Currently wlan_hdd_cfg80211_register_frames falls through to error
handling at the end of the success path. Return success instead.
Change-Id: I28eda918d3d5858cc42ef3958abe918b9ed788e9
CRs-Fixed: 2240003
If DBS opportunistic timer is still running while starting new SAP on
some DFS channel when STA is already present on some 2.4Ghz channel,
wrong hw mode will be set when timer expires as concurrency will only
be updated when VDEV_UP happens which will happen after at least 60
seconds in case of DFS channel. So, while starting SAP, hw mode was set
to DBS mode but on DBS opportunistic timer expiration, as VDEV_UP doesn't
happen yet, concurrency has not been updated till now causing hw mode to
reset to SINGLE_MAC resulting in DFS MCC which is not allowed.
Stop DBS opportunistic timer if running while starting new SAP to avoid
hw mode reset.
Change-Id: I9cef0ab5f6208e1822aff4a68c442d61a3f8a20e
CRs-Fixed: 2222386
The skb buffer is freed after the ealier sanity and then pass along
to connectivity stats tracking. Rearragne the stats tracking into the
scope of the skb sanity to assure the skb poitner is valid before
accessing it and set to NULL after free to make sure it is cleared.
Change-Id: I08106bced48fa8f23955fd261c679a5362ddd7a6
CRs-fixed: 2232362
WLAN wake lock is created during module init but if probe
did not come from platform driver because of Q6 root/user
pd not coming up then during driver unload WLAN wake lock
is not destroyed which is leading to kernel panic because
WLAN driver is unloaded without cleaning up the wake up
source. Add fix to make sure to do hdd_deinit, pld_deinit
dispatcher_deinit and wlan wake up source deinit even if
hdd context is null because if probe did not happen from
platfrom driver then hdd_ctx will be null.
Change-Id: I213bded171b1c69da3332e84ebb0f7cbd940ef44
CRs-Fixed: 2240186
In the function drv_cmd_set_mc_rate, the variable targetRate, used to
store the value parsed from the command from user space, is not
initialized. The variable is assigned value inside a kernel API
kstrtouint that converts the string to an unsigned int. In a certain
case if the kernel API fails, it returns an error code without assigning
any value to the passed parameter. In this scenario, the variable
targetRate, still uninitialized is passed on to function
wlan_hdd_set_mc_rate where it is logged using hdd_debug. As the flow
goes on, it is then again logged using WMA_LOGE(). This might lead to
potential information leak.
Initialize the variable to zero to prevent the mentioned scenario.
Change-Id: Ideec0b1930e994f19ae8a669cd2963db4016eae1
CRs-Fixed: 2226172
In __iw_set_packet_filter_params(), a user controlled length value,
priv_data.length, is used to allocated a buffer. This buffer is then
cast to a struct pointer of struct pkt_filter_cfg type without ensuring
the buffer is large enough to hold the struct. This can lead to a buffer
overread if the user supplied size is smaller than the actual size of the
struct.
Add a sanity check on priv_data.length to ensure that the size is large
enough to hold the struct.
Change-Id: I227856484d4bd7a9b0a16a42e26febbc799f80b5
CRs-Fixed: 2228725
Presently, fw_crash_timeout is disabled as a result of which if
timeout happens for any VDEV related WMI command and device goes
in bad state, no assert or recovery is getting triggered. Due to
this, device crashes at some later point of time at some random
location making debugging difficult.
Enable fw_crash_timeout and trigger recovery if recovery is enabled
otherwise assert.
Change-Id: I9e9e51cba8086bd181be28884c490f0bd77663a8
CRs-Fixed: 2229101
As a result of P2P componentization a variety of P2P-related macros
are no longer used, so remove them.
Change-Id: I693c4f11575968ce0aae9f3ecac9e008b0dbf5b9
CRs-Fixed: 2237726
Replace typos "sucess" and "sucessful" with correct spellings
"success" and "successful"
Change-Id: I30746cbab4533da9b052261fa9bb87214188138b
CRs-Fixed: 2237724
The Linux Coding Style frowns upon mixed-case names and so-called
Hungarian notation, so rename fConnected to align with the Coding
Style.
Change-Id: I5987e3c31aed44448a1f1843d8931daa7dfeb0fe
CRs-Fixed: 2237723
Rename typedef tsap_Config_t and the underlying struct sap_Config to
align with the Linux coding standard.
Change-Id: I6e8f8b5ab9555c7781f1acf12a0475902866300c
CRs-Fixed: 2237721
The following structs are obsolete so remove them:
- struct sQcSapreq_wpaie
- struct sQcSapreq_wscie
Change-Id: Ie9fdba59cf31e9a2c5e468183c1328aee439709a
CRs-Fixed: 2237720
qcacld-2.0 to qcacld-3.0 propagation
Beacon does not show RSN IE after disabling/enabling SAP, hence
stations does not connect back while trying to associate in WPA.
While re-enable SAP from the hostapd_cli, the start_bss is called
twice, first from wps_init which does not populate RSNIE and secondly,
through bss_setup which has RSNIE but getting ignored as the BSS is
already started.
Added the RSN IE in case of BSS already started
Change-Id: I2a67156e66e255e7c007a5925cd81b859009114e
CRs-Fixed: 2144429
wlan_objmgr_get_vdev_by_macaddr_from_psoc and wlan_objmgr_get_peer
API's definitions have been changed to include pdev_id as an
argument.
Modify the callers of these APIs to include pdev_id as argument.
Change-Id: I3d0de6a0bc1dfefbe1b3cad51ec23f703baaf3ad
CRs-Fixed: 2210728
In hdd_send_association_event() and hdd_send_re_assoc_event(), there is
no check of pBssDesc against NULL which can lead to NULL pointer exception.
Circumvent this issue by adding NULL check.
Change-Id: Ic3a0142b4ea3f058e8d3993fb491d418ddb8f70d
CRs-Fixed: 2232721
Takecare to disable TCP delack internally if GRO is enabled.
This is required as enabling both will have adverse effect with
smaller window size and noisy scenario.
Change-Id: If91236620d87287c9b71901e9d8e3d176eebc648
CRs-Fixed: 2172652
wma_is_service_enabled can't get valid service bitmap if call in
hdd_update_pmo_config, use psoc_ctx->caps.unified_wow instead to check
pattern id per vdev and legacy d0 wow capability.
Change-Id: If7bf316f482c49253fc4b95b94e172727b27ffd1
CRs-Fixed: 2225847
When do re-association, there is obsolete data exists in policy manager
until hdd_association_completion_handler updates the active session.
If check sap restart before this, will access the wrong data.
Change-Id: Ie62f86a5d523e70b6bb81b500bd358b63ae90ea8
CRs-Fixed: 2231675
NAPI APIs in HDD are platform specific. So make these API's
conditionally compiled.
Change-Id: Id87c11da13f096d4f9ff735bbd98dcea23d52310
CRs-Fixed: 2235978
Initialize packet filter clear rule on stack properly to avoid
using un-initialized member variables.
Change-Id: I03b129e7660ed761b03ac163fdb3e7eee58ec034
CRs-Fixed: 2119208
GRO aggregations are hanging out of NAPI instance and We cannot use the
hif NAPI instance in Rx thread as gro_list inside NAPI is flushed out
at the start and end of NAPI poll, this will corrupt the gro_list on
which rx_thread is working. Address this concern by creating dummy
NAPI instances mapping to each hif NAPI and not scheduling them.
Change-Id: I6e1de57a669c56206a7863e86c3840bd0a9eabc1
CRs-Fixed: 2128450
Currently api hdd_wmm_free_context free the hdd_wmm context and
this api is getting invoked from several places without disabling
the inactivity timer.
Since the timer is not disabled, when hdd_wmm_inactivity_timer_cb
is invoked it tries to access the hdd_wmm context which is already
freed and the memory is reallocated to someone else resulting to
undefined behaviour.
To address this issue, disable hdd_wmm_inactivity timer before
calling hdd_wmm_free_context function and add a check for
hdd_wmm context in hdd_wmm_inactivity_timer_cb, if hdd_wmm context
is NULL don't proceed further and return.
Change-Id: Ib085c70b3e5a57b5cd494bc2cd21edd0580c16c4
CRs-Fixed: 2233987
During cfg80211 suspend there are couple of redundant log messages
which are logged on console. Change the log level to debug to avoid
logging them on console.
Change-Id: Id45cc261ec7c2c1397afa746216b090fc5d34d7a
CRs-Fixed: 2235798
In case of STA + SAP, when STA connect to a new AP, the SAP
initiate channel switch to the new channel of the STA. During
this transition period DUT will be in MCC scenario and thus
firmware changes the TBTT of SAP to get 50% channel time.
Due to this peer STA connected to DUT-SAP may miss all 5
beacons containing CSA info and when it comes out of hw-dtim
mode after 5 beacon misses the DUT SAP has already moved to
new channel and thus as SAP is not on the old channel the peer
STA will receive Heart beat failure.
To avoid this Heart beat failure on peer STA, increases the
default value of channel switch count (g_sap_chanswitch_beacon_cnt)
to 10 to increase the probability of receiving beacons with
CSA info.
Change-Id: I2e5182c299f2a100c9919ffc123aa6b1290324f9
CRs-Fixed: 2227704
Change data packets printing to one line
Add msdu-id to TX packets
Increase data packets size to include the complete IP packets
Indication/print when the throttling occurs
Change-Id: I4831c517b41f81003b8d28448a3093d911ac978c
CRs-Fixed: 2157768
The return value check logic is wrong in __iw_get_softap_linkspeed.
Which cause get linkspeed always failed.
Change-Id: If37732867f91e2b95996614ab863953d5ed5af93
CRs-Fixed: 2232332
Add GRO support and make it configurable through INI(GROEnable).
GRO and LRO are mutually exclusive. Only one of them can be enabled.
And disable GRO during following conditions
1) Low TPUT scenario
2) Concurrency cases
3) For Rx packets when Peer is not connected
Change-Id: I15535827a03953231670d4138235c4876b16e045
CRs-Fixed: 2098772
INI item "gBmpsModListenInterval" is not used and hence remove
it from source code.
Change-Id: I32cae84a200541e9028667f16736ce5394bae4b7
CRs-Fixed: 2232887
When the SSR is triggered after the modules are closed, uevent
notifier is trying to get the pdev which is alredy destroyed.
Reject the uevent notification if the modules are already closed.
Change-Id: I8074838b18b1e36dd67afa90125f9fbe06d32c34
CRs-Fixed: 2219843
Currently, in __wlan_hdd_cfg80211_set_limit_offchan_param() function after
parsing the 'tos' value from incoming netlink buffer there is no check
against the max value of tos. This tos value is passsed as an argument
to hdd_set_limit_off_chan_for_tos() function where it is using as an array
index for 'limit_off_chan_tbl' variable which is limited to HDD_MAX_AC.
This may cause out of bound access.
Add check to ensure the parsed tos value is not greater than
or equals to its max value.
Change-Id: I661af1f9559d9bec627594d119423be97978ac2c
CRs-Fixed: 2222803
Currently, there is no break statement between switch cases in
function hdd_stop_present_mode(), here first case has to
fall through the next case instead of using break statement.
Add /* fallthrough */ comment between the cases.
Change-Id: I518628fb8ee084a0ea8de5c98c18a9a6c20b2637
CRs-Fixed: 2233177
In hdd_stop_adapter() there is a switch statement where the logic for
case QDF_SAP_MODE falls through to the next case. Add a fallthrough
comment to explicitly document that this is the desired behavior as
opposed to a missing break statement.
Change-Id: I2fc7b2eeb07eb91b57d5008e74267667a5995c6e
CRs-Fixed: 2233181
In hdd_sme_roam_callback() there is a switch statement where the logic
for case eCSR_ROAM_CANCELLED falls through to the next case. Add a
fallthrough comment to explicitly document that this is the desired
behavior as opposed to a missing break statement.
Change-Id: Ic5fd5c92d7ab421a60d56dfc7761578b554349c6
CRs-Fixed: 2233180
In wlan_hdd_update_phymode() when the curr_band is set to BAND_5G the
channelBondingMode5GHz is not correctly set. Fix the logic which
currently has duplicated test for BAND_2G to test for both BAND_2G and
BAND_5G.
Change-Id: I70ed07d12ae891374e511905213ef7fcc7a7f1f1
CRs-Fixed: 2232919
Currently in __iw_get_char_setnone() while processing the
WE_GET_STATES ioctl a call is made to PMAC_STRUCT() to convert a HAL
handle into a tpAniSirGlobal (aka pMac). This is a data hiding
violation since use of pMac should be constrained to the UMAC -- HDD
should not be utilizing this data structure. And it turns out that
after PMAC_STRUCT() is called the resulting pMac actually isn't used
other than testing for NULL. Furthermore this is dead code since the
HAL handle was already tested for NULL and hence the pMac could never
be NULL at the point it is tested. So fix this data hiding violation
by completely removing the code that populates and tests pMac.
Change-Id: I73fc7c388f4482e5c483a4982231f934cc80ef6c
CRs-Fixed: 2232897
In __iw_set_host_offload() there is a switch statement where the logic
for case WLAN_OFFLOAD_ARP_AND_BC_FILTER_ENABLE falls through to the
next case. Add a fallthrough comment to explicitly document that this
is the desired behavior as opposed to a missing break statement.
Change-Id: I9a7d84841d9d059e822ee15ee482111ca9d87caf
CRs-Fixed: 2233178
