target_if expects psoc as part of roam events which is used for
event extraction. Send the same as it's missing currently.
Change-Id: I04729df90d8e64e07427fe47926a8a92a24aaf23
CRs-Fixed: 3027055
For MLO, add support to hdd install key api's
on wdevs without netdev. Add changes under config
flag "CFG80211_KEY_INSTALL_SUPPORT_ON_WDEV".
Change-Id: I67de2ba36dd86f0814c820756d5f49b765defda6
CRs-Fixed: 3017312
Remove STA specific enums from eConnectionState. Also remove
its prints from STA dependent code.
Change-Id: I1a06d16a060bec90147dcf94925582877352d1a8
CRs-Fixed: 3028099
In ucfg_twt_get_all_peer_session_params peer reference is
acquired and its not released leading to the reference leak,
leading to vdev destroy timeout.
Release the peer reference to resolve the issue.
Change-Id: I9992376473c30365f15fcdb8e374c856bc169d6b
CRs-Fixed: 3027189
In some stress test, like doing random on/off sap and random SSR, sometimes we
can hit below test situation:
1. trigger SSR first by calling hdd_crash_inject, then immediately off SAP, the
two actions almost happen at the same time.
2. the action of off SAP calls __wlan_hdd_del_virtual_intf to delete SAP interface
and is waiting for qdf_stop_bss_event completion event.
3. at the moment, sys error interrupt happened, CNSS layer would send PLD_FW_DOWN
event to host driver first. Due to PLD_FW_DOWN event call qdf_complete_wait_events()
to forcedly reset all completion wait event, this action may cause
__wlan_hdd_del_virtual_intf to do vdev destroy before vdev resp mlme state machine
is processed. Once vdev destroy is executed, it will set vdev obj_state to
WLAN_OBJ_STATE_LOGICALLY_DELETED, then vdev resp will not have chances to be flushed,
and then cause peer reference count leakage.
Fix solution: use qdf_wait_single_event instead of qdf_wait_for_event_completion to
avoid stop_bss_event forcedly reset before stop_bss_event is done or timeout.
Change-Id: I45b662fd17ec56bb8fc4453627bdcb41dedf79e0
CRs-Fixed: 2987511
This reverts commit "Change-Id: I847468f0c1839974ea1c6cd31260fb7ef4fbb897"
Revert this change as kernel is not reading cac duration from
wiphy channels, kernel reads the cac duration from reg rules.
Change-Id: I794bb3ddd67f551daf9b4c3b9320909224ac4d76
CRs-Fixed: 3020354
As part of the new GKI model, multiple driver are loaded parallelly
resulting in a error message from the kernel regarding the multiple
registrations of the /dev/wlan.
Register the /dev/wlan file to the kernel once the device is
identified and driver registration is success.
Change-Id: Iaca40abb62be5f59f129604a9a6d74ada6a4ad3a
CRs-Fixed: 3027521
Currently max vdevs are defined as 6 in osif. Increase it to 9
in case of 11be.
Change-Id: If64306cfc55690d7523ee910074064051a1b1d78
CRs-Fixed: 3027377
Update unit test command for policy mgr pcl validation to
support new SBS PCL type.
Change-Id: Ib02ce6c69400d0f47307523d690e1d81a8c0ef83
CRs-Fixed: 2998526
Currently the PS config received from kernel/userspace is set
to mac_ctx globally. This causes PS config set for one vdev to
override on the other vdev.
To avoid this, set the PS config per vdev.
Change-Id: I6a2e2a9d8fb67b94d9d5f1d4164077990bdaf4a5
CRs-Fixed: 3007060
To support DBS-SBS hw mode, populate SBS/SCC/MCC/DBS channel
list for new PCL types.
Change-Id: I5b6bae3daeae0be879d60dc5183574fe63f6a0c9
CRs-Fixed: 2998497
The PER default low rate thresholds configured is 20Mbps. This
will cause PER roam scan always after the 25 secs PER monitor
window since the 11b rates are lower than 20Mbps.
So disable PER roam when the connection/roaming to peer is with
cckm/OFDM rates to avoid frequent roam scans.
Change-Id: Iaae469804547c0f0682b66de3609b5cda4b4b577
CRs-Fixed: 2984973
In the 2.4G 1x1 + 5G 2x2 Hastings specific case,
host will receive below HE MCS info:
1. hw mode id[0], phy id[0]
he_supp_mcs_2G=0x0
he_supp_mcs_5G=0xfffefffa
2. hw_mode_id=1 phy_id=0
he_supp_mcs_2G=0x0
he_supp_mcs_5G=0xfffefffa
3. hw_mode_id=1 phy_id=1
he_supp_mcs_2G=0xfffe
he_supp_mcs_5G=0x0
But in the wma_update_target_ext_he_cap, it always
updates the last mcs=0xfffe info into he_cap,
which will be used in the HE capability of beacon
or probe response frame whatever it running on
2.4G or 5G and it is not right.
So add this fix to update the MCS separately according
to the current operating channel.
Change-Id: I277aa708ae7e13917cf2068f01b05d3864ad7618
CRs-Fixed: 2992253
Optimize STA disconnect handling In LIM, by improving logging
and seperating logic from SAP's peer disconnect handling.
Change-Id: Iad2cc1ac0b5dc3e5ce7294e410d6d8040934ea0d
CRs-Fixed: 3026218
Add new ini item gEnableIcmpOffload to enable/disable
ICMP offload to firmware.
Change-Id: I107aeec4c75ae18fbe37d373e48339a5b76797aa
CRs-Fixed: 3025650
Whenever the host receives the SETROAMSCANFREQUENCIES command,
the host sets ROAMSCANCONTROL to 1 internally. This results
host ignore the ADDROAMSCANFREQUENCIES command if the specific
channel list is already configured.
As per new requirements, the host should not ignore any
ADDROAMSCANFREQUENCIES command which comes after
SETROAMSCANFREQUENCIES.
Change-Id: I9896ff21cdf97ebfcdcf5b198e981f5bd2f501c6
CRs-Fixed: 3022063
Remove ioctl cmd GETROAMSCANCHANNELS in the driver as this
is no longer required from Android S onwards.
Change-Id: I5b57a9a34cbe4b023334855751263fe435d0bdbf
CRs-Fixed: 3021665
Remove ioctl cmd SETROAMSCANCHANNELS in the driver as this
is no longer required from Android S onwards.
Change-Id: I904b7a5407997b07c23a40f14d6c7328ec9de8ea
CRs-Fixed: 3021662
Some AP indicates it support 40M in HTInfo, while not support 40M in
HTCaps, so need intersect HTInfo and HTCaps.
Change-Id: I496b62f50a2e1e19aeef7e77535d0479f31bba44
CRs-Fixed: 3025951
OEM-SS sends ROAMBAND attribute via the vendor command
QCA_WLAN_VENDOR_ROAMING_SUBCMD_CONTROL_SET to host to
get current roam band value in the host.
Fix is to process ROAMBAND attribute and send roam band
value to upper layer.
Change-Id: I56b208d00b40d4d3e6d3658d23fa1073f6fe1146
CRs-Fixed: 3009753
Filter roam scan channel list or frequency list as per the
band received from upper layer via attribute
QCA_WLAN_VENDOR_ATTR_SETBAND_MASK and
QCA_WLAN_VENDOR_ATTR_ROAMBAND_MASK.
Change-Id: I4820bfc487949d8445f215a4779dd671925d001f
CRs-Fixed: 3009758
Whenever host Receives vendor command for SETROAMBAND, Host should
set PCL list as per roam band to fw so that FW can use updated pcl
list for further roaming.
Change-Id: I2f48833801b75f25aaf37ea479867b19477ceecb
CRs-Fixed: 3009751
Process the BAND_MASK attribute which comes to the
host via the QCA_WLAN_VENDOR_ROAMING_SUBCMD_CONTROL_SET
vendor command. Save the value of this band mask into
the RSO configuration structure.
Change-Id: I57ecc10912a38207ef478b38cf3add3c584dd2f4
CRs-Fixed: 3009749
Currently driver cleanup sta connection info at stop
adapter because of which if query for sta connection info
comes after adapter restart, in that case driver is not able
to return this info.
For example according to new design of Android-S, framework
issues stop adapter after every disconnetion to update mac
address dynamically, after that it starts adapter again, now
if driver gets request for connection info, it is not able to
provide the connection info as it is already cleaned up on stop
adapter.
To address above issue, cleanup sta connection info at close
adapter instead of stop adapter.
Change-Id: I54247d4577dd7f8cf1c7d7bc657005a09d69d0df
CRs-Fixed: 3000842
When getting rnr IE information, instead of getting rnr pointer through
input parameter, we can just return rnr pointer.
Change-Id: I7215451d34164d40d6404d0a426f1cc30a143e15
CRs-Fixed: 3025191
This change removes the "delete pmksa on BMISS" logic in commit
I5d8ec9e9d4f8a44178b113d370799fe2bc0373fb.
Will handle the issue in new commit by OUI checking of AP.
Change-Id: I871999f8c473afdc75b774210044fc059fcd07b0
CRs-Fixed: 3021068
Currently the driver dump directory is getting created by appending
the chipset name but wifihal is trying to read the driverdump
file from debugdriver directory.
Don't append the chipset name in the driver dump dir.
CRs-Fixed: 2974459
Change-Id: I7b32a3d0f967d642828a5b5ad4579befe09de652
While Rx buffers are getting umapped from net rx context and
if IPA pipes are enabled at same time from MC thread context,
this is leading to race condition and IPA SMMU map/unmap is
going out of sync, in adrestea targets.
To fix this introducing IPA SMMU map lock and sync map/unmap.
Change-Id: Ida4c321d94f9640135c67210c69edb355d827c86
CRs-Fixed: 3021027
