staging: vt6656: Refactor the assignment of the phy->signal variable

Create a constant array with the values of the "phy->signal" for every
rate. Remove all "phy->signal" assignments inside the switch statement
and replace these with a single reading from the new vnt_phy_signal
array.

The constant array can be of one dimension because the OR mask with
BIT(3) or BIT(4) allow obtain a second value according to the rate,
the preamble_type and the pkt_type.

Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200414153849.5785-5-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Use define instead of magic number for tx_rate

Use the define RATE_11M present in the file "device.h" instead of the
magic number 3. So the code is more clear.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200414153849.5785-4-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Use BIT() macro instead of bit shift operator

Use the BIT() macro instead of the bit left shift operator. So the code
is more clear.

It's safe to remove the casting to u16 type because the value obtained
never exceeds 16 bits. So the casting is unnecessary.

Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200414153849.5785-3-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Return error code in vnt_rf_write_embedded function

Use the error code returned by the vnt_control_out function as the
returned value of the vnt_rf_write_embedded function instead of a
boolean value.

Then, fix all vnt_rf_write_embedded calls removing the "and" operations
and replace with a direct assignment to the ret variable and add a check
condition after every call.

Also replace the boolean values true or false in the vnt_rf_set_txpower
function to 0 or error code EINVAL to follow the coding style guide.

The vnt_rf_set_txpower function is called only in the vnt_rf_setpower
function that already returns error codes. The calls to this function
(vnt_rf_set_txpower) not use the returned values, so they not need to be
fixed.

Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200414153849.5785-2-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vc04_services: bcm2835-audio: Make function arguments alignment match open parenthesis

Make function arguments alignment match open parenthesis.
Found using checkpatch.pl.

Signed-off-by: R Veera Kumar <vkor@vkten.in>
Reviewed-by: Nicolas Saenz Julienne <nsaenzjulienne@suse.de>
Link: https://lore.kernel.org/r/20200413165905.6081-1-vkor@vkten.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: sm750fb: Make function arguments alignment match open parenthesis

Make function arguments alignment match open parenthesis.
Found using checkpatch.pl.

Signed-off-by: R Veera Kumar <vkor@vkten.in>
Link: https://lore.kernel.org/r/20200413160753.3787-1-vkor@vkten.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8712: correct spelling mistake in comment

Correct a spelling mistake discovered by checkpatch.
attatch -> attach

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Link: https://lore.kernel.org/r/20200413215109.15749-1-straube.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci: initialize 'n' variable when it is declared

Variable 'n' in 'mt7621_pcie_init_virtual_bridges' function
can be directly initialized when it is declared instead of
doing it before using it first.

Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200413155018.31921-3-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci: add clarification comment in 'mt7621_pcie_init_virtual_bridges'

Add a comment explaining a bit what is going on in this function.
It would be helfulp for other people for a better understanding
of the code.

Suggested-by: NeilBrown <neil@brown.name>
Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200413155018.31921-2-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci: rename 'PCIE_P2P_MAX' into 'PCIE_P2P_CNT'

Definition 'PCIE_P2P_MAX' is '3'. The value here is not a MAXimum.
It is a count or a number. It is how many masks there are.
The masks are numbered 0, 1, 2 so the maximum is 2. Hence rename
variable into 'PCIE_P2P_CNT' which is a more accurate name.

Suggested-by: NeilBrown <neil@brown.name>
Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200413155018.31921-1-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: update TODO

Update the TODO list associated to the wfx driver with the last
progresses and last remarks made by reviewers.

The items about support for P2P and mesh have also been removed. Indeed,
it seems that the device does not fully support them.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-21-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless attribute 'filter_mcast'

Currently, wfx_update_filtering() apply the value of wvif->filter_mcast
to the hardware. But an attentive reader will note that
wfx_update_filtering() is always called after set wvif->filter_mcast.
Thus, it not necessary to store filter_mcast in the struct wfx_vif. We
can just pass it as parameter.

Also rename wfx_update_filtering() in wfx_filter_mcast() to reflect this
change.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-20-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless filter update when starting AP

mac80211 explicitly call wfx_configure_filters() before/after starting
AP. Thus, it is not necessary to call wfx_update_filtering() from
wfx_start_ap().

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-19-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: split out wfx_filter_beacon()

Currently, wfx_update_filtering() is sometime called only to apply the
value of wvif->filter_beacon to the hardware. It is nicer to have a
specific function for beacon filtering.

In add, an attentive reader would note that wfx_update_filtering() is
always called after change of wvif->filter_beacon. Thus, it not
necessary to store filter_beacon in the struct wfx_vif. We can just pass
it as parameter.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-18-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless attributes 'filter_prbreq' and 'filter_bssid'

It is now useless to store filter_prbreq and filter_bssid in struct
wfx_vif. We can just pass them as parameters to hif_set_rx_filter().

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-17-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless call to hif_set_rx_filter()

hif_set_rx_filter() apply changes on filter_prbreq and filter_bssid to
the hardware. Each time filter_prbreq and filter_bssid are changed,
hif_set_rx_filter() is called.

Currently, on extra call to hif_set_rx_filter() is made from
wfx_update_filtering(). This call is useless. Drop it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-16-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop struct wfx_grp_addr_table

The struct wfx_grp_addr_table is only instantiated on one place. There
is no justifiation for this struct. Moreover, it is not consistent with
other fields related to filtering (filter_bssid, etc...).

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-15-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: align semantic of probe request filter with other filters

Filters provided by HIF API are sometime inclusive, sometime exclusive.

This patch align the behavior and name of the probe request filter with
the other filters.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-14-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: align semantic of beacon filter with other filters

Filters provided by HIF API are sometime inclusive, sometime exclusive.

This patch align the behavior and name of the beacon filter with the
other filters. Also avoid double negation: "disable filter"

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-13-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless wfx_fwd_probe_req()

wfx_fwd_probe_req() is a function of two lines called from only one
place. In order to uniformize all filtering functions, drop
wfx_fwd_probe_req().

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: ensure that probe requests are filtered when AP

Device is able to automatically reply to probe request (therefore, the
probe requests are not forwarded to host). This feature is enabled by
wfx_start_ap(). However, further calls to wfx_configure_filter() can
disable it.

So, relocate the handling of probe request filtering in
wfx_configure_filter().

Note that wfx_configure_filter() is always called by mac80211 when an AP
start.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: stop changing filtering rule in wfx_hw_scan()

It is not very clear why driver ask to firmware to stop filtering probe
request during scan. It is particularly weird because scan is not
supported in AP mode. In add, it does not restore previous filtering
state after scanning.

It seems to be a mistake and should be dropped.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: do not use built-in AUTO_ERP feature

Firmware is able to detect detect changes about ERP protection in
beacons and automatically enable/disable ERP protection. However, it not
bring performance improvements and we are more confident in the ERP
handling of mac80211. So, the patch disable this feature.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-9-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: disabling beacon filtering after hif_reset() is useless

After hif_reset(), device stop to receive any RF data. So, it is
useless to disable beacon filtering. In add, if necessary, mac80211 will
call wfx_configure_filter().

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-8-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: do not wait for a dtim before associate

The only reason we ask to mac80211 to wait for a dtim before to
associate is to get an accurate DTIM value. However, we can temporary
set the dtim period to 1. Then we can waiting for
BSS_CHANGED_BEACON_INFO for the real value. Thus, the association
process is faster.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: wfx_update_filtering_work() is no more used

Since we had simplified the join process, update_filtering_work is no
more used. Drop it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: simplify handling of beacon filter during join process

When the device joins an AP, beacon are forwarded to the host. Once has
retrieved all necessary parameters, we can start to filter the beacon
(and only beacon with changed/new data will be forwarded).

Currently, the driver detect beacons in data Rx process. It is far more
easier to just wait for the BSS_CHANGED_BEACON_INFO event.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: rework wfx_configure_filter()

wfx_configure_filter() lacks of coherency. In add, some corner cases
seems to not been handled properly. Rework the whole function().

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: simplify wfx_update_filtering()

Inline wfx_set_mcast_filter() into wfx_update_filtering() and remove
useless intermediate variables.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: update filtering even if not connected

Currently, filtering are not update if the device is not connected.
There is no obvious rationale for that. Updating filtering while it is
not necessary is harmless. On the other hand, it is difficult to ensure
that the filtering rule has been correctly sent to the firmware.

Just drop this useless condition.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200415161147.69738-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: formulate rspinf values into tables

Four tables can be extracted from RSPINF_A_* based on BB_TYPE_11A or
else being GB rates.

Preamble short or long tables from fixed size len of 14 for RSPINF_B rates.

Remove function vnt_calculate_ofdm_rate and replace with the tables
calling RSPINF_A and RSPINF_B separately.

Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Link: https://lore.kernel.org/r/addca9e1-9ea2-59ee-dec9-3afb3c731c2e@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: r8188eu: replace rtw_malloc/copy_from_user sequence with memdup_user

memdup_user is shorter and expressively.

Signed-off-by: Ivan Safonov <insafonov@gmail.com>
Link: https://lore.kernel.org/r/20200412173716.846469-1-insafonov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Remove unnecessary local variable initialization

Don't initialize the ret variable as it is set a few lines later.

Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200411125104.9625-3-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Remove duplicate code in vnt_vt3184_init function

Remove duplicate code in "if" statements because different branches are
almost the same.

Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200411125104.9625-2-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: Remove unnecessary local variable initialization

Don't initialize the rate variable as it is set a few lines later.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Oscar Carter <oscar.carter@gmx.com>
Link: https://lore.kernel.org/r/20200407163915.7491-3-oscar.carter@gmx.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6656: rxtx remove rate change and current_rate.

There is no longer any need to change power in vnt_tx_packet.

Remove current_rate in vnt_tx_packet and struct vnt_private as
it is no longer used elsewhere.

Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Link: https://lore.kernel.org/r/3ba896aa-5ab2-affb-9ce5-7df8a9b3190a@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vt6556: vnt_rf_setpower convert to use ieee80211_channel.

ieee80211_channel contains all the necessary information to change
power according to tx mode required.

vnt_rf_setpower is moved and so that vnt_rf_set_txpower the only
caller becomes static.

Signed-off-by: Malcolm Priestley <tvboxspy@gmail.com>
Link: https://lore.kernel.org/r/9eab9af8-fde9-1dc6-fced-95c7a36ecc01@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: vc04_services: remove set but not used 'local_entity_uc'

Fix the following gcc warning:

drivers/staging/vc04_services/interface/vchiq_arm/vchiq_arm.c:2356:16:
warning: variable ‘local_entity_uc’ set but not used
[-Wunused-but-set-variable]
  int local_uc, local_entity_uc;
                ^~~~~~~~~~~~~~~

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Link: https://lore.kernel.org/r/20200409085339.47255-1-yanaijie@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8188eu: remove 5 GHz if test

Cleanup a line over 80 characters by removing an if test that is valid
only for 5 GHz. According to the TODO 5 GHz code should be removed.

- find and remove remaining code valid only for 5 GHz. Most of the obvious
  ones have been removed, but things like channel > 14 still exist.

Signed-off-by: Michael Straube <straube.linux@gmail.com>
Link: https://lore.kernel.org/r/20200409080802.16645-1-straube.linux@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci: fix PCIe interrupt mapping

MT7621 has three assigned interrupts for the pcie. This
interrupts should properly being mapped taking into account
which devices are finally connected in which bus according
to link status. So the irq mappings should be as follows
according to link status (three bits indicating which devices
are link up):

* For PCIe Bus 1 slot 0:
  - status = 0x2 || status = 0x6 => IRQ = pcie1_irq (24).
  - status = 0x4 => IRQ = pcie2_irq (25).
  - default => IRQ = pcie0_irq (23).
* For PCIe Bus 2 slot 0:
  - status = 0x5 || status = 0x6 => IRQ = pcie2_irq (25).
  - default => IRQ = pcie1_irq (24).
* For PCIe Bus 2 slot 1:
  - status = 0x5 || status = 0x6 => IRQ = pcie2_irq (25).
  - default => IRQ = pcie1_irq (24).
* For PCIe Bus 3 any slot:
  - default => IRQ = pcie2_irq (25).

Because of this, the function 'of_irq_parse_and_map_pci' cannot
be used and we need to change device tree information from using
the 'interrupt-map' and 'interrupt-map-mask' properties into an
'interrupts' property to be able to get irq information from the
ports using the 'platform_get_irq' and storing an 'irq-map' into
the pcie driver data node to properly map correct irq using a
new 'mt7621_map_irq' function where this map will be read and the
correct irq returned.

Fixes: 46d093124df4 ("staging: mt7621-pci: improve interrupt mapping")
Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200413055942.2714-1-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci-phy: dt: bindings: remove bindings txt file

Yaml file for bindings has been properly added. Hence, remove
the old txt file.

Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200410091836.13068-3-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci-phy: dt: bindings: add mediatek, mt7621-pci-phy.yaml

Add bindings to describe Mediatek MT7621 PCIe PHY.

Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200410091836.13068-2-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: mt7621-pci: properly power off dual-ported pcie phy

Pcie phy for pcie0 and pcie1 is shared using a dual ported
one. Current code was assuming that if nothing is connected
in pcie0 it won't be also nothing connected in pcie1. This
assumtion is wrong for some devices such us 'Mikrotik rbm33g'
and 'ZyXEL LTE3301-PLUS' where only connecting a card to the
second bus on the phy is possible. For such devices kernel
hangs in the same point because of the wrong poweroff of the
phy getting the following trace:

mt7621-pci-phy 1e149000.pcie-phy: PHY for 0xbe149000 (dual port = 1)
mt7621-pci-phy 1e14a000.pcie-phy: PHY for 0xbe14a000 (dual port = 0)
mt7621-pci-phy 1e149000.pcie-phy: Xtal is 40MHz
mt7621-pci-phy 1e14a000.pcie-phy: Xtal is 40MHz
mt7621-pci 1e140000.pcie: pcie0 no card, disable it (RST & CLK)
[hangs]

The wrong assumption is located in the 'mt7621_pcie_init_ports'
function where we are just making a power off of the phy for
slots 0 and 2 if nothing is connected in them. Hence, only
poweroff the phy if nothing is connected in both slot 0 and
slot 1 avoiding the kernel to hang.

Fixes: 5737cfe87a9c ("staging: mt7621-pci: avoid to poweroff the phy for slot one")
Signed-off-by: Sergio Paracuellos <sergio.paracuellos@gmail.com>
Link: https://lore.kernel.org/r/20200409111652.30964-1-sergio.paracuellos@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless update of macaddr

Mac address is set in wfx_add_interface() and removed in
wfx_remove_interface().

Currently, there is also an additional update of mac address in
wfx_do_unjoin(). It has no rationale. Mac address is already present
and nothing has changed it. Therefore, we can drop it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-20-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unused attribute 'beacon_int'

The field beacon_int is never read. Drop it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-19-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: check value of beacon_int

Firmware dislike when beacon_int value is 0. This patch add some
warnings in case it would happen.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-18-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: re-enable BA after reset

Firmware does not support Block Acks when multiple vif are running.
Thus, wfx_add_interface() and wfx_remove_interface() enable and disable
Block Acks as necessary.

Block Ack policy is also reset after hif_reset(). Driver have to
re-enable it after each call to hif_reset().

This patch reflects this behavior.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-17-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: introduce wfx_join_ibss() and wfx_leave_ibss()

Currently, IBSS networks are started by the mean of
wfx_bss_info_changed(). It easier to use use callbacks provided by
mac80211.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-16-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: allow to join IBSS networks

Current code does not permit to join an already existing IBSS network.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-15-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: avoid duplicate updating of beacon template

When ERP changes, mac80211 call wfx_bss_info_changed() with
BSS_CHANGED_ERP_* and with BSS_CHANGED_BEACON.

The driver already update beacon template because of
BSS_CHANGED_BEACON. It is not necessary to also update beacon template
because of BSS_CHANGED_ERP_*.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-14-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: remove unnecessary conditions in wfx_bss_info_changed()

wfx_bss_info_changed() update ERP and CQM related stuff. Thus, it check
the flags BSS_CHANGED_ERP_* and BSS_CHANGED_CQM.

It also update ERP and CQM on join and leave by checking the flag
BSS_CHANGED_ASSOC. This check is useless. Mac80211 already do that job
and set necessary flags as expected.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-13-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: request to send beacons in IBSS mode

Currently, firmware take in charge of start/stop sending beacons while
in IBSS mode. However, this behavior may change in the further releases.

Currently, asking to firmware to send beacon while in IBSS mode return
an error but is harmless.

Therefore, send this request unconditionally.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unnecessary condition checks in wfx_upload_ap_templates()

In former code, wfx_upload_ap_templates() was called in more cases
than necessary. Therefore, it tried to not update the frame templates
if it was not necessary.

Now, wfx_upload_ap_templates() is called only if mac80211 asked to
update the templates. In add, it does not hurt to upload template if
they are not used. So, remove unnecessary conditions at beginning of
wfx_upload_ap_templates()

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: disabling keep alive during unjoin is useless

After a call to hif_reset(), the parameters associated with BSS are
reset. So, it useless to explicitly reset the keep alive period.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: fix support for BSS_CHANGED_KEEP_ALIVE

Chip firmware is able to send periodic null frames to keep the
association with the AP.

The driver arbitrary set this period to 30sec. We prefer to rely on
BSS_CHANGED_KEEP_ALIVE that provide a true value.

Note that if BSS_CHANGED_KEEP_ALIVE is not received, we just disable
keep_alive feature. It is not very disturbing since AP will probably
ping the station before to disconnect it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-9-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: remove useless call to wfx_tx_flush()

wfx_do_join() calls wfx_tx_lock_flush() ate beginning of the function.
Therefore, the subsequent call to wfx_tx_flush() is useless.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-8-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: change the way the station associate to an AP

Chipset need two steps to associate with an AP:
   1. it start receiving beacon from the AP (done with wfx_do_join())
   2. it sent the association request (done with wfx_join_finalize())

The join request (see hif_join()) contains basic rates, beacon interval
and bssid to connect, so we trig on these events for the first step.

The second step is obviously associated to the event BSS_CHANGED_ASSOC.

Note that conf_mutex is now easier to manage. It is held by
wfx_bss_info_changed() and inner functions does not need to lock it.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: set all parameters before starting AP

Current code start AP and then configure the different parameters. Since
all the configuration is sent quickly after AP started, it works.
However, it is not very nice. In add, last firmware releases start to
disallow incorrect settings.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: implement start_ap/stop_ap

Currently, wfx_bss_info_changed() check interface status changes and
guess when the pattern match with an AP start and AP stop (through
wfx_update_beaconing()). It is far easier to rely on start_ap and
stop_ap callbacks provided by mac80211.

wfx_bss_info_changed() keeps only the responsibility of updating the
frame templates.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: call wfx_do_unjoin() synchronously

Currently, wfx_do_unjoin() are called by the mean of work queues.
However, the contexts from where they are called are not atomic. So
there is no reason to not call it synchronously.

This change will simplify the code. Notice two main changes:
   - There no more reason to lock tx queue before to run
     wfx_do_unjoin(). We can lock the tx queue directly from
     wfx_do_unjoin().
   - Most of the time, wfx_do_unjoin_work() was called with conf_mutex
     held. This patch remove lock of conf_mutex in wfx_do_unjoin_work()
     and ensure that conf_mutex is always held whatever the context.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: reduce hold duration of cfg80211_bss

Pointer to cfg80211_bss is held during all duration of wfx_do_join. But,
it is not necessary, We can release it far earlier.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: fix race between configure_filter and remove_interface

wfx_remove_interface() and wfx_configure_filter() can be run
concurrently. Therefore, this patch protect access to the list of
interfaces from wfx_configure_filter().

Notice that wfx_configure_filter() now lock "conf_lock" and "scan_lock".
Beside that, wfx_hw_scan_work() also access to the same locks. So we
have to lock them in same order to avoid any deadlock.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200410133239.438347-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: send just necessary bytes

Size of hif_mib_template_frame must be sufficient to contains bytes
declared by frame_length and cannot exceed 700bytes.

Change the API to reflect that.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: make hif_ie_table_entry const

In wfx_update_filtering(), filter_ies is never modified. So, make it
constant.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: allow to connect an IBSS with an existing SSID

With current code, chip is not able to join an existing IBSS network.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: place hif_tx_mib functions into a .c file

Until now, all functions from hif_tx_mib.h are declared "static inline".
However, they are not time critical. So, it does not make so much sense.

We prefer to place them in a .c file as for other hif functions.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-9-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: align members declarations in hif API

Hardware API headers were not uniformly aligned.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-8-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: fix endianness of hif API

The chip expects little endian in all structs it sends/receives. This
patch fixes the hif API to reflect this fact. Sparse should now report
meaningful errors.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-7-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: remove useless defines

Several values defined in hif API are only here to define length of some
arrays. In most cases, they do not provide any extra information about
the size of the array (ie. "API_FIRMWARE_LABEL_SIZE" is only used to
define the size of member "firmware_label").

Remove these useless definitions.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: remove unused definitions from the hif API

Until now, hif API was directly imported from firmware sources.
However, it does not make sense to keep ton of unused structures in
the driver. Moreover, the hif API is now stable enough to be keep in
sync by importing delta from firmware.

So, drop unused definitions from the hif API.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: relocate TX_RETRY_POLICY_MAX and TX_RETRY_POLICY_INVALID to hif API

The definitions TX_RETRY_POLICY_MAX and TX_RETRY_POLICY_INVALID are
imposed by the hardware. Therefore, they should be located in the
hardware interface API.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-4-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: relocate LINK_ID_NO_ASSOC and MAX_STA_IN_AP_MODE to hif API

The definitions LINK_ID_NO_ASSOC and MAX_STA_IN_AP_MODE are imposed by
the hardware. Therefore, they should be located in the hardware
interface API.

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-3-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unused WFX_LINK_ID_GC_TIMEOUT

The definition WFX_LINK_ID_GC_TIMEOUT is not used anymore since
commit d6aeba575f27 ("staging: wfx: simplify the link-id allocation")

Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200406111756.154086-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: hal: Correct misspelled symbolic name

Correct misspelled symbolic name.
Misspelling found using checkpatch.pl.

Signed-off-by: R Veera Kumar <vkor@vkten.in>
Link: https://lore.kernel.org/r/b893adc31b71722df63d0756f0d360e27f896e51.1586187304.git.vkor@vkten.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: hal: Correct misspelt enum name

Correct misspelt enum name.
Misspelling found using checkpatch.pl.

Signed-off-by: R Veera Kumar <vkor@vkten.in>
Link: https://lore.kernel.org/r/d091fa95d537f613c24e55a2dee638d4493534b5.1586187304.git.vkor@vkten.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: rtl8723bs: Correct misspelt symbolic names

Correct misspelt symbolic names.
Misspelling found using checkpatch.pl.

Signed-off-by: R Veera Kumar <vkor@vkten.in>
Link: https://lore.kernel.org/r/abe807c11a5fda479ed6079a11d528989fe7b63e.1586187304.git.vkor@vkten.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: remove hack about tx_rate policies

Current code contains a weird hack to avoid switch from 54Mbps CTS to
1Mbps. However, we have not been able to reproduce the problem and
hardware team don't know any defect of this kind. So, it seems this hack
is no more necessary.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-33-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: fix case where AP stop with CAB traffic pending

When driver has Content After DTIM Beacon (CAB) in queue, it wait for an
indication from the firmware. However, when we stop to send beacons,
this indication may never happen.

Solve this issue by simply simulate this indication. Firmware will send
data that probably nobody will heard.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-32-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: fix potential deadlock in wfx_tx_flush()

wfx_tx_flush() wait there are no more frame in device buffer. However,
this event may never happens since wfx_tx_flush() don't forbid to
enqueue new frames.

Note that wfx_tx_flush() should only ensure that all frames currently in
hardware queues are sent. So the current code is more restrictive that
it should.

Note that wfx_tx_flush() release the lock before to return while
wfx_tx_lock_flush() keep the lock.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-31-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: wfx_flush() did not ensure that frames are processed

wfx_flush() exited once all frames are retrieved from the device.
However, it did not ensure they were processed by driver before to
return. Therefore, some frame may be processed after the interface has
disappear.

Change the place we signal that the queue is empty to fix that.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-30-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: repair wfx_flush()

Until now, wfx_flush() flushed queue for while device instead of only
the queue of the intended vif. It sometime failed with a timeout, but
this error was not reported.

Moreover, if the device was frozen, wfx_flush didn't do anything and it
results a potential warning (and maybe a resource leak) when the frozen
device was unregistered.

We can also notice that wfx_tx_queues_wait_empty_vif() did only exist to
work around the broken feature of wfx_flush().

This patch repair wfx_flush() and therefore drop
wfx_tx_queues_wait_empty_vif().

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-29-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: relocate wfx_skb_dtor() prior its callers

In a next commit, we would like to mark wfx_skb_dtor as static and stop
to declare it in data_tx.h.

Relocate wfx_skb_dtor() prior its callers to avoid compile error.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-28-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: improve interface between data_tx.c and queue.c

Currently, wfx_pending_remove() (from queue.c) call wfx_skb_dtor()
(from data_tx.c) that forward the tx status to mac80211.

Moreover, there no purpose to retrieve a frame from the pending queue
without dequeuing it. So, the main purpose of wfx_pending_remove() is to
forward the tx status to mac80211.

Let's make the architecture cleaner:
  - merge wfx_pending_remove() into wfx_pending_get()
  - call wfx_skb_dtor() from data_tx.c

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-27-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: simplify usage of wfx_tx_queues_put()

The queue used for wfx_tx_queue_put() can be deducted from the content
of the skb. So drop this parameter from call to wfx_tx_queues_put().

In add, this change uniformizes usage of functions wfx_tx_queues_*.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-26-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop struct wfx_queue_stats

There is no reason to keep the intermediate struct wfx_queue_stats.
Relocate its members to struct wfx_dev.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-25-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop now useless field edca_params

Since we do not rely in QoS parameters to choose which frame to send, it
is no more necessary to keep a copy of EDCA parameters.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-24-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: change the way to choose frame to send

The current code computes itself the QoS policy to choose which frame
should be sent. However, firmware already do that job. Firmware would
prefer to have packets in every queues and be able to choose itself
which queue to use.

So, this patch sort the queues from the emptiest to the fulliest (thanks
to the pending frames counter introduced a few commits earlier). It send
frame to the least full queue.

However, we continue to be careful with frames that have to be sent
after a dtim ("cab": Content After (DTIM) Beacon).

So, this patch splits AC queues in two skb_queues: one for normal frames
and another for cab frames. It cares to send frames from CAB skb_queue
if appropriate.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-23-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: introduce a counter of pending frames

This counter will be useful to know which queue is least full in a
further patch.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-22-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: replace wfx_tx_queues_get_after_dtim() by wfx_tx_queues_has_cab()

It is not necessary to return a skb. Just getting the information if
there is traffic to be sent after DTIM is sufficient.

In add, the acronym "cab" (Content After (DTIM) Beacon) is used in
mac80211 to designate this kind of traffic.

So, make wfx_tx_queues_get_after_dtim() return a boolean and rename
accordingly.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-21-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: rename wfx_tx_get_raw_link_id()

Since concept of "raw_link_id" does not exist anymore, rename
wfx_tx_get_raw_link_id() in wfx_tx_get_link_id().

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-20-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unused raw_link_id field

raw_link_id can be retrieved by wfx_tx_get_raw_link_id(). So, it is not
necessary to keep it in struct wfx_tx_priv.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-19-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unused link_id field

It is not used anymore.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-18-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: do not rely anymore on link_id to choose packet in queue

link_id was expected to contain identifier of a station. It was also
used to mark frames that has to sent after dtim. We do not use the
further purpose. For the last purpose, we can directly check the flag
value in tx_info.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-17-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless link_map_cache

Driver does not use link_map_cache anymore. So we can drop it. In add,
we do not have to keep this counter in sync with skb_queue item, so we
can drop explicit spin_locks.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-16-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: do not use link_map_cache to track CAB

Since we do not track power save status of the stations anymore,
link_map_cache is now only used to track "Content After (DTIM) Beacon".
We prefer to rely on flags from tx_info. So we will be able to drop
link_map_cache.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-15-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop argument tx_allowed_mask since it is constant now

Following the remove of asleep_mask, the tx_allowed_mask argument passed
to various functions is now always the same. Drop this argument and
simplify the code.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-14-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop useless sta_asleep_mask

Currently, the driver tracks power save state of the stations with the
variable sta_asleep_mask. Then, it takes care to not sent data to asleep
stations.

However, this work is already done by mac80211. Normally, there are no
frames for asleep stations in our queues. So, driver do not have to
filter frames in its queues (apart the frames marked "AFTER_DTIM").

Notice that there is a risk of race between state of the station and
data send to the firmware. However, this risk is limited since the
number of frame in queues are small. In add, this race also exists with
the current code. Anyway, the firmware is able to detect the problem and
driver will receive a 'REQUEUE' status (translated in
TX_STAT_TX_FILTERED for mac80211).

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-13-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: simplify wfx_tx_queue_mask_get()

It is far simpler to return a pointer instead of an error. Thus, it is
no more necessary to pass a pointer reference as argument.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-12-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: drop unused argument in wfx_get_prio_queue()

The argument "total" is not used anymore since commit a3c529a83589
("staging: wfx: simplify handling of IEEE80211_TX_CTL_SEND_AFTER_DTIM").

Fixes: a3c529a83589 ("staging: wfx: simplify handling of IEEE80211_TX_CTL_SEND_AFTER_DTIM")
Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-11-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: wfx: simplify wfx_tx_queues_empty()

Thanks to skb_queue_empty_lockless(), it is not necessary to acquire the
spin_lock before to check if the queue is empty.

Reviewed-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200401110405.80282-10-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>