When the chip is power save mode it misses the gpio strobe
from the host platform. The TSF is not latched is not latched
in this case. As a result the reported TSF in the
WMI_VDEV_TSF_REPORT_EVENTID is not correct.
Fix this by adding a retry mechnism when the reported TSF is not proper.
Change-Id: Ica2d5c796e9d021b65e898ee3defe88d62183911
The ANI level determines how well the device is coping with interference
in the wireless environment. Add support in the driver to query the ANI
level from the firmware and populate it for userspace entities to use.
Change-Id: I54934f670aa11737e11eca4d64e12f9dffb4f430
CRs-Fixed: 2554674
Enable Force wake feature through Kbuild
to access address range greater than BAR+4K
at runtime.
Change-Id: Ibe3c3b9b765c256493b1d5f9258e462760d45dc6
CRs-Fixed: 2557978
Since it will takes long time to register spectral to Direct Buffer RX
component, and it will affect wmi ready event handling in pdev open.
So register spectral to dbr after wmi ready event handled.
Change-Id: I51844a708ed69144618e5b0846747601d4d96a9d
CRs-Fixed: 2560822
Currently in SAP+SAP scenario the second ACS request
is moved to pending and is completed when the first
ACS is completed, but since FW serializes the scan
so the reuest for second ACS if not kept in pending
and sent down to FW would automatically be serialized
which would reduce latency in second SAP bringup.
Remove the pending work logic and implementation
to improve latency in second ACS scan.
Change-Id: I439398b406e9371a8e443f5409bd3f25b3062501
CRs-Fixed: 2570259
Currently, userspace is expecting SNR value through
NL80211_STA_INFO_SIGNAL for a disconnected client of SAP. But
the driver sends RSSI value by doing noise floor to the SNR value
received from firmware. Convert this back to SNR and provide to
userspace as per the definition of NL80211_STA_INFO_SIGNAL
and to be in sync with the existing implementation of userspace.
Change-Id: Ia360424a9d44e48fae5a85cf0d2c577e92cd93be
CRs-Fixed: 2552542
hdd_txrx_hist doesn't init when insmod in con_mode 5,
after change back to mission mode, it will cause null
memory access in hdd_bus_bw_work_handler.
test step:
1 insmod wlan.ko con_mode=5
2 echo "0" > con_mode
3 connect with AP
4 issue happens
Move wlan_hdd_init_tx_rx_histogram to start module.
Change-Id: I01b94b9461a87c29cc03b7e3c841bbda6e565188
CRs-Fixed: 2567165
Currently regulatory comnfigs are getting updated from wiphy
init inside regulatory init api which is too late in the driver
init sequence. There are some apis in driver init sequence which
uses these ini configs before wiphy init, one such example
is scm_11d_cc_db_init. since the ini configs are not populated
this may lead to undefined behaviour.
To resolve above issue, populate ini configs at the same place
where all other components inis are getting populated.
Change-Id: I9cb2f069fba3eb76a2c8f7285b6d371cf5fa8de4
CRs-Fixed: 2570148
1) Change the following functions' signatures and/or code
to make them using frequencies instead of channel IDs:
hdd_get_adapter_home_channel
hdd_get_operating_channel
wlan_hdd_request_pre_cac
wlan_hdd_validate_and_get_pre_cac_ch
__wlan_hdd_request_pre_cac
hdd_check_and_disconnect_sta_on_invalid_channel
set_first_connection_operating_channel
sme_is_channel_valid
csr_roam_is_channel_valid
2) Remove the declaration, definition and usage of
csr_roam_is_chan_freq_valid, and replace it with
csr_roam_is_channel_valid as this API does exactly
the same thing.
3) Retain the code/logic for chan ID utilization in
function set_first_connection_operating_channel as
it is used to convert the parameters issued by
user space apps.
4) Fix a few camel cases in csr_api_roam.c
Change-Id: I4f9cfa21a5f40288ad9492ac8be7939ed8eb6484
CRs-Fixed: 2565002
Currently, there is a problem in hdd_for_each_adapter(), as it is
checking for the NDI adapter only in an infinite loop. Create
hdd_adapter_iterate() in order to iterate through all adapters properly.
Change-Id: Ib0e22d0f46d6311d44de3ee48b3202376a42f209
CRs-Fixed: 2568906
gActionOUIConnect1x1 is used to specify action OUIs for 1x1
connection.
Some Access Points are unable to handle SMPS and OMN frames and shows
abnormal behaviour when STA is switching between 2x2 to 1x1 and
vice-versa. To avoid this, restrict connection in 1x1 mode with such APs.
Here the scenario is DUT STA1 connects in 2x2 mode to AP1 in 5G mode then
later on DUT another STA2 iface connects to AP2 in 2G mode. DUT sends OMN
frame to AP1 to inform that it is going to switch to 1x1 mode. AP1 ack
this frame but doesn't change its Tx rates and continues to use 2x2 rates
which DUT cannot ACK. This causes low throughput issue where AP1 Tx rate
falls to 6Mbps due to missing acks.
Hence, 5G band capability bit in "gActionOUIConnect1x1" is set to make
connections in 1x1 with such AP.
Presently, 5G band capability bit is not checked if 2G band capability bit
is not set or 2G band is not enabled. Hence, Vendor AP capability check for
2G and 5G Band is refactored.
Change-Id: Ic6b682451405a2373000e4c64656f50c2fab0ee6
CRs-Fixed: 2558293
Invocation of wrong api hdd_is_valid_mac_address() makes hdd_save_peer()
always return false due to which hdd_roam_register_sta() is skipped in
hdd_ndp_new_peer_handler() and can cause ping failure.
To fix this, use api qdf_is_macaddr_zero() instead of
hdd_is_valid_mac_address() in hdd_save_peer() to get the unused
entry in NDP peer mac table.
Change-Id: Id9aa6bcebb7fd168ee281065ebffe7227f9bc10f
CRs-Fixed: 2554502
As a part of 6GHz support, add support to parse channel frequency
attribute: QCA_WLAN_VENDOR_ATTR_ACS_FREQUENCY_HINT (in MHz)
in QCA_NL80211_VENDOR_SUBCMD_ACS_POLICY command data.
Change-Id: Ic7bee5fe1123b82607e375c71a21b9744d67ea17
CRs-Fixed: 2533998
As a part of 6GHz support, add support to parse and send channel
frequency (in MHz) in QCA_NL80211_VENDOR_SUBCMD_EXTERNAL_ACS
command/event data.
Change-Id: I87879b4f8dd6a5c4ded2e89c527ba91432c9c48a
CRs-Fixed: 2533732
Open/close session callbacks are common for all the device modes
supported, so register once during the start_modules.
Change-Id: Ia94857ce4cc608be30e9974ddde8a6261d45f6e3
CRs-Fixed: 2555919
The current HDD callback in hdd_rx_flush_packet_cbk() does not validate
the adapter context properly. Instead of verifying the adapter magic,
verify the adapter itself is still valid through the adapter list.
Create a new function hdd_get_adapter_by_reference() to verify the
adapter reference.
Change-Id: I468bd55b2318635ad89087e6c6ad6097df68d405
CRs-Fixed: 2563654
Modify the following 3 APIs' code and parameters to make
sure they're using frequency instead of channel ID:
sme_get_reg_info
wlan_hdd_get_channel_info
hdd_update_channel_bw_info
Change-Id: I87ad6fb9cd9fcd7fe4e41e62a32e4954e93f8ba1
CRs-Fixed: 2554710
The existing API/macro hdd_for_each_adapter() iterates over the hdd
adapter list with the iterator pointing to each adapter. This iteration
is not safe when seen with respect to unregistration of netdev going in
parallel in other thread. In case the adapter is removed, the iteration
will result in NULL pointer dereference.
An example of this issue is seen when the following happens in parallel.
1. del_virtual_interface -> hdd_remove_adapter -> removes node
2. hdd_indicate_mgmt_frames -> hdd_for_each_adapter iteration
To make the iteration delete safe, redirect the fetching of the adapter
through objmgr. This reduces the possibility of race window by a large
margin. In this case, even if the adapter is freed as a part of
unregister_netdevice, the reference held ensures that we block the
unregister until the work is done in the current thread.
Change-Id: Ic49aa22a8eef68dc1977fb73f2c1fd920bfeb7a1
CRs-Fixed: 2557890
Motion detection baselining can be started during Motion detection
phase which results in false motion detection
Do not send baselining start/stop during Motion detection phase,
to prevent false motion detection
CRs-Fixed: 2556575
Change-Id: I20de48b3d09cf8adec3a3a2d9585ad3754f12a9e
Do not start motion detection if baselining is not done, or motion
detection config values are not available
CRs-Fixed: 2547648
Change-Id: I23a02dbd21e2b673e625938a35003f3730135327
From android Q, it uses a random MAC to connect to different
SSIDs which causes the interface down/up, the vdev will be
deleted and the latency level info is lost in firmware.
To fix this, save the latency level and reconfig it to fw
after vdev creates.
Change-Id: Ifac93fb869553d0f9eb553d5c41da43c8b014ad8
CRs-Fixed: 2554736
When a particular vdev is deleted, the corresponding rx
packets which have been queued to the rx thread are not
flushed. Hence when such packets are submitted to the
network stack, the dev for this skb will be invalid,
since we have already freed the adapter.
Flush out the packets in the rx thread queues, before
deleting the vdev.
CRs-Fixed: 2552140
Change-Id: Ia49af2c203c64077f7fd87524bb4caa4060e0044
Presently, the driver performs device assert on re-init failure after
SSR. This is not desirable as in some cases the reinit failure can be
recovered from.
Change the default value of the INI controlled variable to 0 to avoid
this assert.
Change-Id: I802921e8602e0648cacc4185f550b72ad2b3af66
CRs-Fixed: 2547295
This change is to support 11ax linkspeed rate
on UI.
It supports different config for gReportMaxLinkSpeed.
Current 11ax rate support to max 80 nss2 mcs11.
Change-Id: Iff8cbafe1354ab50c4b3a90ef8ab698a3350a21d
CRs-Fixed: 2529291
Due to channel number ambiguity with 6ghz, update channel references
in HDD STA connection path to use channel frequency.
Change-Id: I81f3449c9087030e4d98c17a5b12c731f99b39ab
CRs-fixed: 2552009
Remove sta index references in TDLS structures and APIs
and mac address would be used instead to refer the peer.
Change-Id: I861f7ce068f3f545598129b9f3955b84b2a6fbce
CRs-Fixed: 2524512
As a part of the sta_id clean up, remove the usage of the sta_id from
struct hdd_connection_info. Instead of the station id, use peer mac
address to interact with the protocol and DP layers.
Change-Id: I60b939c37ff49b50579791f45e3e81ed81b86967
CRs-Fixed: 2524514
Currently the driver checks for the ACS in progress
in change iface and returns fail if the ACS is in
progress. The return status is not checked by the
kernel and it assumes that the interface change is
successfull, which may lead to out of sync between
driver and kernel.
Fix is to wait for ACS completion if the iface change
or interface down is triggered in between ACS.
Change-Id: Iaabd42fc959a533041b18b181e1b63493f17e0a5
CRs-Fixed: 2541325
As a part of 802.11ax amendment, 6GHz band operation is added.
Since the 6 GHz channel numbers are overlapping with existing 2.4GHz
and 5GHz channel numbers, use frequency to identify unique channel
operation instead of channel number. Channel frequency is unique across
bands.
As part of above requirement add logic to process rx mgmt
packets based on the frequencies instead of channel numbers.
Change-Id: Ib063070738ecdb4f83379eafe50629778a490aae
CRs-fixed: 2522693
Add action_oui ini gActionOUIDisableAggressiveEDCA to identify peers for
which aggressive EDCA configuration causes some IOT issues.
Change-Id: Ifea0c224cef2378727c9f78d8b571ea9c57bb662
CRs-Fixed: 2507191
To support 6Ghz channel, update channel number of struct
sap_StartBssCompleteEvent_s, hdd_ap_ctx and sap_ch_selected_s.
Change-Id: I19e6d7d03072135abed25e077e8573b5326ddba8
CRs-Fixed: 2519308
SDX55 defined both MDM_PLATFORM and FEATURE_WLAN_DP_RX_THREADS,
and this is causing redefined issue.
Change-Id: I1427be477dd3832b98332c9edaa4063b8c573ae4
CRs-Fixed: 2542264
Currently in hdd_adapter, the variable cache_sta_info is an array of
hdd_station_info structured indexed by the sta_id. As sta_id is no
longer used, this array based implementation becomes unnecessary.
Move the sta_info implementation to a hash table implementation.
Change-Id: Iac11ecd2ae4a454649235cf8763e9a5f80e7eb38
CRs-Fixed: 2515316
Currently in hdd_adapter, the variable sta_info is an array of
hdd_station_info structured indexed by the sta_id. As sta_id is no
longer used this array based implementation becomes unnecessary.
Move the sta_info implementation to a hash table implementation.
Change-Id: I7d6fe04b7e0ab22615d431de9a10a4ad38ed65ba
CRs-Fixed: 2514280
HDD indicate mgmt frame API is called from scheduler context to send mgmt
frames to user which can happen in parallel to del virtual interface which
can lead to race conditions and stale netdev pointer usage and hence add
OS_IF DSC vdev op protection to indicate mgmt frame API such that when
management frames are getting indicated to user space del virtual interface
transaction is synchronized.
Change-Id: I620225f7a0a46efa6da6d615e3014a8f4f5e43cf
CRs-Fixed: 2543244
Add new runtime pm wakelock for user such that runtime PM can be
disabled for certain scenarios.
Change-Id: Ibaa64f351393b63559b55827f77cdf126e7038b3
CRs-Fixed: 2534535
Based on the ini "ignore_fw_reg_offload_ind" ignore regulatory
offload indicatin from fw.
Change-Id: I6e20f2100eeac8e93baa5c5d3f3d1c8b33356aed
CRs-Fixed: 2532989
Change the very high bandwidth threshold to 9000 packets. Thus if the
packet length is 1500 bytes and for 100ms bandwidth interval this is
about 1030Mbps. This makes the high bandwidth range 240Mbps - 1030Mbps.
Change-Id: Ia9d5f9e3547148c5dc378a728514c4020649944e
CRs-Fixed: 2527623
Disable the current TWT session and re-enable the TWT if the
current session in FW trigger mode before changing the TWT mode
to host trigger mode.
Change-Id: I7de2126605685c85a240d38356d439cb4be8cce1
CRs-Fixed: 2539353
Identify TCP vs UDP Tx traffic to vote differently for system resource.
Accordingly vote for gold cores only for UDP Tx high TPUT traffic.
Change-Id: I27190b92b1be6ebe1574a9941a5de8c43bb59a0b
CRs-Fixed: 2499147
Currently as part of cfg80211 wlan stats command after fetching the
stats runtime sync suspend is issued which puts the link in low
power state with out waiting for link inactivity timeout. This leads
to throughput degradation in case of rx direction as in rx processing
data path extends the timer by marking last busy to avoid immediate
runtime suspend. But runtime sync suspend does not take in to account
of the link idle timeout value before suspending the link. So in this
change before issuing runtime sync suspend check the duration since
last busy mark by data path rx process. If the duration is less than
run time delay just do runtime put.
Change-Id: I4e071ddda1c6a815f29d5b44b95e699a212b59d6
CRs-Fixed: 2511423
Use separate band - NL80211_BAND_6GHZ for 6GHz channel
list population. And 6GHz channels are reported only
when CFG80211_6GHZ_BAND_SUPPORTED and CONFIG_BAND_6GHZ
feature flags are defined.
Change-Id: Ieec30b304a7c3e8e856f290afdb4177206aeb2c7
CRs-Fixed: 2524262
Currently in STA+STA scenario roaming is enabled on the
first connected STA1 and is enabled on the other STA2 on
STA1 disconnection. This change adds support to dynamically
enable or disable roaming on any STA.
Change-Id: I34dc2730fb38193ea3dd0ec35c79df233f270913
CRs-Fixed: 2520827
