In WMI_TDLS_SET_OFFCHAN_MODE_CMDID the offchannel freq is set to invalid
value due to uninitialized local tdls_channel_switch_params type variable
resulting in fw reading incorrect non zero offchanel freq.
Fix, to zero initialize the local tdls_channel_switch_params variable.
Change-Id: Ia7d6daac098a179a22676620875a2b0d6488a8ad
CRs-Fixed: 2739267
If wma_send_msg_high_priority failed to send msg, msg body is freed,
can't access later.
Move wma_delete_sta_req_sta_mode and log to avoid use after free.
Duplicate MLME_IS_ROAM_SYNCH_IN_PROGRESS is called in 3 functions in
same stack, clean up it.
Change-Id: I28b7cc9f12902aefa37bf1345c94ec6f0b581297
CRs-Fixed: 2748407
As per requirement, send a value of an INI "roam_score_delta" to FW
through WMI_ROAM_ENABLE_DISABLE_TRIGGER_REASON_CMDID.
Change-Id: I9a2087f83c2eb95338d440dc901bd8aa246f7e5f
CRs-Fixed: 2740552
Since sme_ps_enable_auto_ps_timer will be called in
different scenarios, and there is error level output
if trying to start it when it's already started.
To avoid the useless error log, check the state
before starting this timer.
CRs-Fixed: 2747427
Change-Id: If119da266a991b6b5990ad3077e41c2bba11a798
On some cases, dp_rx_thread can not handle nbufs in time,
then a large number of packets are pending in
rx_thread->nbuf_queue, which run out of system memory at
last and failed to malloc new buffers for refill buffer ring,
fw crash is encountered. To avoid this, drop RX packets when
the pending number becomes large.
Change-Id: Ie4e001dc584821cf68253aa3a15a5fd72394c736
CRs-Fixed: 2737181
CLD driver specific time sync feature is no longer used since
core BSP provides more accurate time sync over QMI and hence
disable CLD driver specific time sync feature to avoid periodic
wake ups.
Change-Id: Ic2790e7cdb6cf56f874fb56cda246777bc34d2e3
CRs-Fixed: 2715021
Enable NS offload only when IPv6 address is added. No need to
call it when deleting IPv6 address and ifdown.
Change-Id: I5eb282d1deb1bc101871fa0952ac37796eb16db0
CRs-Fixed: 2747574
Print BTM BSS termination timeout value and BTM MBO assoc
retry timeout value coming from FW through
wmi_roam_trigger_reason_tlv_param tlv.
Print Blacklist reason, Source of adding AP to BL, timestamp
which indicates the time when AP added to blacklist and
Original timeout value in milliseconds when AP added to BL
coming from FW through wmi_roam_ap_info_tlv_param.
Change-Id: Iea5b8c136f83841d679b9ad06721bb0c9349b46a
CRs-Fixed: 2744250
Currently driver sets the NOL state of the channels to false
on every regulatory updated, which indicates the channel is
not in NOL. Which may lead to some issues where the channel
is actually in NOL but host treats it as non-NOL channel.
Ideally NOL list should be maintained throughout the driver
lifetime and across the regulatory changes.
To address this issue add a logic to not update the NOL state
of the channels whenever the regulatory update is received.
CRs-Fixed: 2744916
Change-Id: If96b22ab29a60a2aa752bbf01aaae46fc37362eb
Currently only STA and SAP support PMKSA cache, and return
EOPNOTSUPP for other modes.
Change-Id: Ic8969ffd58815c4f3e2e5e286521421c38eb7f56
CRs-Fixed: 2747485
vdev mlme chan info isn't updated after LFR3 roaming, so
wlan_hdd_cfg80211_get_channel get old center_freq1 from vdev mlme,
center_freq1 isn't same as center_freq for 20M channel,
cfg80211_chandef_valid return false.
Change-Id: I4037f80627d207bdf9ba60cd9c3b7be82bad81e8
CRs-Fixed: 2746246
Enable PM QoS support for 5.x kernel by adopting dev PM QoS APIs to
request device structures for CPUs and then requesting QoS on those CPU
devices.
Change-Id: I9c7565b44a72145e11178fe360f413ff578142cf
CRs-Fixed: 2734025
For 6ghz AP with open security,the STA join req was not going through
due to security check added from CR2715086.
Fix is to move the rmf enabled check inside lim_is_6g_allowed_sec().
Change-Id: I32fb0de34fdf05063dee5f28aac164b1dfcaea5e
CRs-Fixed: 2746745
Currently mac trace msg rx function is passing module
id as a second argument to mac trace function instead of
message id.
As mac trace expects message id as an argument, pass message
id to the mac trace function as message id.
CRs-Fixed: 2719737
Change-Id: I9fa1611d6095975241a01325b798c24ea654df3e
After SAP switches the channel, policy manager needs to check if
the host needs to change HW mode. SAP state machine updates channel
change event to hdd to indicate SAP channel switch. Hence move logic
to change HW mode from start bss success event to channel change
event case in sap event callback.
Problem scenario:
1. DUT is in Single MAC mode by default.
2. SAP is up on Channel-1. So SAP is on MAC-0.
3. Host sets Policy Mgr Hw mode = DBS.
4. STA comes up on Channel-36. As host is in DBS mode, STA came up
on MAC-1.
5. SAP does CSA to Channel-36 (STA’s channel). So SAP moves from
MAC-1 to MAC-0.
6. Host sets Policy Mgr HW mode = Single MAC (As both VDEVs are on
same band).
7. Now, SAP does CSA to Channel-11. As Policy Manager Hw mode is still
Single MAC, SAP is still on MAC-0. So now STA (Channel-36) & SAP
(Channel-1) are in MAC-0 doing MCC.
Fix:
After SAP switches channel, policy manager checks and changes HW mode
change in hdd sap event callback.
Change-Id: If7964057592d85a11dd77197c889a4ed77ed039a
CRs-Fixed: 2719810
After roaming, FW updates host by roam sync indication. Host parses
the reassoc request in roam sync indication to update RSN params of
vdev crypto.
Change-Id: I9638a77150e81ea911e95d294c58d605871630a0
CRs-Fixed: 2746470
The last parameter of stop_roaming_callback is currently an
unsigned int, which does not match the function that is assigned
to this pointer, which has enum wlan_cm_rso_control_requestor.
Update the function pointer to have the correct parameter.
Change-Id: I2761293e6d87e9fcf9586d198a6e60f25ca2b51e
CRs-fixed: 2745188
Currently idle timer does not gets de-initialize on tdls peer
reset because of which there could be a possibility that in
case of multiple tdls connection and disconnection with multiple
peers, this timer might get assigned and re-assigned to different
peers which may lead to an issue where this timer never gets
started for the required peer.
In current scenario whenever the first TDLS connection
(Peer a with sta_idx 0) happens, tdls peer idle timer object is
initialized with “&tdls_soc->tdls_conn_info[0];”.
Now if this peer a gets disconnected and peer b gets connected
and peer gets the same sta_idx 0, then peer will be associated
with timer “&tdls_soc->tdls_conn_info[0];”,hence 2 timer init
happens with the same sta_idx for peer a and peer b.
(here peer a is in disconnected state and peer b is in connected
state)
Now if peer a gets connected it will get the sta_idx 1, as timer
initialization is already done for peer a with sta index 0 so
timer initialization will not be done again.
Now if peer b gets disconnected it will set the tdls_info for
sta_idx 0 to invalid (INVALID_TDLS_PEER_INDEX).
Now for peer a timer will be started and when timer expires in
timer handler function “tdls_ct_idle_handler” will not be invoked
as sta_idx is already set to INVALID_TDLS_PEER_INDEX.
To address above issue, de-initialize the idle timer for tdls
peer on every tdls peer reset.
CRs-Fixed: 2746046
Change-Id: Icc2e64a9299852d5b29d26dec79b7f1efb2594b8
Add configuration feature CONFIG_WAPI_BIG_ENDIAN to
enable or disable BIG ENDIAN WAPI configuration based
on device type.
This was done as newer chipset firmware expects the
WAPI seq in big endian format and to maintain backward
compatibility with the older chipset firmware.
Change-Id: I972ca89e601b706e53408b0751c265d0aa311029
CRs-Fixed: 2746749
Send RSN caps value as in supplicant to firmware instead of
intersection value.
Change-Id: If300836a89329325d906c88dcf06835ad767aa85
CRs-Fixed: 2746357
For qca6750, register hif shutdown notifier to update the
firmware recovery state in case of SSR
Change-Id: Ie6d6e9c909df292c62f89612c08b583321703cfc
CRs-Fixed: 2688921
When rx data stall happens due to ring refill failure
check for refill debt count.
CRS-Fixed: 2740233
Change-Id: Ib88074f0e7ed9a64b522a64ff669af87626ce1da
Enable logging feature which helps to check rx ring refill failure.
Currently rx ring refill count is updated same time from multiple
places without protection, so use atomic variable for refill count.
CRS-Fixed: 2740225
Change-Id: Icd88d351cb15d0f18edf0b986e4de95dbe4e6989
