In case of dense roam scenario, Host should set initial dense
status if roamable AP count is more than minimum dense AP count.
Current implementation checks for the number of channels in roam
cache list instead of AP count.
Add changes to set the flag if roamable AP count is more than
configured value.
Change-Id: I4d5f7c6c69b1a9a527aace1677050373065af213
CRs-Fixed: 1058211
qcacld-2.0 to qcacld-3.0 propagation
If roaming is initiated by firmware, after getting the candidate
list the CSR will start preauth with the candidates and thus
preauth will be the active command in SME.
Now if at the same time connect is issued from supplicant,
HDD will queue disconnect cmd in SME queue and wait for disconnect
to get complete for 2 sec. Now there is a chance that channel
change along with preauth timeout can take more than 2 secs.
In this scenareo the disconnect will timeout in HDD and will
return connect failure to supplicant. Next supplicant will issue
disconnect which is ignored by HDD as disconnect is already been
queued in SME. Now if supplicant again sends the connect command,
as part of connect command CSR will remove the disconnect command
from the SME queue and queue this connect command.
On preauth timeout the preauth state machine checks if disconnect
is queued and if it is queued it will abort preauth. But in this
case disconnect is removed by the connect request and thus preauth
retry continues and if preauth is success handoff will be queued
in SME queue.
Now the connect request is processed and if this connect request's
BSSID is same as the BSSID of the roam candidate, as all the CSR
states for this BSSID session indicate roaming, instead of
join it will try to reassociate and continue changing the CSR states
again. Eventually this connect will fail. Next the handoff command
is processed and as part of handoff it will try to cleanup
original session, but as CSR states are changed by connect request
the original session is not cleaned up properly. This results in
stale PE session entry which does not allow the scan to be sent to
firmware, returning the cached results and thus 0 scan results.
To fix this:
- Increase the WLAN_WAIT_TIME_DISCONNECT to 5 sec to make sure all
sessions are cleaned up before allowing the new connect to proceed.
- Increase the priority of preauth scans.
Change-Id: Id7cc1e6f3c31df8a8c8955eb3ff48cb60e5b5ab2
CRs-Fixed: 1048394
This reverts commit-id: "Ib028c29863d4e95ccac434f7d47bfedd59ef883f"
It caused peer object memory access after it was freed.
CRs-Fixed: 1057347
Change-Id: I99842eb646371e186ade3b126ea0e8f84b7f2187
Auth request is not sent out due to which assoc reject is observed.
As part of the vdev down firmware clears the software tx queue which
is blocked due to the "Ia5394390d376c638b3976d446a6ecc98d0584159"
change. So reverting this change will enable
the vdev down and clears the tx queue.
This reverts commit-id "Ia5394390d376c638b3976d446a6ecc98d0584159"
Change-Id: If7d7ca5939f7586b12d84d6e070e94bb5e7e6c2c
CRs-Fixed: 1057329
While processing ROAM_OFFLOAD_SYNCH_IND, delete the peer entry
even if reference count is non-zero sine firmware has already
deleted that peer. It allows addition of new peer without waiting
for peer_unmap events. Fix the logic error in checking the timeout
in ol_txrx_peer_attach().
Change-Id: Ib028c29863d4e95ccac434f7d47bfedd59ef883f
CRs-Fixed: 1046754
In following conditions RX LDPC support needs to be enabled
1) when STA is coming up in 5G band
2) when IBSS is coming up in 5G band
for anyother cases RX LDPC needs to be disabled. If user has
choosen RX LDPC to be disabled from INI file then it needs to
be disabled for all the cases including above mentioned cases.
To achieve this, take the intersection of global, INI, and harware
specific RX LDPC settings.
Change-Id: Iae10aa4a8c0931cdb796cd9c8ff558d4bc8e0aed
CRs-Fixed: 1050004
Change structure names to fix WIN compilation when used in
coexistence with MCL codebase.
Change-Id: Ife09989e1f0f0c4d619f73c5960968d3a6721b6e
Acked-by: prgandhi@qti.qualcomm.com
CRs-Fixed: 1008872
Propagation from qcacld-2.0 to qcacld-3.0.
1) Add check to find IPV6 mcast packet embedded in a
link layer unicast packet in wma_wow_wake_up_stats.
2) Add extra length check for IPV4 packets.
Change-Id: Ic2af69032262237b8751019817ae181913346818
CRs-Fixed: 1049254
Propagation from qcacld-2.0 to qcacld-3.0.
Correct the updation of various wow wakeup counts.
Change wow_icmpv6_uc_bc_count to wow_icmpv6_count
as there is no broadcast in case of ipv6.
Change-Id: Iceaf1226edbf909ca3881e049c33f7d121860ef2
CRs-fixed: 1049254
Propagation from qcacld-2.0 to qcacld-3.0.
Presently in updating wma_wow_wake_up_stats, icmpv4 and icmpv6
counts are increased just by checking ICMP protocol offset byte
and ICMPV6 protocol offset byte without checking for whether it
is a IPV4 or IPV6 packet. So it is possible that for ICMPV6 packet,
the IPV4 protol offset byte is equal to ICMP protocol or for some
ICMP packet, the IPV6 protocol byte is equal to ICMPV6 protocol and
thus both the icmpv4 and icmpv6 counts will get increased for that packet.
Add ipv4 and ipv6 packet checks as well in addition to the present checks
to avoid wrong increment in the counts.
Change-Id: I2d6e5d095d2f4b2dd474b9338bfc830b04bfa533
CRs-fixed: 1048651
qcacld-2.0 to qcacld-3.0 propagation
The problem with the existing code is:
* WMA_STA_WOW_DEFAULT_PTRN_MAX is set to 4 and in wma_wow_sta()
we make sure atleast WMA_STA_WOW_DEFAULT_PTRN_MAX free slots are
available before configuring the default wow pattern. But after this
check we are actually configuring 5 default wow patterns.
* If BPF is enabled, we are limiting the Max WoW patterns to 2,
because the free slots are 2 which is less than the expected default
free slots WMA_STA_WOW_DEFAULT_PTRN_MAX. we are not configuring
default STA mode wow pattern and hence no wakeups observed for
incoming packets.
Address this issue by changing the WMA_STA_WOW_DEFAULT_PTRN_MAX to 5 and
MAX WoW filters to WMA_STA_WOW_DEFAULT_PTRN_MAX if BPF is enabled.
Change-Id: If433cff18ce511e7fdffadff69ee13b762a1719b
CRs-Fixed: 973054
qcacld-2.0 to qcacld-3.0 propagation
* Set BPF filter only if the station/p2p client is in the connected
state.
* Depending on the BPF service enabled in firmware dynamically
configure the number of wow filters.
- If bpf enabled maxwowfilter is 2.
- If bpf is disabled maxwowfilter is read from ini.
Change-Id: I14c722c9a1189f4ba4fbc2c8a554ae85b7a61fa8
CRs-Fixed: 967535
qcacld-2.0 to qcacld-3.0 propagation
Currently ICMP counter is incremented only for IPV4 multicast packets.
But ICMP count includes total unicast/multicast/broadcast ICMP packets.
Add changes to update ICMP counter for unicast/multicast/broadcast
ICMP packets.
Change-Id: I00f9fb0c8231fd69f6c108ced8028eb19315006e
CRs-Fixed: 1020079
qcacld-2.0 to qcacld-3.0 propagation
Send wakelock stats to HAL layer when it requests through vendor
command QCA_NL80211_VENDOR_SUBCMD_WAKELOCK_STATS. This is a
blocking request in HDD.
Change-Id: Ie043323d90f63c19de2da9d1b8b54bc28d2c8f3d
CRs-Fixed: 958964
KR and CN have different DFS regions than what kernel provides.
Assign the correct DFS regions for KR and CN. Also use "enum
dfs_region" as the parameter type in functions that have
dfs region as parameter.
CRs-Fixed: 1047214
Change-Id: I2ddd67d3c29a448dd2a1d3a63113750783fb6731
Make MAS (MCC Adaptive Scheduler) commands to be PDEV
specific.
Two instances of OCS (Off Channel Scheduler) can exist
in the FW (one per MAC) and FW provides the option of
enabling and disabling MAS on a per MAC basis. But,
Host does not have enable/disable option for individual
MACs. So, the agreement with the FW is for the Host to
send down a ‘pdev id’ of 0. When ‘pdev id’ of 0 is used,
FW treats this as a SOC level command and applies the
same value to both MACs. So, irrespective of the value
of ‘WMI_SERVICE_DEPRECATED_REPLACE’ in the WMI service
bit map, the pdev id needs to be ‘0’ (SOC level) for
the WMI command
WMI_RESMGR_ADAPTIVE_OCS_ENABLE_DISABLE_CMDID.
WMI command WMI_RESMGR_SET_CHAN_TIME_QUOTA_CMDID to set
the MCC quota is sent down as channel – value pairs.
The value being channel time quota and the channel being
sent down is a home channel. The additional requirement
from FW is that all the channels in a single WMI command
belong to the same MAC. FW asserts if the WMI command
mixes and matches home channels associated with
different MACs. So, although the PDEV ID is not part of
the WMI command struct, the cmd is in a sense PDEV
specific.
There is no change in the WMI command
WMI_RESMGR_SET_CHAN_LATENCY_CMDID to make it PDEV
specific since this WMI command always send only one
channel-latency pair to the FW. So, there shouldn’t be
any problem of FW receiving home channels associated
with different MACs, for this WMI command.
Change-Id: Ie22800e07bbeef65c43f9171de828533b982a06b
CRs-Fixed: 1052652
qcacld-2.0 to qcacld-3.0 propagation
Currently extscan_results_available event is posted to user space after a
bucket scan is completed. If the extscan start request is configured for
multiple buckets, then the user space application tries to fetch cached
results on every extscan_results available.
If the get_cached_results is issued with flush=1, then every get_cached_results
will have 0 scan results and since flush=1 is set, due to a bug in firmware
even the temporary scan cache is flushed, there by we miss reporting the
APs found in the previous bucket scans and only the last get_cached_results
will report the APs found in the last bucket scan. Whereas the expectation is
get_cached_results must report all the APs found in the complete scan cycle.
The proposed fix is in both firmware and host:
1. Firmware to ignore get_cached_results flush=1 and not to flush the temporary
scan cache in the middle of extscan cycle.
2. Host to consume bucket_started event within host driver and map
extscan_results_available with cycle_complete instead of bucket_complete.
Change-Id: I399204261fa3d8515146c590630a3bb663ee801c
CRs-Fixed: 975402
qcacld-2.0 to qcacld-3.0 propagation
Make the following enhancements to the EXTSCAN feature:
* Add buckets_scanned param while sending
on_full_scan_result/cached result event to user space.
* Configure REPORT_EVENTS_CONTEXT_HUB to firmware.
* Remove status while sending on_scan_event to user space.
* Add support for SCAN_THRESHOLD_NUM_SCANS /
SCAN_THRESHOLD_PERCENT while sending scan event.
Change-Id: Ib37ce7cef5b3ce9d9ee4ebdc5ea9bcadaad7b695
CRs-Fixed: 960456
Propagation from qcacld-2.0 to qcacld-3.0.
Dump some more info for WOW wakeup packet:
1) ether_type
2) proto_type in case of IPV4 and IPV6 packets
3) src port and dst port in case of TCP/UDP
Change-Id: I1578aa5322af50f0ae43248c48a2f754ea081970
CRs-Fixed: 1043939
Propagation from qcacld-2.0 to qcacld-3.0.
Change print type from %d to %u for the unsigned variables in
wma_wow_dump_mgmt_buffer and wma_wow_parse_data_pkt_buffer.
Change-Id: I2ea9ae2aa5c6f1bbec65b724de35a0071ea26804
CRs-Fixed: 1038593
Propagation from qcacld-2.0 to qcacld-3.0.
Dump soem information such as source MAC address, destination
MAC address, sequence number, packet type, subtype and other
important information for the first mgmt. packet after WOW wakeup.
Change-Id: I4f39d93e8aeb243ec3df5cc1a916aec0a1aa4819
CRs-Fixed: 1019037
Propagation from qcacld-2.0 to qcacld-3.0.
Dump some information such as source MAC address, destination
MAC address, sequence number, packet type, subtype and other
important information for the first data packet after WOW wakeup.
Change-Id: I3c83991f56be34412f3f10adf58bd991ecb536ca
CRs-Fixed: 1019037
as per the latest draft for BSS bandwidth 160MHz, channel frequency segment 2
represents the center channel frequency. Update wma_get_channels to use correct
segment to calculate center channel when operating in 160MHz.
Change-Id: I9a8f2035b590a90415fbc5a0a6608c48bbd5b2e5
CRs-Fixed: 1048793
qcacld-2.0 to qcacld-3.0 propagation
Currently, In SAP mode due to incorrect wep key
index broadcast frames are not encrypted.
Fix is to make sure to update default index for
BSS(group) key.
Change-Id: I328855decbad20dbd88b48022e466e4f78f5d722
CRs-Fixed: 932749
When auth timeout failure happens, as part of clean up process
vdev down is sent to FW but the vdev up happens only after the
assoc response is received. Sending the vdev down to without
vdev up leads to negative vdev up count value in the FW.
Check if vdev is up before sending the vdev down.
Change-Id: Ia5394390d376c638b3976d446a6ecc98d0584159
CRs-Fixed: 1051760
Add support for 5/10 MHz channel width for STA and SAP role. To
enable/disable feature, following ini parameter will be used:
* gSub20ChannelWidth=0: indicates do not use Sub 20 MHz bandwidth
this is also the default value.
* gSub20ChannelWidth=1: Bring up SAP/STA in 5 MHz bandwidth
* gSub20ChannelWidth=2: Bring up SAP/STA in 10 MHz bandwidth
Change-Id: Ic6d534dc1eae60fcd2fb7533c934b9ea28e6dd78
CRs-Fixed: 1013211
qcacld-2.0 to qcacld-3.0 propagation
Take action to delete peer in NDP_END Indication, rather than
NDP_END Response event from firmware. Earlier, device initiating
NDP_END will only get NDP_END_Response event from firmware, but
that behavior is changed now and both self and peer device will get
NDP_END_indication with peer map for ndp end.
Change-Id: I7316cc3864052eef4c7bbe9bbac01f0333909c59
CRs-Fixed: 962367
Pointer pointing to address which used to be start of the
mac header is no longer true after pulling the head of the frame.
Due to above situation, it causes dangling pointer situation. Fix
the error by re-assigning the correct address.
Change-Id: I71693f86b1971e3c6d1ddd9c377ef888122c2bb8
CRs-Fixed: 1048951
Bus link voting needs to vote for the link up before
voting for the link down. The sap module voted for
link down before link up, overriding the default
settings of an always on bus. This confused the suspend
resume system making it send WOW_IFACE_PAUSE_ENABLED
when the bus interface didn't have a "paused" mode.
Sap module should vote for link up when a sta is added
and vote link down when removed.
Change-Id: I5c042a3a4b3d4e6528fdc7edc9e0413c0f811192
CRs-Fixed: 1049469
qcacld-2.0 to qcacld-3.0 propagation
While RTT ranging happens, userspace application initiates FTM
request via host to FW. After sending the FTM request, host
might go to sleep mode sometimes so FW will not be able to send
the OEM response event resulting in poor ranging results.
This change registers OEM response event as WOW wakeable so FW
wakes up host whenever it receives OEM response which helps OEM
response to reach userspace application on time for better
ranging results.
Git-commit: 4de9b661e45a0bde4dfefc760edf03d1b8f4dd68
Change-Id: I71b77cfa09c5039b487e9a93490837a084314fb3
CRs-Fixed: 1018887
Remove the usage of typedef for oem data request and oem
data response data structure.
Change-Id: I94f05833baa9b6baaf2272f63a33236437a96bbf
CRs-Fixed: 1038872
Remove oem data rsp passing over multiple layers.
This change set sends the OEM data response directly to
SME from WMA instead of going through LIM/MLM.
Change-Id: I3cff10ff7bdbcee39b39bd9ba03b5eff8444b017
CRs-Fixed: 1038872
Remove oem data request passing over multiple layers.
This change set sends the OEM data request directly from
SME to WMA instead of going through LIM/MLM.
Change-Id: I151fa771544e9f74b1b69b18d689176752760621
CRs-Fixed: 1038872
Remove handling of deprecated OEM related events since FW and
userspace application will use only TLV based messaging.
Support to handle following events are removed,
WMI_OEM_CAPABILITY_EVENTID
WMI_OEM_MEASUREMENT_REPORT_EVENTID
WMI_OEM_ERROR_REPORT_EVENTID
Change-Id: Ia194f559acde6689a50c56b52078b7dc967c6159
CRs-Fixed: 1038872
Propagation from qcacld-2.0 to qcacld-3.0
- Current length of bpf is not set during the reset filter. set offload
is allocated through malloc, so invalid values for current_length
are recieved during reset. Memset set_bpfload to zero after
allocation.
- update the bpf service to hdd correctly depending on the
wmi service bitmap recieved from the firmware.
Change-Id: Iaf4774865cf0698a2deebac5cea62c873146e838
CRs-Fixed: 985562
In function wma_roam_scan_filter, memory is allocated to local variable
params. And in default case of the switch, we are returning from the function
without freeing the params memory, this cause memory leak.
To fix this, free params memory before return in switch case.
Change-Id: I0a8825b44f4bcd9b05f0e1f0e6cdd0d114af0a01
CRs-Fixed: 1048116
In function wma_send_peer_assoc, logging error when wmi_unified_peer_assoc_send
failed. But error type used in this log is wrong, use "status" instead of "ret".
Change-Id: I8dbdd68e16d665e5a7749e10c7006da9f96a6fd7
CRs-Fixed: 1042968
In function wma_get_mcs_idx variable "match_rate" is not initialized.
Initialize "match_rate" to avoid dereferencing uninitialized variable.
Change-Id: Id4c7e1913628007087c58149e4b7033320d4cc79
CRs-Fixed: 1042968
In function wma_add_bss_sta_mode, variable "peer" is not initialized.
Possibly we may pass this variable to wma_remove_peer function as is
and that can be dereferenced inside wma_remove_peer function.
So initialize peer with NULL to avoid dereferencing uninitialized variable.
Change-Id: Ibc484759b5e92052a3500137464e47287ccad939
CRs-Fixed: 1042968
Update ini configuration parameters for cds modules in the driver,
this helps in avoiding layer violation of using the hdd context in
cds modules.
Change-Id: Iff149cc51ea5eacae4f84d83c51d0ed63834ba50
CRs-Fixed: 1026817
Memory allocation input is not validated correctly in
TDLS Peer assoc command, which triggers intermittent
failures.
Validate the memory allocation input to avoid
failures.
Change-Id: Ibd6fc4d62799783721acee7c9cc339ac6febd2e5
CRs-Fixed: 1044706
wma_enable_wow_in_fw was passing the wrong handle to hif_can_suspend_link.
The wrapper function htc_can_suspend_link will properly extract the hif
context and forward the request.
Change-Id: I40f0c2301b686625ea58ef692f816856bbdcf06b
CRs-Fixed: 1045959
