Situations observed when IP stack schedules lots of
frames for Tx while no connection (connection lost,
for example). In this case, dmesg bloated with error
message "FW not connected", printed for every frame.

Ratelimit this error message to avoid dmesg pollution.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

BAR frames delivered to the host via Rx path; whole BAR frame
get delivered. Advance sequence in the reorder buffer and release
old frames, as per IEEE802.11 spec.

Firmware will reply to BAR, driver responsibility is only reorder
buffer management.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

On Rx, when invalid frame is received and dropped,
reaping of next frames from Rx ring is stopped.

This stops NAPI polling and re-enables the Rx interrupt.

However, in cases where no more frames received,
interrupt will not be triggered and rest of Rx frames
will not be processed.

Skip bad frames and continue to reap Rx packets when
such frames are encountered, and add statistics for
such frames for debug.
Signed-off-by: NHamad Kadmany <qca_hkadmany@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Allow network stack part of Rx processing to run on separate core,
relaxing CPU utilization on the core used for Rx NAPI.

If RXHASH feature is enabled, the driver sets rxhash of each skb to 1
to enable RPS. The core for processing the rx skb is determined by RPS
mechanism according to rx_cpus bit mask which is configured at user level.
For processing skbs on different core from the core which processes
the interrupts, it is recommended not to enable core 0 in rx_cpus bit mask.
Signed-off-by: NVladimir Shulman <QCA_shulmanv@QCA.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Replace macros like "R", "W", "S", "C", defined multiple times,
with inline functions "wil_[rwsc]".

Use "readl" and "writel" instead of "ioread32" and "iowrite32"
since it is granted that memory transactions are used,
not port ones like IN/OUT
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Driver report supported TSO (v4 & v6) and IP checksum offload
in addition to previously supported features. In data path
skbs are checked for non-zero gso_size, and when detected sent
to additional function for processing TSO SKBs. Since HW does not
fully support TSO, additional effort is required from the driver.
Driver partitions the data into mss sized descriptors which are
then DMAed to the HW.
Signed-off-by: NVladimir Shulman <QCA_shulmanv@QCA.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Provide detailed statistics for the Rx frames per MCS
Statistics printed in "stations" debugfs entry
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Introduce 2 types of GTK, Tx (for this STA) and Rx (for each peer).
Now, AP has only Tx GTK, STA - only Rx one. PBSS not supported yet;
for it, continue using pseudo-DMS.
Handle per-vring .1x state, update it from WMI_VRING_EN_EVENTID
event. This allows unification for unicast and broadcast vrings.
This mechanism replaces former per-CID "data_port_open"
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Rx descriptor fields accordingly to the updated
hardware documentation
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Multicast is not yet properly supported for all connection types;
actually, only non-secure AP is supported. For all other cases,
fall back to old "pseudo-DMS" approach. Namely, for:
- PBSS
- secure connection

When re-routing MCAST Rx->Tx on the AP, do not Tx back to the origin
of the frame
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Use dedicated vring for multicast frames; this vring allocated for
AP and PBSS (both P2P GO and client) configurations

For short frames, use MCS0; for long - MCS1
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

For the networking code and for hardware network accelerators,
it is better to have IP header 4*n aligned. On the other side,
DMA on Rx path require buffer to be aligned on 4*n as well.
Having 14 bytes of Ethernet header, these 2 alignment
requests are in contradiction.

To solve this, order hardware offload block to not remove
SNAP header. This adds extra 6 bytes between addresses and
ethertype, making it 20 bytes total. This way, both buffer and
IP header are 4*n aligned. Remaining is only to remove SNAP
by shifting addresses 6 bytes. This involves data copying, so
this feature should be disabled unless required by the platform.

Module parameter "rx_align_2" (bool, default - false)
introduced to control this feature. Feature is completely disabled
when parameter is false.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Use temporal variable for often used vring->swhead;
and use proper index in debug printing - vring->swhead
used before was modified in wil_vring_advance_head
and then increased value was used in debug print
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

For the AP, configuration may say not to bridge traffic between
wireless clients. This is conveyed from user space (ex: hostapd has
ap_isolate parameter) with NL80211_CMD_SET_BSS, to the driver's
cfg80211 ops method change_bss

Add support for this setting.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

This is for AP only. If Rx data frame targeted to one of associated clients,
transmit it back to the wireless media and don't deliver to the host.
For the multicast frames, deliver to both host and wireless media.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Change threshold to be variable debugfs entry from hard-coded 0.
Default threshold value is 16 descriptors because HW is capable
of fetching up to 16 descriptors at once.
Signed-off-by: NVladimir Shulman <qca_shulmanv@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Mark expected branches using likely()/unlikely().
Do it on high performance route - data path and interrupts
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In the old hardware, bug existed that caused DA and SA for every
Rx packet to be swapped in the AP mode.
New hardware has fix for this bug. Enable this fix in the
hardware.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Print Tx descriptors
Print ring index for all Tx related messages
Sort prefixes: Tx for transmit, TxC for completion, added "D"
for descriptor related prints
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Use spinlock, this should not impact Tx as lock is always free
except for de-allocation.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Non-cacheable Tx descriptor for the last fragment of
multi-fragment frame used to be written back twice.

Refactor code to always write non-cacheable descriptor
only once
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In the Tx, work around used to force destination index 0
to be used. This is no more necessary, as firmware supports
multiple destinations
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Implement Tx status reporting using skb_complete_wifi_ack().
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Sync documentation for the Tx/Rx descriptors with the
firmware/hardware documentation.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In the reordering block, Ethernet DA was checked for MCAST, this is wrong.
Check instead MCAST indication from 802.11 MAC header. Hardware saves
this into Rx descriptor.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

In STA mode, all Tx should be directed to the same VRING towards the AP.
Thus, look up for the 1-st eligible VRING and use it.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

wil->status used as bitmap; use DECLARE_BITMAP for it.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

When configuring Tx/Rx VRING's, driver need to specify max. MPDU size
It should take into account all overhead introduced by 802.3->208.11
transformation.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

For manual ADDBA configuration, allow to set desired window size or
disable automatic mechanism.

Introduce module parameter (int) agg_wsize. It can be changed on run time,
will be taken into account on the next connect. Interpretation:
- <0 - disable automatic ADDBA; intended for manual testing through debugfs
- 0 - use automatically calculated window size
- >0 - use this for window size. Clipped by maximum supported by the hardware
with current environment.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Upon Tx vring creation, initiate BACK establishment
with maximum possible window size.

When establishing secure connection, there is EAPOL data exchange
between connection itself and "data port open", where security
is done and non-EAPOL data may be transferred. It is better to
send EAPOL frames using normal ACK because of firmware considerations.

send ADDBA only is 2 conditions met:
- data port open for the corresponded STA
- vring created
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Introduce BACK establishment procedures; decision logic is not implemented
yet; debugfs entry 'addba' used to manually trigger addba/delba for ringid 0.

debugfs usage:
to establish BACK with agg_wsize 16:
  echo 16 > /sys/kernel/debug/ieee80211/phy0/wil6210/addba
to delete BACK:
  echo 0 > /sys/kernel/debug/ieee80211/phy0/wil6210/addba
to change agg_wsize, one need to delete BACK and establish it anew

ADDBA flow for:

- originator

Tx side (initiator) sends WMI_VRING_BA_EN_CMDID providing
agg_wsize and timeout parameters.
Eventually, it gets event confirming BACK agreement - WMI_BA_STATUS_EVENTID
with negotiated parameters. On this event, update Tx vring data
(struct vring_tx_data) and display BACK parameters on debugfs

- recipient

Rx side (recipient) firmware informs driver about ADDBA with
WMI_RCP_ADDBA_REQ_EVENTID, driver process it in service work
queue wq_service. It adjusts parameters and sends response
with WMI_RCP_ADDBA_RESP_CMDID, and final confirmation provided
by firmware with WMI_ADDBA_RESP_SENT_EVENTID. In case of success,
driver updates Rx BACK reorder buffer.

policy for BACK parameters:
- aggregation size (agg_wsize * MPDUsize)) to not exceed 64Kbytes

DELBA flow for:

- originator

driver decides to terminate BACK, it sends WMI_VRING_BA_DIS_CMDID
and updates struct vring_tx_data associated with vring; ignore
WMI_DELBA_EVENTID.

- recipient

firmware informs driver with WMI_DELBA_EVENTID,
driver deletes correspondent reorder buffer

ADDBA request processing requires sending WMI command, therefore
it is processed in work queue context. Same work queue used as for
connect, it get renamed to wq_service
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@codeaurora.org>

Hardware doesn't place any restrictions on the buffer alignment,
consider this TODO resolved.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

In the wil_tx_ring, error message printed when tx attempted
while vring has no space to accommodate all fragments of frame.
Normally, such situation handled by stopping tx queue.
But, if tx queue is by-passed (like pktgen does), this error
will be triggered at high rate and dmesg will be flooded with
this message. Whole system may become unstable and hang with
no possible recover except power cycle.

Rate-limit it to prevent dmesg flooding.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Allow to configure VRING size for both Rx and Tx via module parameters:
rx_ring_order and tx_ring_order. Parameters are ring size orders, i.e.
ring size calculated as 1 << order.
Defaults for both Tx and Rx are order 9, i.e. size 512
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

It is possible to configure driver using mtu_max module parameter
by setting it to value in range of 68..7920 inclusive.
This is sub-optimal performance-wise in case packet is larger than 1 page.
mtu_max default value is 2228.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

HW supports upto 2304 packet size on the air.
HW is responsible for adding (Tx) or removing (Rx) the following headers:
802.11 hdr: 26B
SNAP: 8B
CRC: 4B
Security (optional): 24B
HW adds max 62B to the payload passed from driver. It means driver can use
max packet size of 2304-62 = 2242B
Signed-off-by: NDedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

added misc printouts in some init/uninit functions for better traceability
Signed-off-by: NDedy Lansky <qca_dlansky@qca.qualcomm.com>
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

- parentheses, indentation, typos
- seq_puts() instead of seq_printf() with single argument
- sizeof(var) vs. sizeof(type)
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>

When reading 'bf' file on debugfs, query beam forming status from firmware.
Ignore CID's that return error or return all zeros.

Remove obsolete code that used to maintain statistics on per-device basis,
as now it is reported be per-CID and current.
Signed-off-by: NVladimir Kondratiev <qca_vkondrat@qca.qualcomm.com>
Signed-off-by: NJohn W. Linville <linville@tuxdriver.com>