Facilitates faster data transfers. This provides a basis
to start using the other 4 CEs for additional services.
Only adding structure elements to enable these CEs not
really using them yet.
Change-Id: I4be07413de409195ccc717765c75806a1470941e
CRs-Fixed: 938752
qcacld-2.0 to qcacld-3.0 propagation
When INTRA_BSS_FWD_OFFLOAD is enabled, FW will send all Rx packets to
IPA uC, which needs to be forwarded to other interfaces. IPA driver will
send back non-IP, broadcast and multicast packets to WLAN host driver
in exception pipe with fw_desc field set by FW as:
- FW_RX_DESC_FORWARD_M bit: Forward the packet to TX
- FW_RX_DESC_DISCARD_M bit: Do not send the packet to network stack,
which will be set of for intrabss unicast.
Change-Id: Ie2826c8c7b578e3684133a2554ed086e721f67ae
CRs-Fixed: 895662
Change the range of values for early stop scan feature
configuration items as per firmware expectations.
CRs-Fixed: 939185
Change-Id: I7d893cd1ed44a71bffaab22121f052288b14b640
qcacld-2.0 to qcacld-3.0 propagation
HDD IPA event_lock was originally introduced to protect IPA resources
between HDD_IPA events, but this also used to protect HDD IPA internal
struct now. So, this could cause an unexpected deadlock by guarding
too wide resources.
This change is to separate a lock for IPA resources.
Change-Id: Ie9b676e5d4775e54bddf141b841e0f0daeb22731
CRs-fixed: 934723
qcacld-2.0 to qcacld-3.0 propagation
Change cdf_trace_hex_dump to use hex_dump_to_buffer.
This will print one line even if buffer size is less than 16 bytes.
Change-Id: I2dec78048d52f96d494c9e0ce133c4c18f48e954
CRs-fixed: 933157
The Tx & Rx SS macros aren't called with correct parameter in
wma_update_intf_hw_mode_params. Fix the calls to get the Tx & Rx SS values
correctly.
Change-Id: I4df4db549d4bb0c7d00a521de55165bfccf16a65
CRs-fixed: 935341
This is qcacld-2.0 to qcacld-3.0 propagation
Due to cleanup in function limSendReassocReqWithFTIEsMgmtFrame()
assocReqLen was not being set to zero when the corresponding buffer
was NULL. This caused memory copy in buffers being passed between
PE and SME and HDD to be improper. HDD when reading the buffer
wrongly read it from wrong offset and hence RSN IEs being passed to
supplicant are all zeros. This patch fixes this by setting
assocReqLen = 0 when AssocReq buff is NULL
Change-Id: I0edaff93fcdbb2d5ae80478a76f07cb4dee8aceb
CRs-Fixed: 929797
HTC control messages ready and connect service indications use
CtrlResponseBuffer for processing the HTC messages. Flag
CtrlResponseProcessing is used to protect this buffer from
overwriting before processing. Current implementation has race
condition as setting the flag happens on waiting thread context
instead of receive thread context so this could lead to
overwriting the CtrlResponseBuffer if waiting thread doesn't get
chance to run because of may be another RX interrupt.
Also to catch this kind of issues, add CDF_BUG so that it is
easier to triage the cause for the malformed packet which is not
expected.
CRs-fixed: 881504
Change-Id: Ia279e08c761a3432cf01d4dc129f7b5439d04680
Fix compilation issue when ROAM_OFFLOAD is not enabled. There are
issues with ROAM_OFFLOAD feature conditional compilation.
Change-Id: I068245885ab9f054b4c03307a9ecc0066bc88ebe
CRs-fixed: 784982
- Fix a crash seen due to memory being freed twice
- Deregister the LRO flush callback during LRO disable
CRs-Fixed: 933123
Change-Id: I18edadc620da04b9a571da6e1b143c099c308d90
This change is to address a watchdog bite that is seen when
a peer corresponding to the transmit frame is not found in
the OL TXRX transmit path. The change is to log the error
using a CDF trace warning message.
Change-Id: I8792da5539c75a3041c3c85f5a7e29714b721502
CRs-Fixed: 936374
This commit will fix the incorrect pmf config param for STA DUT
in-case of LFR2/3 roaming scenario.
Change-Id: Ic0d6d4632cfeaa0589b435a6710f3c2380cf5883
CRs-Fixed: 935947
qcacld-2.0 to qcacld-3.0 propagation
Current driver is honoring association response with try again later
as reason code without connection being PMF. Driver has been designed
to handle try again later reason only for PMF connection. For non-pmf
case it should report as failure to upper layer just like any other
failures.
To fix this situation, honor try again later reason code only if the
connection is PMF.
Change-Id: I4b37d072bc92e003fcfe62ffe8f93f7c185eb6c7
CRs-Fixed: 919203
Change "qcacld-3.0: Record data in DP Trace" introduced skb tracing
functionity to the driver. In hdd_softap_hard_start_xmit() logic was
added to trace the skb contents, including logic to trace the skb
contents when an skb was dropped. However some of the code paths to
this "drop_pkt" tracing have a NULL skb. Therefore we must bypass the
dropped skb content tracing when the skb is NULL.
Change-Id: I485c92647355a0f7f420f40640b697a1d3eb5fb0
CRs-Fixed: 938254
Fix layering violation in hdd_update_tgt_services by not accessing
pMac structure directly from HDD.
New SME API - sme_update_tgt_services() introducted to configure the
parameters in pMAC structure.
Change-Id: I7310e51001c5e17ed63730d7a86445548aa3d246
CRs-Fixed: 915267
Fix inappropriate use of Linux-specific workqueue code and
data structures in WMA.
WMA is supposed to be os-agnostic code but it currently has
Linux-specific data declarations and code for ack_cmp_work.
This change set will convert the existing code to os-agnostic
representation.
Change-Id: I3260fa3b4b482c10272ac94ab10e14e40b8950f9
CRs-Fixed: 924421
Fix the usage of list node which was deleted before the usage in
wma_cleanup_hold_req function.
Change-Id: I8ee243fda511de71083ce7e631a21497f4f153c1
CRs-Fixed: 937267
Fix the usage of linux-specific structures and functions by
replacing them with OS-abstracted types for vdev_resp_queue.
Change-Id: If2d4a9a1dffd59ec1db4b4aa572dfa453bfd0167
CRs-Fixed: 931094
qcacld-2.0 to qcacld-3.0 propagation
Add NAN feature follow up TX related dbglog IDs.
Change-Id: Id015d185837e917b3bf026a4a39528e7d80b2588
CRs-fixed: 938151
qcacld-2.0 to qcacld-3.0 propagation
In order to improve driver load time, reduce PM QOS latency tolerance
time to minimum to avoid CPU enters aggressive PS during loading driver.
Change-Id: Ibacf933d379f37063d900d444de20493b94ebd90
CRs-fixed: 935267
This is qcacld-2.0 to qcacld-3.0 propagation
Fix the misspelled entry RIND_ID_DRIVER_DEBUG to
RING_ID_DRIVER_DEBUG in the enum wifi_logging_ring_id
CRs-Fixed: 938286
Change-Id: I0d33128c6d11ecc9ff3d403931bfd6bfdbbb2b4a
Avoid waiting for SME response messages while holding the SME global
lock. Currently a wait for 500ms happens in cds_handle_conc_multiport
after the holding the SME global lock. Because of this, the set hw mode
response posted by LIM is not processed by SME as SME would be waiting
for the lock.
Change-Id: Ib7890cb7124f5879316f46b1bf9f0ad3fb225665
CRs-Fixed: 939026
Add CDS connection manager APIs to manipulate the connection
update event. These APIs can be used to initialize, reset,
set and wait for connection update event.
Change-Id: I5689711d4bdc0e39b2456573158943ebda9cbc2a
CRs-Fixed: 939026
cds_get_bin.h and cds_get_bin.c were derived from qcacld-2.0
vos_get_bin.h and vos_get_bin.c. These files got their names
since at one time they housed the wrappers used to retrieve
binary files nv.bin, cfg.dat, and firmware. But over time the
functionality to actually retrieve binary files (get_bin) has
been removed, and almost all that is left is concurrency
management functions. So, moving the .[ch] functionality to
the right files.
CRs-Fixed: 932777
Change-Id: I4a26c304cc6b6224d0839c365e4006fd32270961
qcacld-2.0 to qcacld-3.0 propagation
An increase in tx_queue_len of netdev to 3000 is giving us better TCP
TX performance. Hence, increasing its value from its default (1000).
This gives us about 30 - 40 Mbps of boost.
Change-Id: I68f26ac0f4a040b0bbf10dd571dae8e5316f16b3
CRs-Fixed: 927673
qcacld-2.0 to qcacld-3.0 propagation
In hdd_cnss_request_bus_bandwidth function, the overall bus bandwidth
and rx bandwidth requirements are being determined based on tx and
rx packets. This change detects tx bandwidth required (high or low)
and sends WLAN_SVC_WLAN_TP_TX_IND indication to cnss-daemon which can
tweak system parameters depending upon TX bandwidth needed.
Change-Id: I3eec2e3f95fa32b191d905209a38cb6d48837c33
CRs-Fixed: 918529
qcacld-2.0 to qcacld-3.0 propagation
Firmware indicates the number of peers connected as part of iface
stats. If there are no peers connected, don't wait for the
peerstats and indicate the stats to host.
Change-Id: I32d04a5711adb5f924d120ee61064a178e753549
CRs-Fixed: 847743
qcacld-2.0 to qcacld-3.0 propagation
Register values are captured in the crash dump with BUG_ON.
Change-Id: I33500aea0769879d63b2e684dc898ba0e3b0b5a2
CRs-Fixed: 813402
qcacld-2.0 to qcacld-3.0 propagation
In the HTT hash de-init function, the network buffers (skb) need to
be unmapped before they are freed.
Change-Id: I879de6671d788f1ab99f7645ab424d56eafdbfd1
CRs-Fixed: 815265
The existing log for dumping concurrency information doesn't consider
DBS & also doesn't cover for various new combinations added newly.
Update logic for dumping DBS info & cover the new concurrency cases.
Change-Id: Id898bdc2c8aabfa70c99e6193c52f2723ded5c65
CRs-fixed: 935453
The Tx & Rx SS & mac id update already been done by
wma_vdev_start_resp_handler. So, no need to update again by calling
wma_update_intf_hw_mode_params.
Change-Id: Ib9804bf0d30c75d95b14925a902fe167fa681be0
CRs-fixed: 935341
qcacld-2.0 to qcacld-3.0 propagation
Currently in hdd_set_app_type2_parser, sscanf writes 4 bytes of data in
tcp_src_port and tcp_dst_port variable. The original size of variable
is 2 byte.
Update code to write appropriate size of data in variable.
Change-Id: Ifdbfa06da6cb7d0bfef4f4e105a6911950f1d6ad
CRs-Fixed: 859493
Issue the connect command immediately when the system is already
in the required hw mode. Currently the connect command is issued
immediately only when the concurrent connection update succeeds,
i.e., only when a set hw mode succeeds. But, when the driver is
already in the required hw mode, the return value would not be
success and the connect command is not getting issued immediately.
Fixed this by checking the return value which indicates that the
driver is already in the required hw mode, and issuing the
connect command immediately.
Change-Id: I1ef2adf4d92953f47d36bdae41a9d7d8d559a203
CRs-Fixed: 936268
qcacld-2.0 to qcacld-3.0 propagation
When the driver is in SoftAP mode when it is unloaded,
and if many clients are connected,
sometimes not all clients can gracefully disconnect.
In this case the IPA HW pipes are not cleaned up properly,
and subsequent pipe setup will usually fail.
To avoid a bad pipe configuration, before unloading driver,
check the pipe status and if the pipes are not closed properly,
shutdown the pipes forcefully.
Change-Id: Icc0543567423cc4b625140280be13ef733d7d67d
CRs-fixed: 886229
Propagation from qcacld-2.0 to qcacld-3.0
Add length sanity check in iw_set_keepalive_params to avoid
stack over flow.
Change-Id: If59794d1b5f8efad2fe54844fab0e7cea9b1b983
CRs-Fixed: 932466
This is a qcacld-2.0 to qcacld-3.0 propagation.
Function hdd_extscan_start_fill_bucket_channel_spec() is used to
transfer EXTSCAN bucket parameters from a netlink message to an
internal representation. A code analysis tool identified a potential
buffer overflow of the per-bucket channel list in this function. In
reality this is a false positive since there is other logic which
limits the total number of channels across all buckets, and this logic
will prevent overflow of the channel list for a single bucket since
they use the same limit.
However this presents a code maintenance issue since a potential
overflow could be introduced in the future if a change is made to
allow the total number of channels across all buckets to exceed the
number of channels allowed in a single bucket.
To protect against this possibility, as well as make the code analysis
tool happy, add an additional check to make sure the per-bucket
channel list does not overflow.
Change-Id: Ifdf2de918d8b2c5a907e833e7bc42315b607e5a5
CRs-Fixed: 922047
This is qcacld-2.0 to qcacld-3.0 propagation
When SETDFSSCANMODE is set to 0, it should now avoid DFS channels
altogether. Original implementation avoided DFS channel in roaming.
Now host code will first disable DFS channels and then follow
the original path.
Modifications to wlan_hdd_cfg80211_disable_dfs_chan_scan()
- Keep validation of vendor command from NL.
- Moved the DFS channel manipulation code to
wlan_hdd_disable_dfs_chan_scan()
Call wlan_hdd_disable_dfs_chan_scan() in SETDFSSCANMODE processing.
CRs-Fixed: 778613
Change-Id: Ia2c00a6eb8d50e7962224375562a5b6417ba8bb2
qcacld-2.0 to qcacld-3.0 propagation
The wait for key timer value in the driver should
be greater than that of supplicant.
Hence increasing it to 15 seconds.
CRs-Fixed: 919403
Change-Id: I5d6e8d445ab9dcef73b923595f41564e0a6e3c6e
qcacld-2.0 to qcacld-3.0 propagation
1) Enable the feature only when connected to 2.4 GHz AP.
2) Modify the default values to reduce the scans and not impact power
CRs-Fixed: 817919
Change-Id: Idaff886cec558a06cf9341e47a8ce17b348ad72b
qcacld-2.0 to qcacld-3.0 propagation
Enable the HI_RSSI feature from the code
instead of WCNSS_qcom_cfg.ini.
CRs-Fixed: 822475
Change-Id: Ic48e49e7042c06b1f0b2bf09d2d7ccae2d7cd9ad
qcacld-2.0 to qcacld-3.0 propagation
The MAC is not started upon FTM start.
Similarly stopping is not needed during FTM stop.
CRs-Fixed: 917306
Change-Id: I72ee4a58c88b54c1dad1b82f749274e0c2e8a9ce
qcacld-2.0 to qcacld-3.0 propagation
In the current design, PMKID is sent in RSN IE in reassoc req for
CCKM roaming.When AP receives CCKM IE and PMKID, it gets confused
and rejects the reassociation.So in case of CCKM+RSN, don't send
PMKID.
Also, in reassoc request, MIC was going 0 in CCKMIE.This MIC is
calculated by calling platform driver API.When this API returns
SUCCESS then result is copied in MIC buffer but if this API
returns EINPROGRESS/EBUSY, then driver wait for the event and once
the event is completed it does not copy the result in MIC buffer.
So changes are done to copy this result in MIC buffer in case of
BUSY/INPROGRESS status.
Change-Id: Ide74db250aa5c129fa223582be830c15f34cfb16
CRs-Fixed: 763969
qcacld-2.0 to qcacld-3.0 propagation
Some of the logs are misleading to believe that there are
some memory related issue which infact is not. Cleanup the
logs to convey the exact message that has to be conveyed for
those scenarios
CRs-Fixed: 907715
Change-Id: Iaab688eb9b8b19705b3d87c4efe1c7ef29ccf613
qcacld-2.0 to qcacld-3.0 propagation
In DFS mode, scan req completed through work item
which is async method and may lead to null
pointer access during driver unload.
So as part of fix null check are put in place
to avoid null data access.
Change-Id: I1f2255c1ad6e3e881626a32384b9badde1b255fc
CRs-Fixed: 894741
qcacld-2.0 to qcacld-3.0 propagation
As a part of unload hdd is stopping the netdev queue.
But still N/W layer 3 is calling the select queue netdev
ops, where error print appears excessively.
As a part of this fix sta select queue netdev ops need
to validate Hdd context to take care of SSR and load
/ unload cases.
Change-Id: I087f50ac9e59c3e181bde54f2041f9dda3177266
CRs-Fixed: 857429
