kernel/linux.git/drivers/net/wireless/ath, branch v3.15.2

ath9k_htc: Stop ANI before doing hw_reset

2014-05-14T19:42:46+00:00

During remain on channel request, ANI worker thread is not stopped before doing hw reset. This is causing kernel crash in hw_per_calibration. This change ensures that ANI is stopped before doing chip reset and it will be rescheduled later when the chip is configured back to home channel and having valid bss. Reported-by: David Herrmann Tested-by: David Herrmann Signed-off-by: Rajkumar Manoharan Signed-off-by: John W. Linville

ath9k: remove tid->paused flag

2014-04-30T15:42:35+00:00

There are some corner cases where the driver could get stuck with a full tid queue that is paused, leading to a software tx queue hang. Since the tx queueing rework, pausing per-tid queues on aggregation session setup is no longer necessary. The driver will assign sequence numbers to buffered frames when a new session is established, in order to get the correct starting sequence number. mac80211 prevents new frames from entering the queue during setup. Signed-off-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k_hw: do not lower ANI setting below default on AR913x

2014-04-30T15:39:54+00:00

When the amount of noise fluctuates strongly, low immunity settings can sometimes disrupt signal detection on AR913x chips. When that happens, no OFDM/CCK errors are reported anymore, and ANI tunes the radio to the lowest immunity settings. Usually rx/tx fails as well in that case. To fix this, keep noise immunity settings at or above ANI default level, which will keep radio parameters at or above INI values. Signed-off-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k: fix race in setting ATH_OP_INVALID

2014-04-25T01:46:22+00:00

The commit "ath9k: move sc_flags to ath_common" moved setting ATH_OP_INVALID flag below ieee80211_register_hw. This is causing the flag never being cleared randomly as the drv_start is called prior to setting flag. Fix this by setting the flag prior to register_hw. Signed-off-by: Rajkumar Manoharan Signed-off-by: John W. Linville

ath9k: add a recv budget

2014-04-22T18:09:37+00:00

Implement a recv budget so that in cases of high traffic we still allow other taskets to get processed. Without this, we can encounter a host of issues during high wireless traffic reception depending on system load including rcu stall's detected (ARM), soft lockups, failure to service critical tasks such as watchdog resets, and triggering of the tx stuck tasklet. The same thing was proposed previously by Ben: http://www.spinics.net/lists/linux-wireless/msg112891.html The only difference here is that I make sure only processed packets are counted in the budget by checking at the end of the rx loop. Signed-off-by: Tim Harvey Acked-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k: fix possible hang on flush

2014-04-22T18:09:37+00:00

If a flush is requested, make sure to clear the descriptor once we've processed it. This resolves a hang that will occur if all RX descriptors are full when a flush is requested. Signed-off-by: Tim Harvey Acked-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k: fix a scheduling while atomic bug in CSA handling

2014-04-09T17:55:55+00:00

Commit "ath9k: prepare for multi-interface CSA support" added a call to ieee80211_iterate_active_interfaces in atomic context (beacon tasklet), which is crashing. Use ieee80211_iterate_active_interfaces_atomic instead. Signed-off-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k_hw: reduce ANI firstep range for older chips

2014-04-09T17:55:55+00:00

Use 0-8 instead of 0-16, which is closer to the old implementation. Also drop the overwrite of the firstep_low parameter to improve stability. Signed-off-by: Felix Fietkau Signed-off-by: John W. Linville

ath9k: Enable DFS only when ATH9K_DFS_CERTIFIED

2014-04-09T17:55:55+00:00

Add DFS interface combination only when CONFIG_ATH9K_DFS_CERTIFIED is set. In other case user can run CAC/beaconing without proper handling of pulse events (without radar detection activated). Reported-by: Cedric Voncken Signed-off-by: Janusz Dziedzic Signed-off-by: John W. Linville

ath9k_htc: set IEEE80211_TX_STAT_AMPDU for acked aggregated frames

2014-04-09T17:55:54+00:00

Frame aggregation requires the IEEE80211_TX_STAT_AMPDU to be set so that mac80211 can report the last_tx_rate correctly. Signed-off-by: Chun-Yeow Yeoh Signed-off-by: John W. Linville