When downloading the reserved page, the first page always contains
a beacon for the firmware to reference. For non-beaconing modes such
as station mode, also put a blank skb with length=1.

And for the beaconing modes, driver will get a real beacon with a
length approximate to the page size. But as the beacon is always put
at the first page, it does not need a tx_desc, because the TX path
will generate one when TXing the reserved page to the hardware. So we
could allocate a buffer with a size smaller than the reserved page,
when using memcpy() to copy the content of reserved page to the buffer,
the over-sized reserved page will violate the kernel memory.

To fix it, add the tx_desc before memcpy() the reserved packets to
the buffer, then we can get SKBs with correct length when counting
the pages in total. And for page 0, count the extra tx_desc_sz that
the TX path will generate. This way, the first beacon that allocated
without tx_desc can be counted with the extra tx_desc_sz to get
actual pages it requires.

Fixes: e3037485 ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Sometimes mac80211 will ask us to flush the hardware queues.
To flush them, first we need to get the corresponding priority queues
from the RQPN mapping table.

Then we can check the available pages are equal to the originally
reserved pages, which means the hardware has returned all of the pages
it used to transmit.

Note that now we only check for 100 ms for the priority queue, but
sometimes if we have a lot of traffic (ex. 100Mbps up), some of the
packets could be dropped.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Based on the mac80211's TXQ implementation, TX-AMSDU can
be used to get higher MAC efficiency. To make mac80211
aggregate MSDUs, low level driver just need to leave skbs
in the TXQ, and mac80211 will try to aggregate them if
possible. As driver will schedule a tasklet when the TX
queue is woke, until the tasklet being served, there will
have some skbs in the queue if traffic is heavy.

Driver can control the max AMSDU size depending on the
current bit rate used by hardware/firmware. The higher
rates are used, the larger AMSDU size can be.

It is tested that can achieve higher T-Put at higher rates.
If the environment is relatively clean, and the bit_rate
is high enough, we can get about 80Mbps improvement.

For lower bit rates, not much gain can we get, so leave
the max_amsdu length low to prevent aggregation.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Whenever the firmware increases/decreases the bit rate used
to transmit to a peer, it sends an RA report through C2H to
driver. Driver can then record the bit rate in the peer's
struct rtw_sta_info, and report to mac80211 when it asks us
for the statistics of the sta by ieee80211_ops::sta_statistics
Signed-off-by: Tzu-En Huang <tehuang@realtek.com>
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

We can indicate IEEE80211_HW_HAS_RATE_CONTROL to mac80211 because
the hardware has its own rate control algorithm. And what driver needs
to do is to choose an RA mask according the peer's capabilities.

But the hardware is not able to setup BA session by itself. So driver
requires to initiate tx BA session for hardware, and tells it if it is
possible to transmit AMPDU. The hardware can then aggregate MPDUs.

And the size of AMPDU is controlled by the TX descriptor and the
register value. Since the TX descriptor will reference the max AMPDU
size from ieee80211_sta::ht_cap::ampdu_factor, just set the register
value to 0x3f, and let it be controlled by TX descriptor.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The mac80211 provides software TX queue for driver, as long as
driver has hooked ieee80211_ops::wake_tx_queue. Each time a
packet is queued onto the TX queue, that queue will be woken
up the inform driver to serve the queue.

Now driver only supports PCI interface ICs, there's no specific
traffic control for each queue, just schedule a tasklet, and
dump all of the packets at once to the DMA ring. Instead of TX
the packets whenever TX queue is woke, tasklet handler can have
more packets dumped to the device, takes advantage of burst
write with DMA engine.

And if the driver is going to support USB/SDIO ICs, the tasklet
can be more flexible for aggregating the packets, enhance the
efficiency of bandwidth usage.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Allows driver to send RTS by filling tx descriptor.

The user may want to set the rts threshold. But since we have not
been taking over rate control from mac80211 to driver by setting flag
IEEE80211_HW_HAS_RATE_CONTROL, there is nothing we can do about it.
So here just store the value, and mac80211 will tell us to use rts
protection by ieee80211_tx_info::control::use_rts.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Driver needs to wait for firmware to restore hardware setting
to active mode after leaving lps.

After getting H2C from driver for leaving lps, firmware will
issue null packet without PS bit to inform AP driver is active,
and then restore REG_TCR Register if AP has receiced null packet.

But the transmission of null packet may cost much more time
in noisy environment. If driver does not wait for firmware,
null packet with PS bit could be sent due to incorrect REG_TCR setting.
And AP will be confused.

In our test, 100ms is enough for firmware to send null packet
to AP. If REG_TCR Register is still wrong after 100ms, we will
modify it directly, force the PS bit to be cleared
Signed-off-by: Chin-Yen Lee <timlee@realtek.com>
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

ath.git patches for 5.5. Major changes:

ath10k

* add support for hardware rfkill on devices where firmware supports it

memcpy() in wmi_set_ie() and wmi_update_ft_ies() is called with
src == NULL and len == 0. This is an undefined behavior. Fix it
by checking "ie_len > 0" before the memcpy() calls.

As suggested by GCC documentation:
"The pointers passed to memmove (and similar functions in <string.h>)
must be non-null even when nbytes==0, so GCC can use that information
to remove the check after the memmove call." [1]

[1] https://gcc.gnu.org/gcc-4.9/porting_to.html

Cc: Maya Erez <merez@codeaurora.org>
Cc: Kalle Valo <kvalo@codeaurora.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: stable@vger.kernel.org
Signed-off-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

memcpy() call with "idata == NULL && ilen == 0" results in undefined
behavior in ar5523_cmd(). For example, NULL is passed in callchain
"ar5523_stat_work() -> ar5523_cmd_write() -> ar5523_cmd()". This patch
adds ilen check before memcpy() call in ar5523_cmd() to prevent an
undefined behavior.

Cc: Pontus Fuchs <pontus.fuchs@gmail.com>
Cc: Kalle Valo <kvalo@codeaurora.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: David Laight <David.Laight@ACULAB.COM>
Cc: stable@vger.kernel.org
Signed-off-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

When hardware rfkill is enabled in the firmware it will report the
capability via using WMI_TLV_SYS_CAP_INFO_RFKILL bit in the WMI_SERVICE_READY
event to the host. ath10k will check the capability, and if it is enabled then
ath10k will set the GPIO information to firmware using WMI_PDEV_SET_PARAM. When
the firmware detects hardware rfkill is enabled by the user, it will report it
via WMI_RFKILL_STATE_CHANGE_EVENTID. Once ath10k receives the event it will
send wmi command WMI_PDEV_SET_PARAM to the firmware to enable/disable the radio
and also notifies cfg80211.

We can't power off the device when rfkill is enabled, as otherwise the
firmware would not be able to detect GPIO changes and report them to the
host. So when rfkill is enabled, we need to keep the firmware running.

Tested with QCA6174 PCI with firmware
WLAN.RM.4.4.1-00109-QCARMSWPZ-1.
Signed-off-by: Alan Liu <alanliu@codeaurora.org>
Signed-off-by: Wen Gong <wgong@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

This patch restores the old behavior that read
the chip_id on the QCA988x before resetting the
chip. This needs to be done in this order since
the unsupported QCA988x AR1A chips fall off the
bus when resetted. Otherwise the next MMIO Op
after the reset causes a BUS ERROR and panic.

Cc: stable@vger.kernel.org
Fixes: 1a7fecb7 ("ath10k: reset chip before reading chip_id in probe")
Signed-off-by: Christian Lamparter <chunkeey@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Offchannel management frames were failing:

[18099.253732] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3780
[18102.293686] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3780
[18105.333653] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3780
[18108.373712] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3780
[18111.413687] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e36c0
[18114.453726] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3f00
[18117.493773] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e36c0
[18120.533631] ath10k_pci 0000:01:00.0: timed out waiting for offchannel skb cf0e3f00

This bug appears to have been added between 4.0 (which works for us),
and 4.4, which does not work.

I think this is because the tx-offchannel logic gets in a loop when
ath10k_mac_tx_frm_has_freq(ar) is false, so pkt is never actually
sent to the firmware for transmit.

This patch fixes the problem on 4.9 for me, and now HS20 clients
can work again with my firmware.

Antonio: tested with 10.4-3.5.3-00057 on QCA4019 and QCA9888
Signed-off-by: Ben Greear <greearb@candelatech.com>
Tested-by: Antonio Quartulli <antonio.quartulli@kaiwoo.ai>
[kvalo@codeaurora.org: improve commit log, remove unneeded parenthesis]
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The "priv->hw_type" is an enum and in this context GCC will treat it
as an unsigned int so the error handling will never trigger.

Fixes: a910e4a9 ("cw1200: add driver for the ST-E CW1100 & CW1200 WLAN chipsets")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The module parameter rtw_fw_support_lps is misleading. It
is not used to represent the firmware's property, but to
determine if driver wants to ask firmware to enter LPS.

However, driver should better enable/disable PS through
cfg80211_ops::set_power_mgmt instead.
For example, one could use iw command to set PS state.

  $ sudo iw wlanX set power_save [on/off]

So rtw_fw_support_lps should be removed because it is
misleading and useless. Instead of checking the parameter,
set PS mode according to IEEE80211_CONF_PS.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Compare with LCLK mode, PG mode saves more power, by turning
off more circuits. Therefore, to recover from PG mode, driver
needs to backup some information into rsvd page. Such as CAM
entries, DPK results.

As CAM entries can change, it is required to re-download CAM
entries after set_key.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Add a module parameter to select deep PS mode. And the mode
cannot be changed after the module has been inserted and probed.
If anyone wants to change the deep mode, should change the mode
and probe the device again to setup the changed deep mode.

When the device is probed, driver will check the deep PS mode
with different IC's PS mode suppotability. If none of the
PS mode is matched, the deep PS mode is changed to NONE,
means deep PS is disabled.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Sometimes LPS is not compatible with COEX's strategy, and
COEX will not allow driver to enter it.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Deep power save allows firmware/hardware to operate in a
lower power state. And the deep power save mode depends on
LPS mode. So, before entering deep PS, driver must first
enter LPS mode.

Under Deep PS, most of hardware functions are shutdown,
driver will not be able to read/write registers and transfer
data to the device. Hence TX path must be protected by each
interface. Take PCI for example, DMA engine should be idle,
and no nore activities on the PCI bus.

If driver wants to operate on the device, such as register
read/write, it must first acquire the mutex lock and wake
up from Deep PS, otherwise the behavior is undefined.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Dynamic mechanism requires BB/RF working to adjust
hardware settings. But PS state periodically turns
off BB/RF, could lead to wrong setting.

So leave PS state before DM to make sure it works.
And then check if we can enter PS state again.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Protect LPS enter/leave routine with rtwdev->mutex.
This helps to synchronize with driver's states correctly.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

This is no more used, remove it.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The original design of LPS enter/leave routines allows
to control the LPS state by each interface. But the
hardware cannot actually handle it that way. This means
the hardware can only enter LPS once with an associated
port, so there is no need to keep tracking the state of
each vif.

Hence the logic of enter/leave LPS state can be simple,
just to check the state of the device's flag. And for
leaving LPS state, it will get the same port id to send
to inform the hardware.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Remove PS related *_irqsafe functions to avoid entering/leaving PS
under interrupt context. Instead, make PS decision in watch_dog.
This could simplify the logic and make the code look clean.

But it could have a little side-effect that if the driver is having
heavy traffic before the every-2-second watch_dog detect the traffic
and decide to leave PS, the thoughput will be lower. Once traffic is
detected by watch_dog and left PS state, the throughput will resume
to the peak the hardware ought to have again.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

If the driver doesn't reset the host's and device's indexes
in a single write, the indexes will become different in a
short period. And it will confuse the DMA engine, make it
start to process non-existed entries.

Better to Write-1-to-reset the indexes, for the DMA engine
to know that this is a reset of the H2C queue, not a kick
off.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

These helper functions seems useless. And in some cases
we want to use test_and_[set/clear]_bit, these helpers
will make the code more complicated. So remove them.
Signed-off-by: Yan-Hsuan Chuang <yhchuang@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/wireless/realtek/rtlwifi/rtl8723be/phy.c: In function rtl8723be_phy_iq_calibrate:
drivers/net/wireless/realtek/rtlwifi/rtl8723be/phy.c:2255:7: warning: variable reg_ecc set but not used [-Wunused-but-set-variable]
drivers/net/wireless/realtek/rtlwifi/rtl8723be/phy.c: In function rtl8723be_phy_iq_calibrate:
drivers/net/wireless/realtek/rtlwifi/rtl8723be/phy.c:2254:34: warning: variable reg_eac set but not used [-Wunused-but-set-variable]

They are not used since commit a619d1ab ("rtlwifi:
rtl8723be: Add new driver")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/wireless/realtek/rtlwifi/rtl8192ee/phy.c: In function rtl92ee_phy_iq_calibrate:
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/phy.c:2805:34: warning: variable reg_ecc set but not used [-Wunused-but-set-variable]
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/phy.c: In function rtl92ee_phy_iq_calibrate:
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/phy.c:2804:34: warning: variable reg_eac set but not used [-Wunused-but-set-variable]

They are not used since commit b1a3bfc9 ("rtlwifi:
rtl8192ee: Move driver from staging to the regular tree")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/wireless/realtek/rtlwifi/rtl8192ee/trx.c: In function rtl92ee_tx_fill_desc:
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/trx.c:656:5: warning: variable short_gi set but not used [-Wunused-but-set-variable]
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/trx.c: In function rtl92ee_tx_fill_desc:
drivers/net/wireless/realtek/rtlwifi/rtl8192ee/trx.c:648:15: warning: variable buf_len set but not used [-Wunused-but-set-variable]

They are not used since commit f1d2b4d3 ("rtlwifi:
rtl818x: Move drivers into new realtek directory")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/wireless/realtek/rtlwifi/efuse.c: In function efuse_pg_packet_write:
drivers/net/wireless/realtek/rtlwifi/efuse.c:937:24: warning: variable dataempty set but not used [-Wunused-but-set-variable]
drivers/net/wireless/realtek/rtlwifi/efuse.c: In function efuse_get_current_size:
drivers/net/wireless/realtek/rtlwifi/efuse.c:1202:5: warning: variable hoffset set but not used [-Wunused-but-set-variable]

They are not used since commit f1d2b4d3 ("rtlwifi:
rtl818x: Move drivers into new realtek directory")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/wireless/realtek/rtlwifi/ps.c: In function rtl_ps_set_rf_state:
drivers/net/wireless/realtek/rtlwifi/ps.c:71:19: warning: variable rtstate set but not used [-Wunused-but-set-variable]

It is not used since commit f1d2b4d3 ("rtlwifi:
rtl818x: Move drivers into new realtek directory")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: zhengbin <zhengbin13@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

In rtl8xxxu_submit_int_urb if usb_submit_urb fails the allocated urb
should be released.
Signed-off-by: Navid Emamdoost <navid.emamdoost@gmail.com>
Reviewed-by: Chris Chiu <chiu@endlessm.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

kmemdup is introduced to duplicate a region of memory in a neat way.
Rather than kmalloc/kzalloc + memcpy, which the programmer needs to
write the size twice (sometimes lead to mistakes), kmemdup improves
readability, leads to smaller code and also reduce the chances of mistakes.
Suggestion to use kmemdup rather than using kmalloc/kzalloc + memcpy.
Signed-off-by: Fuqian Huang <huangfq.daxian@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Sometimes the hardware will push small packets that trigger a WARN_ON
in mac80211. Discard them early to avoid this issue.

This patch ports 2 patches from ath9k to ath9k_htc.
commit 3c0efb74 "ath9k: discard
undersized packets".
commit df5c4150 "ath9k: correctly
handle short radar pulses".

[  112.835889] ------------[ cut here ]------------
[  112.835971] WARNING: CPU: 5 PID: 0 at net/mac80211/rx.c:804 ieee80211_rx_napi+0xaac/0xb40 [mac80211]
[  112.835973] Modules linked in: ath9k_htc ath9k_common ath9k_hw ath mac80211 cfg80211 libarc4 nouveau snd_hda_codec_hdmi intel_rapl_msr intel_rapl_common x86_pkg_temp_thermal intel_powerclamp coretemp snd_hda_codec_realtek snd_hda_codec_generic ledtrig_audio snd_hda_intel snd_hda_codec video snd_hda_core ttm snd_hwdep drm_kms_helper snd_pcm crct10dif_pclmul snd_seq_midi drm snd_seq_midi_event crc32_pclmul snd_rawmidi ghash_clmulni_intel snd_seq aesni_intel aes_x86_64 crypto_simd cryptd snd_seq_device glue_helper snd_timer sch_fq_codel i2c_algo_bit fb_sys_fops snd input_leds syscopyarea sysfillrect sysimgblt intel_cstate mei_me intel_rapl_perf soundcore mxm_wmi lpc_ich mei kvm_intel kvm mac_hid irqbypass parport_pc ppdev lp parport ip_tables x_tables autofs4 hid_generic usbhid hid raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 multipath linear e1000e ahci libahci wmi
[  112.836022] CPU: 5 PID: 0 Comm: swapper/5 Not tainted 5.3.0-wt #1
[  112.836023] Hardware name: MouseComputer Co.,Ltd. X99-S01/X99-S01, BIOS 1.0C-W7 04/01/2015
[  112.836056] RIP: 0010:ieee80211_rx_napi+0xaac/0xb40 [mac80211]
[  112.836059] Code: 00 00 66 41 89 86 b0 00 00 00 e9 c8 fa ff ff 4c 89 b5 40 ff ff ff 49 89 c6 e9 c9 fa ff ff 48 c7 c7 e0 a2 a5 c0 e8 47 41 b0 e9 <0f> 0b 48 89 df e8 5a 94 2d ea e9 02 f9 ff ff 41 39 c1 44 89 85 60
[  112.836060] RSP: 0018:ffffaa6180220da8 EFLAGS: 00010286
[  112.836062] RAX: 0000000000000024 RBX: ffff909a20eeda00 RCX: 0000000000000000
[  112.836064] RDX: 0000000000000000 RSI: ffff909a2f957448 RDI: ffff909a2f957448
[  112.836065] RBP: ffffaa6180220e78 R08: 00000000000006e9 R09: 0000000000000004
[  112.836066] R10: 000000000000000a R11: 0000000000000001 R12: 0000000000000000
[  112.836068] R13: ffff909a261a47a0 R14: 0000000000000000 R15: 0000000000000004
[  112.836070] FS:  0000000000000000(0000) GS:ffff909a2f940000(0000) knlGS:0000000000000000
[  112.836071] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  112.836073] CR2: 00007f4e3ffffa08 CR3: 00000001afc0a006 CR4: 00000000001606e0
[  112.836074] Call Trace:
[  112.836076]  <IRQ>
[  112.836083]  ? finish_td+0xb3/0xf0
[  112.836092]  ? ath9k_rx_prepare.isra.11+0x22f/0x2a0 [ath9k_htc]
[  112.836099]  ath9k_rx_tasklet+0x10b/0x1d0 [ath9k_htc]
[  112.836105]  tasklet_action_common.isra.22+0x63/0x110
[  112.836108]  tasklet_action+0x22/0x30
[  112.836115]  __do_softirq+0xe4/0x2da
[  112.836118]  irq_exit+0xae/0xb0
[  112.836121]  do_IRQ+0x86/0xe0
[  112.836125]  common_interrupt+0xf/0xf
[  112.836126]  </IRQ>
[  112.836130] RIP: 0010:cpuidle_enter_state+0xa9/0x440
[  112.836133] Code: 3d bc 20 38 55 e8 f7 1d 84 ff 49 89 c7 0f 1f 44 00 00 31 ff e8 28 29 84 ff 80 7d d3 00 0f 85 e6 01 00 00 fb 66 0f 1f 44 00 00 <45> 85 ed 0f 89 ff 01 00 00 41 c7 44 24 10 00 00 00 00 48 83 c4 18
[  112.836134] RSP: 0018:ffffaa61800e3e48 EFLAGS: 00000246 ORIG_RAX: ffffffffffffffde
[  112.836136] RAX: ffff909a2f96b340 RBX: ffffffffabb58200 RCX: 000000000000001f
[  112.836137] RDX: 0000001a458adc5d RSI: 0000000026c9b581 RDI: 0000000000000000
[  112.836139] RBP: ffffaa61800e3e88 R08: 0000000000000002 R09: 000000000002abc0
[  112.836140] R10: ffffaa61800e3e18 R11: 000000000000002d R12: ffffca617fb40b00
[  112.836141] R13: 0000000000000002 R14: ffffffffabb582d8 R15: 0000001a458adc5d
[  112.836145]  ? cpuidle_enter_state+0x98/0x440
[  112.836149]  ? menu_select+0x370/0x600
[  112.836151]  cpuidle_enter+0x2e/0x40
[  112.836154]  call_cpuidle+0x23/0x40
[  112.836156]  do_idle+0x204/0x280
[  112.836159]  cpu_startup_entry+0x1d/0x20
[  112.836164]  start_secondary+0x167/0x1c0
[  112.836169]  secondary_startup_64+0xa4/0xb0
[  112.836173] ---[ end trace 9f4cd18479cc5ae5 ]---
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

rs_datalen is be16 so we need to convert it before printing.
Signed-off-by: Masashi Honma <masashi.honma@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Currently, data variable in ar9003_hw_thermo_cal_apply() could be
uninitialized if ar9300_otp_read_word() will fail to read the value.
Initialize data variable with 0 to prevent an undefined behavior. This
will be enough to handle error case when ar9300_otp_read_word() fails.

Fixes: 80fe43f2 ("ath9k_hw: Read and configure thermocal for AR9462")
Cc: Rajkumar Manoharan <rmanohar@qca.qualcomm.com>
Cc: John W. Linville <linville@tuxdriver.com>
Cc: Kalle Valo <kvalo@codeaurora.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: stable@vger.kernel.org
Signed-off-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

In a multiradio board with one QCA9984 and one AR9987
after enabling the crashdump with module parameter
coredump_mask=7, below backtrace is seen.

vmalloc: allocation failure: 0 bytes
 kworker/u4:0: page allocation failure: order:0, mode:0x80d2
 CPU: 0 PID: 6 Comm: kworker/u4:0 Not tainted 3.14.77 #130
 Workqueue: ath10k_wq ath10k_core_register_work [ath10k_core]
 (unwind_backtrace) from [<c021abf8>] (show_stack+0x10/0x14)
 (dump_stack+0x80/0xa0)
 (warn_alloc_failed+0xd0/0xfc)
 (__vmalloc_node_range+0x1b4/0x1d8)
 (__vmalloc_node+0x34/0x40)
 (vzalloc+0x24/0x30)
 (ath10k_coredump_register+0x6c/0x88 [ath10k_core])
 (ath10k_core_register_work+0x350/0xb34 [ath10k_core])
 (process_one_work+0x20c/0x32c)
 (worker_thread+0x228/0x360)

This is due to ath10k_hw_mem_layout is not defined for AR9987.
For coredump undefined hw ramdump_size is 0.
Check for the ramdump_size before allocation memory.

Tested on: AR9987, QCA9984
FW version: 10.4-3.9.0.2-00044
Signed-off-by: Anilkumar Kolli <akolli@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

The IRAM start address in coredump was wrong for QCA9984, QCA4019, QCA9888 and
QCA99x0.

Tested on: QCA9984, QCA4019
FW version: 10.4-3.9.0.2-00044
Signed-off-by: Anilkumar Kolli <akolli@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>

Since ath10k_mac_tx_push_txq() can be called from process context, we
must explicitly disable softirqs before the call into mac80211.

By calling ieee80211_tx_dequeue_ni() instead of ieee80211_tx_dequeue()
we make sure softirqs are always disabled even in the case when
ath10k_mac_tx_push_txq() is called from process context.

Calling ieee80211_tx_dequeue_ni() with softirq's already disabled
(e.g., from softirq context) should be safe as the local_bh_disable()
and local_bh_enable() functions (called from ieee80211_tx_dequeue_ni)
are fully reentrant.
Signed-off-by: Erik Stromdahl <erik.stromdahl@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>