When the op_mode sends an skb whose payload is bigger than
MSS, PCIe will create an A-MSDU out of it. PCIe assumes
that the skb that is coming from the op_mode can fit in one
A-MSDU. It is the op_mode's responsibility to make sure
that this guarantee holds.

Additional headers need to be built for the subframes.
The TSO core code takes care of the IP / TCP headers and
the driver takes care of the 802.11 subframe headers.

These headers are stored on a per-cpu page that is re-used
for all the packets handled on that same CPU. Each skb
holds a reference to that page and releases the page when
it is reclaimed. When the page gets full, it is released
and a new one is allocated.

Since any SKB that doesn't go through the fast-xmit path
of mac80211 will be segmented, we can assume here that the
packet is not WEP / TKIP and has a proper SNAP header.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The transport will need to use the info->driver_data
pointers. Since the op_mode has this memory hot in cache,
clear it there.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The code that handles the TBs that contain the WiFi payload
will be changed for TSO. Move the current code into a
separate function.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Differentiate between the cases where the skb is a large
send and the other cases.
Advertise TSO even if, at this stage, skb_gso_segment will
be called and it will do all the work.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Allow to configure the driver to pretend to have TX CSUM
offload support. This will be useful to test the TSO flows
that will come in further patches.
This configuration is disabled by default.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

ilw@linux.intel.com is not available anymore.
linuxwifi@intel.com should be used instead.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Move constants to the right of binary operators.

Generated by: scripts/coccinelle/misc/compare_const_fl.cocci

type=cleanup
Signed-off-by: NFengguang Wu <fengguang.wu@intel.com>
Signed-off-by: NJulia Lawall <julia.lawall@lip6.fr>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

RFC4862 states that "In all cases, a node MUST NOT respond to
a Neighbor Solicitation for a tentative address".
Currently the driver configures the NS offload and does not wait
for address to become permanent, thus violating the RFC.
Just removing the address from the address list is not good enough
for all cases, since the NS messages are needed for the duplicate
address detection and should not be discarded.

For d0i3 disable NS offload. Put tentative address in the address
list so the NS packet will not be filtered out by ucode.
For D3 the platform will not wake from NS packets - so enable
NS offload while removing the tentative address from the list.

Given that now NS offload might be disabled, and that the ucode
uses the IP data for other puroposes (L3 filtering) add two
independent flags indicating if IPv4\IPv6 data is valid.
Signed-off-by: NSara Sharon <sara.sharon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Today, in order to configure fw in usniffer mode, the ucode
must have the corresponding tlv, which is revealed to the driver
while parsing the ucode.

Expose the mode of the usniffer to other utilities in the driver
(other than the ucode parser) by passing back a pointer to the value.
This can be very useful for allowing configuring the fw dbg data
using an external configuration file, because this configuration
depends on the fw usniffer mode.
Signed-off-by: NGolan Ben-Ami <golan.ben.ami@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

The d0i3_mode variable is used to distinguish between transports that
handle d0i3 entry during suspend by themselves (i.e. the slave
transports) and those which rely on the op_mode layer to do it.  The
reason why the former do it by themselves is that they need to
transition from d0i3 in runtime_suspend into d0i3 in system-wide
suspend and this transition needs to happen before the op_mode's
suspend flow is called.

The wowlan_d0i3 element is also a bit confusing, because it just
reflects the wowlan->any value for the trans to understand.  This is a
bit unclear in the code and not generic enough for future use.

To make it clearer and to generalize the platform power mode settings,
introduce two variables to indicate the platform power management
modes used by the transport.

Additionally, in order not to take too big a step in one patch, treat
this new variables semantically in the same way as the old d0i3_mode
element, introducing a iwl_mvm_enter_d0i3_on_suspend() function to
help with that.

This commit also adds the foundation for a new concept where the
firmware configuration state (i.e. D0, D3 or D0i3) is abstracted from
the platform PM mode we are in (i.e. runtime suspend or system-wide
suspend).
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Remove the vif parameter of iwl_mvm_configure_bcast_filter()
as it's not being used.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

d0i3 commands are not supported in the init image, so take
a reference to ensure we don't enter d0i3 during init image,
and additional checks to prevent d0i3 commands when no
fw image is loaded.

Add a few WARN_ON_ONCE to the d0i3 enter/exit commands
to ensure we send d0i3 commands only when the normal
ucode is loaded.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When we have holes in the BA window, there might be frames
that have been ACKed between the read and the right
pointers. This means that these frames won't be scheduled
again by the SCD and the firwmare won't see them.
This invalidates the number of frames we tell the firmware
to send. When we detect this case, tell mac80211 to close
the SP and to send an EOSP so that the firmware can be in
sync.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Host commands now have a group id, express this in printed messages.
Signed-off-by: NSharon Dvir <sharon.dvir@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This function got too big to be inlined. Uninline it.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

We send an ADD_STA to instruct the firmware to release
frames despite the peer being in PS.
Since the ADD_STA command and the Tx frame that comes
immediately afterwards can be reordered by the DMA engine,
we need to block the Tx queues until the firmware replies
with the ADD_STA response.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

This allows the op_mode to request from the transport to
call a callback when an ASYNC commands is completed by
the firmware. The same callback will be called for all the
commands. Pass the command whose response triggers the
callback as a parameter to the callback itself.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

In certain flows (see next patches), the op_mode may need to
block the Tx queues for a short period. Provide an API for
that. The transport is in charge of counting the number of
times the queues are blocked since the op_mode may block the
queues several times in a row before unblocking them.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

commit 9a4c830007817e ("iwlwifi: mvm: refactor d3 key
update functions") refactored some code into
iwl_mvm_wowlan_config_key_params() function, but the
return value was never checked, and not all the function
flows returned valid values. fix it.

Fixes: ac8ef0ce ("iwlwifi: mvm: refactor d3 key update functions")
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

iwl_mvm_restart_cleanup() calls ieee80211_remain_on_channel_expired()
on cleanup, but it doesn't clean the actual roc time
events, resulting in failure of further ROC attempts.

Refactor iwl_mvm_stop_roc() a bit, and add a new function
to only cleanup the roc time events (without sending further
commands).
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Smatch prints a static checker warning here:

    drivers/net/wireless/intel/iwlwifi/mvm/fw-dbg.c:386 iwl_dump_prph()
    warn: curly braces intended?

Curly braces are NOT intended, the extra tab was added by mistake in
commit 1a616dd2 ('iwlwifi: dump prph registers in a common place
for all transports').

type=cleanup
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If the transport supports it, advertise NETIF_F_SG to mac80211 to
be able to use frag SKBs. This will already improve performance by
allowing software GSO to be used.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If the transport supports it, advertise NETIF_F_SG to mac80211 to
be able to use frag SKBs. This will already improve performance by
allowing software GSO to be used.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Klocwork pointed these out. There is a theoretical possibility
that rate->index might be set to IWL_RATE_INVALID (15).
This could trigger an out of bounds access on ht_vht_rates or
legacy_rates arrays. Fix it by adding some checks.
Signed-off-by: NEyal Shapira <eyalx.shapira@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Change scan configuration (dwell time, suspend time etc.) according
to traffic conditions. This is useful for scans that are managed by
the FW (e.g. scheduled scan).
Signed-off-by: NAvraham Stern <avraham.stern@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When sending the first ADD_STA HCMD for a STA, the %add_modify
field indicates an addition of a STA and not a modification
of one. In such a case, all fields of the HCMD are used to
initialize the corresponding fields in the FW, regardless of
what bits are set in %modify_mask.

Set the %tid_disable_tx field to mvm_sta->tid_disable_agg in
iwl_mvm_sta_send_to_fw(). If the STA is only updated this will
have no effect, but if it is added - it will make sure the
STA starts with the correct queues - if any - configured as
non-aggregated by default (until told otherwise).
Signed-off-by: NLiad Kaufman <liad.kaufman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Extend the configurable option of setting initial rate to RSSI based.
Make the initial rate to be set to VHT/HT SISO or legacy depending on
the AP capabilities.
Signed-off-by: NGregory Greenman <gregory.greenman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

There is no reason to keep a reference when the interface is down,
since we are not really doing anything.  The reference is only needed
when the mac80211 start op (or a hw restart) is running, to prevent
going into runtime or system supend in the meantime.  This will allow
us to support runtime PM when the interface is down (in another
patch).
Signed-off-by: NLuca Coelho <luciano.coelho@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

no need to have a separate debug level for a single
debug print (which is pretty much useless anyway).
remove them both.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When associated to an AP and a ROC event with a long duration is scheduled
the FW may have a hard time scheduling a consecutive time event, since it
has to remain on the connection channel to hear the AP's DTIM.
In addition, when associated and a ROC is requested with a duration
greater than the DTIM interval, the FW will not be able to schedule
the ROC event, since it needs to wake up for the DTIM.

Increasing the "max delay" duration to the DTIM period will allow the FW to
wait until after the DTIM and then schedule the ROC time event.
Limiting the ROC to be less than the DTIM interval will assure that the
time event will be scheduled for at least part of the time (instead of
automatically failing)

Extend the ROC max delay duration to min(dtim_interval * 3, 600TU),
and limit the duration to be less than the DTIM interval.
Signed-off-by: NMatti Gottlieb <matti.gottlieb@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Define the RX API that's used by the 9000 series hardware.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

When the wifi wakes up the system, we need to report it via calling
pm_wakeup_event for lucid sleep. This is so userspace knowns that the
wifi woke up the system via the /sys/power/wakeup_type sysfs interface.
Signed-off-by: NDerek Basehore <dbasehore@chromium.org>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

WARN_ON_ONCE() doesn't take a message, it only takes a condition.  I
have changed this to WARN(1, ...).
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

If the API or capabilities index is bigger than the driver expects,
an error message is printed. Make that message print the index and
distinguish between API and capabilities.
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

ilw@linux.intel.com is not available anymore.
linuxwifi@intel.com should be used instead.
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Fix the gaps between the system requirements and our code.
Signed-off-by: NAvri Altman <avri.altman@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Add a new struct for the 8165 series and a few new
PCI ID entries.
Signed-off-by: NOren Givon <oren.givon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Add a new struct for the 3168 series and a few new
PCI ID entries.
Signed-off-by: NOren Givon <oren.givon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

Add 9000-family configuration to iwl_cfg struct
Add a new struct to define the 5165 series.
Rename the struct that defines the 9000 series to 9260.
Add some new sub-system IDs for the 9260 and 5165 series.
For 9260:
0x0A10, 0x0000, 0x0510, 0x0710, 0x0410, 0x0610.
For 5165:
0x2A10, 0x2010, 0x0310, 0x0210.
Signed-off-by: NHaim Dreyfuss <haim.dreyfuss@intel.com>
Signed-off-by: NOren Givon <oren.givon@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>

We need to reuse the key update logic for d0i3
as well.

Add some parameters to deal with the constraints
implied by the d0i3 flow (specifically, support
non-SYNC commands, and don't take mutexes that
might deadlock).

Change some commands to be ASYNC, in order
to simplify locking a bit.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>