Comment was out-of-date. The headroom is no longer
necessary because HTT Tx fragment list is stored
in dma pool item associated with each sk_buff.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

When re-associating a station, the nss was set back to
maximum value even if user had configured small number
of tx chains.  So, pay attention to user's config in
this case as well.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It appears it takes more than just setting the
hardware's chainmask to make things work well.  Without
this patch, a vdev would only use 1x1 rates when chainmask
was set to 0x3.

Setting the 'nss' (number of spatial streams) on the vdev
helps the firmware's rate-control algorithm work properly.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There are not many of these messages producted by the
firmware, but they are generally fairly useful, so make it easy to print them
with a separate debug level.

kvalo: fix commit log, rename debug level
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There are better ways to get the kernel information, use the
utsname and omit the version code entirely since it's duplicate.
The version magic is rather useless anyway
Signed-off-by: NJohannes Berg <johannes@sipsolutions.net>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fragmented Rx wasn't reporting everything. With
the reworked Rx code it's very easy to fix it.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This makes it more in line with the new Rx path.
It also makes the code more reusable because Rx
descriptor is more accessible.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The original fix has been moved into a different
place in the code.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This creates a single, common path for MSDU,
A-MSDU and fragmented Rx.

Hopefully this will make it easier to understand
Rx path and make it easier to work with.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Make the rx_confused be handled by the rx
indication handlers instead of the buffer popping
function.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Since htt_rx_mpdu_status isn't used anymore
(instead attention flags are used) simplify the
loop.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Instead of using manual sk_buff linking via ->next
use sk_buff_head. It's more robust, cleaner and
there's plenty of helper functions in kernel
already to manage sk_buff_head.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fix to use v10.2 wmi call for firmware v10.2.
It turned out that peer association function was using
v10.1 wmi call for v10.2 firmware during code review.
Signed-off-by: NPeter Oh <poh@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This will enable AP mode to change channel width dynamically
based on 20/40 intolerance report sent by associated client.
Signed-off-by: NRajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This patch would help bring up wifi interface with default board
data in case of failures in otp download. It is useful for initial
calibration.
Signed-off-by: NRajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

For packet log, the transmitted frame 802.11 header alone is sufficient.
Recording entire packet is also consuming lot of disk space. To optimize
this, tx and rx data tracepoints are splitted into header and payload
tracepoints.

To record tx ieee80211 headers

     trace-cmd record -e ath10k_tx_hdr

To record complete packets

     trace-cmd record -e ath10k_tx_hdr -e ath10k_tx_payload

Cc: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: NRajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Drivers might want to know also when mac80211 has
completed reconfiguring after resume (e.g. in order
to know when frames can be passed to mac80211).

Rename restart_complete() to a more-generic reconfig_complete(),
and add a new enum to indicate the reconfiguration type.

Update the current users with the new prototype.
Signed-off-by: NEliad Peller <eliadx.peller@intel.com>
Signed-off-by: NEmmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>

Firmware was crashing when we were trying to warm reset it
after suspend. This was due to the fact that target registeres
can be accessed only if the hardware is awaken.

This patch makes sure to awake the device also on the hif up,
not only in case of probe call.
Signed-off-by: NBartosz Markowski <bartosz.markowski@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In some cases hw recovery was taking an absurdly
long time due to ath10k waiting for things that
would never really complete.

Instead of waiting for inevitable timeouts poke
all completions and wakequeues and check if it's
still worth waiting.

Reading/writing ar->state requires conf_mutex.
Since waiters might be holding it introduce a new
flag CRASH_FLUSH so it's possible to tell waiters
to abort whatever they were waiting for.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Until now it was possible to simulate soft and
hard fw crashes but it wasn't possible to trigger
an immediately hw restart itself (without the fw
crash).

This can be useful when stress testing hw
restarting stability, e.g. during heavy tx/rx
traffic.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

While testing other things I've found that CE
items aren't cleared properly. This could lead to
null dereferences in BMI.

To prevent that make sure CE revoking clears the
nbytes value (which is used as a buffer completion
indication) and memset the entire CE ring data
shared between host and target when
(re)initializing.

Also make sure to check BMI xfer pointer and print
a splat instead of crashing the kernel.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Currently hif_power_up performs effectively a
reset and hif_stop resets the chip as well so
there's no point in resetting here.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The power up procedure was overly complex due to
warm/cold reset workarounds and issues.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

One of the problems with warm reset I've found is
that it must be guaranteed that copy engine
registers are not being accessed while being
reset. Otherwise in worst case scenario the host
may lock up.

Instead of using sleeps and hoping the device is
operational in some arbitrary timeframes use
firmware indication register.

As a side effect this makes driver
boot/stop/recovery faster.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Make ath10k_pci_init_pipes() effectively only
alter shared target-host data.

The per_transfer_context is a host-only thing.
It is necessary to preserve it's contents for a
more robust ring cleanup.

This is required for future warm reset fixes.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This should prevent deadlock predicted by the
following splat:

 ======================================================
 [ INFO: possible circular locking dependency detected ]
 3.17.0-wl-ath+ #67 Not tainted
 -------------------------------------------------------
 kworker/u32:1/7230 is trying to acquire lock:
  (&ar->conf_mutex){+.+.+.}, at: [<ffffffffa040a57d>] ath10k_scan_timeout_work+0x2d/0x50 [ath10k_core]

 but task is already holding lock:
  ((&(&ar->scan.timeout)->work)){+.+...}, at: [<ffffffff8106dae1>] process_one_work+0x151/0x470

 which lock already depends on the new lock.

 the existing dependency chain (in reverse order) is:

 -> #1 ((&(&ar->scan.timeout)->work)){+.+...}:
        [<ffffffff810a12e5>] lock_acquire+0x85/0x100
        [<ffffffff8106cb4d>] flush_work+0x3d/0x270
        [<ffffffff8106e49d>] __cancel_work_timer+0x7d/0x110
        [<ffffffff8106e543>] cancel_delayed_work_sync+0x13/0x20
        [<ffffffffa0409f16>] ath10k_cancel_remain_on_channel+0x36/0x60 [ath10k_core]
        [<ffffffffa028c75c>] ieee80211_cancel_roc+0x1cc/0x2f0 [mac80211]
        [<ffffffffa028c8a2>] ieee80211_mgmt_tx_cancel_wait+0x22/0x30 [mac80211]
        [<ffffffffa0132288>] nl80211_tx_mgmt_cancel_wait+0xa8/0x130 [cfg80211]
        [<ffffffff816654a5>] genl_family_rcv_msg+0x1a5/0x3c0
        [<ffffffff81665749>] genl_rcv_msg+0x89/0xc0
        [<ffffffff81664e91>] netlink_rcv_skb+0xb1/0xc0
        [<ffffffff816650bc>] genl_rcv+0x2c/0x40
        [<ffffffff8166474d>] netlink_unicast+0x18d/0x200
        [<ffffffff81664add>] netlink_sendmsg+0x31d/0x430
        [<ffffffff8161a9ac>] sock_sendmsg+0x9c/0xd0
        [<ffffffff8161b469>] ___sys_sendmsg+0x389/0x3a0
        [<ffffffff8161bed9>] __sys_sendmsg+0x49/0x90
        [<ffffffff8161bf32>] SyS_sendmsg+0x12/0x20
        [<ffffffff8174c456>] system_call_fastpath+0x1a/0x1f

 -> #0 (&ar->conf_mutex){+.+.+.}:
        [<ffffffff810a0bde>] __lock_acquire+0x1b6e/0x1ce0
        [<ffffffff810a12e5>] lock_acquire+0x85/0x100
        [<ffffffff817491eb>] mutex_lock_nested+0x4b/0x370
        [<ffffffffa040a57d>] ath10k_scan_timeout_work+0x2d/0x50 [ath10k_core]
        [<ffffffff8106db41>] process_one_work+0x1b1/0x470
        [<ffffffff8106df63>] worker_thread+0x123/0x460
        [<ffffffff81073f34>] kthread+0xe4/0x100
        [<ffffffff8174c3ac>] ret_from_fork+0x7c/0xb0

 other info that might help us debug this:

  Possible unsafe locking scenario:

        CPU0                    CPU1
        ----                    ----
   lock((&(&ar->scan.timeout)->work));
                                lock(&ar->conf_mutex);
                                lock((&(&ar->scan.timeout)->work));
   lock(&ar->conf_mutex);

  *** DEADLOCK ***
Reported-by: NMarek Puzyniak <marek.puzyniak@tieto.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Rts threshold was being configured instead of
fragmentation threshold.

Keep in mind available firmware binaries don't
seem to support fragmentation anyway so this
doesn't fix fragmentation threshold per se.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

When user space requests survey info, it is useful to know which of the survey
data refers to the channel that is currently actively being used.  One of the
use cases is getting the current channel noise for status output.  Without this
flag you would have to look up the channel separately and then compare it
against the frequency in the survey output in user space.
Signed-off-by: NFelix Fietkau <nbd@openwrt.org>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Instead of defining a completely new tracepoint
use an existing tracepoint class.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fundamentally this was wrong. Tsf is only valid
in last MPDU of a PPDU. This means tsf value was
wrong most of the time during heavy traffic.

Also I don't see much point in exposing a
redundant (and broken) tsf value. Userspace can
already read it from the dumped rx descriptor
buffer.

Cc: Rajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Once driver entered the rx_confused state it would
refuse to rx even after firmware is restarted.
Make sure to clear it so that rx works after, e.g.
hw restart or after all interfaces are stopped.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In theory it was possible to drain entire HTT Rx
ring via fragmented Rx leading to Rx lockup.

In practice non-data traffic would always trigger
replenishment via the regular Rx handler.

For correctness sake make sure to replenish the
ring on fragmented Rx.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The netbuf pop can return NULL. Make sure to check
for that. It shouldn't happen but better safe than
sorry.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There's little point in dropping, e.g. frames with
FCS error early in ath10k.

This simplifies amsdu_allowed() and gets rid of
htt_rx_mpdu_status usage finally.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Treat non-chained and chained popping the same
way. Also this makes netbuf pop fully symmetrical
to (re)filling.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Tracing function was called before buffers were
unmapped from DMA.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Use the globally defined ieee80211 values instead
of re-defining them in the driver again.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The rx descriptor variable was no longer used in
the rx handler.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

HTT_RX_IND_MPDU_STATUS_MGMT_CTRL was pretty greedy
and because of that ath10k ended up dropping
Control Frames as well as Null Func frames.
Reported-by: NOkhwan Lee <ohlee@mwnl.snu.ac.kr>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Calling init to reinit ce pipe state would also
re-set all static structure links and setting
(which don't change over driver lifecycle).

Make it so alloc links structures and initializes
static data and init part to setup state
variables and clear stuff.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>