TWT responder support should be announced, if enabled, in the
Extended capabilities IE.
Change-Id: I1355a9ddf2fe0361cb542b4c49b754d8f8a158fc
CRs-Fixed: 2884862
Currently, the regulatory update event is created in
hdd_regulatory_init. In some cases, the country can be set before this
function, causing a crash. Move the event creation to before the country
setting logic.
Change-Id: I344d0de71f1983b3da7b47b816d6bbb0e402cbe8
CRs-fixed: 2891434
Add support for TWT AP params not in range and IE validation failed
status code.
Change-Id: I939db5388c4e7aac86484b319638a57baf83171e
CRs-Fixed: 2888847
Driver report beacon/probe rsp frame in wlan_hdd_inform_bss_frame
to kernel. "boottime_ns" field needs be filled by "scansystimensec"
in nanosecond, so assign "boottime_ns" (NS) from scan entry to
bss_description instead of "scan_entry_time" (MS).
Change-Id: Id22f119f41f8cf60a8aadc9e04c4f18424c88413
CRs-Fixed: 2870827
Don't allow twt statistics or clear statistics command
if twt statistics or clear statistics is in_progress.
As a Fix, reject the new twt statistic or clear statistic
command till existing twt statistic or clear statistic
command completion.
Change-Id: I710a81ae3409f181689b2e4c9e9d4e583d4aadab
CRs-Fixed: 2870592
Currently, in TWT after posting the msg via sme to WMA
using scheduler thread. Based on the success or failure
driver set the command_in_progress so that subsequent
commands can be rejected till the current command is completed.
There is a chance, context might be broken because of this,
command_in_progress is not set.
Set the command_in_progress before posting
the msg and if there is an error then again set the
command_in_progress as NONE.
Change-Id: If9dc4bd346178d49d5d940ecf8ae008969cb1e73
CRs-Fixed: 2886135
Currently, vdev_id 0 is chosen as default vdev_id in
opportunistic_timer_handler when none of the vdevs are started.
The same is used to send set_hw_mode command. But vdev_id 0 might
not be valid all the time and set_hw_mode command fails in
serialization module in such cases.
Below is one possible scenario,
1. Load the driver. p2p0 gets vdev_id 0 and wlan0 gets vdev_id 1
2. Enable NAN and create two NDIs. p2p0 interface(vdev_id-0) will
be deleted by userspace as part of this.
3. Disable NAN and remove the NDIs. It triggers opportunistic_timer
but hw_mode won't be set to SMM as the vdev corresponds to
vdev_id-0 doesn't exist.
So, choose a valid vdev_id(mostly belong to STA/NAN) as default
vdev_id.
Change-Id: I19bd00a07cb2c818af9ed5021b0ae0aca8c49f2f
CRs-Fixed: 2889404
Don't set CONFIG_FEATURE_WLAN_STA_AP_MODE_DFS_DISABLE for SDX65 to
support AP+AP both with ACS DFS channel list.
Change-Id: Id3d9fc6ef2d8b285e89c3c4e0f34badd2c63da5f
CRs-Fixed: 2857810
Currently, WoW wakeup host event is handled in tasklet context and
it is taking more than 800ms to execute. This is due to excessive
logging to the console from the tasklet. As a result of this, IRQs
are getting blocked during this tasklet execution.
To avoid this, add a sysfs entry to enable/disable console logging
from host wakeup event based on requirement.
Change-Id: If91a8bd5f783f358dabf9dcf46bf4c50d05d7909
CRs-Fixed: 2881972
As a result of long path names and a large number of source files,
combined with upstream kernel build infrastructure changes, the
qcacld-3.0 build may encounter E2BIG errors when linking the final
wlan.o. To address this problem support for nested composite objects
is being added to the kernel. Update the Kbuild with the logic
required to take advantage of the nested composite support.
Change-Id: Ic6748517a922fc03a43acab2b44133b37e02707a
CRs-Fixed: 2861273
When the country is being updated via WMI, FW responds with a channel
list after some time. During this time, another event can come, blocking
the channel list from being updated. Add synchronization so that these
requests are blocked until the channel list is updated.
Change-Id: I4c1b7d33242226b8153821fdc14db54fff764c17
CRs-fixed: 2884840
Currently, there are two ini option 0/1 to disable/enable ILP
profiles in HW.
Extend this option upto 3 where
0 indicates disable
1 indicates perf settings
2 indicates max power saving
3 indicates balanced settings
Change-Id: Iff41b92b84309a670e2ad301ee72fb3caaba3df3
CRs-Fixed: 2886526
During suspend if the netif queue pause failed for some reason,
network stack may send the pkt for hard_start_xmit and will
crash due to device is in suspend state, so adding a check to
drop the pkt if device was already in system suspended.
Change-Id: I5b8f5511157f10e28297be4366fafe6e683c392b
CRs-Fixed: 2880892
While processing the roam synch indication reassoc response frame,
the new roamed peer object is not created. So call to
mlme_set_twt_peer_capabilities() to update the new peer capabilities
will fail with null peer object for new bssid. So after roaming,
update the peer TWT capabilities once the new peer object is
created.
Update peer TWT capabilities to new peer object after roaming to
avoid invalid peer capabilities.
Change-Id: Ic37bd4d96bf122b2ec3a60c1ae41650a672b9020
CRs-Fixed: 2870589
Remove g_prefer_5g_scc_to_dbs, and update PCL table
pm_second_connection_pcl_dbs_2x2_table.
For STA + GO concurrency if the second connection is GO,
perfer SCC to DBS, which makes GO get high throughput.
Change-Id: I0bf68662fc363a66c778904f9c12714407cd738a
CRs-Fixed: 2881383
