Change the code to allow making all the rate control ops
const, nothing ever needs to change them. Also change all
drivers to make use of this and mark the ops const.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Discovered by klocwork

Array 'iwl_rate_mcs' of size 15 may use index value(s) -1
  * rs.c:2562: index = iwl_hwrate_to_plcp_idx(rate)
    * rs.c:2562: Result of function call 'iwl_hwrate_to_plcp_idx(rate)' is '[-1,14]'
  * rs.c:2565: Array 'iwl_rate_mcs' size is 15.
  * rs.c:2565: Possible attempt to access element -1 of array 'iwl_rate_mcs'.

While at it stop using index = -1 and always use IWL_RATE_INVALID
Signed-off-by: NEyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If we can't switch to a column because no rates are supported
in that column this led to a state where the search cycle
got stuck and never ended. This in turn also led to aggregation
not being turned on. Fix this by marking a column as
visited if we can't switch to it.
Reported-and-tested-by: NKarl Beldan <karl.beldan@gmail.com>
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

None of these files are actually using any __init type directives
and hence don't need to include <linux/init.h>.  Most are just a
left over from __devinit and __cpuinit removal, or simply due to
code getting copied from one driver to the next.
Signed-off-by: NPaul Gortmaker <paul.gortmaker@windriver.com>
Acked-by: NLarry Finger <Larry.Finger@lwfinger.net>
Acked-by: NChristian Lamparter <chunkeey@googlemail.com>
Acked-by: NGertjan van Wingerde <gwingerde@gmail.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Happy new year!
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Found by klocwork analysis.
mvm could be NULL which may cause a NULL dereference
in a theoretical call flow

rs_fill_lq_cmd(mvm = NULL, ...)
rs_build_rates_table
rs_fill_rates_for_column
ucode_rate_from_rs_rate
IWL_ERR(mvm,...)

No real reason for passing NULL to rs_fill_lq_cmd so fix that.
Reported-by: NEytan Lifshitz <eytan.lifshitz@intel.com>
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Don't check if mvm->fw->cs is NULL since it can't be.
cs is an array member of iwl_fw, it can't be NULL.

Use memset(ptr, 0, sizeof(*ptr)); instead of
memset(ptr, 0, sizeof(struct ptr_type));
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Reviewed-by: NEyalX Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

MCS9 introduces some corner cases in the current rs
algorithm which may lead to non optimal throughput and
instability in the throughput. Until all the corner
cases are resolved disable MCS9 for Tx as a workaround
which yields better throughput results as MCS8 is much
more stable.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Extract the scale action decision to a different function
in preparation of modifying it. While at it also convert
the scale action values from hardcoded values to a clear enum.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The logs are emitted in a flow in which there were retries
and the rates in the rate table entry didn't match the active
or search table. This doesn't indicate a problem and is
expected in most cases where there will be retries for some
reason. Remove the logs.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The new logic will attempt more rates with less retries
per rate. Also when starting off with MIMO it will
fallback to SISO with the same MCS and only then to Legacy.
Previously we fell back directly to Legacy.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The supported legacy rate mask is initialized when rs
is initialized based on the remote peer supported rates.
There's no need to re mask it repeatedly with the supported
remote peer rates.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Simplify the code a bit more by extracting the rates table
building logic into a separate function and handle setting
a fixed rate for debug in a separate flow.
Also avoid using and saving ucode rate format in different
places. Instead use rs_rate struct and convert to ucode format
only when filling the rates table in the LQ command.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Move rs_program_fix_rate right before it's caller where we're
already in the context of an ifdef CPTCFG_MAC80211_DEBUGFS so
we can get rid of the extra ifdefs surrounding the original
location.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This fixes compilation issues with CONFIG_MAC80211_DEBUGFS
not enabled.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Rewrite the search cycle state machine to use a more data
oriented approach where the different Tx columns (configs)
limitations and next columns to search are reprsented in
tables which are easy to change. This overhaul also includes
several major fixes:

1. Prevent going back to a specific Tx column in a search
cycle if it was already explored.

2. Avoid switching to a Tx column that doesn't have any chance
if it performs perfectly to beat the current throughput we're
getting.

These issues were degrading throughput as they were causing
switching to "bad" Tx columns.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Use the 20Mhz channel width define instead of just the number
zero for legacy rates. Note that the define has the same value
so this is just a minor cleanup.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The table rs_ht_to_legacy is used to get the next legacy rate
following the last HT or VHT rate in the LQ rates table.
The mapping wasn't correct as well as didn't include entries
for MCS8/9 which led to out of bounds access. This didn't
trigger a crash but led to legacy rate entries using 1Mbps
rate.
In 5Ghz this probably caused the Tx to fail completely given
that Tx attempt would have reached the legacy entries and 1Mbps
isn't valid.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Drop code which was relevant when there were chips with
3 antennas. Setting to ANT_AB should be ok with all mvm
supported chips.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

No use of the ieee80211 header. Drop it.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Introduce rs_rate which represents a rate. Use this structure
instead of iwl_scale_tbl_info where we're dealing with a single
rate.
This avoids allocating the big iwl_scale_tbl_info structure
on the stack in several cases like converting to ucode rate
format or from ucode rate format.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remove this as it's unused.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remain in the same Tx modulation (i.e. column) for a longer
time before starting a search cycle for a better modulation.
This has been shown to give better results.
Also change the name of the timeout define to better match its
description.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Rename for clearer names.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Expected TPT table was updated only when switching to a new
modulation. This is wrong as toggling aggregation changes
the expected TPT signficantly.
This leads to scenarios where turning aggregation on after
being in MIMO sends us back to SISO despite a perfect success
ratio.
This occurred because the TPT of the SISO mode was being
estimated based on aggregation while the MIMO one wasn't.

Also remove an error print which isn't an error anymore
since we might be updating the expected TPT table due
to aggregation changes.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This terminates a test window of a new rate quicker in case
we've hit a bad rate and should recover better.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Enable better tracking of different decisions made
by the rate scale algorithm.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remote peer can publish a different number of supported nss via the
operating mode notification IE or action frame. If it limits to 1
then we don't want mimo rates configured in the rate table.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

MCS9 in NSS=1 or NSS=2 isn't valid for 20Mhz so don't enable
it in case we're dealing with a 20Mhz sta. Trying to configure an
MCS9 rate with 20Mhz in the LQ rate table would lead to a FW assert.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Switch to using the new btcoex decision api regarding MIMO and stop
accessing the internal btcoex structs.
In case MIMO should be disabled it would detect this upon the next
Tx and force a search. The search will switch to SISO on a antenna A
which isn't used by BT.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Add more bit field definitions used in LQ flags.
Also rename the first bit to a shorter name.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Re-initialize rs when we get a rate_update hook called which
happens when remote STA changes bandwidth or sends a VHT opmode
notification.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Refactor setting of VHT enabled rates in preparation of adding
some more logic there.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remove the flags parameter which should be set to sync or async
according to whether this is called during sta init or not.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Info about VHT/HT rates wasn't printed properly when dumping
the rate scale table. Fix that and print more info. While at it
fix some other minor issues in the printing and prevent
overflowing the print buffer.
Signed-off-by: NEyal Shapira <eyal@wizery.com>
Reviewed-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remove code that is not needed and always allow MIMO when
in tight mode. In loose mode, we should avoid MIMO since BT
can use the other antenna to Rx while we Tx.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

The code limiting the AMPDU length due to BT traffic was
duplicated. Remove the redundant code.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Tests have shown that we should start AMPDU even when BT
is active. So remove that constraint.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

In loose BT Coex scheme, the aggregation size doesn't need to
be limited.

To avoid triggering it, remove a lockdep assertion - we need
to compute the AMPDU size limit from rate control code which
can't take mvm->mutex. This means that there is a race but in
the worst case, we will have a wrong AMPDU size limit which
is not a big issue.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>