In function p2p_process_scan_complete_evt roc timer state is checked
using the api qdf_mc_timer_get_current_state, this api does not take
timer lock while returning the state. In other thread, if the timer
state is changed then the wrong timer state is returned in current thread.
And wrong timer state can lead to destroying roc timer without stopping
it synchronously.
Call qdf_timer_stop_sync api without checking the timer state
as the timer already checking the state by taking a timer lock.
Change-Id: Idb22e405178f2e0f1ed80b7c8aac04170e5b69bc
CRs-Fixed: 2511768
Object 'tx_conf_event' was dereferenced after being freed by calling
'qdf_mem_free(tx_conf_event)' function. Object 'tx_conf_event->nbuf'
can't be freed.
Free 'tx_conf_event->buf' before 'tx_conf_event' free.
Change-Id: Id5a8f17db4d58395837db9878c6c7d87e9fe40c4
CRs-Fixed: 2503761
It just support to tx probe request/response and action frames, and
reject to tx other frames. So add this change to support other mgmt
frames.
Change-Id: I6d99f1c1b0187105fc8eec1f170483980819d0ae
CRs-Fixed: 2485734
It just support to tx probe respnse and action frames currently. So
add this change to support to tx probe reqest frame.
Change-Id: I4899eeca80958990aff2ec5eac3226f9a37ccbf3
CRs-Fixed: 2481346
In P2P GO mode, if the P2P client device is not VHT capable and
only HT capable, but the DUT P2P GO is VHT capable and advertises
2x2 NSS and sends packets with 2x2 rates, it results in interop
issues with HT only capable P2P client devices.
When GO is operating in DBS mode, GO beacons advertise 2x2
capability but include OMN IE to indicate current operating mode
of 1x1. But here peer device is only HT capable and will not
understand OMN IE.
Check if the P2P client device is only HT capable and has
"\x00\x50\xf2\x04". Then downgrade the P2P GO operation to 1x1
nss. Modify rx mcs map value in association response to enable
only mcs 0-7.
Change-Id: I7177c5aee9a560a20c50a63886c0ee77f7d617ae
CRs-Fixed: 2464925
It won't send mgmt frame with NULL channel information currently,
which is different to other branches. Add this change because up layer
pass NULL channel information when tx mgmt frame in some platform.
Change-Id: I689d64789187fe0df03ed57ef0ff10c5157aeeb6
CRs-Fixed: 2459485
Mc timer is used to initialize the p2p roc timer.
And the actual timer runs in the soft irq thread and
when the timer exipres it posts the message to mc thread.
Currently, qdf_mc_timer_stop is called to stop the timer.
It calls the del_timer internally to delete the timer.
del_timer() ensures that the given timer is not queued
to run anywhere in the system. But the callback may
be running on another CPU core can create race conditions.
So use del_timer_sync to delete the roc timer.
Change-Id: I2c0fd6e335fc342a3acf06ede534c84d40e19346
CRs-Fixed: 2445705
Currently in the case of concurrent sessions running,
the driver updates the active dwell time for the scan
request to the default value, overwriting the already
filled active dwell time which the DUT got from the
AP as part of RRM request, which results in violation
of protocol.
Fix is to not update the concurrency params if the scan
request is of type RRM.
Change-Id: I09ebfbee0d282391be17aed7eaf56e3c53c2a5e2
CRs-Fixed: 2438535
1. The original mgmt send path changed by converged p2p component, now
wma_tx_packet() will not invoke when sending station mode action
frames, it invokes p2p_mgmt_tx() instead. But for tx buffers of PMF
frames sending to some target require to do bi-direction dma map to
pass smmu sanity check, which is missed in p2p_mgmt_tx() path.
2. The offchan flag passed to wlan_hdd_mgmt_tx() is not always accurate.
There have been cases where the flag is set even though the operation is
taking place on the home channel, and blindly following the flag leads
to an unnecessary ROC.
Fix it by adding QDF_NBUF_CB_TX_DMA_BI_MAP to new p2p mgmt tx path and
bring back home channel check to override the offchan flag if
necessary.
Change-Id: I9589157e57e95d0c61a432f87a07630c4b77f757
CRs-Fixed: 2423341
Driver uses cipher stored in vdev to get the MIC length, which
may get updated if multiple peer(TDLS peer in STA case) get
connected to the vdev. Thus depending on latest peer cipher type
the MIC length will be calculated for all peers.
To fix store cipher info in peer and use it to calculate MIC length
for the frame.
Change-Id: I8afbf9a3bb43c294dbacbbaa7fa0746600937d11
CRs-Fixed: 2428482
Per the Linux Kernel coding style, as enforced by the kernel
checkpatch script, pointers should not be explicitly compared to
NULL. Therefore within p2p replace any such comparisons with logical
operations performed on the pointer itself.
Change-Id: I3f987cc2603bd3e4d41881db8c847046884d7e76
CRs-Fixed: 2418399
When forming P2P connection, mgmt tx timer timeout when tx ack event
queued in scheduler but not processed, and then wild pointer access
in tx timer timeout callback since processed tx ack event and tx
context is freed.
Change-Id: I598e611a9e1d72bbaddbbe3541f37479f3cb9bbb
CRs-Fixed: 2372686
In p2p go mode the peer type in wlan peer common object is always
set to P2P_CLI. This leads to non-cancellation of NoA after connecting
to legacy stations. Correct the peer type for legacy stations.
Change-Id: Ib706f9a80d0ad367e27fd21eca6cf026cba63f57
CRs-Fixed: 2406127
Up layer tx probe response frame with wait/duration 0, which may be
start ROC with duration 0 and then tx mgmt frame fail since of very
short ROC and no enough time to wait for tx ack.
Change-Id: I5243ff10726b642d8cf18992a230d741f9d2655f
CRs-Fixed: 2397719
Currently, INI gSkipDfsChannelInP2pSearch is initialized in P2P
component and later copied to scan component for usage.
Since this INI is only used by scan component and not the P2P
component, move the initialization part also to the scan
component.
Change-Id: I44a6208ab5f3ee062073550143e23e5800342376
CRs-Fixed: 2405395
Driver multiplies the supplicant requested roc duration by either
6 or 10 times and then requests FW roc of that much duration.
This is because to extend the roc after rx frame is received.
Currently, host is initializing the internal roc timer for the
duration same as requested to FW.
In such case we cannot extend the roc as currently there is only check
for timer running, but not the remaining time. And FW may give
scan completion event immediately after sending the tx frame.
Change-Id: Ib864949d232c31422f83d78bd6b318cadbf13366
CRs-Fixed: 2397924
Currently many modules use the scheduler_post_msg API which
doesn't allow the driver to know about the src, and destination
modules.
Fix is to call scheduler_post_message which has src, dest, and
the que id.
Change-Id: Ia5c43fe3ceeb7db9dde6241cbabac0ce48f88061
Required sta interface to support to tx p2p action frames, this is
part of customer requirement.
Change-Id: I1ebebce4d70dcba34427bb2276cd3e1cc23c2bd8
CRs-Fixed: 2388414
Change I9b0126d9cc71f7de0d255981ba5fe88f3da05ce2 ("qcacmn: Define
configures about P2P") introduced struct p2p_param. One of the fields,
ignore_dynamic_dtim_in_p2p_mode, has apparently never been used, so
remove it.
Change-Id: I1c7c34d72f22c89f243d2639ec2baccb76be7dc2
CRs-Fixed: 2379074
P2P component related files moved from hostcmn to CLD, this change
adds initial functions, enables P2P in Kbuild.
Change-Id: I04102a5f3e59ded13ada15b1b417ae81c7a10e90
CRs-Fixed: 2361500
This is mirror change for 0f9f01950
Currently if the number of APs in the STA environment
are many, then the STA will receive many beacons, whose
beacon process path can take long time, in the kernel
work queue, hence the other processes have to wait
for them to complete, and may get timeout, if the
the time to process the beaocns is larger than their
process timeout.
Fix is to :-
1. Add rate limit to failure conditions of memory
not allocated
2. Make memory allocation in path of beacon process
atomic.
Change-Id: I488b446c23fd01c993f7dd9bd989867fda2331d8
CRs-Fixed: 2363307
