ath10k driver is using dma_pool_alloc per packet and dma_pool_free
in coresponding at Tx completion.
Use of pre-allocated DMA buffer in Tx will improve saving CPU resource
by 5% while it consumes about 56KB memory more as trade off.
Signed-off-by: NPeter Oh <poh@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

drivers/net/wireless/ath/ath10k/htt_tx.c:457: braces {} are not necessary for single statement blocks
drivers/net/wireless/ath/ath10k/htt_tx.c:545: braces {} are not necessary for single statement blocks
drivers/net/wireless/ath/ath10k/mac.c:200: braces {} are not necessary for single statement blocks
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In the case of raw mode without nohwcrypt parameter, we
should still make sure the frame is protected before
adding MIC_LEN to avoid skb_under_panic errors.
Signed-off-by: NBob Copeland <me@bobcopeland.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In a noisy environment, when multiple interfaces are created,
the management tx descriptors are fully occupied by the probe
responses from all the interfaces. This prevents a new station
from a successful association.

Fix this by limiting the probe responses when the specified
threshold limit is reached.
Signed-off-by: NVivek Natarajan <nataraja@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The function returns 1 when DMA mapping fails. The
driver would return bogus values and could
possibly confuse itself if DMA failed.

Fixes: 767d34fc ("ath10k: remove DMA mapping wrappers")
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Initial QCA99X0 support has a known issue with TCP Tx throughput.
All other path such as UDP Tx/Rx and TCP Rx meet their expectation
(> 900Mbps), but TCP Tx marked as low as 5Mbps when single pair is
used on iperf.

The root cause is turned out because TSO flag is not initialized
properly so that firmware configures TSO in wrong way.
TSO flags in msdu extension descriptor is required to be reset
to indicate firmware there is no TSO is enabled, otherwise it
could act as TSO is enabled which causes huge throughput drop.

In fact, it's enough by resetting TSO flags only to prevent the
unexpected behavior, but initializing whole msdu ext. descriptor
will help to clear uncertainty of firmware could bring on as it
constantly updated.
Signed-off-by: NPeter Oh <poh@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This patch enables raw Rx/Tx encap mode to support software based
crypto engine. This patch introduces a new module param 'cryptmode'.

 cryptmode:

   0: Use hardware crypto engine globally with native Wi-Fi mode TX/RX
      encapsulation to the firmware. This is the default mode.
   1: Use sofware crypto engine globally with raw mode TX/RX
      encapsulation to the firmware.

Known limitation:
   A-MSDU must be disabled for RAW Tx encap mode to perform well when
   heavy traffic is applied.

Testing: (by Michal Kazior <michal.kazior@tieto.com>)

     a) Performance Testing

      cryptmode=1
       ap=qca988x sta=killer1525
        killer1525  ->  qca988x     194.496 mbps [tcp1 ip4]
        killer1525  ->  qca988x     238.309 mbps [tcp5 ip4]
        killer1525  ->  qca988x     266.958 mbps [udp1 ip4]
        killer1525  ->  qca988x     477.468 mbps [udp5 ip4]
        qca988x     ->  killer1525  301.378 mbps [tcp1 ip4]
        qca988x     ->  killer1525  297.949 mbps [tcp5 ip4]
        qca988x     ->  killer1525  331.351 mbps [udp1 ip4]
        qca988x     ->  killer1525  371.528 mbps [udp5 ip4]
       ap=killer1525 sta=qca988x
        qca988x     ->  killer1525  331.447 mbps [tcp1 ip4]
        qca988x     ->  killer1525  328.783 mbps [tcp5 ip4]
        qca988x     ->  killer1525  375.309 mbps [udp1 ip4]
        qca988x     ->  killer1525  403.379 mbps [udp5 ip4]
        killer1525  ->  qca988x     203.689 mbps [tcp1 ip4]
        killer1525  ->  qca988x     222.339 mbps [tcp5 ip4]
        killer1525  ->  qca988x     264.199 mbps [udp1 ip4]
        killer1525  ->  qca988x     479.371 mbps [udp5 ip4]

      Note:
       - only open network tested for RAW vs nwifi performance comparison
       - killer1525 (qca6174 hw2.2) is 2x2 device (hence max 866mbps)
       - used iperf
       - OTA, devices a few cm apart from each other, no shielding
       - tcpX/udpX, X - means number of threads used

      Overview:
       - relative Tx performance drop is seen but is within reasonable and
         expected threshold (A-MSDU must be disabled with RAW Tx)

     b) Connectivity Testing

      cryptmode=1
       ap=iwl6205 sta1=qca988x crypto=open     topology-1ap1sta          OK
       ap=iwl6205 sta1=qca988x crypto=wep1     topology-1ap1sta          OK
       ap=iwl6205 sta1=qca988x crypto=wpa      topology-1ap1sta          OK
       ap=iwl6205 sta1=qca988x crypto=wpa-ccmp topology-1ap1sta          OK
       ap=qca988x sta1=iwl6205 crypto=open     topology-1ap1sta          OK
       ap=qca988x sta1=iwl6205 crypto=wep1     topology-1ap1sta          OK
       ap=qca988x sta1=iwl6205 crypto=wpa      topology-1ap1sta          OK
       ap=qca988x sta1=iwl6205 crypto=wpa-ccmp topology-1ap1sta          OK
       ap=iwl6205 sta1=qca988x crypto=open     topology-1ap1sta2br       OK
       ap=iwl6205 sta1=qca988x crypto=wep1     topology-1ap1sta2br       OK
       ap=iwl6205 sta1=qca988x crypto=wpa      topology-1ap1sta2br       OK
       ap=iwl6205 sta1=qca988x crypto=wpa-ccmp topology-1ap1sta2br       OK
       ap=qca988x sta1=iwl6205 crypto=open     topology-1ap1sta2br       OK
       ap=qca988x sta1=iwl6205 crypto=wep1     topology-1ap1sta2br       OK
       ap=qca988x sta1=iwl6205 crypto=wpa      topology-1ap1sta2br       OK
       ap=qca988x sta1=iwl6205 crypto=wpa-ccmp topology-1ap1sta2br       OK
       ap=iwl6205 sta1=qca988x crypto=open     topology-1ap1sta2br1vlan  OK
       ap=iwl6205 sta1=qca988x crypto=wep1     topology-1ap1sta2br1vlan  OK
       ap=iwl6205 sta1=qca988x crypto=wpa      topology-1ap1sta2br1vlan  OK
       ap=iwl6205 sta1=qca988x crypto=wpa-ccmp topology-1ap1sta2br1vlan  OK
       ap=qca988x sta1=iwl6205 crypto=open     topology-1ap1sta2br1vlan  OK
       ap=qca988x sta1=iwl6205 crypto=wep1     topology-1ap1sta2br1vlan  OK
       ap=qca988x sta1=iwl6205 crypto=wpa      topology-1ap1sta2br1vlan  OK
       ap=qca988x sta1=iwl6205 crypto=wpa-ccmp topology-1ap1sta2br1vlan  OK

      Note:
       - each test takes all possible endpoint pairs and pings
       - each pair-ping flushes arp table
       - ip6 is used

     c) Testbed Topology:

      1ap1sta:
        [ap] ---- [sta]

        endpoints: ap, sta

      1ap1sta2br:
        [veth0] [ap] ---- [sta] [veth2]
           |     |          |     |
        [veth1]  |          \   [veth3]
            \   /            \  /
            [br0]            [br1]

        endpoints: veth0, veth2, br0, br1
        note: STA works in 4addr mode, AP has wds_sta=1

      1ap1sta2br1vlan:
        [veth0] [ap] ---- [sta] [veth2]
           |     |          |     |
        [veth1]  |          \   [veth3]
            \   /            \  /
          [br0]              [br1]
            |                  |
          [vlan0_id2]        [vlan1_id2]

        endpoints: vlan0_id2, vlan1_id2
        note: STA works in 4addr mode, AP has wds_sta=1

Credits:

    Thanks to Michal Kazior <michal.kazior@tieto.com> who helped find the
    amsdu issue, contributed a workaround (already squashed into this
    patch), and contributed the throughput and connectivity tests results.
Signed-off-by: NDavid Liu <cfliu.tw@gmail.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Tested-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

During tx completion, tx_lock is held for longer than required, preventing
efficient refill of htt->pending_tx. Refactor the code so that only MSDU
related operations are protected by the lock.

Improves downstream performance on a dual-core ARM Freescale LS1024A
(f.k.a. Mindspeed Comcerto 2000) AP with a 3x3 client from 495 to 580 Mbps.
Other CPU bound multicore systems may also benefit.
Signed-off-by: NDenton Gentry <dgentry@google.com>
Signed-off-by: NAvery Pennarun <apenwarr@google.com>
[mfaltesek@google.com: removed conflicting code for tracking msdu_ids.]
Signed-off-by: NMarty Faltesek <mfaltesek@google.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

HTT_H2T_MSG_TYPE_MGMT_TX msg in 10.4 firmware carries additional
4 byte in htt_mgmt_tx_desc where it tells to firmware that at what
rate mgmt frame has to go out in the air. It's an optional parameter,
setting this field to zero will force firmware to choose auto rate
and send the frame out.

Those 4 byte info is missed out in the current code and 10.4 firmware
ended up reading some junk in those 4 byte and sometime malfunctioning.

Fix it by adding 4 byte in struct htt_mgmt_tx_desc. Non 10.4 firmware
will not process those four byte. So, adding 4 byte at the end of
struct htt_mgmt_tx_desc will not create any impact on other chipset.
Signed-off-by: NRaja Mani <rmani@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The patch adds support to offload TCP/UDP checksum
calculations for QCA99x0.
Signed-off-by: NManikanta Pubbisetty <c_mpubbi@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Since QCA99X0 uses fragmentation descriptor differently from
other ones on tx path, we need to handle it separately.

QCA99X0 is using 48 bits for address and 16 bits for length
out of 2 dword and each values have to be programmed by frag
desc base addr + msdu id, so that hardware can retrieve
corresponding frag data.
Signed-off-by: NPeter Oh <poh@qca.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Pre qca99X0 chipsets follows the model where dynamically allocate
memory for frag desc on getting new skb for TX. But, this is not
going to be the case in qca99X0. It expects frag desc memory to be
allocated at boot time and let the driver to reuse allocated memory
after every TX completion. So there won't be any dynamic frag memory
memory allocation in qca99X0 during data transmission.

qca99X0 hardware doesn't need fragment desc address to be programmed
in msdu descriptor for every data transaction. It needs to know only
starting address of fragment descriptor at the time of the boot.
During data transmission, qca99X0 hardware can retrieve corresponding
frag addr by adding programmed frag desc base addr + msdu id.

Allocate continuous fragment descriptor memory (same size as number of
descriptor) at the time of target initialization and configure allocated
dma address to the target via HTT_H2T_MSG_TYPE_FRAG_DESC_BANK_CFG.

How this is allocated continuous memory is going to be used is not
covered in this patch. It just allocates memory and hand over to firmware.
If we don't do it at init time, qca99X0 will stall when firmware tries
to do TX.
Signed-off-by: NRaja Mani <rmani@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Tx queue locking was very simple until now.
Multi-channel support will require a more flexible
and fine grained control.

This introduces a per-hw and per-vif (each with a
bitmask of reasons) tx queue locking.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There are a few different tx paths depending on
firmware and frame itself.

Creating a uniform decision will make it possible
to switch between different txmode easier, both
for testing and for future features as well.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NMarek Puzyniak <marek.puzyniak@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Otherwise ath10k will just checksum everything even if it did not
go through the TCP/IP stack (for example bridged frames). In the worst
case this could mean recreating the checksum for incorrect data.
Signed-off-by: NHelmut Schaa <helmut.schaa@googlemail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

New wmi-tlv firmware uses HTT 3.0 protocol which
uses TX_FRM command for management frames (instead
of a dedicated command). To support PMF it is
necessary to provide explicit tailroom.
Signed-off-by: NMarek Kwaczynski <marek.kwaczynski@tieto.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

HTT Tx protocol uses arbitrary host assigned ids
too associate with MSDUs when delivering
completions.

Instead of rolling out own id generation scheme
use the tools provided in kernel.

This should have little to no effect on
performance.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Return a negative error code on failure.

A simplified version of the semantic match that finds this problem is as
follows: (http://coccinelle.lip6.fr/)

// <smpl>
@@
identifier ret; expression e1,e2;
@@
(
if (\(ret < 0\|ret != 0\))
 { ... return ret; }
|
ret = 0
)
... when != ret = e1
    when != &ret
*if(...)
{
  ... when != ret = e2
      when forall
 return ret;
}
// </smpl>
Signed-off-by: NJulia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Better to have this in same place as other firmware interface handling.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

New firmware revisions don't need peer creation
when doing offchannel tx. Earlier revisions would
queue and never release frames without a peer.

This prevent new firmware revisions from stopping
replenishing wmi-htc tx credits and improves
reliability of offchannel tx which would sometimes
silently fail.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.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>

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>

Add tracing support to forward management and data frames to
user space for packet inspection.
Signed-off-by: NRajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The tx info such as msdu_id, frame len, vdev id and tid are reported
to user space by tracepoint. This is useful for collecting tx
statistics.
Signed-off-by: NRajkumar Manoharan <rmanohar@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fixes a checkpatch warning:

WARNING: void function return statements are not generally useful
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fixes checkpatch warnings:

CHECK: Please don't use multiple blank lines
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This makes it a lot easier to log and debug
messages if there's more than 1 ath10k device on a
system.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Firmware could request inspection of some
submitted tx requests. Since the callback wasn't
implemented it was possible to bleed tx msdu_ids
which could translate to tx flushing timeouts.

There's nothing ath10k can do to help firmware
with tx processing now so just report all tx
frames as already inspected to prevent firmware
from sending up inspection events and force it to
report regular tx completion indications with
discard status.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Allow to setup maximum subframes for AMSDU and AMPDU aggregation
via debugfs htt_max_amsdu_ampdu file.

Eg.
echo "2 64" > htt_max_amsdu_ampdu
will setup maximum amsdu subframes equal 2 and
maximum ampdu subframes equal to 64.
Signed-off-by: NJanusz Dziedzic <janusz.dziedzic@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It was possible to call hif_stop() 2 times through
ath10k_htc_connect_init() timeout failpath which
could lead to double free_irq() kernel splat for
multiple MSI interrupt case.

Re-order init sequence to avoid this problem. The
HTC stop shouldn't stop HIF implicitly since it
doesn't implicitly start it. Since the re-ordering
required some functions to be split/removed/renamed
rename a few functions to make more sense while at
it.
Reported-By: NBen Greear <greearb@candelatech.com>
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

It is inefficient to grab irqsave spinlocks for
skb lists for each queue/dequeue action.

Using rx_ring.lock and tx_lock allows to use less
heavy bh spinlock functions and moving locking
upwards allows to toggle spinlocks less often.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Going through full htc tx path for htt tx is a
waste of resources. By skipping it it's possible
to easily submit scatter-gather to the pci hif for
reduced host cpu load and improved performance.

The new approach uses dma pool to store the
following metadata for each tx request:
 * msdu fragment list
 * htc header
 * htt tx command

The htt tx command contains a msdu prefetch.
Instead of copying it original mapped msdu address
is used to submit a second scatter-gather item to
hif to make a complete htt tx command.

The htt tx command itself hands over dma mapped
pointers to msdus and completion of the command
itself doesn't mean the frame has been sent and
can be unmapped/freed. This is why htc tx
completion is skipped for htt tx as all tx related
resources are freed upon htt tx completion
indication event (which also implicitly means htt
tx command itself was completed).

Since now each htt tx request effectively consists
of 2 copy engine items CE_HTT_H2T_MSG_SRC_NENTRIES
is updated to allow maximum of
TARGET_10X_NUM_MSDU_DESC msdus being queued. This
keeps the tx path resource management simple.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There's no real benefit from using them. DMA-API
already provides debugging. Some skbuffs are
already mapped directly with DMA-API since wrapper
arguments were insufficient and extending them
would be pointless.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Make it easier to grep for htt rx errors.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

10.1.389 firmware has some differences in
calculation of number of outstanding HTT TX
completions. This led to FW crashes of 10.1.389
while main firmware branch was unnaffected.

The patch makes sure ath10k doesn't queue up more
MSDUs than it should.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This is still the only way to submit mgmt frames in case
of 10.X firmware.

This patch introduces wmi_mgmt_tx queue, because of the
fact WMI command can block. This is a problem for
ath10k_tx_htt(), since it's called from atomic context.
The skb queue and worker are introduced to move the mgmt
frame handling out of .tx callback context and not block.
Signed-off-by: NBartosz Markowski <bartosz.markowski@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Instead of allocating sk_buff for a mere 16-byte
tx fragment list buffer use headroom of the
original msdu sk_buff.

This decreases CPU cache pressure and improves
performance.

Measured improvement on AP135 is 560mbps ->
590mbps of UDP TX briding traffic.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Use a saner goto scheme for failure handling. Also
group operations more sensibly.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Until now the all MSDU transfer related structures
were freed when all resources were unreferenced.

Now HTC transfer is freed independently and HTT
transfer is so too.

This yields a way more simpler ath10k_skb_cb and
should possibly enable parallel pipe processing
(which is now serialized in
ath10k_pci_process_ce routine).
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This reduces number of memory accesses and
hopefully contributes to better performance in the
future.
Signed-off-by: NMichal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>