In dp_txrx_peer_find_inact_timeout_handler ignore pdev
which is already detached
Change-Id: I0349d5f9c8d249cf332727bff3c120ddf38a6807
CRs-fixed: 2312279
Set search type only when peer map v2 messaging is enabled
as we will have the search index (AST hash) only when v2 is
enabled
Change-Id: I6c1ab4a4d7519278755d93537a8430755d708a3c
CRs-fixed: 2316418
When a BSS peer gets associated with vap configured in sta mode,
the ast type associated with that entry should be different from
that of regular static entries, The reason being bridge on root AP
can pick up the mac address of AP VAP.
Change-Id: Ie322a015d883e2712f41623f71ccbc255b99baf3
When multiple STA-PEERs are connected to SAP-DUT device and if SAP
is getting shutdown then as part of shutdown:
1) PEER delete happens first and as part of it,
firmware sends peer_unmap event in interrupt-context.
2) entire peer_unmap handling happens in interrupt_context on one
of the available CPU-core.
3) as part of peer_unmap handling, driver removes the peer from
peer_list first and then cleans up its peer data structure.
4) so between peer getting removed from peer_list and
peer data structure getting cleaned up;
control thread (hdd_vdev_destroy/IOCTL/mcthread),
which is running on different CPU core in process-context,
takes control.
5) control thread checks if peer_list is empty. If it is empty then
cleans dp_vdev structure.
6) so when interrupt_context is cleaning up peer data-structure queue,
it gets stability issue while accessing dp_vdev structure which is
already cleaned up by control thread.
Fix the situation by cleaning-up the peer data-struct first and
then remove peer from peer_list
CRs-Fixed: 2291861
Change-Id: I39675bbe4448df57911963942800cbcede508917
While freeing the AST table traverse through table and
free any AST entries present in the table
Change-Id: I87382da2425380938d91ee1980a9c356c0f647ef
CRs-fixed: 2275212
WDS entry should be removed before adding peer with same mac address.
iIn DBDC mode, this can be ensured only by waiting for response for WDS
delete from FW before creating peer. Add logic to defer AUTH until WDS
is removed from FW.
Change-Id: Ie76d08c4817f953504913ae6cc49fc5388169e4a
CRs-Fixed: 2270592
Host changes to enable HTT version 2 messaging for
PEER map and unmap in FW and changes to handle these
messages in host
Change-Id: Ifbe478212bbbc9c9ea1c1e4791c7a78407c376cc
Currently aggregate vdev/pdev is missing Rx nawds mcast drop
counter, add nawds mcast drop counter from peer
Change-Id: I10650639bf5c56410c0f5311090c8a544df89d5d
Add support to send ast hash during peer create.
This value will be used for packet enqueue to transmit block.
Change-Id: Iea188a4b60f96d7770037265d12e534ee2b55bba
Reducing min value of DP configuration parameter to support
low memory emulation platform.
Change-Id: I524dca73b3ce52e138527c7abd1a9d8d0cbe9ac1
CRs-Fixed: 2309295
In HKv1 HW there is a buffersize setting per peer, hence
once we set buffersize we have to stick to this for all
subsequent requests for all tids. If buffersize in Addba
request is more than first session's buffersize, we restrict
buffersize to old value, else we send respond with new size
and send delba to previous sessions not compatible with new
buffersize
Change-Id: Ia71d994ee06ca7b37d93273f5fa69547f816dd20
Crs-fixed: 2247816
Set monitor direct filter when smart monitor is
enabled. Deliver header from NAC if md bit is set
and smart mesh feature is used.
Change-Id: If1ef865e013c21bbb58bcb89a87856b12d6c7278
Crs-fixed: 2246660
Set the frag list of the nbuf of a RX defragmented pkt to NULL.
This list will be used to stich the fragmentation list.
Change-Id: I341b0085c123a72dcd74d7f76a4979286e75dc8e
CRs-Fixed: 2297483
Currently wds enable for vdev conditional check is there for
intra bss rx forward. This check will break the intra bss for
the platforms which does not support WDS feature. In this change
remove the wds enable check for dp vdev.
Change-Id: Ib61c53734254b768902704ccb0e2a5c9b53f35f1
CRs-Fixed: 2303905
Remove SW WAR in rawmode for qca8074v2 target. Added flag in
vdev structure which will be filled during vdev attach time.
Based on the flag war is applied.
Change-Id: Id91d6359b0a575b1f2b1222e56ac93a0e08a3c73
CRs-Fixed: 2296584
To avoid excessive logging to console, control the log rate when
RX data peer is invalid in function dp_rx_err_deliver, use API
QDF_TRACE_ERROR_RL instead of original QDF_TRACE.
CRs-Fixed: 2300774
Change-Id: If1bf3fd93801e0ef37c7fdb1a32cda16aa429464
Some of the macro names defined in qca8074v1, are defined with
a slightly different name in qca8074v2, and few macros have the
same name in both headers but are defined with different values.
Fixed the same.
Change-Id: I5e948baf5326d1d8fdfa2bd7ee8aa072c710d17c
There are many instances where outside DP peer reference is
used. This may lead to potential invalid access as the peer
reference used outside DP can be deleted in peer unmap. So
implement DP ops to give peer handle by incrementing peer ref
count and also to decrement the peer ref count.
Change-Id: I54b131e4ce4fde888e5d0e69f81c6da8633e2787
CRs-Fixed: 2307529
Remove DP_INTR_POLL FLAG as polling is now handled runtime using
hif_is_polled_mode_enabled. This is needed as 8074 works on interrupts
whereas 6290 is still working on polling mode
Change-Id: I0e7fdc338d462735dc216c929b1b5a99dc984435
Regular allocs usually result in allocation of size aligned to
slab boundary resulting in more memory than the requested. Use
multi-page allocation to avoid allocation of more memory than
requested.
Change-Id: Id6562f2fe419179e27206ff5b1d543090b931607
CRs-Fixed: 2294213
Currently gi value was not considered for ratekbps calculation
if sgi is enabled, include gi parameter for ratekbps calculation
Change-Id: Idd8b36375493c26f22f2f4f0c7ae922df37e65a1
1) This reverts commit
I7025fdd7540f47bb4d3f95ef6d9f2ef92af5f1f1
2) Reverted changes have been taken care through:
Idd7617782e71ee187eef7fcb3523c05b49f82094
CRs-Fixed: 2300055
Change-Id: Icc47ded9a585e356b7eae1ad53ffea6668510308
Tie increment and decrement PM usage count to descriptor allocation
and free instead of HW enqueue and TX completion since allocation
and free will be the most accurate places to track if the packet
has been sent or not.
Change-Id: Ia61c5bb26386d20590e87149d0fe88dba02d0fb7
CRs-fixed: 2302214
Reduce the log level for TX/RX path stats
as excessive logging is causing timeouts.
Change-Id: I3c7ef1f75373bfd319b02aab98f5bef099086bac
CRs-Fixed: 2298368
Currently there is no callback for MCL DP to do further action
when receiving data with invalid peer, add it for further processing.
Change-Id: I07c93b5119f0c1a46abba587b6b18c7b691683c7
CRs-Fixed: 2299489
Set 'need_status' while sending REO queue update command
for deleting REO queues to get status from HW, which will
trigger callback to cleanup and free the queues.
Also fix issue in initializing REO command descriptor, causing
command number to be cleared.
Change-Id: Ie30fc80f76d835e239dd0b3d3e78c7f32bfa09a4
When running dptrace dump on Napier, observed that eapol TX counters
are not incremented. Reason is that in Lithium DP nbuf->cb is reset
before dp_tx_trace_pkt() is invoked while dptrace relies on nbuf->cb.
Fix is to move qdf_nbuf_reset_ctxt() after dp_tx_trace_pkt() so that
TX counters that rely on nbuf->cb can be captured.
Change-Id: I5048113e66e8742b8f1f321c07494a6dcf84256a
CRs-Fixed: 2290500
Currently host is taking long time to process the peer unmap
event from FW which is executed in softirq context. Reason is
as part of unref delete spin lock is taken and error level logs
are logged and error level logs take more time to process when
some other printks are already in progress to write to console.
This is leading to change interface failure from SAP to STA as
due to longer time processing of peer unmap event.
Reduce the severity of log level such that logs are directed to
wifi logger logs instead to console.
Change-Id: I543619036aa06c9ac4d97d18e44465b3c9b71806
CRs-Fixed: 2271260
Add support for DP RX Threads as a part of the FR. Multiple RX
threads can be enabled from the ini. The code is added in a new DP
module outside of the cmn project.
Change-Id: Id966c46c2799d23c2a4fa1c884610955afed3565
CRs-Fixed: 2256470
In some cases, error level logs take more time to process when
some other printks are already in progress to write to console.
As this error log inside spinlock, so when some other core is also
competing for this spinlock, that core keeps iterating in a loop.
After some time spinlock panic may trigger as it is suspected that
this core is locked on this spinlock.
This fix moves the error log outside spinlock so that the other
core don't have to wait to acquire spinlock due to this.
Change-Id: I38be80ba3cb65dd77520ca39bf6210e9711f56da
CRs-Fixed: 2299628
Update stats request value if statistics request
from user level is greater that CDP_TXRX_STATS_HTT_MAX value.
Change-Id: I0e36200021fb70662481bc823a95f9e64a8b1e14
CRs-Fixed: 2291842
Using common function to replace the duplicated rate info code in
statistics of data path for TX, RX and peer info printing.
Change-Id: I80d296001b95d7a5ff11d7ce538e16f86e7e67e8
CRs-Fixed: 2291395
If mon vap is down, pdev->monitor_vdev will be set to NULL.
Then function returns, this return will causing mon dest ring not reaped,
causing mon dest ring full FW assert.
And it'll causing mon status ring & dest ring out of sync.
Change-Id: Ie10792da2a4a78d959c676c4b4fba3d9ee05c6e6
When a ppdu id on a ppdu in status ring find a matching ppdu id
in destination ring. The old logic can cause ppdu id is destination
ring stuck on high value before wrap around. The new logic check
ppdu id between status ring and destination ring is no more than
20000.
Change-Id: Ia601403b0a4fc6cdfff980c7745e055c814137b7
