Felix Fietkau [Sat, 26 Sep 2020 20:40:41 +0000 (22:40 +0200)]
mt76: mt7915: support 32 station interfaces
When looking for a MAC address slot, start by using main BSSID slots 1-3,
afterwards use 16 repeater mode BSSID slots, then start using the slots
usually used for AP mode.
This search order should prevent unnecessary conflicts with AP mode interfaces
on the same PHY
Felix Fietkau [Fri, 25 Sep 2020 13:56:49 +0000 (15:56 +0200)]
mt76: mt7615: add debugfs knob for setting extended local mac addresses
This is primarily for testing and can be used in combination with monitor
mode to make the card respond to packets sent to a specific MAC address.
For now this is only exposed as a debug/testing feature, later on the
approach might be used to support more concurrent station interfaces
Lorenzo Bianconi [Sat, 26 Sep 2020 15:06:22 +0000 (17:06 +0200)]
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>
Ajay Singh [Wed, 25 Nov 2020 11:41:10 +0000 (11:41 +0000)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:59 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:58 +0000 (10:10 +0800)]
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).
Ching-Te Ku [Thu, 26 Nov 2020 02:10:57 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:56 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:55 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:54 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:53 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:51 +0000 (10:10 +0800)]
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.
Ching-Te Ku [Thu, 26 Nov 2020 02:10:50 +0000 (10:10 +0800)]
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.
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.
Bryan O'Donoghue [Tue, 24 Nov 2020 15:59:19 +0000 (17:59 +0200)]
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.
Rikard Falkeborn [Tue, 24 Nov 2020 15:59:18 +0000 (17:59 +0200)]
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.
Rakesh Pillai [Tue, 24 Nov 2020 15:59:17 +0000 (17:59 +0200)]
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.
Sven Eckelmann [Tue, 24 Nov 2020 15:59:16 +0000 (17:59 +0200)]
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.
Sven Eckelmann [Tue, 24 Nov 2020 15:59:15 +0000 (17:59 +0200)]
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):
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.
[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)
Sven Eckelmann [Tue, 24 Nov 2020 15:59:15 +0000 (17:59 +0200)]
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.
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.
Ritesh Singh [Tue, 24 Nov 2020 15:59:14 +0000 (17:59 +0200)]
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.
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.
Dmitry Safonov [Mon, 16 Nov 2020 03:06:35 +0000 (03:06 +0000)]
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
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/
Remi Depommier [Mon, 16 Nov 2020 00:16:39 +0000 (19:16 -0500)]
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
Wang Hai [Sat, 14 Nov 2020 12:33:47 +0000 (20:33 +0800)]
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
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.
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.
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.
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.
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.
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.
Zhang Changzhong [Fri, 13 Nov 2020 06:28:16 +0000 (14:28 +0800)]
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
Ching-Te Ku [Thu, 12 Nov 2020 03:14:30 +0000 (11:14 +0800)]
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.)
Ching-Te Ku [Thu, 12 Nov 2020 03:14:29 +0000 (11:14 +0800)]
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.
Ching-Te Ku [Thu, 12 Nov 2020 03:14:25 +0000 (11:14 +0800)]
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.
Ching-Te Ku [Thu, 12 Nov 2020 03:14:23 +0000 (11:14 +0800)]
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.
Ching-Te Ku [Thu, 12 Nov 2020 03:14:22 +0000 (11:14 +0800)]
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.
Ching-Te Ku [Thu, 12 Nov 2020 03:14:21 +0000 (11:14 +0800)]
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.
Ching-Te Ku [Thu, 12 Nov 2020 03:14:20 +0000 (11:14 +0800)]
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.
Ping-Ke Shih [Mon, 9 Nov 2020 09:01:22 +0000 (17:01 +0800)]
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.
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.
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
P Praneesh [Mon, 9 Nov 2020 09:41:55 +0000 (15:11 +0530)]
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.
Bhaumik Bhatt [Wed, 18 Nov 2020 18:20:25 +0000 (10:20 -0800)]
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.
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>
Ching-Te Ku [Mon, 9 Nov 2020 08:59:09 +0000 (16:59 +0800)]
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.
Ching-Te Ku [Mon, 9 Nov 2020 08:59:08 +0000 (16:59 +0800)]
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.
Ching-Te Ku [Mon, 9 Nov 2020 08:59:07 +0000 (16:59 +0800)]
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.
Ching-Te Ku [Mon, 9 Nov 2020 08:59:06 +0000 (16:59 +0800)]
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.