deliverable/linux.git
9 years agobrcmfmac: cleanup a sizeof()
Dan Carpenter [Thu, 7 May 2015 09:59:19 +0000 (12:59 +0300)] 
brcmfmac: cleanup a sizeof()

"flowrings" and "*flowrings" are both pointers so this always returns
sizeof(void *) and the current code works fine.  But "*flowrings" is
intended here and static checkers complain, so lets change it.

Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agowl18xx: fallback to default conf in case of invalid conf file
Eliad Peller [Wed, 6 May 2015 08:29:54 +0000 (11:29 +0300)] 
wl18xx: fallback to default conf in case of invalid conf file

If the wl18xx-conf.bin file is missing or invalid (e.g. due
to recent driver change), fallback to default configuration
instead of failing driver load.

Reported-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agomwifiex: parsing aid while receiving association response
Xinming Hu [Mon, 4 May 2015 16:40:08 +0000 (22:10 +0530)] 
mwifiex: parsing aid while receiving association response

cfg80211 would complain about invalid AID in during TDLS setup.
This was happening because we use same AID value as provide by AP
while forming setup messages.
Parse AID from Association response and filter BIT14 and BIT15
of AID to get valid AID.

Signed-off-by: Xinming Hu <huxm@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agomwifiex: do not return success when command times out
Dmitry Torokhov [Thu, 30 Apr 2015 23:52:02 +0000 (16:52 -0700)] 
mwifiex: do not return success when command times out

wait_event_interruptible_timeout() returns 0 upon timeout. We should
convert it to a negative error code (such as -ETIMEDOUT) instead of
returning it directly, as return code of 0 indicates that command was
executed.

Signed-off-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Enable short repeat by default on ath9k_htc
Nick Kossifidis [Wed, 29 Apr 2015 23:51:22 +0000 (23:51 +0000)] 
ath9k: Enable short repeat by default on ath9k_htc

Normal FFT reports generate a LOT of data, to save resources
enable short repeat by default also on ath9k_htc (it's already
enabled by default on ath9k). This can always be changed from
userspace if needed.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Mix the received FFT bins to the random pool
Nick Kossifidis [Wed, 29 Apr 2015 23:51:21 +0000 (23:51 +0000)] 
ath9k: Mix the received FFT bins to the random pool

E/M noise is a pretty good source of entropy so we mix the FFT
measurements of the E/M spectrum to /dev/random pool. Note that
this doesn't increase the pool's entropy count but it still helps
on improving the output of /dev/(u)random.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: No need for that extra memset
Nick Kossifidis [Wed, 29 Apr 2015 23:51:20 +0000 (23:51 +0000)] 
ath9k: No need for that extra memset

Temp buffer is only used for fixing malformed frames, there is no
need to memset it every time.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Skip FFT reports if we are out of output buffers
Nick Kossifidis [Wed, 29 Apr 2015 23:51:19 +0000 (23:51 +0000)] 
ath9k: Skip FFT reports if we are out of output buffers

There is no reason to keep processing FFT reports if there is no
space left on the relayfs buffers for the results. This saves
lots of CPU cycles, especially on normal (non-short) reports.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: No need for that extra memcpy
Nick Kossifidis [Wed, 29 Apr 2015 23:51:18 +0000 (23:51 +0000)] 
ath9k: No need for that extra memcpy

No need to copy the frame to the temporary buffer when its length
is ok.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Skip malformed frames on normal FFT report
Nick Kossifidis [Wed, 29 Apr 2015 23:51:17 +0000 (23:51 +0000)] 
ath9k: Skip malformed frames on normal FFT report

Since we have lots of frames on a normal FFT report don't bother
processing the malformed ones. Only try to fix malformed frames
in case of a short FFT report (only a single frame on the report).

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Support processing of multiple FFT frames per report.
Nick Kossifidis [Wed, 29 Apr 2015 23:51:16 +0000 (23:51 +0000)] 
ath9k: Support processing of multiple FFT frames per report.

Since we can identify the FFT frames on the report by checking the
consistency of their magnitude info, we can process all of them,
even when some of them are corrupted.

This patch introduces two functions to quickly verify the integrity
of the mag_info fields and some further tweaks to detect the frames
in the report and process them.

Note that in case of missing bytes we don't duplicate them, instead
we leave them as zeroes. This way we get less noise on the FFT plot.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Perform integrity checks when processing FFT frames
Nick Kossifidis [Wed, 29 Apr 2015 23:51:15 +0000 (23:51 +0000)] 
ath9k: Perform integrity checks when processing FFT frames

a) Check that the maximum magnitude is at the specified index

b) Check if the maximum magnitude index is at dc_pos and if so
calculate a new one (value at dc_pos is invalid)

c) Check if the specified maximum magnitude is indeed the maximum

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Move processing of FFT frames to different functions
Nick Kossifidis [Wed, 29 Apr 2015 23:51:14 +0000 (23:51 +0000)] 
ath9k: Move processing of FFT frames to different functions

Since more checks and fixes will be added later and ath_cmn_process_fft is
already big enough.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Fix hanlding of maximum magnitude index
Nick Kossifidis [Wed, 29 Apr 2015 23:51:13 +0000 (23:51 +0000)] 
ath9k: Fix hanlding of maximum magnitude index

Maximum magnitude index is a 5bit signed integer,
convert to an 8bit signed integer and then "shift" it so that it can be used
as an array index. Note that the current implementation adds +1 to the index
value (so it can't be used as an array index) and it's only valid for HT20
channels.

Note that the maximum magnitude index is not being used by
the userspace tools that parse FFT samples (they just use
maximum magnitude) so this doesn't break userspace compatibility.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoath9k: Add a new debug flag for FFT spectral scan
Nick Kossifidis [Wed, 29 Apr 2015 23:51:12 +0000 (23:51 +0000)] 
ath9k: Add a new debug flag for FFT spectral scan

FFT spectral scan is a functionality that makes sense to have its own
debug flag.

Signed-off-by: Nick Kossifidis <mickflemm@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agowl18xx: wlan_irq: support platform dependent interrupt types
Eyal Reizer [Wed, 29 Apr 2015 14:50:03 +0000 (17:50 +0300)] 
wl18xx: wlan_irq: support platform dependent interrupt types

* Interrupt request need to happen when the wilink chip is powered on and
  driving the wlan_irq line. This avoids spurious interrupt issues that
  are a result of different external pulls configuration on different
  platforms
* Allow working with wl18xx level-low and falling edge irqs by configuring
  wl18xx to invert the device interrupt

Signed-off-by: Eyal Reizer <eyalr@ti.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agowireless: fix typos in mwifiex and rtlwifi
Masanari Iida [Mon, 27 Apr 2015 14:14:58 +0000 (23:14 +0900)] 
wireless: fix typos in mwifiex and rtlwifi

This patch fix some spelling typo in drivers/net/wirless

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agomwifiex: reduce spinlock usage in main process
Cathy Luo [Fri, 17 Apr 2015 11:18:29 +0000 (04:18 -0700)] 
mwifiex: reduce spinlock usage in main process

We will reset more_task_flag at the end of main process
routine. This optimization will help to remove spinlock
operations at couple of places in the code.

Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Amitkumar Karwar <akarwar@marvell.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobcma: enable 32 GPIO pins for BCM4707
Felix Fietkau [Wed, 15 Apr 2015 13:07:53 +0000 (15:07 +0200)] 
bcma: enable 32 GPIO pins for BCM4707

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobcma: use absolute base for SoC GPIO pins
Felix Fietkau [Wed, 15 Apr 2015 13:07:52 +0000 (15:07 +0200)] 
bcma: use absolute base for SoC GPIO pins

On some BCM5301x ARM devices, user space still needs to control some
system GPIO pins for which no driver exists. This is a lot easier to do
with a predictable GPIO base.

Signed-off-by: Felix Fietkau <nbd@openwrt.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: Add support for multiple PCIE devices in nvram.
Hante Meuleman [Tue, 14 Apr 2015 18:10:33 +0000 (20:10 +0200)] 
brcmfmac: Add support for multiple PCIE devices in nvram.

With PCIE it is possible to support multiple devices with the
same device type. They all load the same nvram file. In order to
support this the nvram can specify which part of the nvram is
for which pcie device. This patch adds support for these new
types of nvram files.

Reviewed-by: Arend Van Spriel <arend@broadcom.com>
Reviewed-by: Franky (Zhenhui) Lin <frankyl@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Daniel (Deognyoun) Kim <dekim@broadcom.com>
Signed-off-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: add additional 43602 pcie device id.
Hante Meuleman [Tue, 14 Apr 2015 18:10:32 +0000 (20:10 +0200)] 
brcmfmac: add additional 43602 pcie device id.

Reviewed-by: Arend Van Spriel <arend@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Hante Meuleman <meuleman@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: add support for BCM4358 PCIe device
Arend van Spriel [Tue, 14 Apr 2015 18:10:31 +0000 (20:10 +0200)] 
brcmfmac: add support for BCM4358 PCIe device

This patch adds support for the BCM4358 2x2 11ac device.

Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: fix sdio suspend and resume
Arend van Spriel [Tue, 14 Apr 2015 18:10:30 +0000 (20:10 +0200)] 
brcmfmac: fix sdio suspend and resume

commit 330b4e4be937 ("brcmfmac: Add wowl support for SDIO devices.")
changed the behaviour by removing the MMC_PM_KEEP_POWER flag for
non-wowl scenario, which needs to be restored. Another necessary
change is to mark the card as being non-removable. With this in place
the suspend resume test passes successfully doing:

 # echo devices > /sys/power/pm_test
 # echo mem > /sys/power/state

Note that power may still be switched off when system is going
in S3 state.

Reported-by: Fu, Zhonghui <<zhonghui.fu@linux.intel.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Franky (Zhenhui) Lin <frankyl@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Acked-by: Ulf Hansson <ulf.hansson@linaro.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: process interrupt regardless sdiod state
Arend van Spriel [Tue, 14 Apr 2015 18:10:29 +0000 (20:10 +0200)] 
brcmfmac: process interrupt regardless sdiod state

When the sdio bus state is not ready to process we abort the
interrupt service routine. This is not wanted as it keeps the
interrupt source active. Better clear the interrupt source.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Reviewed-by: Franky (Zhenhui) Lin <frankyl@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: add support for BCM4324 rev B5 chipset
Arend van Spriel [Tue, 14 Apr 2015 18:10:28 +0000 (20:10 +0200)] 
brcmfmac: add support for BCM4324 rev B5 chipset

This patch adds support for the BCM4324 B5 revision. This device
is similar to BCM43241 from driver and firmware perspective. It
is known to be used in Lenovo Thinkpad Tablet devices.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: make scheduled scan support conditional
Arend van Spriel [Tue, 14 Apr 2015 18:10:27 +0000 (20:10 +0200)] 
brcmfmac: make scheduled scan support conditional

The scheduled scan support depends on firmware supporting the PNO
feature. This feature is optional so add a feature flag for this
in the driver and announce scheduled scan support accordingly.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: add description for feature flags
Arend van Spriel [Tue, 14 Apr 2015 18:10:26 +0000 (20:10 +0200)] 
brcmfmac: add description for feature flags

Some feature flags were not described in the header file. Adding
the description.

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: update wiphy band information upon updating regulatory domain
Arend van Spriel [Tue, 14 Apr 2015 18:10:25 +0000 (20:10 +0200)] 
brcmfmac: update wiphy band information upon updating regulatory domain

When change the country code the available channels may change. So
the wiphy bands should be updated accordingly.

Reviewed-by: Daniel (Deognyoun) Kim <dekim@broadcom.com>
Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agobrcmfmac: use static superset of channels for wiphy bands
Arend van Spriel [Tue, 14 Apr 2015 18:10:24 +0000 (20:10 +0200)] 
brcmfmac: use static superset of channels for wiphy bands

The driver was constructing a list of channels per wiphy band
by querying the device. This list is not what the hardware is
able to do as it is already filtered by the country setting in
the device. As user-space may change the country this would
require updating the channel list which is not recommended [1].
This patch introduces a superset of channels. The individual
channels are disabled appropriately by querying the device.

[1] http://mid.gmane.org/1426706320.3001.21.camel@sipsolutions.net

Reviewed-by: Hante Meuleman <meuleman@broadcom.com>
Reviewed-by: Daniel (Deognyoun) Kim <dekim@broadcom.com>
Reviewed-by: Pieter-Paul Giesberts <pieterpg@broadcom.com>
Signed-off-by: Arend van Spriel <arend@broadcom.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agomwifiex: fix invalid HT IE configuration in FW
Zhaoyang Liu [Mon, 13 Apr 2015 16:02:25 +0000 (21:32 +0530)] 
mwifiex: fix invalid HT IE configuration in FW

This patch fixes an issue where it was discovered that driver is setting
invalid HT IEs to FW. This was happening because bug in parsing HT IE.
Driver would incorrectly point to start of HT IE while FW only needs actual
HT configuration - excluding EID and length.

Signed-off-by: Zhaoyang Liu <liuzy@marvell.com>
Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agorevert "mwifiex: enable aggregation for TID 6 and 7 streams"
Avinash Patil [Mon, 13 Apr 2015 16:02:24 +0000 (21:32 +0530)] 
revert "mwifiex: enable aggregation for TID 6 and 7 streams"

This patch reverts 'commit c5534844cdee ("mwifiex:
enable aggregation for TID 6 and 7 streams")'.
TID6 & TID7 is voice traffic which is quite time sensitive.
Do not enable any type of aggregation on these TIDs.

Signed-off-by: Avinash Patil <patila@marvell.com>
Signed-off-by: Cathy Luo <cluo@marvell.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
9 years agoMerge tag 'iwlwifi-next-for-kalle-2015-05-03' of https://git.kernel.org/pub/scm/linux...
Kalle Valo [Sat, 9 May 2015 13:21:49 +0000 (16:21 +0300)] 
Merge tag 'iwlwifi-next-for-kalle-2015-05-03' of https://git./linux/kernel/git/iwlwifi/iwlwifi-next

* major rework of the scan code (Luca)
* some work on the thermal code (Chaya Rachel)
* some work on the firwmare debugging infrastructure

9 years agoiwlwifi: mvm: include wildcard SSID in scans
David Spinadel [Tue, 28 Apr 2015 15:06:45 +0000 (18:06 +0300)] 
iwlwifi: mvm: include wildcard SSID in scans

Fix a copy paste bug that didn't copy wildcard SSIDs to
scan requests. This bug causes scan with only wildcard
SSID to be passive, and scans with more than one SSID to
send only the direct probes.

Fixes: 2a28ac14c518 ("iwlwifi: mvm: add common scan params to thw iwl_mvm_scan_params struct")
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 agoiwlwifi: mvm: remove some unused stuff from scan.c
Luciano Coelho [Tue, 21 Apr 2015 15:32:25 +0000 (18:32 +0300)] 
iwlwifi: mvm: remove some unused stuff from scan.c

Some more tidbits left over from the legacy scan removal.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: make thermal throttling values configurable per NIC family
Chaya Rachel Ivgi [Sun, 19 Apr 2015 09:26:39 +0000 (12:26 +0300)] 
iwlwifi: mvm: make thermal throttling values configurable per NIC family

The thermal throttling parameters were constant and hardcoded, not allowing
changes for different NIC families.
Change this so that the values are part of the NIC family configuration and
are not constant (so they can be changed dynamically in the future).

Signed-off-by: Chaya Rachel Ivgi <chaya.rachel.ivgi@intel.com>
Reviewed-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: don't reset key index on HW restart
Johannes Berg [Mon, 20 Apr 2015 15:31:10 +0000 (17:31 +0200)] 
iwlwifi: mvm: don't reset key index on HW restart

When a firmware restart is done, don't try to reprogram the keys to new
slots but rather just keep the old key index, while skipping keys that
weren't programmed before.

Not only does this restore the state more faithfully, but it will also
allow using the HW key index for internal purposes as an array index.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rename generic_scan_cmd functions to dwell
Luciano Coelho [Mon, 20 Apr 2015 12:29:03 +0000 (15:29 +0300)] 
iwlwifi: mvm: rename generic_scan_cmd functions to dwell

The generic scan command functions are now irrelevant, since both
sched and regular scans are in the same code.  So rename this
functions to dwell and isolate all the dwell-related setting to them.
Keeping the dwell code separate makes it easier to compare the LMAC
and UMAC versions.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: move all LMAC scan flags into a single funtion
Luciano Coelho [Mon, 20 Apr 2015 11:47:49 +0000 (14:47 +0300)] 
iwlwifi: mvm: move all LMAC scan flags into a single funtion

Having an LMAC counterpart for the existing UMAC flags function makes
things more consistent and easy to compare and spot the differences.
The flags are the same, but are in different bits, so unfortunately we
can't use a single function for both APIs.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: move all UMAC scan flags setting into the relevant function
Luciano Coelho [Mon, 20 Apr 2015 11:37:16 +0000 (14:37 +0300)] 
iwlwifi: mvm: move all UMAC scan flags setting into the relevant function

Clean the code a little by moving all the flags into the right
function.  Additionally, rename the function, since "common" is now
irrelevant.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: remove deprecated command IDs
David Spinadel [Tue, 14 Apr 2015 06:17:43 +0000 (09:17 +0300)] 
iwlwifi: mvm: remove deprecated command IDs

Signed-off-by: David Spinadel <david.spinadel@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: allow to limit the size of the external buffer for firmware debugging
Emmanuel Grumbach [Tue, 14 Apr 2015 20:14:48 +0000 (23:14 +0300)] 
iwlwifi: allow to limit the size of the external buffer for firmware debugging

When we use an external buffer, it is allocated from the
t DRAM and can be as big as 64MB. This buffer is huge and
might not be needed for the specific issue being chased.
Especially if lots of dumps are going to be created.
Allow to limit the size of the buffer in the configuration.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: avoid use-after-free on iwl_mvm_d0i3_enable_tx() [BUGFIX]
Eliad Peller [Tue, 14 Apr 2015 08:36:23 +0000 (11:36 +0300)] 
iwlwifi: mvm: avoid use-after-free on iwl_mvm_d0i3_enable_tx() [BUGFIX]

qos_seq points (to a struct) inside the command response data.

Make sure to free the response only after qos_seq is not
needed anymore.

type=bugfix
bug=not-tracked
fixes=I78c07110b59dcd389207bd2b168b0a2760839fe0

Change-Id: I619ce2c17e064dc98c7be9abc1e23175fdc8fb9a
Reported-by: Heng Luo <heng.luo@intel.com>
Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Reviewed-on: https://gerrit.rds.intel.com/r/55381
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Tested-by: IWL Jenkins
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine UMAC scans into one
Luciano Coelho [Fri, 17 Apr 2015 11:49:58 +0000 (14:49 +0300)] 
iwlwifi: mvm: combine UMAC scans into one

With just a few differences left in the UMAC scan functions now, we
can merge them into one, taking care of the small difference according
to the total number of iterations required.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: trim sched scan delay down to 16-bit for LMAC as well
Luciano Coelho [Fri, 17 Apr 2015 08:18:46 +0000 (11:18 +0300)] 
iwlwifi: mvm: trim sched scan delay down to 16-bit for LMAC as well

In theory, LMAC scans can handle a 32-bit delay, but since waiting for
over 18 hours to start the scan is a bit silly and to keep it aligned
with UMAC scans (which only support 16-bit delays), trim it down to
16-bits.

This makes the LMAC vs. UMAC and the UMAC reg scan vs. UMAC sched scan
code more similar.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine LMAC scans into one
Luciano Coelho [Fri, 17 Apr 2015 06:53:55 +0000 (09:53 +0300)] 
iwlwifi: mvm: combine LMAC scans into one

The last remaining difference between the regular scan and scheduled
scan flows for LMAC is the FW capabilities check for EBS scans.

Merge these checks into a new function and then combine the LMAC scan
functions into a single one.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add number of scan iterations and multiplier to params
Luciano Coelho [Thu, 16 Apr 2015 17:13:28 +0000 (20:13 +0300)] 
iwlwifi: mvm: add number of scan iterations and multiplier to params

As another step towards combining the scan and sched scan functions,
add parameters that tell the scan function how many iterations we want
(i.e. 1 for normal scan, more for scheduled scan) and that set the
full scan multiplier (only meaningful for LMAC).

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine LMAC and UMAC preq generation
Luciano Coelho [Wed, 15 Apr 2015 13:34:13 +0000 (16:34 +0300)] 
iwlwifi: mvm: combine LMAC and UMAC preq generation

The probe request to be added to both LMAC and UMAC scan commands are
identical, so move the generation out of the LMAC/UMAC-specific code.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: rename scan_calc_params to scan_calc_dwell
Luciano Coelho [Wed, 15 Apr 2015 13:21:16 +0000 (16:21 +0300)] 
iwlwifi: mvm: rename scan_calc_params to scan_calc_dwell

To make things clearer, rename the iwl_mvm_scan_calc_params() function
to iwl_mvm_calc_dwell() and make it calculate and fill in only
dwell-related parameters.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine SSID functions for sched and regular scans
Luciano Coelho [Thu, 2 Apr 2015 14:49:04 +0000 (17:49 +0300)] 
iwlwifi: mvm: combine SSID functions for sched and regular scans

Now that both scheduled scan and regular scan SSID populating
functions do the same thing, they can be combined into a single
function.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: revert order of SSIDs for sched scans
Luciano Coelho [Thu, 2 Apr 2015 13:31:42 +0000 (16:31 +0300)] 
iwlwifi: mvm: revert order of SSIDs for sched scans

The firmware inverts the order of the SSIDs sent out in probe requests
(for some reason).  For regular scans, we've been passing the SSIDs in
the opposite order so they go out in the order we want.  With
scheduled scans, we were not doing that, so they were sent out in
reverse order of priority.  Fix that by using the reverse order when
populating the SSIDs array for scheduled scans as well.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine ssid_bitmap setting for regular scans
Luciano Coelho [Thu, 2 Apr 2015 12:37:05 +0000 (15:37 +0300)] 
iwlwifi: mvm: combine ssid_bitmap setting for regular scans

The only difference in the ssid_bitmap between LMAC and UMAC scans is
that in LMAC bits 1 to 20 are used, while in UMAC bits 0 to 19 are
used (*sigh*).  So we can combine the bitmap creation into a single
function and simply shift left if LMAC is used.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add common scan params to thw iwl_mvm_scan_params struct
Luciano Coelho [Wed, 1 Apr 2015 21:08:35 +0000 (00:08 +0300)] 
iwlwifi: mvm: add common scan params to thw iwl_mvm_scan_params struct

Many parameters are common for all scans.  Instead of parsing the
cfg80211 scan and sched scan requests differently in each flow, move
the parsing outside of the API/scan-type specific functions.  In this
way, we only need to differentiate between scan types once.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine parts of UMAC and LMAC sched scans
Luciano Coelho [Wed, 1 Apr 2015 14:37:44 +0000 (17:37 +0300)] 
iwlwifi: mvm: combine parts of UMAC and LMAC sched scans

Similarly to the regular scan patch, a lot of the UMAC and LMAC sched
scan code is also almost identical.  Grab the low hanging fruits by
combining the obvious parts.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: combine parts of UMAC and LMAC scans
Luciano Coelho [Wed, 1 Apr 2015 14:09:56 +0000 (17:09 +0300)] 
iwlwifi: mvm: combine parts of UMAC and LMAC scans

A lot of the UMAC and LMAC scan code is almost identical.  Grab the
low hanging fruits by combining the obvious parts.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: clarify the device / firmware mapping in Kconfig
Emmanuel Grumbach [Sun, 19 Apr 2015 06:57:09 +0000 (09:57 +0300)] 
iwlwifi: clarify the device / firmware mapping in Kconfig

The lists of the devices supported by either MVM or DVM
firmwares was incomplete. Point to
https://wireless.wiki.kernel.org/en/users/drivers/iwlwifi#firmware
instead of maintaining the lists.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: iterate all interfaces during HW recovery cleanup
Arik Nemtsov [Mon, 23 Feb 2015 12:42:41 +0000 (14:42 +0200)] 
iwlwifi: mvm: iterate all interfaces during HW recovery cleanup

Usually during HW recovery the state of all active interfaces is cleaned
up during drv_start(). There's a special case where a HW restart is
requested when an interface is going down. In this case the iface-iterator
won't see this interface and we won't clean it. This has bad consequences
once the interface is legitimately brought up again.

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: combine scan size checks into a common function
Luciano Coelho [Fri, 27 Mar 2015 07:28:26 +0000 (10:28 +0300)] 
iwlwifi: mvm: combine scan size checks into a common function

Instead of repeating the same code in 4 different places, combine the
comparisons into a new function.  Additionally, this change fixes UMAC
scans where the RRM IEs were not taken into consideration when
calculating the IE length.

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: always use iwl_mvm_scan_size to calculate the scan size
Luciano Coelho [Mon, 30 Mar 2015 12:09:24 +0000 (15:09 +0300)] 
iwlwifi: mvm: always use iwl_mvm_scan_size to calculate the scan size

We have a function (iwl_mvm_scan_size()) that can calculate the scan
size for both UMAC and LMAC scans.  Use that function instead of
calculating manually for LMAC scan and sched scan.

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: allow scheduled scan for all the firmwares
Emmanuel Grumbach [Mon, 16 Mar 2015 12:49:55 +0000 (14:49 +0200)] 
iwlwifi: mvm: allow scheduled scan for all the firmwares

We don't support -9.ucode so, all the released firmwares
support scheduled scan properly.

Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: some clean ups in fw-api-scan.h
Luciano Coelho [Wed, 1 Apr 2015 21:25:44 +0000 (00:25 +0300)] 
iwlwifi: mvm: some clean ups in fw-api-scan.h

Remove unused struct iwl_scan_offload_req and enum
iwl_scan_offload_flags which are not used anymore.  Rename
iwl_scan_offload_schedule to iwl_scan_schedule_lmac to make it clear
that this is for LMAC only.  And fix a small typo.

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: rs: remove unneeded check of average tpt in window
Liad Kaufman [Thu, 2 Apr 2015 13:54:27 +0000 (16:54 +0300)] 
iwlwifi: rs: remove unneeded check of average tpt in window

Previously there was a check that compared window->average_tpt
to some value, and if it was different - it set it to that
value. However, this value was already calculated and set in
_rs_collect_tx_data(), so the entire check is unneeded.

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: pcie: support marbh fw dbg mode
Liad Kaufman [Mon, 5 Jan 2015 12:06:14 +0000 (14:06 +0200)] 
iwlwifi: pcie: support marbh fw dbg mode

This adds support for configuring and retrieving the FW
monitor in MARBH mode.

Signed-off-by: Liad Kaufman <liad.kaufman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: differentiate net-detect from sched scan
Luciano Coelho [Fri, 20 Mar 2015 14:11:28 +0000 (16:11 +0200)] 
iwlwifi: mvm: differentiate net-detect from sched scan

Net-detect scans were using the same type as sched scan, which was
causing the driver to return -EBUSY and prevent the system from
suspending if there was an ongoing scheduled scan.  To avoid this, add
a new type for net-detect and don't stop anything when it is
requested, so that the existing scheduled scan will be resumed when
the system wakes up.

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: move scan code from mac80211.c to scan.c
Luciano Coelho [Fri, 20 Mar 2015 13:51:36 +0000 (15:51 +0200)] 
iwlwifi: mvm: move scan code from mac80211.c to scan.c

Move all the scan code that was in mac80211.c to scan.c where it
belongs, leaving only the parts that are specific to mac80211 ops.

Change some function definitions slightly to improve consistency.

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: rename unified_scan symbols to just scan
Luciano Coelho [Fri, 20 Mar 2015 11:35:47 +0000 (13:35 +0200)] 
iwlwifi: mvm: rename unified_scan symbols to just scan

All scans are using the unified APIs now, so using "unified" in the
symbols is useless and just make them much longer and the main
difference between scans now is LMAC vs. UMAC.  Remove "unified" from
all relevant symbols.

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: generalize the other-scan stopping code
Luciano Coelho [Thu, 19 Mar 2015 20:58:33 +0000 (22:58 +0200)] 
iwlwifi: mvm: generalize the other-scan stopping code

Instead of hardcoding the differences between UMAC scans and LMAC
scans (which in this case is the number of simultaneous scans that can
run), introduce a max_scans variable and stop scans of the other type
(i.e. stop sched scan if regular scan is being attempted and
vice-versa) if the number of running scans reached the maximum.

Add a function that checks if the maximum number of scans was reached
and stops the appropriate scan to make room for the new scan.

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 if scan can be started before cancelling other scans
Luciano Coelho [Fri, 20 Mar 2015 07:29:42 +0000 (09:29 +0200)] 
iwlwifi: mvm: check if scan can be started before cancelling other scans

If a new scan cannot be run for some reason, we shouldn't cancel other
ongoing scans.  Move the checks to before the code that cancels other
scans.

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: don't wait for scan stopped work when cancelling scans
Luciano Coelho [Fri, 20 Mar 2015 06:59:56 +0000 (08:59 +0200)] 
iwlwifi: mvm: don't wait for scan stopped work when cancelling scans

Now that we have separate flags for stopping scans, we don't need to
wait for the scan stopped work to complete before starting the new
scan.  Previously we needed it because we had no way of distinguishing
the scan that was being stopped from the scan that was currently
running.  With the new flags there won't be any confusions and we are
able to handle the stop for the correct type of scan.

Thus we can remove the iwl_mvm_cancel_scan_wait_notif() function.

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: convert scan_status to a bitmap
Luciano Coelho [Tue, 10 Feb 2015 08:42:26 +0000 (10:42 +0200)] 
iwlwifi: mvm: convert scan_status to a bitmap

LMAC scans cannot handle more than one scan at a time, but UMAC scans
can.  To avoid confusion we should combine the states of these two
types of scans.  To do so, we need to support mutliple scans at the
same time for UMAC.

This commit changes the scan_status element from a single value to a
bitmask of running scan types for LMAC.  Later, we will modify UMAC
scans to use the same state bitmask.

Additionally, add stopping scan flags for scheduled and regular scans.
This makes it easier to differentiate and handle stop requests
triggered by the driver and spontaneous stops generated by the
firmware.

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: don't increase max_out_time when low priority scan is requested
Luciano Coelho [Thu, 9 Apr 2015 11:48:11 +0000 (14:48 +0300)] 
iwlwifi: mvm: don't increase max_out_time when low priority scan is requested

In some cases, max_out_time value is smaller than 200 and having the
NL80211_SCAN_FLAG_LOW_PRIORITY flag was actually causing the
max_out_time to be increased.  To avoid that, set max_out_time to 200
only if it's greater than 200.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Alexander Bondar <alexander.bondar@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: add scan parameters debugging info
Luciano Coelho [Thu, 9 Apr 2015 09:18:56 +0000 (12:18 +0300)] 
iwlwifi: mvm: add scan parameters debugging info

Add scan parameters information to make it easier to debug scan dwell
times and fragmentation.

Signed-off-by: Luciano Coelho <luciano.coelho@intel.com>
Reviewed-by: Alexander Bondar <alexander.bondar@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: rs: cleanup last_txrate_idx
Eyal Shapira [Sun, 12 Apr 2015 20:45:27 +0000 (23:45 +0300)] 
iwlwifi: rs: cleanup last_txrate_idx

last_txrate_idx isn't used anymore and can be dropped as this info
exists already somewhere else.

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: rs: remove code duplication when filling lq cmd
Liad Kaufman [Sun, 5 Apr 2015 10:52:50 +0000 (13:52 +0300)] 
iwlwifi: rs: remove code duplication when filling lq cmd

Same code appear a few lines later while the position has
no effect on the actual flow.

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: ROC: Reduce the aux roc max delay
Matti Gottlieb [Sun, 29 Mar 2015 10:38:16 +0000 (13:38 +0300)] 
iwlwifi: mvm: ROC: Reduce the aux roc max delay

When user space requests mac80211 to transmit a frame off
channel, mac80211 notifies the driver, and the driver requests
a time event from the ucode, and then transmits the frame.
When the driver requests a time event, it can specify what is the allowed
max delay for starting the time event.

When the max delay is too big, this can cause a timeout in the user space,
that is waiting for the frame to be transmitted.

Currently the max delay is extremely long.

Reduce the max delay for the AUX ROC time event that is sent to the ucode.

Signed-off-by: Matti Gottlieb <matti.gottlieb@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoMerge remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next
Emmanuel Grumbach [Wed, 29 Apr 2015 10:12:18 +0000 (13:12 +0300)] 
Merge remote-tracking branch 'iwlwifi-fixes/master' into iwlwifi-next

9 years agoiwlwifi: mvm: force quota update update after FW restart
Johannes Berg [Tue, 21 Apr 2015 07:21:46 +0000 (09:21 +0200)] 
iwlwifi: mvm: force quota update update after FW restart

During firmware restart, the quota command isn't calculated multiple
times, but after the firmware restart it has to be sent, so force it.
Otherwise the firmware crashes.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: fix typo in CONFIG option
Emmanuel Grumbach [Tue, 28 Apr 2015 12:00:23 +0000 (15:00 +0300)] 
iwlwifi: mvm: fix typo in CONFIG option

I forgot to rename the CPTCFG_ prefix...

Fixes: 484b3d13b4ac ("iwlwifi: mvm: add debugfs entry with the number of net-detect scans")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoiwlwifi: mvm: don't power off the device between INIT and OPER firmwares
Eran Harary [Mon, 27 Apr 2015 07:29:31 +0000 (10:29 +0300)] 
iwlwifi: mvm: don't power off the device between INIT and OPER firmwares

Our device needs two different firmwares: the INIT firmware
and the operational (OPER) firmware. The first one is run
when the driver loads and it returns calibrations results
as well as the NVM. The second one implements the WiFi
protocol.

If the wlan interface is not brought up, the device is put
to low power state: no firmware will be running. When the
interface is brought up, we would run the OPER firmware
only and reuse the results of the run of the INIT firmware
when the driver was loaded. This is changing with this
patch.
We now run the INIT firmware every time mac80211 calls
start(). The penalty for that is minimal since the INIT
firwmare run fast. I now also avoid to power down the device
between the INIT and OPER firmware on certains buses.

The motivation for this change is that there are components
on the device (MFUART) that are triggered by the INIT
firmware and need the device to be powered up in order to
keep running. Powering the device down between the INIT and
OPER firmware would stop these components and prevent them
from running again since they are triggered by the INIT
firmware only.
The new flow allows this and also allows to trigger these
components again when the interface is brought up after
it has been brought down.

Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
9 years agoMerge ath-next from ath.git
Kalle Valo [Tue, 28 Apr 2015 11:43:54 +0000 (14:43 +0300)] 
Merge ath-next from ath.git

Major changes in ath10k:

* enable channel 144 on 5 GHz
* enable Adaptive Noise Immunity (ANI) by default
* add Wake on Wireless LAN (WOW) patterns support
* add basic Tunneled Direct Link Setup (TDLS) support
* add multi-channel support for QCA6174
* enable IBSS RSN support
* enable Bluetooth Coexistance whenever firmware supports it
* add more versatile way to set bitrates used by the firmware

9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net
Linus Torvalds [Mon, 27 Apr 2015 21:05:19 +0000 (14:05 -0700)] 
Merge git://git./linux/kernel/git/davem/net

Pull networking fixes from David Miller:

 1) mlx4 doesn't check fully for supported valid RSS hash function, fix
    from Amir Vadai

 2) Off by one in ibmveth_change_mtu(), from David Gibson

 3) Prevent altera chip from reporting false error interrupts in some
    circumstances, from Chee Nouk Phoon

 4) Get rid of that stupid endless loop trying to allocate a FIN packet
    in TCP, and in the process kill deadlocks.  From Eric Dumazet

 5) Fix get_rps_cpus() crash due to wrong invalid-cpu value, also from
    Eric Dumazet

 6) Fix two bugs in async rhashtable resizing, from Thomas Graf

 7) Fix topology server listener socket namespace bug in TIPC, from Ying
    Xue

 8) Add some missing HAS_DMA kconfig dependencies, from Geert
    Uytterhoeven

 9) bgmac driver intends to force re-polling but does so by returning
    the wrong value from it's ->poll() handler.  Fix from Rafał Miłecki

10) When the creater of an rhashtable configures a max size for it,
    don't bark in the logs and drop insertions when that is exceeded.
    Fix from Johannes Berg

11) Recover from out of order packets in ppp mppe properly, from Sylvain
    Rochet

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (41 commits)
  bnx2x: really disable TPA if 'disable_tpa' option is set
  net:treewide: Fix typo in drivers/net
  net/mlx4_en: Prevent setting invalid RSS hash function
  mdio-mux-gpio: use new gpiod_get_array and gpiod_put_array functions
  netfilter; Add some missing default cases to switch statements in nft_reject.
  ppp: mppe: discard late packet in stateless mode
  ppp: mppe: sanity error path rework
  net/bonding: Make DRV macros private
  net: rfs: fix crash in get_rps_cpus()
  altera tse: add support for fixed-links.
  pxa168: fix double deallocation of managed resources
  net: fix crash in build_skb()
  net: eth: altera: Resolve false errors from MSGDMA to TSE
  ehea: Fix memory hook reference counting crashes
  net/tg3: Release IRQs on permanent error
  net: mdio-gpio: support access that may sleep
  inet: fix possible panic in reqsk_queue_unlink()
  rhashtable: don't attempt to grow when at max_size
  bgmac: fix requests for extra polling calls from NAPI
  tcp: avoid looping in tcp_send_fin()
  ...

9 years agobnx2x: really disable TPA if 'disable_tpa' option is set
Michal Schmidt [Mon, 27 Apr 2015 15:20:38 +0000 (17:20 +0200)] 
bnx2x: really disable TPA if 'disable_tpa' option is set

bnx2x's 'disable_tpa=1' module option is not respected properly and TPA
(transparent packet aggregation) remains enabled. Even though the
module option causes LRO to be disabled, TPA is enabled in GRO mode.

Additionally, disabling GRO via ethtool then has no effect. One can
still observe tpa_* statistics increase and large packets being received
in tcpdump.

The bug was an unintended consequence of commit aebf6244cd39 "bnx2x: Be
more forgiving toward SW GRO".

Fix it by following the bp->disable_tpa flag when initializing fp's.

Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet:treewide: Fix typo in drivers/net
Masanari Iida [Mon, 27 Apr 2015 14:14:57 +0000 (23:14 +0900)] 
net:treewide: Fix typo in drivers/net

This patch fix spelling typo in printk.

Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/mlx4_en: Prevent setting invalid RSS hash function
Amir Vadai [Mon, 27 Apr 2015 10:40:56 +0000 (13:40 +0300)] 
net/mlx4_en: Prevent setting invalid RSS hash function

mlx4_en_check_rxfh_func() was checking for hardware support before
setting a known RSS hash function, but didn't do any check before
setting unknown RSS hash function. Need to make it fail on such values.
In this occasion, moved the actual setting of the new value from the
check function into mlx4_en_set_rxfh().

Fixes: 947cbb0 ("net/mlx4_en: Support for configurable RSS hash function")
Signed-off-by: Amir Vadai <amirv@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agomdio-mux-gpio: use new gpiod_get_array and gpiod_put_array functions
Rojhalat Ibrahim [Mon, 27 Apr 2015 08:37:31 +0000 (10:37 +0200)] 
mdio-mux-gpio: use new gpiod_get_array and gpiod_put_array functions

Use the new gpiod_get_array and gpiod_put_array functions
(added to mainline in the v4.1 merge window) for obtaining and
disposing of GPIO descriptors.

Cc: David Miller <davem@davemloft.net>
Cc: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Rojhalat Ibrahim <imr@rtschenk.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonetfilter; Add some missing default cases to switch statements in nft_reject.
David S. Miller [Mon, 27 Apr 2015 17:20:34 +0000 (13:20 -0400)] 
netfilter; Add some missing default cases to switch statements in nft_reject.

This fixes:

====================
net/netfilter/nft_reject.c: In function ‘nft_reject_dump’:
net/netfilter/nft_reject.c:61:2: warning: enumeration value ‘NFT_REJECT_TCP_RST’ not handled in switch [-Wswitch]
  switch (priv->type) {
  ^
net/netfilter/nft_reject.c:61:2: warning: enumeration value ‘NFT_REJECT_ICMPX_UNREACH’ not handled in switch [-Wswi\
tch]
net/netfilter/nft_reject_inet.c: In function ‘nft_reject_inet_dump’:
net/netfilter/nft_reject_inet.c:105:2: warning: enumeration value ‘NFT_REJECT_TCP_RST’ not handled in switch [-Wswi\
tch]
  switch (priv->type) {
  ^
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoath10k: implement more versatile set_bitrate_mask
Michal Kazior [Tue, 21 Apr 2015 17:42:58 +0000 (20:42 +0300)] 
ath10k: implement more versatile set_bitrate_mask

Until now only a single fixed tx rate or nss was
allowed to be set.

The patch attempts to improve this by allowing
most bitrate masks. The limitation is VHT MCS
rates cannot be expressed separately using
existing firmware interfaces and only the
following VHT MCS ranges are supported: none, 0-7,
0-8, and 0-9.

This keeps the old behaviour when requesting
single tx rate or single nss. The new bitrate mask
logic is only applied to other cases that would
return -EINVAL until now.

Depending on firmware revisions some combinations
may crash firmware so use with care, please.

This depends on "ath10k: don't use reassoc flag".
Without it key cache would effectively be
invalidated upon bitrate change leading to
communication being no longer possible.

These work:

  iw wlan0 set bitrates legacy-5 6 12 ht-mcs-5 1 2 3
  iw wlan0 set bitrates legacy-5 ht-mcs-5 7 8 9
  iw wlan0 set bitrates legacy-5 24 ht-mcs-5 vht-mcs-5 1:0-9

These won't work:

  iw wlan0 set bitrates legacy-5 ht-mcs-5 vht-mcs-5 1:0-5
  iw wlan0 set bitrates vht-mcs-5 2:7-9

(note the invalid VHT MCS ranges)

Signed-off-by: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: Kalle Valo <kvalo@qca.qualcomm.com>
9 years agoMerge branch 'ppp_mppe_desync'
David S. Miller [Mon, 27 Apr 2015 03:25:13 +0000 (23:25 -0400)] 
Merge branch 'ppp_mppe_desync'

Sylvain Rochet says:

====================
ppp: mppe: fixes MPPE desync on links which don't guarantee packet ordering

I am currently having an issue with PPP over L2TP (UDP) and MPPE in
stateless mode (default mode), UDP does not guarantee packet ordering so
we might get out of order packet. MPPE needs to be continuously synched
so we should drop late UDP packet.

I added a printk on the number of time we rekeyed in MPPE decompressor,
this is what we currently have if we receive a slightly out of order UDP
packet:

[1731001.049206] mppe_decompress[1]: ccount 1559
[1731001.049216] mppe_decompress[1]: rekeyed 1 times

[1731001.049228] mppe_decompress[1]: ccount 1560
[1731001.049232] mppe_decompress[1]: rekeyed 1 times

[1731001.050170] mppe_decompress[1]: ccount 1562
[1731001.050182] mppe_decompress[1]: rekeyed 2 times

[1731001.050191] mppe_decompress[1]: ccount 1561
[1731001.062576] mppe_decompress[1]: rekeyed 4095 times
                                             ^^^^
This is obviously wrong, we missed packet 1561 and we already rekeyed 2
times for 1562 we previously received, we can't recover the decryption
key we need for 1561, we should drop it instead of rekeying 4095 times.

This patch series drop any packet with are not within the 4096/2 forward
window.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoppp: mppe: discard late packet in stateless mode
Sylvain Rochet [Sun, 26 Apr 2015 18:40:53 +0000 (20:40 +0200)] 
ppp: mppe: discard late packet in stateless mode

When PPP is used over a link which does not guarantee packet ordering,
we might get late MPPE packets. This is a problem because MPPE must be
kept synchronized and the current implementation does not drop them and
rekey 4095 times instead of 0, which is wrong.

In order to prevent rekeying about a whole count space times (~ 4095
times), drop packets which are not within the forward 4096/2 window and
increase sanity error counter.

Signed-off-by: Sylvain Rochet <sylvain.rochet@finsecur.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoppp: mppe: sanity error path rework
Sylvain Rochet [Sun, 26 Apr 2015 18:40:52 +0000 (20:40 +0200)] 
ppp: mppe: sanity error path rework

We are going to need sanity error path a little further, rework to be
able to use the sanity error path anywhere in decompressor.

Signed-off-by: Sylvain Rochet <sylvain.rochet@finsecur.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agonet/bonding: Make DRV macros private
Matan Barak [Sun, 26 Apr 2015 12:55:57 +0000 (15:55 +0300)] 
net/bonding: Make DRV macros private

The bonding modules currently defines four macros with
general names that pollute the global namespace:
DRV_VERSION
DRV_RELDATE
DRV_NAME
DRV_DESCRIPTION

Fixing that by defining a private bonding_priv.h
header files which includes those defines.

Signed-off-by: Matan Barak <matanb@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoLinux 4.1-rc1
Linus Torvalds [Mon, 27 Apr 2015 00:59:10 +0000 (17:59 -0700)] 
Linux 4.1-rc1

9 years agox86_64, asm: Work around AMD SYSRET SS descriptor attribute issue
Andy Lutomirski [Sun, 26 Apr 2015 23:47:59 +0000 (16:47 -0700)] 
x86_64, asm: Work around AMD SYSRET SS descriptor attribute issue

AMD CPUs don't reinitialize the SS descriptor on SYSRET, so SYSRET with
SS == 0 results in an invalid usermode state in which SS is apparently
equal to __USER_DS but causes #SS if used.

Work around the issue by setting SS to __KERNEL_DS __switch_to, thus
ensuring that SYSRET never happens with SS set to NULL.

This was exposed by a recent vDSO cleanup.

Fixes: e7d6eefaaa44 x86/vdso32/syscall.S: Do not load __USER32_DS to %ss
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Peter Anvin <hpa@zytor.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Cc: Brian Gerst <brgerst@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Mon, 27 Apr 2015 00:55:26 +0000 (17:55 -0700)] 
Merge branch 'drm-fixes' of git://people.freedesktop.org/~airlied/linux

Pull intel drm fixes from Dave Airlie.

* 'drm-fixes' of git://people.freedesktop.org/~airlied/linux:
  drm/i915: vlv: fix save/restore of GFX_MAX_REQ_COUNT reg
  drm/i915: Workaround to avoid lite restore with HEAD==TAIL
  drm/i915: cope with large i2c transfers

9 years agoMerge git://git.infradead.org/intel-iommu
Linus Torvalds [Mon, 27 Apr 2015 00:47:46 +0000 (17:47 -0700)] 
Merge git://git.infradead.org/intel-iommu

Pull intel iommu updates from David Woodhouse:
 "This lays a little of the groundwork for upcoming Shared Virtual
  Memory support — fixing some bogus #defines for capability bits and
  adding the new ones, and starting to use the new wider page tables
  where we can, in anticipation of actually filling in the new fields
  therein.

  It also allows graphics devices to be assigned to VM guests again.
  This got broken in 3.17 by disallowing assignment of RMRR-afflicted
  devices.  Like USB, we do understand why there's an RMRR for graphics
  devices — and unlike USB, it's actually sane.  So we can make an
  exception for graphics devices, just as we do USB controllers.

  Finally, tone down the warning about the X2APIC_OPT_OUT bit, due to
  persistent requests.  X2APIC_OPT_OUT was added to the spec as a nasty
  hack to allow broken BIOSes to forbid us from using X2APIC when they
  do stupid and invasive things and would break if we did.

  Someone noticed that since Windows doesn't have full IOMMU support for
  DMA protection, setting the X2APIC_OPT_OUT bit made Windows avoid
  initialising the IOMMU on the graphics unit altogether.

  This means that it would be available for use in "driver mode", where
  the IOMMU registers are made available through a BAR of the graphics
  device and the graphics driver can do SVM all for itself.

  So they started setting the X2APIC_OPT_OUT bit on *all* platforms with
  SVM capabilities.  And even the platforms which *might*, if the
  planets had been aligned correctly, possibly have had SVM capability
  but which in practice actually don't"

* git://git.infradead.org/intel-iommu:
  iommu/vt-d: support extended root and context entries
  iommu/vt-d: Add new extended capabilities from v2.3 VT-d specification
  iommu/vt-d: Allow RMRR on graphics devices too
  iommu/vt-d: Print x2apic opt out info instead of printing a warning
  iommu/vt-d: kill bogus ecap_niotlb_iunits()

9 years agoMerge branch 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa...
Linus Torvalds [Mon, 27 Apr 2015 00:44:09 +0000 (17:44 -0700)] 
Merge branch 'i2c/for-next' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "This has a mixture of merge window cleanups and bugfixes"

* 'i2c/for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: st: add include for pinctrl
  i2c: mux: use proper dev when removing "channel-X" symlinks
  i2c: digicolor: remove duplicate include
  i2c: Mark adapter devices with pm_runtime_no_callbacks
  i2c: pca-platform: fix broken email address
  i2c: mxs: fix broken email address
  i2c: rk3x: report number of messages transmitted

9 years agoMerge branch 'for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mason...
Linus Torvalds [Mon, 27 Apr 2015 00:40:30 +0000 (17:40 -0700)] 
Merge branch 'for-linus-4.1' of git://git./linux/kernel/git/mason/linux-btrfs

Pull btrfs fixes from Chris Mason:
 "Filipe hit two problems in my block group cache patches.  We finalized
  the fixes last week and ran through more tests"

* 'for-linus-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Btrfs: prevent list corruption during free space cache processing
  Btrfs: fix inode cache writeout

9 years agoMerge tag 'drm-intel-next-fixes-2015-04-25' of git://anongit.freedesktop.org/drm...
Dave Airlie [Mon, 27 Apr 2015 00:35:15 +0000 (10:35 +1000)] 
Merge tag 'drm-intel-next-fixes-2015-04-25' of git://anongit.freedesktop.org/drm-intel into drm-fixes

three fixes for i915.

* tag 'drm-intel-next-fixes-2015-04-25' of git://anongit.freedesktop.org/drm-intel:
  drm/i915: vlv: fix save/restore of GFX_MAX_REQ_COUNT reg
  drm/i915: Workaround to avoid lite restore with HEAD==TAIL
  drm/i915: cope with large i2c transfers

9 years agoMerge tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs
Linus Torvalds [Mon, 27 Apr 2015 00:33:59 +0000 (17:33 -0700)] 
Merge tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs

Pull NFS client updates from Trond Myklebust:
 "Another set of mainly bugfixes and a couple of cleanups.  No new
  functionality in this round.

  Highlights include:

  Stable patches:
   - Fix a regression in /proc/self/mountstats
   - Fix the pNFS flexfiles O_DIRECT support
   - Fix high load average due to callback thread sleeping

  Bugfixes:
   - Various patches to fix the pNFS layoutcommit support
   - Do not cache pNFS deviceids unless server notifications are enabled
   - Fix a SUNRPC transport reconnection regression
   - make debugfs file creation failure non-fatal in SUNRPC
   - Another fix for circular directory warnings on NFSv4 "junctioned"
     mountpoints
   - Fix locking around NFSv4.2 fallocate() support
   - Truncating NFSv4 file opens should also sync O_DIRECT writes
   - Prevent infinite loop in rpcrdma_ep_create()

  Features:
   - Various improvements to the RDMA transport code's handling of
     memory registration
   - Various code cleanups"

* tag 'nfs-for-4.1-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (55 commits)
  fs/nfs: fix new compiler warning about boolean in switch
  nfs: Remove unneeded casts in nfs
  NFS: Don't attempt to decode missing directory entries
  Revert "nfs: replace nfs_add_stats with nfs_inc_stats when add one"
  NFS: Rename idmap.c to nfs4idmap.c
  NFS: Move nfs_idmap.h into fs/nfs/
  NFS: Remove CONFIG_NFS_V4 checks from nfs_idmap.h
  NFS: Add a stub for GETDEVICELIST
  nfs: remove WARN_ON_ONCE from nfs_direct_good_bytes
  nfs: fix DIO good bytes calculation
  nfs: Fetch MOUNTED_ON_FILEID when updating an inode
  sunrpc: make debugfs file creation failure non-fatal
  nfs: fix high load average due to callback thread sleeping
  NFS: Reduce time spent holding the i_mutex during fallocate()
  NFS: Don't zap caches on fallocate()
  xprtrdma: Make rpcrdma_{un}map_one() into inline functions
  xprtrdma: Handle non-SEND completions via a callout
  xprtrdma: Add "open" memreg op
  xprtrdma: Add "destroy MRs" memreg op
  xprtrdma: Add "reset MRs" memreg op
  ...

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 26 Apr 2015 22:48:49 +0000 (15:48 -0700)] 
Merge branch 'for-linus' of git://git./linux/kernel/git/viro/vfs

Pull fourth vfs update from Al Viro:
 "d_inode() annotations from David Howells (sat in for-next since before
  the beginning of merge window) + four assorted fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  RCU pathwalk breakage when running into a symlink overmounting something
  fix I_DIO_WAKEUP definition
  direct-io: only inc/dec inode->i_dio_count for file systems
  fs/9p: fix readdir()
  VFS: assorted d_backing_inode() annotations
  VFS: fs/inode.c helpers: d_inode() annotations
  VFS: fs/cachefiles: d_backing_inode() annotations
  VFS: fs library helpers: d_inode() annotations
  VFS: assorted weird filesystems: d_inode() annotations
  VFS: normal filesystems (and lustre): d_inode() annotations
  VFS: security/: d_inode() annotations
  VFS: security/: d_backing_inode() annotations
  VFS: net/: d_inode() annotations
  VFS: net/unix: d_backing_inode() annotations
  VFS: kernel/: d_inode() annotations
  VFS: audit: d_backing_inode() annotations
  VFS: Fix up some ->d_inode accesses in the chelsio driver
  VFS: Cachefiles should perform fs modifications on the top layer only
  VFS: AF_UNIX sockets should call mknod on the top layer only

This page took 0.047122 seconds and 5 git commands to generate.