Refactor connect complete code and move common code
for connection manager and legacy path to a common function
to remove the duplication of the code.
Change-Id: I25e39635005efd9ad19174e9f26d009946912cdb
CRs-Fixed: 2842413
Problem scenario:
1) Host gets roam event from FW and RSO state is switched to
ROAM_IN_PROGRESS.
2) Host gets reassoc response followed by FW down indication.
3) Host processes reassoc response in HDD and sends disassoc to
FW as FW down indication received.
4) As a part of disassoc host tries to switch RSO state to STOP.
But as present RSO state is ROAM_IN_PROGRESS and FW is down,
host fails to send RSO stop command to FW and doesn't switch
RSO state to STOP.
5) Lim disassoc send fails due to FW is down and host proceeds
with lim_send_disassoc_cnf and lim_cleanup_rx_path.
6) In lim_cleanup_rx_path() host checks for present RSO state is
ROAM_IN_PROGRESS and returns without calling lim_del_sta();
7) Thus peer delete and vdev delete is not send to FW and host
vdev State machine stuck at SP-SUSPEND_DOWN which leads to
peer and vdev reference leaks.
Fix:
Allow switch to RSO stop state while present RSO state is
ROAM_IN_PROGRESS.
Change-Id: I4a838ecd98a7ca377cd557d7a01a93ac53449595
CRs-Fixed: 2844647
Enable 10th bit of CFG_BTM_ENABLE ini to forward BTM request
frame to upper layers when frame contains MBO assoc retry
attribute. when supplicant receives this MBO frame it will
use the frame info for blacklisting the AP so for the next
connection framework will avoid this AP.
Change-Id: I9a675cc01769c38ad5be26ea5fd28ada8dc03546
CRs-Fixed: 2843323
Currently when the FW is down, runtime resume return success
status but does not send the power exit and WOW wake up from
sleep command to FW. So WMI_WOW_SET_ACTION_WAKE_UP_CMDID
command is sent to FW which leads to panic.
To avoid this issue, send the correct error status for runtime
resume if FW is down.
CRs-Fixed: 2830614
Change-Id: Id27d2648427b32d32c15370fbb170fafec9450d2
FW considers 9th bit of CFG_BTM_ENABLE for btm query
with candidate list which is not mentioned in ini
documentation.
Change-Id: I4fefdaa464e14c350d885107a28051fa0ad309f0
CRs-Fixed: 2839587
Fix the compilation error about redefinition of 'ipa_is_ready'.
This API is defined in <linux/ipa.h>. Rename the function name.
Change-Id: I206a221a5df95494e947483f2d133df6e5053220
CRs-Fixed: 2817997
It doesn't free resources in failed cases like getting vdev or tdls
vdev object when handling tdls commands. This causes reference
count of psoc leak or memory leak. Following are the functions,
tdls_set_responder
tdls_notify_sta_connect
tdls_notify_sta_disconnect
tdls_set_operation_mode
tdls_reset_all_peers
Change-Id: I8b5d5fa8a137a1b07f0fa6e4667047db95d7b99e
CRs-Fixed: 2839099
Remove IPA related TX completion memory not used in lithium
and helium platforms.
Change-Id: Ibdce6169a1b5b95289fee80a36e2041cf785ad86
CRs-Fixed: 2808513
Periodically request pdev stats from FW to receive channel noise floor
information after monitor vdev has been brought up, based on the band
currently being sniffed. Use the band information as a secondary parameter
when sending periodic request to fw so a single noise floor value is sent
to host. Pass this noise floor information to dp_pdev to be populated in
radiotap header.
Change-Id: I488e4985392cb646ffdee914359fa995dbeb025e
CRs-Fixed: 2833458
Currently, Stats request commands are sent over qmi right from the
target suspend request is sent to FW. This is leading to a crash in
FW since it is trying to access PCI when it is in suspend state.
To address this, send stats request over QMI only after the ack is
received for the target suspend command.
Change-Id: I1b2b6a60c5d8c7ec4375d328e58a66d749794e3f
CRs-Fixed: 2835903
Add ini to set init chain mode for BTC, default value should
be set to the same as macro WLAN_COEX_BTC_CHAIN_MODE_UNSETTLED
that means no need to send init chain mode for BTC to firmware
in wma_vdev_pre_start() process.
Change-Id: I3051a7f8a0c708c7df40d1b95b46e0bfa6892f6c
CRs-Fixed: 2837173
There are cases where only a few TWT roles may be required
to be enabled. For example TWT Requestor role alone or
TWT Respondor role alone. Also, it could be unicast TWT only
and disable broadcast TWT.
Support broadcast TWT configuration in requestor and responder
role by adding new INI.
Extend the use of existing requestor and responder INIs to further
enable requestor and responder roles itself.
Change-Id: I3dfa196edfda9cafd01fa56c0bb6ce73e27b4144
CRs-Fixed: 2828718
In the customer platform, the chain0 for 2.4G is disabled,
it just only support 1x1 2.4G + 2x2 5G for hasting. With
such kind of configuration, 2.4G SAP start failure due to
policy_mgr_is_hw_dbs_required_for_band return false with
following config, which block switching hw mode from single
mac(single mac just only support 5G) to dual mac and then
make start 2.4G SAP failure on single mac configuration
[0]-MAC0: tx_ss:2 rx_ss:2 bw_idx:5 band_cap:2
[0]-MAC1: tx_ss:0 rx_ss:0 bw_idx:0
[0] DBS:0 SBS:0 hw_mode_id:0
[1]-MAC0: tx_ss:2 rx_ss:2 bw_idx:5 band_cap:2
[1]-MAC1: tx_ss:1 rx_ss:1 bw_idx:4
[1] DBS:1 SBS:0 hw_mode_id:1
Change-Id: If9e76fb47743c32c313eacf150146ba8fa60eb2d
CRs-Fixed: 2833620
It doesn't handle failed case about getting vdev or tdls vdev object
when teardown tdls connection. Which causes reference count of psoc
leak.
Change-Id: Ie4e7966c0a07f378483cd5e6344599645c75ac68
CRs-Fixed: 2836949
Currently, STA reconnect (e.g. reconnect to same BSSID) is handled
as STA+STA as part of NAN concurrency checks. This results in NAN
disable when reconnect is issued.
Check if the current incoming connection is on same vdev as
previous sta connection instance and disable NAN only if it's
different.
Handle NAN+SAP reenable in similar way.
Also, remove the redundant usage of NDI cleanup API
ucfg_nan_check_and_disable_unsupported_ndi in
if_mgr_connect_start.
Change-Id: Ia063a69bb2efdf1d51c6988b8905ceac0f454dab
CRs-Fixed: 2821352
To avoid roaming enable failed for operations bitmap set/clear mismatch for
SAP+STA or STA+STA concurrency, clear whole operations bitmap when RSO
deinit, then roaming can be enabled again after STA disconnect and
reconnect.
Add log when set/get operations bitmap to help debug.
Change-Id: Ie1f1a213dcdefb393ec3e1e2c6a8a71ab17b226c
CRs-Fixed: 2834199
Currently, Host sends classifier type and classifier
mask part of TCLAS Mask OPIE only with tag length of
2 bytes.
As per spec tag length for TCLAS Mask OPIE should be 18.
Ext Tag: TCLAS Mask
Ext Tag Number: TCLAS Mask (89)
Ext Tag length: 18
Classifier Type: 4
Classifier Mask: 0x5f
Reserved:00000000000000000000000000000000
Fix is to add a reserved parameter in TCLAS Mask OPIE
of length 16 bytes and set it to NULL.
Change-Id: I54cd896a4ffad16581c14fc254575bcec8e6f6d7
CRs-Fixed: 2822616
Currently the connection manager queries the
blacklist manager about the action to be taken
and if all the candidates of a certain SSID are
blacklisted then the connection would be rejected.
there are various types of blacklisting, some
can be overided and some have to be strictly
followed.
Fix is to overide the blacklist candidate and
connect to it if no other candidates are suitable
for connection.
Change-Id: Id47fd214130411179d9a53005841786d6d14ce81
CRs-Fixed: 2831535
Previously, if force SCC is enabled and Standalone SAP starts
on non-DFS channel, it starts with master mode Disable. If it
is switched to DFS channel it will switch without CAC.
As a fix, if SAP is started on non-DFS channel with force SCC,
don't allow it to switch to DFS channel if master mode is disabled.
Change-Id: I1822f67a5480d6c16fa980c9c6b262341af7c2de
CRs-Fixed: 2827237
Currently, host decrements beacon switch count and sends
channel switch frame every time until count becomes 0 this
may introduce sync issue between fw and host in case fw has
not sent beacons but host decremented switch count due to
timeout and sent new channel switch frame or vdev restart in
case count beacomes 0.
Fix is to offload CSA to fw to avoid any sync issue.
Change-Id: Idb48d419ec4117ff39cb823ca8af75d7a0abc3c0
CRs-Fixed: 2828045
The event WMI_ROAM_STATS_EVENTID is received after
every roam, once the roam synch complete is sent by the host.
This event contains details regarding the roam RSSI TH reset.
This helps in debugging/understanding the scenario when roam
failure happens.
Print the info received related to roam RSSI TH reset into kmsg.
Change-Id: Ibee9fc6e9220511d6679efb24080c79bcd807ab8
CRs-Fixed: 2826321
Currently max value of gEnableModulatedDTIM is 5 which is
incorrect as FW supports maximum no. of modulated DTIM as 10.
To address above issue update max value of gEnableModulatedDTIM
ini.
CRs-Fixed: 2830240
Change-Id: I7f1cab6f0cb2b2a60e7cc1bb37a9971c1752a1bb
Currently when Adaptive 11R feature is disable via ini
then user space parse Vendor Specific IE and trigger
connection in FT-PSK/EAP based on STA configured key
mgmt, but due to feature disabled via ini it failed
to connect AP and trigger Assoc reject to user space.
To overcome this advertise Adaptive 11R supported feature
bit to user space and based on this feature bit user space
parse, trigger connection in FT-PSK/EAP other wise fallback to
PSK/EAP.
Change-Id: I5d67a0cb77c3f39328467a6313a7a795e6a7a4d8
CRs-Fixed: 2785878
At present when user trigger SAP move to unsafe channel,
driver doesn't reject the request since "strict = false"
and unsafe checking is skipped.
Fix by check SAP channel switch target channel safe or
unsafe with API policy_mgr_is_sap_freq_allowed before
perform channel switch.
Change-Id: I2950fb31346df8705c8fc608fd79e1a44f4d4947
CRs-Fixed: 2826619
Stop/Reject SAP connection if STA is/comes up
on a 6ghz freq and SAP is not capable of 6ghz
freq to avoid MCC situation in a non-DBS capable
HW.
Fix is to stop/reject the SAP startup in above
mentioned case.
Change-Id: I451c95929f378cd0790bd5fc647235fc2dd0071a
CRs-Fixed: 2818030
Currently the driver uses the API
policy_mgr_is_dbs_enable to check
whether the DBS is enable or not which
does not check the service capability of
DBS, and thus would return true if ini
and FW HW modes contain DBS. In this case
if STA is on 5ghz and SAP comes on 2.4ghz
then force SCC would not happen as the
driver would consider the HW to be DBS
capable which actually is not, and thus
the overall concurrency would be MCC which
is not favourable.
Fix is to use policy_mgr_is_hw_dbs_capable
API instead of DBS enable API which checks
for the service bit also, so that the driver
forces SAP to STA's channel.
Change-Id: I9319293ecfb54fe3e1ad0eaff15542c671514bf4
CRs-Fixed: 2820377
Currently the driver checks whether hw is
already in DBS in the function
policy_mgr_check_and_set_hw_mode_for_channel_switch
and it does not set hw mode if it is already
in DBS mode. It can happen that SMM is already
requested and is present in serialization and it
would change the HW mode from DBS->SMM and if
driver starts any vdev on 2.4ghz in HST/HSP
which requires the DBS mode for 2.4ghz , the FW
would send a vdev start failure.
Fix is to remove the check of current hw mode and
let DBS be posted to serialization. If hw mode is
already in DBS it would return from serialization
with E_ALREADY but would ensure that HW mode is
always in DBS for a 2.4ghz connection.
Change-Id: Ica660058175cb1edc2931dd2d28ee6cdd9bab2f4
CRs-Fixed: 2823238
Currently, there is no INI parameter to control NAN feature in 6GHz
frequency band. Because of this, there is no provision to disable the
NAN in 6GHz when there is no requirement.
To address this, add INI parameter to configure NAN feature in 6GHz.
Change-Id: I6b4faa8e8aa1a3df72539fed1d7ff3cc8287debd
CRs-Fixed: 2817920
Use WMI_RX_SERIALIZER_CTX context for thermal event handler.
The default context is tasklet, it may cause issue when interact
with OS interface.
Change-Id: I349412c2afee9a081f0d38704541f77b5e4569f4
CRs-Fixed: 2824461
IPA MCC work is initialiazed only when uc code loaded successfully,
before canceling MCC work check for uc load success state.
Change-Id: I9af501d1b974d0da21891634bde2beafd69701b2
CRs-Fixed: 2822382
Firmware sends peer create confirmation to detect duplicate peer
on other vdev. If the status field in the
wmi_peer_create_conf_event_fixed_param is 0, peer creation is
successful.
Add support to handle the peer create confirmation command for
sta mode. For roaming scenario, the peer will be internally
created by the firmware, so bail out after creating object
manager peer/datapath peer without sending the peer create
command similar as the current implementation.
For SAP/P2P Go mode the legacy peer create path will be used.
Proceed to do post linkstate change if status is success, else
do vdev stop and indicate failure to upper layers
Change-Id: I0abbd70dd509f2b9afa8d4d7db1774e289d1e021
CRs-Fixed: 2747870
Add a check for passive channels when checking the restricted
bands on SAP. This change blocks SAP from restarting on a passive
channel.
Implement the function to filter out passive channels from the PCL.
Change-Id: I80a4b78c1af77f5bfa68be3163f9e9a78cc6425a
CRs-fixed: 2817589
Modify connect start code in interface manager by
moving disable roam and tdls link teardown notify
APIs from HDD to if mgr. Alongside, move tdls link
teardown notify API from HDD to if mgr in start bss API.
Also, move the tdls link teardown API to TDLS module from
HDD module as TDLS should manage the wait logic.
Change-Id: I09fa31878563a3daaa7c5fde46327475829317b3
CRs-Fixed: 2811807
Add new ini "monitor_mode_concurrency" to support this feature
and introduce policy manager api's for concurrency checks.
Change-Id: I35ee1fece0a6f9ae8fe340b0598c4a3e20b17e82
CRs-Fixed: 2814523
In mission mode, add support for virtual add/del monitor
mode interface. In addition to STA vdev, create and maintain
a new vdev for monitor interface.
Change-Id: I91ea8e56976cce92f9b1ae308d4bd0fb30815366
CRs-Fixed: 2818499
On some third party platform, the in-band bus suspend-resume is
not supported well, instead add gpio irq as wakeup source to
support out-of-band suspend-resume.
Change-Id: I7621f04d093013e15719b5c84dc244604be238df
CRs-Fixed: 2815754
Check the factors as below to decide whether the channel is
allowed or not:
* If the channel is in LTE coex channel avoidance list;
* If it's STA+SAP SCC;
* If STA+SAP SCC on LTE coex channel is allowed.
Replace policy_mgr_is_safe_channel() with this new function
for sap channel selection and acs channel filtering, to allow
unsafe channels when it's STA+SAP SCC and STA+SAP SCC on LTE
coex channel is allowed.
CRs-Fixed: 2743042
Change-Id: Ic5a84b2628200fe9decf6972f00706f190f04722
Add logic to
- Check if connect req freq is 6Ghz and security
is not allowed for 6Ghz, reject connect.
- Ignore 6Ghz APs if connect req security is invalid
for 6Ghz
Also added user config key_mgmt_mask_6ghz mask
to allow specific AKMs, by default all are allowed.
Add added user config check_6Ghz_security to enable
security checks as per spec.
Change-Id: Iada7bb685b272e26cff7e89cf64d181ed6eb48a4
CRs-Fixed: 2820358
cds_get_context() function already takes care of logging the
caller function name in case of any error.
Hence there is no need to add the error log again.
Getting rid of these unnecessary logs reduces driver memory footprint.
Change-Id: I5bf676c79e9e12a797dc6792ee8014fb30b0dff3
CRs-Fixed: 2821005
The SHO feature is used to offload beacon/probe response to HW.
So a new vdev param is required to enable/disable this feature.
Change-Id: I83d71d63485dee8d3433bf967a7feb56c1039f93
CRs-Fixed: 2815270
Add check in function 'policy_mgr_get_sbs_channels' to restrict the
array index out of bounds.
Change-Id: I387b666095107faf284e35f073dfe856d38323d3
CRs-Fixed: 2776966
A race condition is observed when runtime suspend is aborted
due to pending TX packets but context switch happened and pm_state
did not revert to active state yet. In parallel, WMI_PEER_DELETE
command is queued but not sent as pm_state is still in suspending
state.
To fix this, schedule HTC queue kicker work to resend pending
commands again after dp_runtime_suspend aborts due
to pending TX packets.
Change-Id: I6d2c4140a4635c0aec304f48b15fb2b539ac39ea
CRs-Fixed: 2813657
