The per rate stats should be cleared when aggregation state changes
to avoid making rate scale decisions based on throughput figures which
were collected prior to the aggregation state change and are now stale.
While at it make sure any clearing of the per rate stats will get logged.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Don't search columns which are unlikely to succeed as previous
columns searched with less aggressive modulation failed.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Allow switching back to legacy Tx columns so we'll stop doing
HT/VHT in case we're far from the AP. Stop active aggregation when
making a deciding to stay in a legacy column.
Despite having low legacy rates in the LQ table lower entries
it doesn't help much in case we're doing aggregations as the
aggregation was being transmitted in the initial rate of the table.
This should help traffic stalls when far from the AP.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
After being idle for a long time (>5sec) the rs statistics
will be stale so we prefer to reset rs and start from legacy
rates again. This gives better results when the attenuation
increased signficantly (e.g. we got further from the AP) and
after a while we start Tx
Note that the first Tx after the idle period will still go out
in the old modulation and rate but this seemed a simpler approach
compared to adding a timer or modifying mac80211 for this.
The negative impact is negligble as we'll recover quickly.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Change the down/upscale decision logic a bit to be based
on different success ratio thresholds. This fixes the implementation
compared to the rate scale algorithm which was planned to yield
optimal results. Also fix a case where a lower rate wasn't explored
despite being a potential for better throughput.
While at it rewrite rs_get_rate_action to be more clear and clean.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The selection of the max expected throughput for a column
didn't take into account the maximal allowed rate for the current
peer. This can cause unnecessary switches during the search cycle
to columns which have no chance of beating the current throughput.
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This firmware is not supported any more.
A few code paths specific to old firmware can be removed.
We can now assume that a few TLV flags are always set since
we won't load firmware that didn't support the corresponding
features. This will be done in a separate patch.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
While the scan_cmd should really be allocated in init (and
we do fail init in case the allocation failed), it doesn't
mean we should lock up the machine if something really bad
happened.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Collect accumulated stats of tx attempts and successes
per rate and column for debugging purposes.
These stats can be read via debugfs file drv_tx_stats
and can also be reset by writing to this file.
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
rcu_assign_pointer() ensures that the initialization of a structure is
carried out before storing a pointer to that structure. However, in the
case that NULL is assigned there's no structure to initialize so using
RCU_INIT_POINTER instead is safe and more efficient.
Signed-off-by: Monam Agarwal <monamagarwal123@gmail.com>
[rewrite commit log]
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The firmware doesn't support broadcast filtering on P2P
client. Trying to enable it makes the firmware assert.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When NIC is exiting from thermal throttling state (i.e. after
heating and then cooling down), tx-backoff values are assigned
to be zero, instead of being restored to the minimal value. Fixed.
Signed-off-by: Eytan Lifshitz <eytan.lifshitz@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When NIC exit thermal throttling while having minimal tx-backoff
restriction, the warning message that indicate about that state
won't show up. Fixed.
Signed-off-by: Eytan Lifshitz <eytan.lifshitz@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The Rx FIFO includes valuable data - dump it when the FW
asserts. Also - free the SRAM and Rx FIFO when we create
the file, and don't collect new SRAM / Rx FIFO if the
previous file hasn't been collected through debugfs yet.
Also - add a comment to saying that the ASSERT output should
not be modified since we have automatic scripts that monitor
this output.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This must not happen - otherwise we might keep flushing
forever.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This will be used later to flush / wait for queues that are
related to a specific vif.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Now that mac80211 calls pre_rcu_remove and we set the
fw_id_to_mac_id pointer to -ENOENT before the station is
removed, we don't need to set fw_id_to_mac_id to -EINVAL
when the station is really removed.
Leave fw_id_to_mac_id to be -ENOENT which will let the
drain worker (iwl_mvm_sta_drained_wk) know that this station
is not to be drained.
We don't need to drain this station since it is our AP in
managed mode and we flush all its frames synchronously
anyway.
Setting the AP station to -EINVAL could lead to confusion
since internal stations are also reserved with -EINVAL,
this confusion showed up in the logs as:
Drained sta 0, but it is internal?
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The firmware tells us if an Rx packet was beamformed or not.
Propagate this data to mac80211's rx_status.vht_flags.
The main user of this data is the radiotap header.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This define is related to the firmware packaging and is
needed by more than one transport.
Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Besides the different allocation flags, they are really
the same. Pass the gfp_t flags as a parameter, and unify
them.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Similar to the recent RX queue patch, this changes the need_update
handling for the TX queues to be clearer and only done when needed.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When updating the write pointer, the TX queue should be locked
to get consistent state, fix that in the interrupt handler.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Instead of using u8 to hold logic values, use bool.
Also fix a comment, the return value is no longer relevant.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
When shadow registers are enabled, then need_update never needs
to be set, so move the need_update handling into the function
that really needs to do it (iwl_pcie_rxq_inc_wr_ptr) and also
separate the check when it woke up. While at it, convert it to
bool.
This also clarifies the locking and means the irq_lock needs to
no longer be held for any such updates.
The irq_lock also doesn't have to be held for restocking since
everything else locks the RX queue properly, so remove that and
finally disentangle the two locks entirely so there aren't any
dependencies between the two left.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Use the new NAPI infrastructure added to mac80211 to get
GRO. We don't really implement NAPI since we don't have
a real poll function and we never schedule a NAPI poll.
Instead of this, we collect all the packets we got from a
single interrupt and then call napi_gro_flush().
This allows us to benefit from GRO. In half duplex medium
like WiFi, its main advantage is that it reduces the number
of TCP Acks, hence improving the TCP Rx performance.
Since we call the Rx path with a spinlock held, remove
the might_sleep mention from the op_mode's API.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Reviewed-by: Ido Yariv <ido@wizery.com>
[Squash different patches and rewrite the commit message]
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Allow fixing the tx power reduction through debugfs
file.
The change doesn't take affect immediately, but
will be considered the next time tpc is evaluated.
Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Implement Adaptive Tx Power Control algorithm.
ATPC basically tries to decrease the tx power
as much as possible while the throughput is
not being hurt.
Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
I forgot to send the new Look Up Table to the firmware and
I also forgot to free the command which is kzalloc'ed.
This code is relevant for 7265 device only.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
mimo_delim was always set to 0 instead of pointing to
the first SISO entry after MIMO rates.
This can cause keep transmitting in MIMO even when we shouldn't.
For example when the peer is requesting static SMPS.
Cc: <stable@vger.kernel.org> [3.14]
Signed-off-by: Eyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
If we have no beacon data before association, delay smart FIFO
enablement until after we have this data.
Not doing so can cause association failures in extremely silent
environments (usually only a shielded box/room) as beacon RX is
not sent to the host immediately, and then the association time
event ends without the host receiving any beacon even though it
was on the air - it's just stuck on the FIFO.
Cc: <stable@vger.kernel.org> [3.14]
Fixes: 1f3b0ff8ec ("iwlwifi: mvm: Add Smart FIFO support")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The fw recently added an option to set tx power reduction
per station (in the lq_cmd command), and get the tx power
reduction used (in the tx_resp struct).
Use them and propogate this value up to mac80211's tx
response (in order to use it later in the rate-scaling
algorithm).
Signed-off-by: Eliad Peller <eliadx.peller@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This will allow the low level driver to make decision based
on the vif such as queues etc...
Since the vif might be NULL, we can't add it to the tracing
functions.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
[fix staging rtl8821ae driver]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
A hardware bug had been discovered on 7260 / 3160 and 7265
and the workaround for this bug is to force the NIC to stay
awake as long as we have host commands in flight. This
workaround has been introduced for all NICs in a previous
patch:
b943949105 ("iwlwifi: pcie: keep the NIC awake when commands are in flight")
In newer NICs, this bug is solved, so we can let the NIC go
to sleep even when we send commands. The hardware will wake
up when we increment the scheduler write pointer.
Make the workaround conditional to only use it on affected
hardware.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Number of channels changed from 40 to 50
Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Today, the driver logic looks for the conditions to disable
power management albeit power management should be enabled
in a very few distinct cases.
This patch changes the driver logic to enable power
management once the required conditions met.
While at it, make some housekeeping and support a few
additional use cases:
a) Add support for a standalone p2p client:
Power management should be enabled for a P2P client
MAC only if the firmware supports it (TLV flag is set).
Instead we used the DCM flag, therefore we didn't cover
use cases that did not include the DCM flag.
b) Add support to Same-Channel-Mode (SCM):
If both clients share the same channel (SCM), and there
are no other active vifs in the system, power management
should be enabled only if the firmware supports this
(TLV flag is set).
c) Fix power management logic for GO/AP:
Today, when we detect an active GO / AP MAC - we disable
power management for all the vifs altogether.
Actually, the correct behavior is to enable power
management on a client if on a different channel
(based on the firmware capabilities).
d) Housekeeping - Along with that, this patch includes some
code-reorganizing: Today the logic of disabling power is
scattered across several functions, specifically in the
iterator. For the sake of both readability and
scalability, we moved this logic to its applicable
function, leaving the iterator gather information only.
Furthermore, as power management is a MAC-related
attribute, we moved the power management member to the
iwl_mvm_vif structure.
Signed-off-by: Avri Altman <avri.altman@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This fixes the build for powerpc:
After merging the final tree, today's linux-next build (powerpc allyesconfig)
failed like this:
drivers/net/wireless/iwlwifi/mvm/debugfs.c: In function 'iwl_dbgfs_fw_error_dump_release':
drivers/net/wireless/iwlwifi/mvm/debugfs.c:161:2: error: implicit declaration of function 'vfree' [-Werror=implicit-function-declaration]
vfree(file->private_data);
^
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The current firmware advertises support for uAPSD, but
critical bugs force us to disable the feature.
When a fixed firmware will be available, we will be able to
re-enable uAPSD.
Cc: <stable@vger.kernel.org> [3.13+]
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
If the low-latency update is called but there's no change then
ignore the update instead of triggering all the required work.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Remove a spurious blank line in the quota code.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
In case of system low latency configure passive scan to be fragmented.
Set the following scan parameters for both immediate and scheduled scan:
- passive scan fragment duration = 20ms
- out-of-channel time = 70ms
- suspend time = 105ms
Restructure channel's active/passive dwell time configuration to better
suit the above change.
The idea is that under low latency traffic passive scan is fragmented,
i.e. that dwell on a particular channel will be fragmented. Each
fragment dwell time is 20ms and fragments period is 105ms. Skipping to
next channel will be delayed by the same period (105ms). So suspend_time
parameter describing both fragments and channels skipping periods is set
to 105ms. This value is chosen so that overall passive scan duration
will not be too long. Max_out_time in this case is set to 70ms, so for
active scanning operating channel will be left for 70ms while for
passive still for 20ms (fragment dwell).
Signed-off-by: Alexander Bondar <alexander.bondar@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Some scan parameters should be dependent on traffic conditions.
Centralize conditions verification in one function and obtain
scan max out-of-channel and suspend time in that new function.
Rely on bound interfaces indication instead of association state
to calculate scan parameters. If no bound interfaces use default
values for out-of-channel and suspend time parameters.
Additionally, get rid of NL80211_SCAN_FLAG_LOW_PRIORITY checks
since no use case for this exists so far.
Signed-off-by: Alexander Bondar <alexander.bondar@intel.com>
[reword commit log a bit]
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The firmware decided to not implement this API in this way,
so for now remove setting the field completely. This will
allow the firmware to change how to use this field later.
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
In order to debug the firmware, we need to be able to set
the BT priority of WiFi packets. This priority is set based
on the type of the packet (control frames, EAPOL etc...).
For debugging purposes, allow to override this priority by
a debugfs controlled value.
Enable this feature that needs this priority to be able to
test it.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
