mt76: mt7915: add 802.11 encap offload support

It is currently limited to 3-address mode AP and STA interfaces

Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: sdio: convert {status/net}_work to mt76_worker

In order to improve driver throughput, convert status_work and net_work
to mt76 worker APIs.
Remove txrx_wq sdio workqueue since it is no longer used

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7663s: disable interrupt during txrx_worker processing

In order to reduce sdio bus contention, disable interrupts during
txrx_worker processing

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7663s: convert txrx_work to mt76_worker

In order to improve driver throughput, convert txrx_work to mt76 worker
threads APIs.

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7663s: move tx/rx processing in the same txrx workqueue

Move mt7663 tx and rx processing in the same workqueue in order to
reduce jitter that can hit tcp performances. This is a preliminary patch
to switch to mt76 workers APIs

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7615: retry if mt7615_mcu_init returns -EAGAIN

mt7615_load_patch in mt7615/mcu.c sometimes fails with:
mt7622-wmac 18000000.wmac: Failed to get patch semaphore
and returns -EAGAIN. But this error is returned all the way up to
mt7615_init_work with no actual retrial performed, leaving a
broken wireless phy.
Wait a bit and retry for up to 10 times before giving up.

Signed-off-by: Chuanhong Guo <gch981213@gmail.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7915: fix VHT LDPC capability

The MCU field should contain a boolean 0/1, not the flag itself.

Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

mt76: mt7915: measure channel noise and report it via survey

Read per-stream measurements every 100 ms and build a simple moving average.

Tested-by: Shayne Chen <shayne.chen@mediatek.com>
Signed-off-by: Ryder Lee <ryder.lee@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>

Merge ath-next from git://git./linux/kernel/git/kvalo/ath.git

ath.git patches for v5.11. Major changes:

ath11k

* Fast Initial Link Setup (FILS) discovery and unsolicited broadcast
  probe response support

* qcom,ath11k-calibration-variant Device Tree setting

* cold boot calibration support

* new DFS region: JP

wnc36xx

* enable connection monitoring and keepalive in firmware

ath10k

* firmware IRAM recovery feature

mhi

* merge mhi-ath11k-immutable branch to make MHI API change go smoothly

wl1251: remove trailing semicolon in macro definition

The macro use will already have a semicolon.

Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201127180835.2769297-1-trix@redhat.com

airo: remove trailing semicolon in macro definition

The macro use will already have a semicolon.

Signed-off-by: Tom Rix <trix@redhat.com>
Reviewed-by: Kieran Bingham <kieran.bingham+renesas@ideasonboard.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201127175821.2756988-1-trix@redhat.com

wilc1000: added queue support for WMM

Added multiple queues[BK,BE,VI,VO] to handle different priority data
packets. Before adding a packet to the queue, checked its priority from
the header, and then add to the suitable queue. The limit for each queue
is maintained separately. Also while passing the packets to the firmware
via VMM take care to select data packets based on priority and available
space.

Signed-off-by: Ajay Singh <ajay.kathat@microchip.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201125114059.10006-6-ajay.kathat@microchip.com

wilc1000: call complete() for failure in wilc_wlan_txq_add_cfg_pkt()

Added complete() call for failure case in wilc_wlan_txq_add_cfg_pkt().

Signed-off-by: Ajay Singh <ajay.kathat@microchip.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201125114059.10006-5-ajay.kathat@microchip.com

wilc1000: free resource in wilc_wlan_txq_add_mgmt_pkt() for failure path

Before returing from wilc_wlan_txq_add_mgmt_pkt() invoke tx_complete_fn()
callback to free up allocated memory for failure case.

Signed-off-by: Ajay Singh <ajay.kathat@microchip.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201125114059.10006-4-ajay.kathat@microchip.com

wilc1000: free resource in wilc_wlan_txq_add_net_pkt() for failure path

Before returing from wilc_wlan_txq_add_net_pkt() invoke tx_complete_fn()
callback to free up allocated memory for failure cases.

Signed-off-by: Ajay Singh <ajay.kathat@microchip.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201125114059.10006-3-ajay.kathat@microchip.com

wilc1000: added 'ndo_set_mac_address' callback support

Added support for 'ndo_set_mac_address call' callback to allow change of
interface MAC address.

Signed-off-by: Ajay Singh <ajay.kathat@microchip.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201125114059.10006-2-ajay.kathat@microchip.com

brcmfmac: expose firmware config files through modinfo

Apart from a firmware binary the chip needs a config file used by the
FW. Add the config files to modinfo so that they can be read by
userspace.

Signed-off-by: Matthias Brugger <mbrugger@suse.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201124120018.31358-1-matthias.bgg@kernel.org

wlcore: Switch to using the new API kobj_to_dev()

Switch to using the new API kobj_to_dev().

Signed-off-by: Tian Tao <tiantao6@hisilicon.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1606096062-32251-1-git-send-email-tiantao6@hisilicon.com

rtw88: coex: add feature to enhance HID coexistence performance

Add toggle table related function to enhance WL throughput when WL coexist
with 4/18 HID.
The toggle table feature will toggle WL/BT priority table during WL slot,
it can decrease the impact from HID's frequently packets and prevent HID
lag.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-11-pkshih@realtek.com

rtw88: coex: upgrade coexistence A2DP mechanism

Add modifications for A2DP interoperability issue of TWS earphones.
Some TWS earphones has low buffer size, so it's necessary to reduce
the slot period to let it get data more frequently.
(e.g. use 4-slot to replace the original 2-slot method).

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-10-pkshih@realtek.com

rtw88: coex: add action for coexistence in hardware initial

The original setting misses some necessary configuration.
Make coexistence slot align to beacon and check the connection situation
to know if we need to extend WLAN slot or not.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-9-pkshih@realtek.com

rtw88: coex: add function to avoid cck lock

Some AP will not follow the power save request, or it cannot stop
transmission until its queue is empty. It may bring the decreasing of
data rate.

WLAN firmware will count is the AP still leaked packet after power save
handshake was done or not to enable WLAN slot extend mechanism.

The extend WLAN slot mechanism will extend the WLAN slot after power save
handshake, 5 ms per times, maximum is 5 times to received the leaked packet
to avoid the rate lower down.

And if the transmission was already locked at CCK rate.
The extended WLAN slot can also increase the opportunity that we can
received the CCK's long packet and be released from CCK rate.

While BT multi-link status was finished, there is possible that it still
has some packet remained for seconds. Add a timer to remain the multi-link
mechanism to protect WLAN Rx.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-8-pkshih@realtek.com

rtw88: coex: change the coexistence mechanism for WLAN connected

Add a flag to make decision whether the mechanism
should go into free-run mode or not.
For now, it is always false, the flag assignment will
be implemented later.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-7-pkshih@realtek.com

rtw88: coex: change the coexistence mechanism for HID

Add TDMA slot type setting for later usage.
Since the transmission of HID profile is very frequently,
it may bring a big impact to WLAN performance.
To change slot type, it can make mechanism be more flexible.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-6-pkshih@realtek.com

rtw88: coex: update AFH information while in free-run mode

In free run mode, this WLAN info updating is only related to AFH
information, not related to the connection state, which is the current
implementation.
Always update WLAN info in free run mode now.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-5-pkshih@realtek.com

rtw88: coex: update the mechanism for A2DP + PAN

Update A2DP+PAN+WL mechanism for CPT(Coexistence Performance Test) to
enhance WL/BT performance at the environment that some specific AP are
existing.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-4-pkshih@realtek.com

rtw88: coex: add debug message

Add message for debugging usage and the program flow is no change.
Add a variable reserved for WLAN firmware synchronize.
Add a group of variable to save BT packet counter, it will be
assigned as mechanism judgment in the future. Now it is just for
debug usage.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-3-pkshih@realtek.com

rtw88: coex: run coexistence when WLAN entering/leaving LPS

When WLAN entering or leaving, it's necessary to run coexistence mechanism
to ensure the setting matched current status.
Without calling rtw_coex_run_coex(), WLAN poor throughput or bad A2DP
quality may happen.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201126021059.11981-2-pkshih@realtek.com

Revert "rtl8xxxu: Add Buffalo WI-U3-866D to list of supported devices"

This reverts commit 28606150768a20b291a35dbbbb0ab4dd7d4739db.
Since actually Buffalo WI-U3-866D is falied to power on with the change.

The rtl8812au driver code to power on is same with the rtl8821a code.
But the rtl8821ae included rtl8821a does not support USB interface.
So seems that rtl8812au should be supported by rt1l8821a code with USB.

Signed-off-by: Tokunori Ikegami <ikegami.t@gmail.com>
Cc: linux-wireless@vger.kernel.org
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201121003411.9450-1-ikegami.t@gmail.com

ath11k: dp_rx: fix monitor status dma unmap direction

After enabling CONFIG_DMA_API_DEBUG there was a warning about using
dma_unmap_single() in wrong direction from ath11k_dp_rx_process_mon_status().

[  140.279477] ------------[ cut here ]------------
[  140.279908] DMA-API: ath11k_pci 0000:06:00.0: device driver syncs DMA memory with different direction [device address=0x00000000fac08a40] [size=2176 bytes] [mapped with DMA_FROM_DEVICE] [s
[  140.279925] WARNING: CPU: 7 PID: 97 at kernel/dma/debug.c:1120 check_sync+0x494/0x730
[  140.279939] Modules linked in: ath11k_pci ath11k mac80211 libarc4 cfg80211 qmi_helpers qrtr_mhi mhi qrtr ns nvme nvme_core
[  140.279958] CPU: 7 PID: 97 Comm: kworker/u16:1 Not tainted 5.10.0-rc4+ #262
[  140.279968] Hardware name: Intel(R) Client Systems NUC8i7HVK/NUC8i7HVB, BIOS HNKBLi70.86A.0049.2018.0801.1601 08/01/2018
[  140.279995] Workqueue: phy0 ieee80211_scan_work [mac80211]
[  140.280009] RIP: 0010:check_sync+0x494/0x730
[  140.280022] Code: 8b 4c 24 10 4c 8b 44 24 18 4c 8b 54 24 20 48 89 c6 4c 89 54 24 10 4c 89 f9 4c 89 ea 48 c7 c7 40 b9 74 9c 41 56 e8 2f a0 ab 00 <0f> 0b 48 89 ef e8 e5 17 ac 00 41 58 4c 8b
[  140.280033] RSP: 0018:ffff9f588024cbd8 EFLAGS: 00010086
[  140.280046] RAX: 0000000000000000 RBX: ffff9f588024cc40 RCX: ffff8eed18dd9f98
[  140.280057] RDX: 00000000ffffffd8 RSI: 0000000000000027 RDI: ffff8eed18dd9f90
[  140.280067] RBP: ffff8eebc1407800 R08: 00000000ffffffea R09: 0000000000000000
[  140.280082] R10: 0000000000000003 R11: 3fffffffffffffff R12: ffffffff9e081060
[  140.280093] R13: ffff8eebc43908b0 R14: ffffffff9c74c104 R15: 00000000fac08a40
[  140.280104] FS:  0000000000000000(0000) GS:ffff8eed18c00000(0000) knlGS:0000000000000000
[  140.280115] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  140.280127] CR2: 00007f6feafb12a0 CR3: 00000001604ca001 CR4: 00000000003706e0
[  140.280138] Call Trace:
[  140.280149]  <IRQ>
[  140.280161]  debug_dma_sync_single_for_cpu+0x79/0x80
[  140.280173]  ? mark_held_locks+0x50/0x80
[  140.280185]  ? lockdep_hardirqs_on_prepare.part.0+0x65/0x130
[  140.280197]  ? __local_bh_enable_ip+0x6f/0xb0
[  140.280215]  ? ath11k_dp_rx_reap_mon_status_ring+0x202/0x340 [ath11k]
[  140.280231]  ath11k_dp_rx_reap_mon_status_ring+0x22c/0x340 [ath11k]
[  140.280249]  ? ath11k_dp_rx_process_mon_rings+0x1a0/0x1a0 [ath11k]
[  140.280265]  ath11k_dp_rx_process_mon_status+0x83/0x3c0 [ath11k]
[  140.280278]  ? __lock_acquire+0x3bd/0x6d0
[  140.280296]  ? ath11k_dp_rx_process_mon_rings+0x1a0/0x1a0 [ath11k]
[  140.280311]  ? ath11k_dp_rx_process_mon_rings+0x1a0/0x1a0 [ath11k]
[  140.280326]  ? ath11k_hw_mac_id_to_srng_id_qca6390+0x10/0x10 [ath11k]
[  140.280341]  ? ath11k_dp_rx_process_mon_rings+0x4a/0x1a0 [ath11k]
[  140.280353]  ? timer_fixup_init+0x30/0x30
[  140.280367]  ? ath11k_dp_rx_process_mon_rings+0x1a0/0x1a0 [ath11k]
[  140.280385]  ath11k_dp_service_mon_ring+0x2b/0x50 [ath11k]
[  140.280400]  ? ath11k_dp_rx_process_mon_rings+0x1a0/0x1a0 [ath11k]
[  140.280413]  call_timer_fn+0xb1/0x2d0
[  140.280426]  __run_timers.part.0+0x205/0x2f0
[  140.280439]  run_timer_softirq+0x21/0x50
[  140.280450]  __do_softirq+0xc2/0x454
[  140.280463]  asm_call_irq_on_stack+0x12/0x20
[  140.280476]  </IRQ>
[  140.280488]  do_softirq_own_stack+0x56/0x60
[  140.280500]  irq_exit_rcu+0x9a/0xd0
[  140.280511]  sysvec_apic_timer_interrupt+0x43/0xa0
[  140.280526]  asm_sysvec_apic_timer_interrupt+0x12/0x20
[  140.280540] RIP: 0010:_raw_spin_unlock_irqrestore+0x25/0x40
[  140.280551] Code: 80 00 00 00 00 55 48 89 fd 48 83 c7 18 53 48 89 f3 48 8b 74 24 10 e8 ca 8f 4b ff 48 89 ef e8 22 dc 4b ff f6 c7 02 75 0c 53 9d <65> ff 0d 04 51 20 64 5b 5d c3 e8 9c 08 56
[  140.280563] RSP: 0018:ffff9f58802e7878 EFLAGS: 00000246
[  140.280578] RAX: 00000000000435ef RBX: 0000000000000246 RCX: 0000000000000040
[  140.280592] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffffffff9be13e84
[  140.280603] RBP: ffff8eed18dde480 R08: 0000000000000001 R09: ffff8eebc2292760
[  140.280614] R10: 0000000000000005 R11: ffff8eebc2292760 R12: 0000000000000000
[  140.280625] R13: ffff9f58802e7900 R14: ffff8eed18dde480 R15: ffff8eed18dde480
[  140.280637]  ? _raw_spin_unlock_irqrestore+0x34/0x40
[  140.280649]  __mod_timer+0x274/0x400
[  140.280661]  ? wait_for_completion_timeout+0x76/0x110
[  140.280675]  schedule_timeout+0xa8/0x140
[  140.280687]  ? __next_timer_interrupt+0x100/0x100
[  140.280698]  wait_for_completion_timeout+0xa2/0x110
[  140.280714]  ath11k_start_scan+0x4c/0xf0 [ath11k]
[  140.280730]  ath11k_mac_op_hw_scan+0x1e9/0x2c0 [ath11k]
[  140.280763]  drv_hw_scan+0x79/0x260 [mac80211]
[  140.280789]  __ieee80211_scan_completed+0x379/0x440 [mac80211]
[  140.280816]  ieee80211_scan_work+0x12f/0x330 [mac80211]
[  140.280830]  process_one_work+0x279/0x5b0
[  140.280842]  worker_thread+0x49/0x300
[  140.280854]  ? process_one_work+0x5b0/0x5b0
[  140.280868]  kthread+0x135/0x150
[  140.280880]  ? __kthread_bind_mask+0x60/0x60
[  140.280891]  ret_from_fork+0x22/0x30
[  140.280903] irq event stamp: 275961
[  140.280918] hardirqs last  enabled at (275960): [<ffffffff9b270f1f>] __local_bh_enable_ip+0x6f/0xb0
[  140.280931] hardirqs last disabled at (275961): [<ffffffff9be13ca3>] _raw_spin_lock_irqsave+0x63/0x80
[  140.280946] softirqs last  enabled at (275944): [<ffffffffc055288d>] ath11k_ce_send+0x14d/0x260 [ath11k]
[  140.280958] softirqs last disabled at (275953): [<ffffffff9c000f72>] asm_call_irq_on_stack+0x12/0x20
[  140.280971] ---[ end trace 31cb94e18d401398 ]---

Tested-on: QCA6390 hw2.0 PCI WLAN.HST.1.0.1-01740-QCAHSTSWPLZ_V2_TO_X86-1

Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1606156046-24764-1-git-send-email-kvalo@codeaurora.org

wcn36xx: Send NULL data packet when exiting BMPS

This commit updates the BMPS exit path to be consistent with downstream in
terms of exiting BMPS mode. Downstream sets the flag to send a NULL data
frame to the host on exiting BMPS.

This will tell the AP to send any queued frames to the STA immediately.
Verified the relevant bit toggle in wireshark.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201120021403.2646574-2-bryan.odonoghue@linaro.org

ath10k: Constify static qmi structs

qmi_msg_handler[] and ath10k_qmi_ops are only used as input arguments
to qmi_handle_init() which accepts const pointers to both qmi_ops and
qmi_msg_handler. Make them const to allow the compiler to put them in
read-only memory.

Signed-off-by: Rikard Falkeborn <rikard.falkeborn@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201122234031.33432-3-rikard.falkeborn@gmail.com

ath10k: Release some resources in an error handling path

Should an error occur after calling 'ath10k_usb_create()', it should be
undone by a corresponding 'ath10k_usb_destroy()' call

Fixes: 4db66499df91 ("ath10k: add initial USB support")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201122170358.1346065-1-christophe.jaillet@wanadoo.fr

ath10k: Fix an error handling path

If 'ath10k_usb_create()' fails, we should release some resources and report
an error instead of silently continuing.

Fixes: 4db66499df91 ("ath10k: add initial USB support")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201122170342.1346011-1-christophe.jaillet@wanadoo.fr

ath10k: Fix the parsing error in service available event

The wmi service available event has been
extended to contain extra 128 bit for new services
to be indicated by firmware.

Currently the presence of any optional TLVs in
the wmi service available event leads to a parsing
error with the below error message:
ath10k_snoc 18800000.wifi: failed to parse svc_avail tlv: -71

The wmi service available event parsing should
not return error for the newly added optional TLV.
Fix this parsing for service available event message.

Tested-on: WCN3990 hw1.0 SNOC WLAN.HL.3.2.2-00720-QCAHLSWMTPL-1

Fixes: cea19a6ce8bf ("ath10k: add WMI_SERVICE_AVAILABLE_EVENT support")
Signed-off-by: Rakesh Pillai <pillair@codeaurora.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605501291-23040-1-git-send-email-pillair@codeaurora.org

ath11k: Fix an error handling path

If 'kzalloc' fails, we must return an error code.

While at it, remove a useless initialization of 'err' which could hide the
issue.

Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201122173943.1366167-1-christophe.jaillet@wanadoo.fr

ath11k: Build check size of ath11k_skb_cb

It is rather easy to add more entries to ath11k_skb_cb while forgetting the
size limit of ieee80211_tx_info->driver_data. So just check this during the
build to reduce the change of accidental buffer overflow in the skbuff->cb.

Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119154235.263250-3-sven@narfation.org

ath11k: Reset ath11k_skb_cb before setting new flags

It was observed that the codepath for the ATH11K_SKB_HW_80211_ENCAP was
used even when the IEEE80211_TX_CTRL_HW_80211_ENCAP was not enabled for a
an skbuff. This became even more prominent when the QCAs wlan-open patchset
for ath11k [1] was applied and a sane looking fix just caused crashes when
injecting frames via a monitor interface (for example with ratechecker):

  [   86.963152] Unable to handle kernel NULL pointer dereference at virtual address 00000338
  [   86.963192] pgd = ffffffc0008f0000
  [   86.971034] [00000338] *pgd=0000000051706003, *pud=0000000051706003, *pmd=0000000051707003, *pte=00e800000b000707
  [   86.984292] Internal error: Oops: 96000006 [#1] PREEMPT SMP
  [...]
  [   87.713339] [<ffffffbffc802480>] ieee80211_tx_status_8023+0xf8/0x220 [mac80211]
  [   87.715654] [<ffffffbffc98bad4>] ath11k_dp_tx_completion_handler+0x42c/0xa10 [ath11k]
  [   87.722924] [<ffffffbffc989190>] ath11k_dp_service_srng+0x70/0x3c8 [ath11k]
  [   87.730831] [<ffffffbffca03460>] 0xffffffbffca03460
  [   87.737599] [<ffffffc00046ef58>] net_rx_action+0xf8/0x288
  [   87.742462] [<ffffffc000097554>] __do_softirq+0xfc/0x220
  [   87.748014] [<ffffffc000097900>] irq_exit+0x98/0xe8
  [   87.753396] [<ffffffc0000cf188>] __handle_domain_irq+0x90/0xb8
  [   87.757999] [<ffffffc000081ca4>] gic_handle_irq+0x6c/0xc8
  [   87.763899] Exception stack(0xffffffc00081bdc0 to 0xffffffc00081bef0)

Problem is that the state of ath11k_skb_cb->flags must be considered
unknown and could contain anything when it is not manually initialized. So
it could also contain ATH11K_SKB_HW_80211_ENCAP. And this can result in the
code to assume that the ath11k_skb_cb->vif is set - even when this is not
always the case for non ATH11K_SKB_HW_80211_ENCAP transmissions.

Tested-on: IPQ8074 hw2.0 WLAN.HK.2.4.0.1.r1-00026-QCAHKSWPL_SILICONZ-2

[1] https://source.codeaurora.org/quic/qsdk/oss/system/feeds/wlan-open/tree/mac80211/patches?h=NHSS.QSDK.11.4.r3
    (162 patches at the moment which are often not upstreamed but essential
     to get ath11k working)

Fixes: e7f33e0c52c0 ("ath11k: add tx hw 802.11 encapsulation offloading support")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119154235.263250-2-sven@narfation.org

ath11k: Don't cast ath11k_skb_cb to ieee80211_tx_info.control

The driver_data area of ieee80211_tx_info is used in ath11k for
ath11k_skb_cb. The first function in the TX patch which rewrites it to
ath11k_skb_cb is already ath11k_mac_op_tx. No one else in the code path
must use it for something else before it reinitializes it. Otherwise the
data has to be considered uninitialized or corrupt.

But the ieee80211_tx_info.control shares exactly the same area as
ieee80211_tx_info.driver_data and ath11k is still using it. This results in
best case in a

  ath11k c000000.wifi1: no vif found for mgmt frame, flags 0x0

or (slightly worse) in a kernel oops.

Instead, the interesting data must be moved first into the ath11k_skb_cb
and ieee80211_tx_info.control must then not be used anymore.

Tested-on: IPQ8074 hw2.0 WLAN.HK.2.4.0.1.r1-00026-QCAHKSWPL_SILICONZ-2

Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices")
Signed-off-by: Sven Eckelmann <sven@narfation.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119154235.263250-1-sven@narfation.org

ath11k: remove "ath11k_mac_get_ar_vdev_stop_status" references

Unused structure ath11k_vdev_stop_status is removed.
'ath11k_mac_get_ar_vdev_stop_status' api has been replaced
with 'ath11k_mac_get_ar_by_vdev_id' inside vdev_stopped_event.

Signed-off-by: Ritesh Singh <ritesi@codeaurora.org>
Signed-off-by: Maharaja Kennadyrajan <mkenna@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605514143-17652-4-git-send-email-mkenna@codeaurora.org

ath11k: peer delete synchronization with firmware

Peer creation in firmware fails, if last peer deletion
is still in progress.
Hence, add wait for the event after deleting every peer
from host driver to synchronize with firmware.

Signed-off-by: Ritesh Singh <ritesi@codeaurora.org>
Signed-off-by: Maharaja Kennadyrajan <mkenna@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605514143-17652-3-git-send-email-mkenna@codeaurora.org

ath11k: vdev delete synchronization with firmware

When the interface is added immediately after removing the
interface, vdev deletion in firmware might not have been
completed.

Hence, add vdev_delete_resp_event and wait_event_timeout
to synchronize with firmware.

Signed-off-by: Ritesh Singh <ritesi@codeaurora.org>
Signed-off-by: Maharaja Kennadyrajan <mkenna@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605514143-17652-2-git-send-email-mkenna@codeaurora.org

Merge mhi-ath11k-immutable into ath-next

This is an immutable branch from mhi tree for handling an API change in MHI
subsystem and minimise conflicts between ath and mhi trees.

cw1200: fix missing destroy_workqueue() on error in cw1200_init_common

Add the missing destroy_workqueue() before return from
cw1200_init_common in the error handling case.

Fixes: a910e4a94f69 ("cw1200: add driver for the ST-E CW1100 & CW1200 WLAN chipsets")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Qinglang Miao <miaoqinglang@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119070842.1011-1-miaoqinglang@huawei.com

mwifiex: Remove duplicated REG_PORT definition

The REG_PORT is defined twice, so remove one of them.

Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119101204.72fd5f0a@xhacker.debian

rtlwifi: rtl8723ae: avoid accessing the data mapped to streaming DMA

In rtl8723e_tx_fill_cmddesc(), skb->data is mapped to streaming DMA on
line 531:
  dma_addr_t mapping = dma_map_single(..., skb->data, ...);

On line 534, skb->data is assigned to hdr after cast:
  struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)(skb->data);

Then hdr->frame_control is accessed on line 535:
  __le16 fc = hdr->frame_control;

This DMA access may cause data inconsistency between CPU and hardwre.

To fix this bug, hdr->frame_control is accessed before the DMA mapping.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119015218.12220-1-baijiaju1990@gmail.com

rtlwifi: rtl8192de: avoid accessing the data mapped to streaming DMA

In rtl92de_tx_fill_cmddesc(), skb->data is mapped to streaming DMA on
line 667:
  dma_addr_t mapping = dma_map_single(..., skb->data, ...);

On line 669, skb->data is assigned to hdr after cast:
  struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)(skb->data);

Then hdr->frame_control is accessed on line 670:
  __le16 fc = hdr->frame_control;

This DMA access may cause data inconsistency between CPU and hardwre.

To fix this bug, hdr->frame_control is accessed before the DMA mapping.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119015205.12162-1-baijiaju1990@gmail.com

rtlwifi: rtl8192ce: avoid accessing the data mapped to streaming DMA

In rtl92ce_tx_fill_cmddesc(), skb->data is mapped to streaming DMA on
line 530:
  dma_addr_t mapping = dma_map_single(..., skb->data, ...);

On line 533, skb->data is assigned to hdr after cast:
  struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)(skb->data);

Then hdr->frame_control is accessed on line 534:
  __le16 fc = hdr->frame_control;

This DMA access may cause data inconsistency between CPU and hardwre.

To fix this bug, hdr->frame_control is accessed before the DMA mapping.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119015151.12110-1-baijiaju1990@gmail.com

rtlwifi: rtl8188ee: avoid accessing the data mapped to streaming DMA

In rtl88ee_tx_fill_cmddesc(), skb->data is mapped to streaming DMA on
line 677:
  dma_addr_t mapping = dma_map_single(..., skb->data, ...);

On line 680, skb->data is assigned to hdr after cast:
  struct ieee80211_hdr *hdr = (struct ieee80211_hdr *)(skb->data);

Then hdr->frame_control is accessed on line 681:
  __le16 fc = hdr->frame_control;

This DMA access may cause data inconsistency between CPU and hardwre.

To fix this bug, hdr->frame_control is accessed before the DMA mapping.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201119015127.12033-1-baijiaju1990@gmail.com

mwifiex: Fix fall-through warnings for Clang

In preparation to enable -Wimplicit-fallthrough for Clang, fix multiple
warnings by explicitly adding multiple break statements instead of
letting the code fall through to the next case.

Link: https://github.com/KSPP/linux/issues/115
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201117160958.GA18807@embeddedor

rsi: fix error return code in rsi_reset_card()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: 17ff2c794f39 ("rsi: reset device changes for 9116")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zhang Changzhong <zhangchangzhong@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605582454-39649-1-git-send-email-zhangchangzhong@huawei.com

brcmsmac: ampdu: Check BA window size before checking block ack

bindex can be out of BA window (64):
  tid 0 seq 2983, start_seq 2915, bindex 68, index 39
  tid 0 seq 2984, start_seq 2915, bindex 69, index 40
  tid 0 seq 2985, start_seq 2915, bindex 70, index 41
  tid 0 seq 2986, start_seq 2915, bindex 71, index 42
  tid 0 seq 2879, start_seq 2915, bindex 4060, index 63
  tid 0 seq 2854, start_seq 2915, bindex 4035, index 38
  tid 0 seq 2795, start_seq 2915, bindex 3976, index 43
  tid 0 seq 2989, start_seq 2924, bindex 65, index 45
  tid 0 seq 2992, start_seq 2924, bindex 68, index 48
  tid 0 seq 2993, start_seq 2924, bindex 69, index 49
  tid 0 seq 2994, start_seq 2924, bindex 70, index 50
  tid 0 seq 2997, start_seq 2924, bindex 73, index 53
  tid 0 seq 2795, start_seq 2941, bindex 3950, index 43
  tid 0 seq 2921, start_seq 2941, bindex 4076, index 41
  tid 0 seq 2929, start_seq 2941, bindex 4084, index 49
  tid 0 seq 3011, start_seq 2946, bindex 65, index 3
  tid 0 seq 3012, start_seq 2946, bindex 66, index 4
  tid 0 seq 3013, start_seq 2946, bindex 67, index 5

In result isset() will try to dereference something on the stack,
causing panics:
  BUG: unable to handle page fault for address: ffffa742800ed01f
  #PF: supervisor read access in kernel mode
  #PF: error_code(0x0000) - not-present page
  PGD 6a4e9067 P4D 6a4e9067 PUD 6a4ec067 PMD 6a4ed067 PTE 0
  Oops: 0000 [#1] PREEMPT SMP PTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Not tainted 5.8.5-arch1-1-kdump #1
  Hardware name: Apple Inc. MacBookAir3,1/Mac-942452F5819B1C1B, BIOS    MBA31.88Z.0061.B07.1201241641 01/24/12
  RIP: 0010:brcms_c_ampdu_dotxstatus+0x343/0x9f0 [brcmsmac]
  Code: 54 24 20 66 81 e2 ff 0f 41 83 e4 07 89 d1 0f b7 d2 66 c1 e9 03 0f b7 c9 4c 8d 5c 0c 48 49 8b 4d 10 48 8b 79 68 41 57 44 89 e1 <41> 0f b6 33 41 d3 e0 48 c7 c1 38 e0 ea c0 48 83 c7 10 44 21 c6 4c
  RSP: 0018:ffffa742800ecdd0 EFLAGS: 00010207
  RAX: 0000000000000019 RBX: 000000000000000b RCX: 0000000000000006
  RDX: 0000000000000ffe RSI: 0000000000000004 RDI: ffff8fc6ad776800
  RBP: ffff8fc6855acb00 R08: 0000000000000001 R09: 00000000000005d9
  R10: 00000000fffffffe R11: ffffa742800ed01f R12: 0000000000000006
  R13: ffff8fc68d75a000 R14: 00000000000005db R15: 0000000000000019
  FS:  0000000000000000(0000) GS:ffff8fc6aad00000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: ffffa742800ed01f CR3: 000000002480a000 CR4: 00000000000406e0
  Call Trace:
   <IRQ>
   brcms_c_dpc+0xb46/0x1020 [brcmsmac]
   ? wlc_intstatus+0xc8/0x180 [brcmsmac]
   ? __raise_softirq_irqoff+0x1a/0x80
   brcms_dpc+0x37/0xd0 [brcmsmac]
   tasklet_action_common.constprop.0+0x51/0xb0
   __do_softirq+0xff/0x340
   ? handle_level_irq+0x1a0/0x1a0
   asm_call_on_stack+0x12/0x20
   </IRQ>
   do_softirq_own_stack+0x5f/0x80
   irq_exit_rcu+0xcb/0x120
   common_interrupt+0xd1/0x200
   asm_common_interrupt+0x1e/0x40
  RIP: 0010:cpuidle_enter_state+0xb3/0x420

Check if the block is within BA window and only then check block's
status. Otherwise as Behan wrote: "When I came back to Dublin I
was courtmartialed in my absence and sentenced to death in my absence,
so I said they could shoot me in my absence."

Also reported:
https://bbs.archlinux.org/viewtopic.php?id=258428
https://lore.kernel.org/linux-wireless/87tuwgi92n.fsf@yujinakao.com/

Reported-by: Yuji Nakao <contact@yujinakao.com>
Signed-off-by: Dmitry Safonov <dima@arista.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201116030635.645811-1-dima@arista.com

brcmfmac: Fix incorrect type in assignment

The left-hand side of the assignment from cpu_to_le32() should be of
type __le32. This commit clears the warning reported by sparse when
building with C=1 CF="-D__CHECK_ENDIAN__".

Fixes: d56fd83cf99c ("brcmfmac: fix SDIO access for big-endian host")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Remi Depommier <rde@setrix.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201116001639.31958-1-rde@setrix.com

qtnfmac: fix error return code in qtnf_pcie_probe()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: b7da53cd6cd1 ("qtnfmac_pcie: use single PCIe driver for all platforms")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201114123347.29632-1-wanghai38@huawei.com

rtlwifi: rtl8192de: remove the useless value assignment

The variable u4tmp is overwritten by the following call and the assignment
is useless, so remove it.

Reported-by: Tosk Robot <tencent_os_robot@tencent.com>
Signed-off-by: Kaixu Xia <kaixuxia@tencent.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605332735-9648-1-git-send-email-kaixuxia@tencent.com

orinoco: Annotate ezusb_read_ltv()

ezusb_read_ltv() is always invoked via the ->read_ltv() callback. This
callback is mostly invoked under orinoco_lock() which disables BH.

There are a few invocations during probe which occur in preemptible
context via:
   ezusb_probe() -> orinoco_init() -> determine_fw_capabilities()

Extend `hermes_ops' with the ->read_ltv_pr callback which is implemented
with the same callback like ->read_ltv on `hermes_ops_local'.
On `ezusb_ops' ->read_ltv is used for callbacks under the lock which
need to poll.
The new ->read_ltv_pr() is used in the preemptible context in which it
is possible to wait for the completion. Provide HERMES_READ_RECORD_PR()
and hermes_read_wordrec_pr() which behave like their non _pr equivalents
and invoke ->read_ltv_pr().

This removes the last user of ezusb_req_ctx_wait() and can now be
removed.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-11-bigeasy@linutronix.de

orinoco: Annotate ezusb_docmd_wait()

All invocations of ezusb_docmd_wait() happen via ->cmd_wait(). This
callback is always invoked under the orinoco_lock() which disables BH.

Use ezusb_req_ctx_wait_poll() for ezusb_docmd_wait() because it must not
sleep.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-10-bigeasy@linutronix.de

orinoco: Remove ezusb_doicmd_wait()

ezusb_doicmd_wait() is invoked via ->init_cmd_wait() callback.
This callback is only invoked hermesi_program_init() and
hermesi_program_end() which are the ->program_init() and ->program_end()
callbacks as assigned by `hermes_ops_local'. They are never used by the
USB interface since the USB interface provides its own set of callbacks
by `ezusb_ops'.

Replace ezusb_doicmd_wait() with a warning in case I missed the obvious.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-9-bigeasy@linutronix.de

orinoco: Annotate ezusb_write_ltv()

All invocation of ezusb_write_ltv() happen via ->write_ltv() and are
performed under the orinoco_lock() which disables BH.

Use ezusb_req_ctx_wait_poll() for ezusb_write_ltv() because it must not
sleep.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-8-bigeasy@linutronix.de

orinoco: Annotate ezusb_read_pda()

ezusb_read_pda() is invoked via ->read_pda() while firmware is loaded in
preemtible context.

Use ezusb_req_ctx_wait_compl() in ezusb_read_pda().

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-7-bigeasy@linutronix.de

orinoco: Annotate firmware loading

The ezusb_program() is invoked via ->program() in preemptible
context during firmware loading. This is also true for the
->program_init() and ->program_end() callback.

Use ezusb_req_ctx_wait_compl() in ezusb_program_init(),
ezusb_program_bytes(), ezusb_program_end() which are part of firmware
loading during device probe.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-6-bigeasy@linutronix.de

orinoco: Annotate ezusb_init()

ezusb_init() is always invoked in preemptible context during device
probe. Only orinoco_up() -> orinoco_reinit_firmware() may invoke the
function from atomic context but this is never used for the USB
interface.

Use ezusb_req_ctx_wait_compl() for the ezusb_write_ltv() and
ezusb_docmd_wait() invocations from within ezusb_init().
Preserve the generic versions which have still other user via the
callback.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-5-bigeasy@linutronix.de

orinoco: Annotate ezusb_xmit()

ezusb_xmit() sets ->in_rid in its request which means it does not wait
for an answer.

Use the ezusb_req_ctx_wait_skip() to denote that an answer is not
expected.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-4-bigeasy@linutronix.de

orinoco: Prepare stubs for in_interrupt() removal

ezusb_access_ltv() sends the prepared request to the USB device.
Requests which have ->in_rid set expect an answer from the USB device
and the function has to wait until the URB with the answer arrives.
The function uses in_interrupt() to determine if it can simply sleep on
the completion and be woken up once the answer arrives or if it needs to
poll on the completion.

The usage of in_interrupt() in drivers is phased out and Linus clearly
requested that code which changes behaviour depending on context should
either be separated or the context be conveyed in an argument passed by the
caller, which usually knows the context.

Aside of that in_interrupt() is not correct as it does not catch preempt
disabled regions in which sleeping is also not allowed.

Provide stubs which can be used as a replacement. The current default is
the current behaviour which sleeps/polls depending on in_interrupt().
The goal is to audit all callers and use either the poll or sleep
version.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-3-bigeasy@linutronix.de

orinoco: Move context allocation after processing the skb

ezusb_xmit() allocates a context which is leaked if
orinoco_process_xmit_skb() returns an error.

Move ezusb_alloc_ctx() after the invocation of
orinoco_process_xmit_skb() because the context is not needed so early.
ezusb_access_ltv() will cleanup the context in case of an error.

Fixes: bac6fafd4d6a0 ("orinoco: refactor xmit path")
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201113212252.2243570-2-bigeasy@linutronix.de

brcmfmac: fix error return code in brcmf_cfg80211_connect()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: 3b1e0a7bdfee ("brcmfmac: add support for SAE authentication offload")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Zhang Changzhong <zhangchangzhong@huawei.com>
Reviewed-by: Chi-hsien Lin <chi-hsien.lin@infineon.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605248896-16812-1-git-send-email-zhangchangzhong@huawei.com

cw1200: replace a set of atomic_add()

a set of atomic_inc() looks more readable

Signed-off-by: Yejune Deng <yejune.deng@gmail.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1604991491-27908-1-git-send-email-yejune.deng@gmail.com

rtw88: wow: print key type when failing

It's much easier to spot what went wrong when you print out the
unhandled key type.

Signed-off-by: Brian Norris <briannorris@chromium.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112181432.2958021-1-briannorris@chromium.org

rtw88: coex: change the decode method from firmware

Fix sometimes FW information will be parsed as wrong value,
do a correction of sign bit to show the correct information.
(Ex, Value should be 20, but it shows 236.)

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-12-pkshih@realtek.com

rtw88: coex: remove unnecessary WLAN slot extend

Remove extend 5 ms related operation at WLAN media status notify.
The mechanism should be enable only while DUT connect to a AP that it
doesn't follow our power save control or its rate is decreasing unusually
sensitive. So we extend a extra slot to save the transmission rate.

The original logic will always extend WLAN slot.
It will lead to BT slot always decrease 5 ms.

Remove write scoreboard action at WLAN media status notify.
It has already done in WLAN info notify.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-11-pkshih@realtek.com

rtw88: coex: remove write scan bit to scoreboard in scan and connect notify

Remove some action which were writing scan bit to scoreboard,
since these behaviors are redundant.
It is already done in WLAN info notify.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-10-pkshih@realtek.com

rtw88: coex: fix BT performance drop during initial/power-on step

Force set the coexistence to BT high priority during
WLAN initial/power-on step. Since the duration the related setting
may be not ready yet.

The score board is not related to scan when initialing, remove the scan
parameter.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-9-pkshih@realtek.com

rtw88: coex: Change antenna setting to enhance free-run performance

While the mechanism goes to 2G_free_run or WL5G_BT2G,
set BT to hardware PTA mode to get a more efficiently performance.

Add a flag to indicate antenna switch is supported or not so
that the IC serials which has different antenna structure can
set antenna correctly.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-8-pkshih@realtek.com

rtw88: coex: update the TDMA parameter when leave LPS

The updated parameter and original parameter all means TDMA-OFF.
The original setting write 0x8 to firmware, but it only works for
some old IC series.
To avoid the confusing, update a proper parameter.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-7-pkshih@realtek.com

rtw88: coex: add the mechanism for RF4CE

Add the related coexistence mechanism about RF4CE and WLAN connected
flag to decide control packet is whether higher or normal priority.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-6-pkshih@realtek.com

rtw88: coex: Add force flag for coexistence table function

Because sometimes hardware action may be unpredictable.
(Ex: after WLAN LPS...etc)
Add a check mechanism if the action is triggered by
some concerned case, it can force to write the table again.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-5-pkshih@realtek.com

rtw88: coex: add write scoreboard action when WLAN in critical procedure

When WLAN in critical procedure, such as connecting, scan and so on,
coexistence will notify BT by scoreboard.
While BT firmware received the scoreboard interrupt, it will adjust
BT page priority lower to avoid WLAN critical procedure suffering impact.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-4-pkshih@realtek.com

rtw88: coex: remove unnecessary feature/function

Because the WLAN RSSI report feature had been implemented at WLAN link info
update function, it is redundant to update WLAN RSSI while BT is updating
link info, so remove the update WLAN RSSI part.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-3-pkshih@realtek.com

rtw88: coex: update TDMA settings for different beacon interval

Add considering for different WLAN beacon interval in coexistence
mechanism.
Because the WLAN beacon period may be not 100 ms, so it's necessary
to consider any beacon period and set timer according to the interval.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201112031430.4846-2-pkshih@realtek.com

rtw88: add CCK_PD debug log

In order to help debugging in field, we add some log messages.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109090123.9254-2-pkshih@realtek.com

rtw88: 8723d: add cck pd seetings

CCK PD can reduce the number of false alarm of the CCK rates.
It dynamically adjusts the power threshold and CS ratio.
The values are compared to the values of the previous level, if
the level is changed, set new values of power threshold and CS
ratio.

Implement rtw_chip_ops::cck_pd_set() for 8723d.

Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109090123.9254-1-pkshih@realtek.com

ath11k: Fix the rx_filter flag setting for peer rssi stats

Set the rx_filter in ath11k_mac_config_mon_status_default(),
only when the rx_filter value exists in ath11k_debug_rx_filter().

Without this change, rx_filter gets set to 0 and peer rssi stats
aren't updating properly from firmware.

Tested-on: IPQ8074 WLAN.HK.2.1.0.1-01230-QCAHKSWPL_SILICONZ-4

Fixes: ec48d28ba291 ("ath11k: Fix rx_filter flags setting for per peer rx_stats")

Signed-off-by: Maharaja Kennadyrajan <mkenna@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1605091117-11005-1-git-send-email-mkenna@codeaurora.org

ath10k: add target IRAM recovery feature support

This target IRAM recovery feature support is to copy target
IRAM contents available at ATH10K_MEM_REGION_TYPE_REG to host
memory for back up after firmware loaded. Target IRAM contents
are copied to wmi memory chunks allocated for the
WMI_IRAM_RECOVERY_HOST_MEM_REQ_ID and provide the wmi chunks
address to the firmware through wmi init command.

If firmware detects andy IRAM corruption through periodic
checksum validation, It will download the IRAM contents back
from the provided wmi memory chunks address using hif_memcpy.

This IRAM recovery feature prevent target assert in case of
unexpected target IRAM corruptions.

This patch also introduce a new feature flag 'iram-recovery' for
backward compatibility.

Tested-on: QCA9888 10.4-3.9.0.2-00094

Signed-off-by: Venkateswara Naralasetty <vnaralas@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1604657442-23674-1-git-send-email-vnaralas@codeaurora.org

ath11k: Fix beamformee STS in HE cap

STS Beamformee capability sent from firmware is overwritten as
num_tx_chains - 1.
When num_tx_chains is 2, then STS is set to invalid value 1.
Since STS is not limited by number of TX chain, not overwriting the
capability received from firmware.

Tested on: IPQ6018 WLAN.HK.2.4.0.1-00303-QCAHKSWPL_SILICONZ-1

Signed-off-by: Ramya Gnanasekar <rgnanase@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1604933229-15815-1-git-send-email-rgnanase@codeaurora.org

ath11k: add processor_id based ring_selector logic

Current ring selector logic follows round-robin scheduling,
when a process switch from one processor to another each time,
the data in the cache would have to be invalidated
and re-loaded from RAM leads to poor cache utilization.

Therefore smp_processor_id() attempt to keep processes
on the same processor via processor affinity which improves
cache utilization. Also, ring selection could be chosen round
robin when the processor_id based ring selection does not
have the resources.

Tested-on : IPQ8074 hw2.0 AHB WLAN.HK.2.4.0.1-00874-QCAHKSWPL_SILICONZ-1

Signed-off-by: P Praneesh <ppranees@codeaurora.org>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1604914915-12831-1-git-send-email-ppranees@codeaurora.org

net: qrtr: Unprepare MHI channels during remove

Reset MHI device channels when driver remove is called due to
module unload or any crash scenario. This will make sure that
MHI channels no longer remain enabled for transfers since the
MHI stack does not take care of this anymore after the auto-start
channels feature was removed.

Signed-off-by: Bhaumik Bhatt <bbhatt@codeaurora.org>
Acked-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

net: qrtr: Start MHI channels during init

Start MHI device channels so that transfers can be performed.
The MHI stack does not auto-start channels anymore.

Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

bus: mhi: Remove auto-start option

There is really no point having an auto-start for channels.
This is confusing for the device drivers, some have to enable the
channels, others don't have... and waste resources (e.g. pre allocated
buffers) that may never be used.

This is really up to the MHI device(channel) driver to manage the state
of its channels.

While at it, let's also remove the auto-start option from ath11k mhi
controller.

Signed-off-by: Loic Poulain <loic.poulain@linaro.org>
Acked-by: Kalle Valo <kvalo@codeaurora.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
[mani: clubbed ath11k change]
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

rtw88: coex: simplify the setting and condition about WLAN TX limitation

Life time related setting is related to WLAN active port number,
not its mode. The original setting may cause poor WLAN performance.

In most case, WLAN TX limitation is helpful for BT in coexistence
scenario, especially for A2DP quality.
So this is necessary to fix the setting and constraint it.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-12-pkshih@realtek.com

rtw88: coex: add debug message

Since coexistence issue is related to WL/BT and each digital/analog/rf,
and these issues are often critical with low failure rate, add more
debugging information is helpful to clarify issues.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-11-pkshih@realtek.com

rtw88: coex: update WLAN 5G AFH parameter for 8822b

Adaptive Frequency Hopping(AFH) is a method of transmitting
radio signals by rapidly changing the carrier frequency among
many distinct frequencies, which can avoid interference.
Add this feature to update AFH parameter for 8822b.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-10-pkshih@realtek.com

rtw88: coex: change the parameter for A2DP when WLAN connecting

The original mechanism may cause A2DP glitch during WiFi connecting AP.
Because the original TDMA may decrease too much A2DP slot.
This patch add a timer and variable to let the case A2DP + WL_Connecting
performed more well.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-9-pkshih@realtek.com

rtw88: coex: modified for BT info notify

Add counter to count BT info C2H command for debug usage.
It could present the C2H channel situation.

Fix BT IQK state decision condition for counting.
The original condition is wrong, it would cause coexistence mechanism
going to the wrong strategy.

New format supports getting TDMA parameter from WL firmware, since
coexistence needs current TDMA parameter to decide strategy.

Remove unnecessary operations about scoreboard, it is not a must for
current version.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-8-pkshih@realtek.com

rtw88: coex: add separate flag for manual control

The original stop_dm flag is shared with the wifi and coexistence
mechanism internal usage, which represent two meanings, and makes
developers hard to debug.
Add a new flag manual_control for coexistence mechanism usage only.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-7-pkshih@realtek.com

rtw88: coex: Modify the timing of set_ant_path/set_rf_para

To make sure the related ANT/RF_PARA setting will be expected result in COEX.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-6-pkshih@realtek.com

rtw88: coex: coding style adjustment

Adjust space and comments, and fix "line over 80 characters" warnings
reported by checkpatch.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-5-pkshih@realtek.com

rtw88: coex: reduce magic number

Use macro definition to replace magic number.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-4-pkshih@realtek.com

rtw88: coex: update coex parameter to improve A2DP quality

Update COEX parameters and logic to enhance WL/BT performance
while WL_Busy + A2DP in a less interference environment.
It can avoid the interference comes cross from each other
and earned more performance.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-3-pkshih@realtek.com

rtw88: coex: fixed some wrong register definition and setting

Some register definition and bit definition were incorrect.
e.g. REG_BT_COEX_V2 should be word alignment to meet the coding style.
e.g. set REG_BT_TDMA_TIME[5:0]=0x5,
But the original is to set REG_BT_TDMA_TIME[7:0]=0x5.
This will cause unexpected hardware behavior.

Signed-off-by: Ching-Te Ku <ku920601@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/20201109085909.9143-2-pkshih@realtek.com

rtlwifi: fix spelling typo of workaround

workarould -> workaround

Signed-off-by: Wang Qing <wangqing@vivo.com>
Signed-off-by: Kalle Valo <kvalo@codeaurora.org>
Link: https://lore.kernel.org/r/1604738439-24794-1-git-send-email-wangqing@vivo.com