wlan_hdd_update_txq_timestamp tries to grab txq spinlock
in __dev_queue_xmit context. Issue is txq spinlock is
already held before calling into driver's ndo_start_xmit
callback. On UP system, kernel throws a BUG and system
crashes.
Fix is to detect spinlock recursion and only tries
to grab the lock when it is not held on the same
CPU or not held.
Change-Id: I3956ef00c4f4a563155dc82ed8f95f097129fb0c
CRs-Fixed: 2050134
The WMA vdev structure contains wakelocks which must be deinitialized
and initialized any time these structures are mem-zeroed. Add
deinit/init calls to vdev attach error cases.
Change-Id: I36012606d4cc49c076ac6903ff49530c45433374
CRs-Fixed: 2102645
The host roam timers are deactivated and deleted in
lim_delete_timers_host_roam but again in lim_deactivate_timers
they are deactivated which results in assert.
Fix this by deactivating roam timers before they are destroyed.
Change-Id: I03f069edaf4361111dc5240fce2003c04c9c6509
CRs-Fixed: 2095265
WLAN_NL80211_TESTMODE need to be controlled by CONFIG_NL80211_TESTMODE,
or compiler will throw error, like - API cfg80211_testmode_* not defined.
CRs-Fixed: 2103007
Change-Id: I84536b8527656d8d0d4a0a7961ca8a83e8dd733c
HDD_CFG prints were being dropped from the qxdm log.
Initialize the nl server before the logging service
to guarantee all logs are delivered over the nl sockets.
Change-Id: I5719f8e2791e37a8c5e61ccf5c025afd1084ed99
CRs-Fixed: 2094161
To aid in debugging memory leaks, and improve the robustness of log
recovery from crash dumps, statically allocate pglog_msg in the global
scope. This allows logs to be inspected even in cases where a crash
came late in the unload process. It also improves log recovery
robustness by removing the possibility of using a NULL pointer offset
under certain memory corruptions scenarios.
Change-Id: I7301254d3d3ff811b322e435962e78f997937c2b
CRs-Fixed: 2074611
qcacld-2.0 to qcacld-3.0 propagation
FW has limitation to support old WMI_REQUEST_STATS_CMDID and
WMI_UPDATE_STATS_EVENTID interface on AP vDev. FW suggest to use new
WMI_REQUEST_PEER_STATS_INFO_CMDID and WMI_PEER_STATS_INFO_EVENTID to
get tx/rx rate.
This checkin is about HDD layer, provide APIs/callbacks to upper layer
to query rssi/tx/rx statistics of peer station associated with SAP,
and provide a default disabled ini setting and checking FW's
WMI_SERVICE_PEER_STATS_INFO capability to control this get peer info
interface. Add/change iwpriv/QCA_NL80211_VENDOR_SUBCMD_GET_STATION/
cfg80211_ops.get_station interfaces to upper layer.
Change-Id: Ib1571219faa16e3bc8c3eb1260137ae08b8b6016
CRs-Fixed: 2059507
Add changes to enforce start and end channel limit in do_acs
if driver force acs and vendor based acs are enabled.
Also add changes to fix memory leak in set vendor channel API
Change-Id: Id45aaf3eb3c3dc1606a249136d3801a63ee854c4
CRs-Fixed: 2088567
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: Iebaa140de17abf4face710245c8431680663a0a0
CRs-Fixed: 2100153
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: Ia1b151d2590493b15422bd600dedb427a6eb39db
CRs-Fixed: 2100152
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: Id69553b8ef9cc8daad1e1c136c0edfccfb820860
CRs-Fixed: 2100150
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: If336b3f168d6e4b5cd1925030cd65e171f8100b3
CRs-Fixed: 2100148
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: Icf9c75c9b99c7f94b14d0f70597ab05eebadc4f5
CRs-Fixed: 2100145
When printing all of the stations connected to a SAP adapter, the ini
value for max number of station clients is used as the iteration limit.
Instead, use the max number of stations constant used to allocate the
array. This ensures all connected stations are printed in all cases, and
removes the theoretical possibility of running off the end of the array
in some configurations.
Change-Id: I3a506905f40272efe472028391e8080d764cb390
CRs-Fixed: 2100833
qcacld-2.0 to qcacld-3.0 propagation
Currently, few hdd driver commands which expect input are not checked
for space termination of command name. Skip of this check can result
in buffer overread error.
To fix this, add check based on the length of command name.
Change-Id: I92a4efd7e729a9d792af166b4a09963886c9942f
CRs-Fixed: 2090558
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: I51019e186fff585146ff79928c00dfc6a854e78a
CRs-Fixed: 2100191
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: I5ac8829606eda6940deebe13037e15e47bed6f9b
CRs-Fixed: 2100188
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: If3ba3e869a9dddd14432c6d9416d7b386f23eb3d
CRs-Fixed: 2100182
The Linux Coding Style enumerates a few special cases where typedefs
are useful, but stresses "NEVER EVER use a typedef unless you can
clearly match one of those rules." The hdd_context_t typedef does not
meet any of those criteria, so replace references to it with a
reference to the underlying struct.
Change-Id: I5f846c4b234bd55ff3120fd69ea13332e3e80d74
CRs-Fixed: 2100180
Currently wherever cds_trigger_recovery() is called in atomic
context, work is being scheduled.
This change is to modify cds_trigger_recovery() itself detect
that it is being called in atomic context and then
itself schedule work to its own handler.
Change-Id: Id37e97c9b838ad18c355234aa13f2b9c05503029
CRs-Fixed: 2081666
Compiler is throwing undefined WLAN_HDD_VENDOR_IE_ACCESS_NONE error after
WLAN_NL80211_TESTMODE disabled.
Move enum wlan_hdd_vendor_ie_access_policy out of WLAN_NL80211_TESTMODE.
CRs-Fixed: 2100717
Change-Id: I09e28eeff236006486087911064691339e59fc84
qcacld-2.0 to qcacld-3.0 propagation
Setting CSA IE if not all the STA devices support ECSA while
configure the macro CHANNEL_HOPPING_ALL_BANDS. Otherwise, SAP
will miss CSA IE when switch band from 5G to 2G.
Change-Id: I7aad2bb0d967e73c0ea87acb9820bebdc0455410
CRs-Fixed: 2024425
OS_IF layer function recieving NDP peer departed indication expects
different structure than what was supplied, causing wrong values to
be interprated and leading to crash. Fix this by passing correct
structure data type.
Change-Id: I1ad6860e5c55b0efe65691262bd563e84f523a7f
CRs-Fixed: 2100848
If vendor acs support is enabled, driver sends a list of channels
through a vendor event when it receives do_acs. Currently it sends
list of all valid channels even if hostapd or INI has specified
specific channels which is not expected.
If available send list of channels specified by hostapd or INI.
Change-Id: I9b931c69ce72a18c7e1ec603143091da6abca70c
CRs-Fixed: 2100742
Currently we are not getting any event for ASSOC/AUTH
completion.To debug issues where AP does not respond
to our AUTH\ASSOC request or the response packet from
AP is lost we would like to have clear indication whether
the request from DUT has reached AP or not.
Add diag event for both auth and assoc completion to
inform whether we got ack for the auth and assoc request
pkt.
Change-Id: Ia01da4a9f0d5435733844178643a29fad58d2a5c
CRs-Fixed: 2001968
Some vendor APs are not able to handle the SMPS frames causing the
data rate to drops to very low rates and before it recovers the next
SMPS frame is triggered.
Add new vendor AP OUI and vendor AP data, compare the vendor AP OUI
and data with AP capabilities and if there is a match, associate in
1x1 to that AP.
Change-Id: I753582fa6a0db9fc24587f053be34d44d5604944
CRs-Fixed: 2093741
If DUT connects to 1x1 AP in DBS mode, the nss value
will be reduced to 0 in current logic.
Fix by checking the nss > 1 before reduce.
Change-Id: I908ba358411ec7c99f57bb472a0d52a24e0422c3
CRs-Fixed: 2089211
Add support to use user provided BW for 2.4Ghz HT connection.
If HT capability in connect req indicate connection in HT20, force
the connection in 2.4Ghz to HT20 only even if AP support HT40.
Added ini override_ht20_40_24g to enable/disable this feature.
Change-Id: Ia03b67b9d6487e3e27f1cf8875afe95af1300b4f
CRs-Fixed: 2084929
When DUT connects to specific AP, it receives the frame whose
SA is self MAC and MIC failure flag is set at the same time.
This will cause disconnection with the AP because MIC failure.
To fix this unexpected disconnection issue, do not check MIC
failure if discard flag is set by fw.
Change-Id: I9029da9d329679bf0fd61be234690c764cceb57c
CRs-Fixed: 2090575
Driver starts a vendor ACS timer which times out if vendor app doesn't
respond with commands within stipulated time. But in do_acs this timer
is started before all the channel list parsing, regulatory checks and
buffer allocation code. In some cases this takes up enough time to
expire the timer before sending the vendor event. Similar situation will
arise if a buffer allocation fails in between.
To avoid this start timer after the vendor event is sent.
Change-Id: If9bea4643e03f0541ac8d8af2183c1e8d3a6cc40
CRs-Fixed: 2096246
