A-Team_Digital_Solutions/android_kernel_motorola_sm6375
Watch
2
0
Fork
You've already forked android_kernel_motorola_sm6375
 0
mirror of https://github.com/BobTheBlinker/android_kernel_motorola_sm6375.git synced 2026-10-11 07:03:09 -04:00
Code Issues Projects Releases Packages Wiki Activity

qcacld-3.0: Fix TDLS teardown issue

Browse source 
TDLS teardown happens in high throughput
scenario as rx pkt count increment API
"tdls_update_rx_pkt_cnt" is not called
if hdd_is_current_high_throughput API
returns TRUE. The reason for this is due
to high cpu cycles consumed by
tdls_update_rx_pkt_cnt API.
As part of fix optimize tdls_update_rx_pkt_cnt,
tdls_update_tx_pkt_cnt API's for high
throughput scenario. In high throughput
scenario mostly the use-case is to increment
the pkt counter so move that logic to
the top of the API's.

Change-Id: I4eb0da3571c45b5ee91e051e251158af15f73973
CRs-Fixed: 2888746
This commit is contained in:
Utkarsh Bhatnagar 2021-03-30 12:30:56 +05:30 • committed by snandini
commit 3e70c71cb1
2 changed files with 57 additions and 28 deletions
Download patch file Download diff file Expand all files Collapse all files

73
components/tdls/core/src/wlan_tdls_ct.c
View file

@ -1,5 +1,5 @@
/*
* Copyright (c) 2017-2020 The Linux Foundation. All rights reserved.
* Copyright (c) 2017-2021 The Linux Foundation. All rights reserved.
*
* Permission to use, copy, modify, and/or distribute this software for
* any purpose with or without fee is hereby granted, provided that the
@ -249,6 +249,24 @@ void tdls_update_rx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
if (!tdls_soc_obj->enable_tdls_connection_tracker)
return;
/* Here we do without lock to ensure that in high throughput scenarios
* its fast and we quickly check the right mac entry and increment
* the pkt count. Here it may happen that
* "tdls_vdev_obj->valid_mac_entries", "tdls_vdev_obj->ct_peer_table"
* becomes zero in another thread but we are ok as this will not
* lead to any crash.
*/
valid_mac_entries = tdls_vdev_obj->valid_mac_entries;
mac_table = tdls_vdev_obj->ct_peer_table;
for (mac_cnt = 0; mac_cnt < valid_mac_entries; mac_cnt++) {
if (qdf_mem_cmp(mac_table[mac_cnt].mac_address.bytes,
mac_addr, QDF_MAC_ADDR_SIZE) == 0) {
mac_table[mac_cnt].rx_packet_cnt++;
return;
}
}
if (qdf_is_macaddr_group(mac_addr))
return;
@ -269,16 +287,15 @@ void tdls_update_rx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
wlan_objmgr_peer_release_ref(bss_peer, WLAN_TDLS_NB_ID);
}
qdf_spin_lock_bh(&tdls_soc_obj->tdls_ct_spinlock);
valid_mac_entries = tdls_vdev_obj->valid_mac_entries;
mac_table = tdls_vdev_obj->ct_peer_table;
for (mac_cnt = 0; mac_cnt < valid_mac_entries; mac_cnt++) {
if (qdf_mem_cmp(mac_table[mac_cnt].mac_address.bytes,
mac_addr, QDF_MAC_ADDR_SIZE) == 0) {
mac_table[mac_cnt].rx_packet_cnt++;
goto rx_cnt_return;
}
}
/* when we take the lock we need to get the valid mac entries
* again as it may become zero in another thread and if is 0 then
* we need to reset "mac_cnt" to zero so that at zeroth index we
* add new entry
*/
valid_mac_entries = tdls_vdev_obj->valid_mac_entries;
if (!valid_mac_entries)
mac_cnt = 0;
/* If we have more than 8 peers within 30 mins. we will
* stop tracking till the old entries are removed
@ -290,7 +307,6 @@ void tdls_update_rx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
mac_table[mac_cnt].rx_packet_cnt = 1;
}
rx_cnt_return:
qdf_spin_unlock_bh(&tdls_soc_obj->tdls_ct_spinlock);
return;
}
@ -312,6 +328,24 @@ void tdls_update_tx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
if (!tdls_soc_obj->enable_tdls_connection_tracker)
return;
/* Here we do without lock to ensure that in high throughput scenarios
* its fast and we quickly check the right mac entry and increment
* the pkt count. Here it may happen that
* "tdls_vdev_obj->valid_mac_entries", "tdls_vdev_obj->ct_peer_table"
* becomes zero in another thread but we are ok as this will not
* lead to any crash.
*/
mac_table = tdls_vdev_obj->ct_peer_table;
valid_mac_entries = tdls_vdev_obj->valid_mac_entries;
for (mac_cnt = 0; mac_cnt < valid_mac_entries; mac_cnt++) {
if (qdf_mem_cmp(mac_table[mac_cnt].mac_address.bytes,
mac_addr, QDF_MAC_ADDR_SIZE) == 0) {
mac_table[mac_cnt].tx_packet_cnt++;
return;
}
}
if (qdf_is_macaddr_group(mac_addr))
return;
@ -329,17 +363,15 @@ void tdls_update_tx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
}
qdf_spin_lock_bh(&tdls_soc_obj->tdls_ct_spinlock);
mac_table = tdls_vdev_obj->ct_peer_table;
/* when we take the lock we need to get the valid mac entries
* again as it may become zero in another thread and if is 0 then
* we need to reset "mac_cnt" to zero so that at zeroth index we
* add new entry
*/
valid_mac_entries = tdls_vdev_obj->valid_mac_entries;
for (mac_cnt = 0; mac_cnt < valid_mac_entries; mac_cnt++) {
if (qdf_mem_cmp(mac_table[mac_cnt].mac_address.bytes,
mac_addr, QDF_MAC_ADDR_SIZE) == 0) {
mac_table[mac_cnt].tx_packet_cnt++;
goto tx_cnt_return;
}
}
if (!valid_mac_entries)
mac_cnt = 0;
/* If we have more than 8 peers within 30 mins. we will
* stop tracking till the old entries are removed
@ -351,7 +383,6 @@ void tdls_update_tx_pkt_cnt(struct wlan_objmgr_vdev *vdev,
tdls_vdev_obj->valid_mac_entries++;
}
tx_cnt_return:
qdf_spin_unlock_bh(&tdls_soc_obj->tdls_ct_spinlock);
return;
}

12
core/hdd/src/wlan_hdd_tx_rx.c
View file

@ -2593,13 +2593,11 @@ QDF_STATUS hdd_rx_packet_cbk(void *adapter_context,
dest_mac_addr = (struct qdf_mac_addr *)(skb->data);
mac_addr = (struct qdf_mac_addr *)(skb->data+QDF_MAC_ADDR_SIZE);
if (!hdd_is_current_high_throughput(hdd_ctx)) {
vdev = hdd_objmgr_get_vdev(adapter);
if (vdev) {
ucfg_tdls_update_rx_pkt_cnt(vdev, mac_addr,
dest_mac_addr);
hdd_objmgr_put_vdev(vdev);
}
vdev = hdd_objmgr_get_vdev(adapter);
if (vdev) {
ucfg_tdls_update_rx_pkt_cnt(vdev, mac_addr,
dest_mac_addr);
hdd_objmgr_put_vdev(vdev);
}
skb->dev = adapter->dev;