17 +0000 (13:43 +0200)]
mac80211: skip disabled channels in VHT check

The patch "40a11ca mac80211: check if channels allow 80 MHz for VHT
probe requests" considered disabled channels as VHT enabled, and
mistakenly sent out probe-requests with the VHT IE.

Signed-off-by: Arik Nemtsov <arikx.nemtsov@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: define multicast group names in header
Johannes Berg [Tue, 23 Dec 2014 16:17:38 +0000 (17:17 +0100)]
nl80211: define multicast group names in header

Put the group names into the userspace API header file so that
userspace clients can use symbolic names from there instead of
hardcoding the actual names. This doesn't really change much,
but seems somewhat cleaner.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agocfg80211: add extensible feature flag attribute
Gautam Kumar Shukla [Tue, 23 Dec 2014 15:55:19 +0000 (16:55 +0100)]
cfg80211: add extensible feature flag attribute

With the wiphy::features flag being used up this patch adds a
new field wiphy::ext_features. Considering extensibility this
new field is declared as a byte array. This extensible flag is
exposed to user-space by NL80211_ATTR_EXT_FEATURES.

Cc: Avinash Patil <patila@marvell.com>
Signed-off-by: Gautam (Gautam Kumar) Shukla <gautams@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agonl80211: document NL80211_BSS_STATUS_AUTHENTICATED isn't used
Johannes Berg [Mon, 5 Jan 2015 10:16:42 +0000 (11:16 +0100)]
nl80211: document NL80211_BSS_STATUS_AUTHENTICATED isn't used

The flag is no longer used (and hasn't been for a long time)
since trying to track authentication (and make decisions based
on state) was just causing issues all over - see commit
95de817b9034d50860319f6033ec85d25024694c.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoRevert "mac80211: Fix accounting of the tailroom-needed counter"
Johannes Berg [Mon, 5 Jan 2015 09:28:49 +0000 (10:28 +0100)]
Revert "mac80211: Fix accounting of the tailroom-needed counter"

This reverts commit ca34e3b5c808385b175650605faa29e71e91991b.

It turns out that the p54 and cw2100 drivers assume that there's
tailroom even when they don't say they really need it. However,
there's currently no way for them to explicitly say they do need
it, so for now revert this.

This fixes https://bugzilla.kernel.org/show_bug.cgi?id=90331.

Cc: stable@vger.kernel.org
Fixes: ca34e3b5c808 ("mac80211: Fix accounting of the tailroom-needed counter")
Reported-by: Christopher Chavez <chrischavez@gmx.us>
Bisected-by: Larry Finger <Larry.Finger@lwfinger.net>
Debugged-by: Christian Lamparter <chunkeey@googlemail.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
9 years agoiwlwifi: bump firmware API for mvm devices to 12
Emmanuel Grumbach [Mon, 5 Jan 2015 07:42:40 +0000 (09:42 +0200)]
iwlwifi: bump firmware API for mvm devices to 12

This allows 3160 / 7260 / 7265 / 7265D / 8000 devices to
use the latest version of the firmware.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: scan dwell time corrections
David Spinadel [Mon, 29 Dec 2014 13:43:48 +0000 (15:43 +0200)]
iwlwifi: mvm: scan dwell time corrections

Use only basic dwell time (10 ms for active scan and 110 for passive),
regardless of the number of the probes and the band, if it is
supported by the FW. The FW will add 3 ms for each probe sent and 10
ms for low band channels.
Add a TLV flag to indicate such support in FW.

This fix is needed to fix few bugs regarding scans that take too much time.

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add a flag to enable match found notification
David Spinadel [Tue, 23 Dec 2014 12:38:09 +0000 (14:38 +0200)]
iwlwifi: mvm: add a flag to enable match found notification

Add a flag that enables match found notification to align with
FW API change.

Cc: <stable@vger.kernel.org> [3.17+]
Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: fix out of bounds access to tid_to_mac80211_ac
Eyal Shapira [Wed, 31 Dec 2014 15:58:23 +0000 (17:58 +0200)]
iwlwifi: mvm: fix out of bounds access to tid_to_mac80211_ac

When tid_tspec was set to IWL_TID_NON_QOS (8) this led to an
out of bounds access to the tid_to_mac80211_ac array whose size
is 7. Fix this.

Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: fix Rx with both chains
Eyal Shapira [Wed, 31 Dec 2014 17:30:15 +0000 (19:30 +0200)]
iwlwifi: mvm: fix Rx with both chains

commit 5c90422439d6
"iwlwifi: mvm: don't allow diversity if BT Coex / TT forbid it"
broke Rx with 2 chains for diversity.
This had an impact on throughput where we're using only a single
stream (11a/b/g APs, single stream APs, static SMPS).

Fixes: 5c90422439d6 ("iwlwifi: mvm: don't allow diversity if BT Coex / TT forbid it")
Cc: Stable <stable@vger.kernel.org> [3.16+]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>