Here are a number of small staging and IIO driver fixes for 5.6-rc7
Nothing major here, just resolutions for some reported problems:
- iio bugfixes for a number of different drivers
- greybus loopback_test fixes
- wfx driver fixes
All of these have been in linux-next with no reported issues.
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-----BEGIN PGP SIGNATURE-----
iG0EABECAC0WIQT0tgzFv3jCIUoxPcsxR9QN2y37KQUCXnTRmQ8cZ3JlZ0Brcm9h
aC5jb20ACgkQMUfUDdst+ykkCwCfS8zLibsZtuv655yAK17C/6YghPQAoJ3iSew+
EkzEDWQ9j2NLE+lpWtKQ
=fZ2X
-----END PGP SIGNATURE-----
Merge tag 'staging-5.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging
Pull staging/IIO fixes from Greg KH:
"Here are a number of small staging and IIO driver fixes for 5.6-rc7
Nothing major here, just resolutions for some reported problems:
- iio bugfixes for a number of different drivers
- greybus loopback_test fixes
- wfx driver fixes
All of these have been in linux-next with no reported issues"
* tag 'staging-5.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
staging: rtl8188eu: Add device id for MERCUSYS MW150US v2
staging: greybus: loopback_test: fix potential path truncations
staging: greybus: loopback_test: fix potential path truncation
staging: greybus: loopback_test: fix poll-mask build breakage
staging: wfx: fix RCU usage between hif_join() and ieee80211_bss_get_ie()
staging: wfx: fix RCU usage in wfx_join_finalize()
staging: wfx: make warning about pending frame less scary
staging: wfx: fix lines ending with a comma instead of a semicolon
staging: wfx: fix warning about freeing in-use mutex during device unregister
staging/speakup: fix get_word non-space look-ahead
iio: ping: set pa_laser_ping_cfg in of_ping_match
iio: chemical: sps30: fix missing triggered buffer dependency
iio: st_sensors: remap SMO8840 to LIS2DH12
iio: light: vcnl4000: update sampling periods for vcnl4040
iio: light: vcnl4000: update sampling periods for vcnl4200
iio: accel: adxl372: Set iio_chan BE
iio: magnetometer: ak8974: Fix negative raw values in sysfs
iio: trigger: stm32-timer: disable master mode when stopping
iio: adc: stm32-dfsdm: fix sleep in atomic context
iio: adc: at91-sama5d2_adc: fix differential channels in triggered mode
Remove spaces after type casting operators in order to comply codding
standards.
Issue found with checkpatch.
Signed-off-by: Lourdes Pedrajas <lu@pplo.net>
Link: https://lore.kernel.org/r/20200315044922.14249-1-lu@pplo.net
Reviewed-by: Stefano Brivio <sbrivio@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Access to result of ieee80211_bss_get_ie() is protected by RCU. In other
hand, function hif_join() can sleep and cannot be called with RCU
locked.
Provide a copy of "ssidie" to hif_join() to solve this behavior.
Fixes: 9ced9b5937 ("staging: wfx: simplify hif_join()")
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200310101356.182818-6-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Access to sta->ht_cap is protected by RCU. However,
hif_set_association_mode() may sleep, so it can't be called in RCU.
This patch fix this behavior by handling sta and its RCU directly from
function hif_set_association_mode().
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Fixes: d001490110 ("staging: wfx: fix RCU usage")
Link: https://lore.kernel.org/r/20200310101356.182818-5-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
After hif_shutdown(), communication with the chip is no more possible.
It the only request that never reply. Therefore, hif_cmd.lock is never
unlocked. hif_shutdown() unlock itself hif_cmd.lock to avoid a potential
warning during disposal of device. hif_cmd.key_renew_lock should also
been unlocked for the same reason.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200310101356.182818-2-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
struct hif_cnf_reset is not used anywhere in wfx. It's parent file
hif_api_cmd.h is imported form some other driver, hence some functions
in it are not used. This is one of them. Hence remove it.
Signed-off-by: Kaaira Gupta <kgupta@es.iitr.ac.in>
Link: https://lore.kernel.org/r/20200310142509.25632-7-kgupta@es.iitr.ac.in
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Blank line is not necessary before a close brace '}', remove it.
Signed-off-by: Payal Kshirsagar <payalskshirsagar1234@gmail.com>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200309165528.5721-1-payalskshirsagar1234@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
There are several DT doc references that require manual fixes.
I found 3 cases fixed on this patch:
- directory named "binding/" instead of "bindings/";
- .txt to .yaml renames;
- file renames (still on txt format);
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Miquel Raynal <miquel.raynal@bootlin.com>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Acked-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Rob Herring <robh@kernel.org>
Lockdep is complaining about recursive locking, because it can't make
a difference between locked skb_queues. Annotate nested locks and avoid
double bh_disable/enable.
[...]
insmod/815 is trying to acquire lock:
cb7d6418 (&(&list->lock)->rlock){+...}, at: wfx_tx_queues_clear+0xfc/0x198 [wfx]
but task is already holding lock:
cb7d61f4 (&(&list->lock)->rlock){+...}, at: wfx_tx_queues_clear+0xa0/0x198 [wfx]
[...]
Possible unsafe locking scenario:
CPU0
----
lock(&(&list->lock)->rlock);
lock(&(&list->lock)->rlock);
Cc: stable@vger.kernel.org
Fixes: 9bca45f3d6 ("staging: wfx: allow to send 802.11 frames")
Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Link: https://lore.kernel.org/r/5e30397af95854b4a7deea073b730c00229f42ba.1581416843.git.mirq-linux@rere.qmqm.pl
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Current code races in init/exit with interrupt handlers. This is noticed
by the warning below. Fix it by using devres for ordering allocations and
IRQ de/registration.
WARNING: CPU: 0 PID: 827 at drivers/staging/wfx/bus_spi.c:142 wfx_spi_irq_handler+0x5c/0x64 [wfx]
race condition in driver init/deinit
Cc: stable@vger.kernel.org
Fixes: 0096214a59 ("staging: wfx: add support for I/O access")
Signed-off-by: Michał Mirosław <mirq-linux@rere.qmqm.pl>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/f0c66cbb3110c2736cd4357c753fba8c14ee3aee.1581416843.git.mirq-linux@rere.qmqm.pl
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/staging/wfx/queue.c: In function wfx_tx_queues_get:
drivers/staging/wfx/queue.c:484:28: warning: variable tx_priv set but not used [-Wunused-but-set-variable]
commit 2e57865e79 ("staging: wfx: pspoll_mask make no sense")
left behind this unused variable.
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Reviewed-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200211140334.55248-1-yuehaibing@huawei.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_tx_queue_get_num_queued() can take advantage of BIT() instead of
maintaining one variable for a counter and another for a mask.
In add, wfx_tx_queue_get_num_queued() has no real reason to return a
size_t instead of an int.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-64-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Since enum action has now only two cases, it can be dropped. Then
hif_handle_tx_data() can be simplified.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-63-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Obviously, the value of wvif->state as no reason to be wrong. At least,
if it the case, dropping the frame is probably not the bast thing to do.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-62-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In current code, packet_id is derivated from mac802.11 packet sequence
number, but this number is only unique for a station. It is not
sufficient.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-60-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When mac80211 ask for a frame to be sent after a DTIM, driver should:
1. Update TIM with multicast bit set (using update_ie). This function
can be called whenever.
2. Keep buffered all frames marked "after dtim"
3. When it receive a suspend_resume indication (see
wfx_suspend_resume_mc()), send all the buffered frames. This
indication is sent by the firmware 4ms before the dtim.
4. If one of the frames returns status "REQUEUE", it means that the
DTIM period was ended before to be able to send the frame.
5. When all the buffered frames were sent or if DTIM period was ended,
driver should update the TIM with multicast bit reset.
All the mess with the asynchronous works can be dropped.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-58-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Indeed, it is not necessary to pass whole hif_ind_suspend_resume_tx to
wfx_suspend_resume_mc().
In add, the structure hif_ind_suspend_resume_tx come from hardware API.
It is not intended to be manipulated in upper layers of the driver.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-57-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
At the beginning, firmware could send suspend_resume indication to
notify that a station wake up or sleep down. However, mac80211 already
handles power save status of stations and this behavior has been removed
from the firmware. So now, when suspend_resume indication is received,
it is always to notify that a DTIM is about to be sent.
So, it is possible to simply wfx_suspend_resume().
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-56-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_ps_notify() is called for any changes in the TIM. However,
association ID 0 is a very special case that should be handled
independently.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-55-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
pspoll_mask is here to send data buffered in driver. But since station
is marked buffered, TIM for this station is 1 and mac80211 will call
sta_notify when a ps-poll is received. So pspoll_mask is useless and
sta_alseep_mask is sufficient.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-54-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_tx_get_tid() was used as a wrapper around ieee80211_get_tid(). It
did sometime return WFX_MAX_TID to ask to upper layers to not include
the frame in "buffered" counter. The objective of this behavior is not
clear, but tests has shown that wfx_tx_get_tid() can be replaced by
ieee80211_get_tid() without any regressions.
BTW, it is not necessary to save the tid in tx_rpiv since it can be
retrieved from the 802.11 header.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-53-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The "link-id" is a slot number provided to the chip. A link-id is
allocated to every station associated with the chip (mainly when the
chip is in AP mode). It is more or less the same thing than the
association ID, but it is limited to 14 values.
Firmware uses the link-id to track the power save status of the
stations.
The current code try to associate a link-id as soon as data are
exchanged with station. It is far easier to rely on sta_add() and
sta_remove().
Until now the value WFX_LINK_ID_NO_ASSOC, was only used when no more
link-id was available. Now, we also use this value for
not-yet-associated stations (that was its primary behavior).
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-51-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Argument provided to wfx_set_tim_impl() is always wvif->aid0_bit_set and
there is no reason to provide another argument.
Also rename wfx_set_tim_impl() into wfx_update_tim() to reflect the new
behavior.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-50-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_tx_confirm_cb() retrieves the station associated with a frame using
the MAC address from the 802.11 header. In the other side wfx_tx()
retrieves the station using sta field from the ieee80211_tx_control
argument.
In wfx_cqm_bssloss_sm(), wfx_tx() was called directly without valid sta
field, but with a valid MAC address in 802.11 header. So there the
processing of this packet was unbalanced and may produce weird bugs.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-48-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It simplify the code if field buffered is hosted in the struct sta_priv
instead of in the struct wfx_link_entry. More globally, struct
wfx_link_entry has no real reasons to exist and should be dropped soon.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-47-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It seems that purpose of "early_data" was to prevent sending data to
mac80211 before station was completely associated. It is a useless
precaution.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-46-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
It is recommended to use function time_*() to compare jiffies.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-45-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
wfx_conf_tx() is called for each queue. On every call, the function
updates UAPSD mask and PM mode for all queues. It is a pity since the
UAPSD configuration very rarely changes and it makes exchanges between
the host and the chip more difficult to track.
This patch avoid to update UAPSD and Power Mode in most usual cases.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-43-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When multiple vif are in use (the only supported configuration is one
station and one AP), the driver force power save flag on station.
This behavior allows the station to leave the station channel and make
its business on AP channel.
However, this has a big impact on station performances (especially since
only legacy PS is supported).
When both vifs use the same channel, it is not necessary to keep this
restriction. This greatly improve station performances.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-42-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
mac80211 and the device are both able to control power-save per vif.
But, the current code retrieve power-save from wfx_config(). So, it does
not allow to setup power-save independently for each vif. Driver just
has to rely on wfx_bss_info_changed() instead of wfx_config().
wfx_config() has nothing to do anymore, but we keep it since it is
mandatory for mac80211.
Signed-off-by: Jérôme Pouiller <jerome.pouiller@silabs.com>
Link: https://lore.kernel.org/r/20200115135338.14374-41-Jerome.Pouiller@silabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Greg Kroah-Hartman