For devices where the FW supports WoWLAN but user-space has not
configured it, we don't do any PCI-specific suspend/resume operations,
because mac80211 doesn't call drv_suspend() when !wowlan. This has
particularly bad effects for some platforms, because we don't stop the
power-save timer, and if this timer goes off after the PCI controller
has suspended the link, Bad Things will happen.

Commit 32faa3f0 ("ath10k: add the PCI PM core suspend/resume ops")
got some of this right, in that it understood there was a problem on
non-WoWLAN firmware. But it forgot the $subject case.

Fix this by moving all the PCI driver suspend/resume logic exclusively
into the driver PM hooks. This shouldn't affect WoWLAN support much
(this just gets executed later on).

I would just as well kill the entirety of ath10k_hif_suspend(), as it's
not even implemented on the USB or SDIO drivers. I expect that we don't
need the callback, except to return "supported" (i.e., 0) or "not
supported" (i.e., -EOPNOTSUPP).

Fixes: 32faa3f0 ("ath10k: add the PCI PM core suspend/resume ops")
Fixes: 77258d40 ("ath10k: enable pci soc powersaving")
Signed-off-by: NBrian Norris <briannorris@chromium.org>
Cc: Ryan Hsu <ryanhsu@qti.qualcomm.com>
Cc: Kalle Valo <kvalo@qca.qualcomm.com>
Cc: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Make them const as they are not modified in the file referencing
them. They are only stored in the const field 'hw_ce_reg' of an ath10k
structure. Also, make the declarations in the header const.
Signed-off-by: NBhumika Goyal <bhumirks@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

ACPI will rely on device driver to tell it if the device could support
wakeup function when system in D3 state.

This has caused some platform can't support remote wakeup correctly,
because the ACPI wakeup GPE is not enabled, hence registers the .set_wakeup
callback to handle it if device supports wakeup.

Tested with QCA6174 hw3.0, firmware ('WLAN.RM.4.4.1-00008-QCARMSWP-1')
Signed-off-by: NRyan Hsu <ryanhsu@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The actual PCI suspend/resume in ath10k has been handled in wow.c,
but in the case of the device doesn't support remote wakeup,
the .hif_suspend() and .hif_resume() will never be handled.

  ath10k_wow_op_suspend()
  {
	if (WARN_ON(!test_bit(ATH10K_FW_FEATURE_WOWLAN_SUPPORT,
		    ar->running_fw->fw_file.fw_features))) {
		ret = 1;
		goto exit;
	}

	....

	ret = ath10k_hif_suspend(ar);
  }

So register the PCI PM core to support the suspend/resume if the device
doesn't support remote wakeup.
Signed-off-by: NRyan Hsu <ryanhsu@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In commit 9f5bcfe9 ("ath10k: silence firmware file probing
warnings") the firmware loading was changed from request_firmware() to
request_firmware_direct() to silence some warnings in case it fails.
request_firmware_direct() directly searches in the file system only and
does not send a hotplug event to user space in case it could not find
the firmware directly.
In LEDE we use a user space script to extract the calibration data from
the flash memory which gets triggered by the hotplug event. This way the
firmware gets extracted from some vendor specific partition when the
driver requests this firmware. This mechanism does not work any more
after this change.

Fixes: 9f5bcfe9 ("ath10k: silence firmware file probing warnings")
Signed-off-by: NHauke Mehrtens <hauke@hauke-m.de>
Cc: Michal Kazior <michal.kazior@tieto.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

irq_wq in struct ath10k_sdio is a remnant from an earlier
version of the sdio patchset.

Its use was removed as a result of Kalle's review, but somehow
the struct member survived.

It is not used and can therefore safely be removed.
Signed-off-by: NErik Stromdahl <erik.stromdahl@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

In napi_poll, the budget number is used to control the amount of packets
we should handle per poll to balance the resource in the system.

In the list of the amsdu packets reception, we check if there is budget
count left and handle the complete list of the packets, that it will have
chances the very last list will over the budget leftover.

So adding one more parameter - budget_left, this would help while
traversing the list to avoid handling more than the budget given.
Reported-by: NAndrey Ryabinin <aryabinin@virtuozzo.com>
Fix-suggested-by: NIgor Mitsyanko <igor.mitsyanko.os@quantenna.com>
Link: https://lkml.kernel.org/r/26670dce-4dd2-f8e4-0e14-90d74257e739@virtuozzo.comSigned-off-by: NRyan Hsu <ryanhsu@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

There are new types and helpers that are supposed to be used in new code.

As a preparation to get rid of legacy types and API functions do
the conversion here.
Signed-off-by: NAndy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The rx ring buffers are added to a hash table if
firmware support full rx reorder. If the full rx
reorder support flag is not set before allocating
the rx ring buffers, none of the buffers are added
to the hash table.

There is a race condition between rx ring refill and
rx buffer replenish from napi poll. The interrupts are
enabled in hif start, before the rx ring is refilled during init.
We replenish buffers from napi poll due to the interrupts which
get enabled after hif start. Hence before the entire rx ring is
refilled during the init, the napi poll replenishes a few buffers
in steps of 100 buffers per attempt. During this rx ring replenish
from napi poll, the rx reorder flag has not been set due to which
the replenished buffers are not added to the hash table

Set the rx full reorder support flag before we allocate
the rx ring buffer to avoid the memory leak.
Signed-off-by: NRakesh Pillai <pillair@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Each time we get disconnected from AP we get flooded with messages like:

...
ath10k_pci 0000:03:00.0: no channel configured; ignoring frame(s)!
<until ratelimit kicks in>
ath10k_warn: 155 callbacks suppressed
...

Use ath10k_dbg() here too.
Signed-off-by: NGabriel Craciunescu <nix.or.die@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Commit a53e35db ("reset: Ensure drivers are explicit when requesting
reset lines") started to transition the reset control request API calls
to explicitly state whether the driver needs exclusive or shared reset
control behavior. Convert all drivers requesting exclusive resets to the
explicit API call so the temporary transition helpers can be removed.

No functional changes.

Cc: Kalle Valo <kvalo@qca.qualcomm.com>
Cc: ath10k@lists.infradead.org
Cc: linux-wireless@vger.kernel.org
Cc: netdev@vger.kernel.org
Signed-off-by: NPhilipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

WMI interface for all the firmwares(except QCA6174) does not include the
type of peer(default/bss/tdls) requested during peer creation, therefore
target creates a default peer.

TDLS implementation on 10.4 firmware requires host to configure the
peer type(tdls) for TDLS peers. This patch adds peer type parameter to the
existing WMI interface for peer creation to accommodate this requirement.

Tested this change on QCA9888(10.4-3.5.1-00018) and QCA988x(10.2.4.70.9-2)
with ping tests for AP/STA modes.
Signed-off-by: NManikanta Pubbisetty <mpubbise@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This patch adds the support of TDLS feature for 10.4 firmware
versions.

A new WMI service is added to advertise the support of TDLS for
10.4 firmwares.
Signed-off-by: NManikanta Pubbisetty <mpubbise@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Though there is room to accommodate 512 services in wmi service
ready event, target uses only first 4-bits of each 32-bit word for
advertising wmi services thereby limiting max wmi services to 64.

TDLS implementation for 10.4 firmwares introduces new wmi services by
making use of remaining unused bits of each 32-bit word, therefore the
wmi service mapping in host needs to be extended.

This patch adds the logic to extend the wmi SVCMAP to accommodate new
wmi services.
Signed-off-by: NManikanta Pubbisetty <mpubbise@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

As suggested by Arnd Bergmann, replace
"while (time_before_...) {}"
with
"do {} while (time_before_...)"

This fixes the following warnings detected by gcc 4.1.2:

drivers/net/wireless/ath/ath10k/sdio.c: In function
‘ath10k_sdio_mbox_rxmsg_pending_handler’:
drivers/net/wireless/ath/ath10k/sdio.c:676: warning: ‘ret’ may be used uninitialized in this function

...

drivers/net/wireless/ath/ath10k/sdio.c: In function
‘ath10k_sdio_irq_handler’:
drivers/net/wireless/ath/ath10k/sdio.c:1331: warning: ‘ret’ may be used uninitialized in this function
Signed-off-by: NErik Stromdahl <erik.stromdahl@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Chipsets like QCA9377 have support for USB so add initial USB bus
support to ath10k. With this patch we have the low level HIF and
HTC protocol working and it's possible to boot the firmware,
but it's still not possible to connect or anything like.

More changes are needed for full functionality. For that reason
we print during initialisation:

WARNING: ath10k USB support is incomplete, don't expect anything to work!
Signed-off-by: NErik Stromdahl <erik.stromdahl@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Definitions for USB based chipsets
Signed-off-by: NErik Stromdahl <erik.stromdahl@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Set the a-mpdu reference number in ath10k to make it accessible in the
receivers radiotap header. Implemented as in ath9k.  The reference number is
needed for troubleshooting and research at the receivers site (e.g. to identify
mpdu's that were aggregated in an a-mpdu)
Signed-off-by: NMatthias Frei <mf@frei.media>
[kvalo@qca.qualcomm.com: fix checkpatch warning, commit log cleanup]
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

All wmi_services are not printing when we give below command.

cat /sys/kernel/debug/ieee80211/phyX/ath10k/wmi_services

This patch increases the buffer_len to 8192 to print all the wmi_services.
Signed-off-by: NTamizh chelvam <c_traja@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Copy engine is a host to target communication interface
between wlan firmware and wlan wcn3990 platform driver. Add copy
engine register map for wcn3990 wlan module. This add support
for the copy engine source/destination ring configuration for
wcn3990 chipset.
Signed-off-by: NGovind Singh <govinds@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Remove bus specific dependencies from CE layer
to have common CE layer across multiple targets.
This is required for adding support for WCN3990
chipset support as WCN3990 chipset uses SNOC
bus interface with Copy Engine endpoint.
Signed-off-by: NGovind Singh <govinds@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Commit 59ae1d12 ("networking: introduce and use skb_put_data()") introduced
a new checkpatch warning:

drivers/net/wireless/ath/ath10k/wmi.c:3308: code indent should use tabs where possible

Fix that.
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Fix the following spelling mistakes in messages:
  syncronise -> synchronize
  unusally -> unusually
  addrress -> address
  inverval -> interval
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Declare thermal_cooling_device_ops structure as const as it is only passed
as an argument to the function thermal_cooling_device_register and this
argument is of type const. So, declare the structure as const.
Signed-off-by: NBhumika Goyal <bhumirks@gmail.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

QCA9888 supports VHT80 with 2x2. But it only support 1x1 with VHT160 or
VHT80+80. Inform userspace and the the QCA firmware about that limitation
whenever VHT80+80 or VHT160 is configured.
Signed-off-by: NSven Eckelmann <sven.eckelmann@openmesh.com>
[kvalo@qca.qualcomm.com: use hw_params]
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

QCA9984 hardware can do 4x4 at 80Mhz, but only 2x2 at 160Mhz.

First, report this to user-space by setting the max-tx-speed
and max-rx-speed vht capabilities.

Second, if the peer rx-speed is configured, and if we
are in 160 or 80+80 mode, and the peer rx-speed matches
the max speed for 2x2 or 1x1 at 160Mhz (long guard interval),
then use that info to set the peer_bw_rxnss_override appropriately.

Without this, a 9984 firmware will not use 2x2 ratesets when
transmitting to peer (it will be stuck at 1x1), because
the firmware would not have configured the rxnss_override.
Signed-off-by: NBen Greear <greearb@candelatech.com>
[sven.eckelmann@openmesh.com: rebase, cleanup, drop 160Mhz workaround cleanup]
Signed-off-by: NSven Eckelmann <sven.eckelmann@openmesh.com>
[kvalo@qca.qualcomm.com: use hw_params, rename the title]
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The ath10k firmware doesn't announce its VHT channel width capabilities in
the vht_cap information from the "service ready event" arguments. The
driver must therefore check whether the 160MHz short GI bit is set and
whether the driver still doesn't set the bits for the 160/80+80 MHz
capabilities.

The two bits for the channel width are a two bit integer and not two
separate bits which cannot be parsed without the knowledge of the other
bit. Using IEEE80211_VHT_CAP_SUPP_CHAN_WIDTH_160_80PLUS80MHZ (b10..) as a
mask for this task doesn't make any sense. The correct mask for the VHT
channel width should be used instead to make this check more readable.
Signed-off-by: NBen Greear <greearb@candelatech.com>
[sven.eckelmann@openmesh.com: separate 160Mhz workaround cleanup, add commit
 message]
Signed-off-by: NSven Eckelmann <sven.eckelmann@openmesh.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

A common pattern with skb_put() is to just want to memcpy()
some data into the new space, introduce skb_put_data() for
this.

An spatch similar to the one for skb_put_zero() converts many
of the places using it:

    @@
    identifier p, p2;
    expression len, skb, data;
    type t, t2;
    @@
    (
    -p = skb_put(skb, len);
    +p = skb_put_data(skb, data, len);
    |
    -p = (t)skb_put(skb, len);
    +p = skb_put_data(skb, data, len);
    )
    (
    p2 = (t2)p;
    -memcpy(p2, data, len);
    |
    -memcpy(p, data, len);
    )

    @@
    type t, t2;
    identifier p, p2;
    expression skb, data;
    @@
    t *p;
    ...
    (
    -p = skb_put(skb, sizeof(t));
    +p = skb_put_data(skb, data, sizeof(t));
    |
    -p = (t *)skb_put(skb, sizeof(t));
    +p = skb_put_data(skb, data, sizeof(t));
    )
    (
    p2 = (t2)p;
    -memcpy(p2, data, sizeof(*p));
    |
    -memcpy(p, data, sizeof(*p));
    )

    @@
    expression skb, len, data;
    @@
    -memcpy(skb_put(skb, len), data, len);
    +skb_put_data(skb, data, len);

(again, manually post-processed to retain some comments)
Reviewed-by: NStephen Hemminger <stephen@networkplumber.org>
Signed-off-by: NJohannes Berg <johannes.berg@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Report per chain RSSI to mac80211.
Signed-off-by: NNorik Dzhandzhapanyan <norikd@gmail.com>
[kvalo@qca.qualcomm.com: fix conflicts and style]
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

Define structures for the copy engine ctrl/misc registers,
that includes CE CMD halt, watermark source, watermark destination,
host IE ring, source, destination and dmax ring.

This adds support to avoid the conditional compilation,
code optimization and dynamic configuration of the copy engine
register map for respective hardware bus interface.
Signed-off-by: NSarada Prasanna Garnayak <c_sgarna@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The original idea is to limit the maximum TDLS peer link, but the logic
is always false, and never be able to restrict the number of TDLS peer
creation.

Fix the logic here and also move the checking earlier, so that it could
avoid to handle the failure case, e.g disable the tdls peer, delete the
peer and also vdev count cleanup.
Signed-off-by: NRyan Hsu <ryanhsu@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

QCA99X0, QCA9888, QCA9984 supports calibration data in
either OTP or DT/pre-cal file. Current ath10k supports
Calibration data from OTP only.

If caldata is loaded from DT/pre-cal file, fetching board id
and applying calibration parameters like tx power gets failed.

error log:
[   15.733663] ath10k_pci 0000:01:00.0: failed to fetch board file: -2
[   15.741474] ath10k_pci 0000:01:00.0: could not probe fw (-2)

This patch adds calibration data support from DT/pre-cal
file.  Below parameters are used to get board id and
applying calibration parameters from cal data.

		EEPROM[OTP]	FLASH[DT/pre-cal file]
Cal param	0x700		0x10000
Board id	0x10		0x8000

Tested on QCA9888 with pre-cal file.
Signed-off-by: NAnilkumar Kolli <akolli@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

ath10k firmware checks nbytes == 0 as part of determining if DMA
has completed successfully.  To help make this work more often,
have the driver initialize nbytes to zero when freeing the descriptor
slot.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This lets one have a clue that maybe timeouts are happening
when we just aren't waiting long enough.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

When testing a 9888 chipset NIC, I notice it often takes
almost 2 seconds, and then many times OTP fails, probably due
to the two-second timeout.

[ 2269.841842] ath10k_pci 0000:05:00.0: bmi cmd took: 1984 jiffies (HZ: 1000), rv: 0
[ 2273.608185] ath10k_pci 0000:05:00.0: bmi cmd took: 1986 jiffies (HZ: 1000), rv: 0
[ 2277.294732] ath10k_pci 0000:05:00.0: bmi cmd took: 1989 jiffies (HZ: 1000), rv: 0

So, increase the BMI timeout to 3 seconds.
Signed-off-by: NBen Greear <greearb@candelatech.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

This reverts commit b0578865 ("ath10k: do not use coherent memory for
allocated device memory chunks") in 2015 which converted this allocation from
dma_map_coherent() to kzalloc() / dma_map_single().

The current problem manifests when using later model NICs with larger
(>700KiB) scratch spaces in memory.  Although the kzalloc call
succeeds, the software IOMMU TLB code (via dma_map_single()) panics
because it can't find 700KiB of linear physmem bounce buffers for DMA.
Now, this is a bit of a silly failure mode for the dma map API,
but it's what we currently have to play with.

In these cases, doing kzalloc() works fine, but the dma_map_single()
call fails.

After chatting with Linus briefly about this, it indeed should be
using dma_alloc_coherent() for doing larger device memory allocation
that requires some kind of physical address mapping.

You're not supposed to be using kzalloc and dma_map_* calls for
large memory regions, and I'm guessing not for long-held mappings
either.  Typically dma mappings should be temporary for DMA,
not long held like these.

Now, since hopefully the major annoying underlying problem has also been
addressed (ie, ath10k is no longer tears down all of these allocations
and reallocates them every time the vdevs are brought down) fragmentation
should stop being such a touchy issue.  If it is though, using
dma_alloc_coherent() use gets us access to the CMB APIs too relatively
easily and ideally we would be allocating memory early in boot for
exactly these reasons.
Signed-off-by: NAdrian Chadd <adrian@FreeBSD.org>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The QCA4019 firmware 10.4-3.2.1-00050 reports only HT MCS rates between
0-9. But 802.11n MCS rates can be larger than that. For example a 2x2
device can send with up to MCS 15.

The firmware encodes the higher MCS rates using the NSS field. The actual
calculation is not documented by QCA but it seems like the NSS field can be
mapped for HT rates to following MCS offsets:

 * NSS 1: 0
 * NSS 2: 8
 * NSS 3: 16
 * NSS 4: 24

This offset therefore has to be added for HT rates before they are stored
in the rate_info struct.

Fixes: cec17c38 ("ath10k: add per peer htt tx stats support for 10.4")
Signed-off-by: NSven Eckelmann <sven.eckelmann@openmesh.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

The array fields in struct wmi_start_scan_arg that are checked here are
fixed size arrays so they can never be NULL.

Addresses-Coverity-ID: 1260031
Cc: Arend Van Spriel <arend.vanspriel@broadcom.com>
Cc: Kalle Valo <kvalo@qca.qualcomm.com>
Signed-off-by: NGustavo A. R. Silva <garsilva@embeddedor.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

QCA9xxx and QCA61x4/QCA93xx are using different wmi operation, in order
for userspace to differentiate it, appends the wmi_op_version information
alone with the get_version command.
Signed-off-by: NRyan Hsu <ryanhsu@qti.qualcomm.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>

During write to debugfs file simulate_fw_crash, fixed-size local buffer
'buf' is accessed and modified at index 'count-1', where 'count' is the
size of the write (so potentially out of bounds).
This patch fixes this problem.
Signed-off-by: NMichael Mera <dev@michaelmera.com>
Signed-off-by: NKalle Valo <kvalo@qca.qualcomm.com>