Presently, the reason codes for driver to trigger recovery on are
limited in their scope; most of the scenarios are not correctly
capture.
Add new reason codes to cover all the possible scenarios where the
driver has need to trigger recovery.
Change-Id: Ia257855bc30cd0bc7e81b9d0e21e4b84427d546b
CRs-Fixed: 2630951
To undergo self recovery, the driver uses qdf_trigger_self_recovery to
initiate the sequence. Currently, this framework is valid for only a
single psoc driver.
Extend this framework to include support for multiple psoc driver by
providing the psoc on which the recovery has to be undertaken.
Change-Id: I782b505c03819223a914dabe7673b369aa175b7c
CRs-Fixed: 2617707
Add more hang reason codes for the hang reason in the
QCA_NL80211_VENDOR_SUBCMD_HANG events. This also introduces the
attribute QCA_WLAN_VENDOR_ATTR_HANG_REASON_DATA to carry the required
data for the respective hang reason. This data is expected to contain
the required dump to analyze the reason for the hang
Change-Id: I4912f896d394667055937faf68c9ceb32a469978
CRs-Fixed: 2650292
Currently qdf mc timer stop api updates mc timer state
to QDF_TIMER_STATE_STOPPED before it deletes the timer which
may lead to race condition where if any other thread tries to
read the state of the timer and re-start the timer considering
it is stopped which may get deleted after that.
To address above issue, update the timer state after timer
delete.
Change-Id: I302b89bdedf23f8277eacb6d42b5bd8e6daa4d05
CRs-Fixed: 2643254
Currently, mgmt_txrx_get_action_frm_subtype does not handle
ACTION_CATEGORY_RVS. In case action_category is
ACTION_CATEGORY_RVS, mgmt_txrx_get_action_frm_subtype
returns MGMT_FRM_UNSPECIFIED due to default handling. Function
tgt_mgmt_txrx_rx_frame_handler returns QDF_STATUS_E_FAILURE
on receiving MGMT_FRM_UNSPECIFIED frame type and drops mgmt
frames.
Fix is to add support for ACTION_CATEGORY_RVS in
mgmt_txrx_get_action_frm_subtype to avoid drop of MCSC
frames. Add new management frames MGMT_ACTION_SCS_REQ,
MGMT_ACTION_SCS_RSP, MGMT_ACTION_GROUP_MEMBERSHIP_REQ,
MGMT_ACTION_GROUP_MEMBERSHIP_RSP, MGMT_ACTION_MCSC_REQ,
MGMT_ACTION_MCSC_RSP and new element ACTION_CATEGORY_RVS
in mgmt_action_category. mgmt_txrx_get_action_frm_subtype
Finds rvs frame subtype and returns subtype instead of returning
MGMT_FRM_UNSPECIFIED.
Change-Id: Iad1abac120770724f9ae6d86f524c4d722fb0aaa
CRs-Fixed: 2649900
Add qdf API support to check if the packet should be
tracked as per the configured protocol types.
Change-Id: Ia9e937865d3d440f1350307a67e32b20ea218e54
CRs-Fixed: 2642156
When packet tx fails at transport layer, packet is re-queued and given to
HIF layer again. In this change, print the warning message about re-queue
only when tx fails for 5 consecutive attempts due to no resources
available at HIF layer.
Since re-queuing is not fatal, it need not be printed on every attempt.
Change-Id: I00082470420e29dfe4fb7a745ac709a682ba032e
CRs-Fixed: 2621763
Remove redundant data structure at wmi for vdev delete all peer
response extraction, instead use mlme host data structure.
Change-Id: Ice91f55e0bf8a23ea1a639a74f6a8ac3b2779a7b
CRs-Fixed: 2641261
In struct weight_config, "oce_ap_tx_pwr_weightage" added newly.
Station uses "oce_ap_tx_pwr_weightage" as one of the parameter for
selecting AP candidate during initial connection. Host sends
"oce_ap_tx_pwr_weightage" ini value to firmware over WMI command
WMI_ROAM_AP_PROFILE.
Add oce_ap_tx_pwr_weightage in scan object scoring param to calculate
weightage based on AP tx power and select candidate accordingly.
Change-Id: I7f717e9cc398e8b9307f2f464f972b095c06364b
CRs-Fixed: 2647116
reg_get_psoc_obj may return NULL.
Add check if regulatory psoc is NULL and return QDF_STATUS_FAILURE.
CR-Fixed:2648152
Change-Id: I9637fa367873cd62fa18d1f5ecd80f6fe454558b
Use qdf streamfs APIs to indicate data to user space. This is part of
moving CFR component to hostmn.
Change-Id: I2abc46fdfb36d24eef201ab21f5744d951595ea1
CRs-Fixed: 2637128
Disable the PCIE scratch reg read/write to wake up PCI bus.
This fixes the PCNOC error seen during boot up on writing to
the PCIE scratch register.
Enable DEVICE_FORCE_WAKE_ENABLE feature flag to enable
PCIE sracth reg read/write.
Change-Id: If4efd0afbddf97462bf9d0f06d0085c0a5b37a64
CRs-Fixed: 2636160
Currently whenever host gets an indication for the cc event
list, it starts 11d machine and inside 11d state machine,
it decides that whether host needs to send start 11d scan
command to fw or stop 11d scan command to fw, based on that it
updates the enable_11d_supp variable of regulatory psoc private
object and posts the message to scheduler.
FW expects that host should not send back to back 11d start
scan command, there should be one 11d stop scan command before
next 11d start scan command is sent if already one 11d start
scan command is sent.
Host 11d state machine takes care of the above sequence
of 11d scan command by using enable_11d_supp. 11d state machine
sets enable_11d_supp whenever 11d scan start command needs to
be sent and resets enable_11d_supp whenever 11d scan stop
command needs to be sent and posts the message in the scheduler.
There is a case where if host receives the 11d cc event from FW
it sends the set country command to FW and at the same time if
user command comes to set the country, host again sends the set
country command to FW. FW processes both of these commands and
sends the cc list event to host. when host receives cc events
it starts the 11d machine. For the first cc event which is for
11d country which is processed as user country because of
incorrect logic in reg process master channel list where it is
assumed that at a time only one type of country set command can
be pending and first priority is given to the user country pending,
host resets enable_11d_supp and post the command to scheduler.
When host receives the cc list event for the user country it is
processed as 11d country and inside 11d state machine it sets
enable_11d_supp and posts the command to scheduler.
Now if scheduler schedules these commands it treats both the
commands as start 11d scan commands as enable_11d_supp is set
and sends start 11d scan command back to back.
To avoid above issue, add enable_11d_supp also as part of
the scheduler msg body pointer along with the psoc, in she
scheduler callback use this local variable to decide whether
11d start scan needs to be sent or 11d stop scan needs to be
sent.
Change-Id: I60150da1475251a1c22778a0f924bdfbe1bb1140
CRs-Fixed: 2641803
The HE GI and HE LTF values are obtained from phy
TLVs in monitor path. The stats used in host have an
enum defined used by other upper layer code.
The radiotap spec defines different set of enums.
Add translation function to convert host stats enum
to radiotap defined enum for sniffer capture purpose.
Change-Id: I8c76adb42b0623e83958f7a37147b74e72b1ce74
CRs-Fixed: 2636051
Add cases for NAN and TDLS in qdf_opmode_str to return the mode
in string format.
Change-Id: Ibdc22a337005d56485e5ee3b392f4d5d0089304a
CRs-Fixed: 2648590
Currently QDF_MAX_NUM_CHAN and NUM_CHANNELS aren't aligned, this unalignment
may cause many potential OOB access. So replace QDF_MAX_NUM_CHAN with
NUM_CHANNELS to keep unified.
Change-Id: I7bf7829d776f7caf5b2afbd2c9fd0c20d608e630
CRs-Fixed: 2644073
Remove redundant data structure at wmi for vdev start
response extraction, instead use mlme host data structure.
Change-Id: Ie0d3e6fa0acb1fea8bc0f50095491dbe803f6810
CRs-Fixed: 2641085
Support RX 2K jump/OOR frame handling from REO2TCL ring.
(a) configure REO error destination ring register to route 2K jump
/OOR frame to REO2TCL ring.
(b) for 2K jump RX frame, only accept ARP frame and drop others,
meanwhile, send delba action frame to remote peer once receive first
2K jump data.
(c) for OOR RX frame, accept ARP/EAPOL/DHCP/IPV6_DHCP frame, otherwise
drop it.
Change-Id: I7cb33279a8ba543686da4eba547e40f86813e057
CRs-Fixed: 2631949
Fix macro name to enable delayed register writes.
Donot use delayed register writes for non fastpath access.
Change-Id: I235116ab3df5cb26bbfbb72de4ac6ed4b363a13a
CRs-Fixed: 2645865
As a part of enabling IPA pipes, the WBM2SW2 head
pointer register is written with the number of
buffers which have been allocated initially. This
register write is a critical one and failure in
writing this register can be fatal.
Confirm the written value, when initializing
the HP register for WBM2SW2 (for IPA).
Change-Id: Ib2da3a7aa6096375cf64857721619f47c50658de
CRs-Fixed: 2620750
Per HW team's analysis, we find a TXDMA HW limitation:
ADDR0&0x1FFFFFFF8 should not equal ADDR1&0x1FFFFFFF8.
Otherwise, TXDMA will run into exception, which cause TX fail.
ADDR0: the address of last words in previous buffer;
ADDR1: the address of first words in next buffer;
We hit this limitation in NAT forward TSO jumbo case whose buffer
address of two fragments like below:
tso_frags = (
(length = 0x42, vaddr = 0xFFFFFFD0F5FA2F82, paddr = 0x1F5FA2F82),
(length = 0x05A8, vaddr = 0xFFFFFFD0F5FA2FC4, paddr = 0x1F5FA2FC4)
In this case, ADDR0 = 0x1F5FA2F82 + 0x42 -2 = 0x1F5FA2FC2,
ADDR1 = 0x1F5FA2FC4, then
ADDR0&0x1FFFFFFF8 = ADDR1&0x1FFFFFFF8.
To avoid this, shift server bytes for ADDR0.
Change-Id: Ib09afd7e4c14a33bb5347c68b602b360a9e36619
CRs-Fixed: 2585868
Enhanced channel frequency response (CFR) based on monitor status
ring, need to enable mon reap timer before starting CFR and disable
the timer after stopping CFR capture. This change adds interface to
enable/disable monitor status ring's reap timer.
Change-Id: I843433ac07c4d55e14b42855ee22779eb7ecb678
CRs-Fixed: 2639307
WMI commands for DBR and CFR are based on pdev id, so need to convert
host pdev id and target pdev id when tx wmi commands or rx wmi events.
Change-Id: I54bf4a28a285e5b4dfdd125aeaabb9968a6d052a
CRs-Fixed: 2634729
The FW has updated the wmi command wmi_vdev_adfs_ch_cfg_cmd_fixed_param to
configure the Agile channel to enable the 165MHz channel as an Agile
channel by enabling 2 center frequencies.
For a 160MHz and 80MHz channel, the parameter center_freq1 will be the
center of the respective channels. For the 165MHz channel(restricted 80p80)
the parameter center_freq1 will be the center of the left 80MHz channel
(5690MHz) and the parameter center_freq2 will be the center of the right
80MHz channel (5775MHz)..
Update the ADFS parameters across UMAC, TARGET_IF and WMI layes to
update the Agile channel configuration command.
Change-Id: Ie480af9a6bb8dad87dd783ab06e17b449605b1c9
CRs-Fixed: 2642555
Introduce constants for building the 160MHz precac tree.
Add support to insert the 165MHz channel into the precac tree.
The precac entry is not restricted to be 80MHz bandwidth, it can be
20/40/80/160/165MHz, therefore introduce a new variable bw to
accommodate different bandwidths.
Change-Id: I57d1e2173712b54238f89079783f27ef591921d1
CRs-Fixed: 2647246
Avoid memzeo of seg_info and msdu_info removing from default standard path
qdf_ktime_to_msg(qdf_ktime_get()); -- moved under a flag
Eth_hdr access made inside the Mcast enhancement check.
Change-Id: I12b6cc397d7eb626abdcdae001e92ef0d980e8a6
Restrict lock hold time in transmit path to only ring access time.
Also store lmac_id in vdev to avoid multi indirection.
Arrange fields in data structure for better alignment in data path access.
Change-Id: I1d32880ed88be486171be46281ec180d2a4906bf
Provide support for bangradar with detector ID as one of the parameters.
Add the parameter as part of the packed arguments to be sent to FW.
Also add APIs for basic sanity check of bangradar params and packing
bangradar params inside an 32 bit unsigned integer.
CRs-Fixed: 2646549
Change-Id: Ie781bc9421b7ac0d407eb01814c9242c7f988884
In QCN9000, wbm release ring has msdu continuation bit
support for invld peer MPDUs. Host needs to form SG
buffer for packets with msdu continuation bit set
Change-Id: Ica03c78068d32d2c8dc609b9a50298b91dd48c0a
According to Annex E of 802.11 specification (Table E-1—Operating classes
in the United States) add opclass support for HT40 channel 140 and
144 in us_op_class map. Currently for HT40 channels 140 and 144,
us_op_class map does not have a valid entry and hence opclass is
fetched as 0.
CRs-Fixed: 2626714
Change-Id: I848d6266c829c363f353de9911359349c379126b
Currently vdev private data osif_priv gets allocation
separately and requires to be de-alloc separately.
This de-alloc needs to be taken care at different places
based on the vdev create failure or whenever vdev is freed.
As vdev private data is associated to the vdev and needs to
be freed if vdev is not there, it is good to allocate this
private data along with vdev so that there is no need
to maintain allocation and de-allocation of vdev priv data
separately.
As part of this change alloc the vdev priv data along with
the vdev allocation as a single memory and free of the vdev priv
data will be taken care as part of vdev free.
Change-Id: Ib442958c2d8684706830782d339c589c5f5daddf
CRs-Fixed: 2636222
Channel Frequency Response(CFR) feature requires PPDU information
for correlation with CFR data. Host subscribes for the relevant PPDU
status TLVs via the Host RX monitor status ring. During monitor status
ring reap, all information needed for CFR correlation is accumulated
in a HAL PPDU structure and delivered to WDI event subscribers. Add
this change for HSP.
Change-Id: If3060271cf2923171487d30f99f2ab5c55370601
CRs-Fixed: 2634766
In HSP platform, ring index 0 is for spectral scan, and ring idex 1 is
for CFR capture. Extend max rings to 2 and support CFR capture.
Change-Id: I02b39dd626b67172123bc5e77173d50a132fc735
CRs-Fixed: 2634636
When argument '_len' is equal to 0 or less than 0, 'ascii'
array element will be used uninitialized. To fix this case,
add arguemnt sanity check before use.
Change-Id: I2e2a4c199fac72466f831bb4261a6a03ac116e11
CRs-Fixed: 2643354
When 'err' is null pointer, QDF_ASSERT(0) will not return
directly, then null pointer reference will occur. To fix this
issue, return failure when 'err' is null pointer.
Change-Id: I5912571795ae55c6729ae06d5e06496598c21fbb
CRs-Fixed: 2642581
Host driver processes cld80211 vendor subcmd
CLD80211_VENDOR_SUB_CMD_OEM_DATA and send
response in NLA format. Peer status indication
is also sent in nla format
CRs-Fixed: 2597282
Change-Id: I94f1fb7939141952ef92dbb7d3a130ba20d31608
Host driver processes cld80211 vendor sub command
CLD80211_VENDOR_SUB_CMD_GET_CH_INFO and respond with
NLA type CH_INFO response
CRs-Fixed: 2595374
Change-Id: Ibe6f2431ac1cae6e4560ec11424434b4f1cf8b76
Host driver processes cld80211 vendor sub command
CLD80211_VENDOR_SUB_CMD_GET_CAPS and respond with
NLA type CAP response
CRs-Fixed: 2595140
Change-Id: I5a6aa1fac537ca618404520b81c541a20dafd6fb
Sometimes if the user changes his location in the
connected state it may happen that the scan does not
happen and thus the AP cache channel map would not be
updated.
Now if partial scan happens due to a roaming trigger then
the candidate would not be found and thus would effect
roaming.
To avoid this set a full scan in roam invoke if the APs
are not found in the partial scan.
Change-Id: I2d2fd94760de78eaf495effc9caf8cca66c50ff6
CRs-Fixed: 2636431
