Currently, per-user information in Tx PPDU completion is allocated for
37 users though FW may send information for fewer number of users. In
new scheme, FW provides number of users in USERS_INFO TLV and host can
allocate the per-user struct based on this number. This can save memory
while processing each completion message.
Make related changes in tx capture files to use pointer instead of fixed
size member.
CRs-Fixed: 2730933
Change-Id: I8709665a481d64dc4ab086ce9b27e8837c38b53f
Debugfs is added to store the last 30 ppdu desc and display
as readable format to user. Feature of storing ppdu desc is
enabled through debugfs and it is not enabled by default.
Change-Id: I85ff52760462abd9a6e33d082db45d8179e67c0d
Currently, same user can be processed multiple times, which leads to
release ref count wrongly, with this, some of the users in PPDU
desc are never processed and can possibly cause memory leak.
Added a flag in PPDU user to indicate that the user is already
processed.
CRs-Fixed: 2718150
Change-Id: Ia00e0ad12999578157e9f8a6df0ad4763f828c88
For some peers, tx capture tid init is not invoked,
It is resulting in ppdu_desc leak.
Invoking tid init on tid peer update, if tid is not
initialized.
Change-Id: Ia49c7e6d093a42e57e8d9239a60ad0b8dabfcc5b
CRs-Fixed: 2718150
for MU frame, rts and cts are transmitted only for first user.
fix other place to use index from caller instead of 0.
Change-Id: I672d91c3df2fac358942304ad6613ee32313e442
deauth frames are treated as speacial frame where peer id
point to bss peer id but payload RA points to station
mac address. Peer filter is done based on payload.
Change-Id: I4238a108a62accbf3a53f74332a10377675870b8
CRs-Fixed: 2692363
Before adding a ppdu to the retry queue, check if the
sched_cmdid matches that of the first ppdu in the queue.
If it does not, then the retry queue should be freed as
the received ppdu is not a retry of the ppdu in the queue.
Change-Id: I96a097f57d86539c8395af28f6fd57c13f4cad49
delete redefinition of mcs from rx_user_status and keep mcs
derived from preamble which came from rts.
Change-Id: I5323d7944abd474a82d34b812caac1290168ea94
Add support
a. Drop msdu on queue exceed threshold of 4096
b. Add support to print consolidated peer tid queue.
Change-Id: I2b91b151531c839657716ac52987cf5e4a62e7cc
preamble value for mubar case is stored in phy_mode variable. this needs
to get carried over to preamble variable in mpdu_info so its used
correctly in radiotap setup.
Change-Id: If149f7b58a3d8b788bab706f8943d66f95e8517c
For BRP frame ppdu, frame control is not valid in all user
completion tlv, which is causing to drop ppdu as wrong ctrl mgmt
queue checked for payload.
To fix this, updated frame control based on ppdu frame type
Change-Id: Ia5eb67da14647cccfe2ade7c0626a97171454446
CRs-Fixed: 2677252
all probe responses and disassoc from bss peer will be accepted and
later filtered by
mac. Additionally retry queue is having a treshold for this new
mechanism to be able to flush queue in case it gets filled up.
this is meant to work for every control frame that has reached such a
state in its retry q.
Change-Id: Ie49d00d9b9422f02f5d9512a891ceb546cdd432d
a. ppdu desc in pending ppdu queue is dropped on queue length exceeds
threshold which is 32.
b. configuring ring map for soc only once.
Change-Id: I7398a478ce9e4d974d9ecc2a06d30821f151a1b5
For packets from exception ring will have extra HTT header, that
is adjusted to avoid wrong LLC header
Change-Id: I36adaa6ab0c3ba96a5eec9bf05747576e3938028
CRs-Fixed: 2661034
The current implementation of the TX Capture
feature assumes a non QoS 3 address frame for
each frame. If a frame is a QoS frame, the extra
fields are manually added, but if the frame is a
4 address frame, there is no handling. Add QoS,
4 address and 4 address QoS frame types to the
TX Capture feature and update accordingly.
Change-Id: Idd7f8f55a5543718f52bc38be396d671b87b54bd
CRs-Fixed: 2636684
filter pass seems to not be working for control frames
so we are using mac check address comparison for non-bss freames
when tx_monitor is set
Change-Id: I3a003636381f73191081e821dbe8cf00a67cb042
If tx capture, sniffer are not enabled, mgmt nbuf is freed.
It is causing use-after-free in bpr enabled case
Added change to free only when bpr is disabled
CRs-Fixed: 2662214
Change-Id: I0d889f371cf47047200f70563b589fac99733c49
use mac addr compare function with dummy mac addr set to all 0 and dest
mac addr. if they are similar then just ignore frame.
Change-Id: I8c892d0ca4a210865720b94b580b6e24dad0382b
Beacon frames are FW generated frames, FW indicated them as DATA_SU,
with this, frame is not marked as self gen. It gets added to retry_q.
Check ppdu frame type and subtype to mark frame as selfgen if it beacon
Change-Id: If0bfd9bf2cc321afe207a4db7529e24baebfd376
CRs-Fixed: 2643848
if last_msdu is not received then allocated
mpdu_nbuf and mpdu_q is not freed,
This change frees the buffers in last_msdu not seen case.
Change-Id: Icaaa6e227f3ea1d314c6d15a9f76c2f622d5d93f
CRs-Fixed: 2643848
If tx_capture is disabled, the frames notified to tx_capture
module needs to be freed.
This change frees mgmt frames in tx capture module, in case feature
is disabled
Change-Id: I71f102ba2f25f23bdcef63dddb525d5f7b0a692a
CRs-Fixed: 2643848
1) Avoid removing wrong nbuf from queue
In corner case, the current flow can remove first nbuf instead of
actual nbuf from queue, this can lead to having the actual pointer
stored at two array indexs.
2) tsf wraparound handling for mgmt frame
3) Mpdu tried is higher than ba_sisze
mpdus[] is allocated with ba_size, it is observed that mpdu_tried
is higher than ba_size, which can lead to buffer overlow.
Added the check of ba_size along with mpdu_tried.
To avoid it, copied nbuf in the corner case, and left the actual nbuf
in the xretries queue.
Change-Id: Ifa57eaf62cf50aa2c356a6d1f41ef4b7283be52e
CRs-Fixed: 2643848
Fix to support Tx capture for below issues
a. intermittent change to drops all MU frames in Tx capture.
b. Fix to support, FW indicates enqueue bitmap with more bits than
actual MPDUs tried by HW tx capture holds more pending PPDUs, it causes
higher memory usage.
c. holding peer reference on get and release reference on delete
during tx capture.
Change-Id: I0ac5dce7995413585164b6c15dd8b647f0886fb1
Currently, there is no check for QoS NULL when
handling data frames in the TX Cap feature. This
leads to QoS NULL frames not being seen OTA. Add
a check in dp_tx_ppdu_stats_process to ensure data
frame is not a QoS NULL data frame before handling,
since QoS NULL payload comes in management path.
Change-Id: I9fa2e68115327a283b241214ff62534bcb1ed74a
CRs-Fixed: 2632712
