Sven Eckelmann <s...@narfation.org> writes:

> From: Sven Eckelmann <seckelm...@datto.com>
>
> The survey counters from firmwares like 10.2.4 are not actually using the
> full 64 bit. Instead, they only use the lower 31 bit and overflow ever
> 14-30s. The driver must frequently fetch the survey data and add it to the
> survey data storage to avoid this problem and to present meaningful values
> to the caller of .get_survey.
>
> It is assumed for now that only the current rx_channel retrieves relevant
> updates for the survey data. This should avoid that the bss channel survey
> request times out too often.

Please clarify in the commit log on which hardware this workaround is
enabled. It should be all hardware with ATH10K_HW_CC_WRAP_SHIFTED_ALL
enabled so both QCA988X versions and QCA9887, right?

> Tested on QCA988x hw2.0 10.2.4-1.0-00043
>
> Signed-off-by: Sven Eckelmann <seckelm...@datto.com>

[...]

> --- a/drivers/net/wireless/ath/ath10k/mac.h
> +++ b/drivers/net/wireless/ath/ath10k/mac.h
> @@ -40,6 +40,9 @@ void ath10k_offchan_tx_purge(struct ath10k *ar);
>  void ath10k_offchan_tx_work(struct work_struct *work);
>  void ath10k_mgmt_over_wmi_tx_purge(struct ath10k *ar);
>  void ath10k_mgmt_over_wmi_tx_work(struct work_struct *work);
> +void ath10k_survey_dwork(struct work_struct *work);
> +int ath10k_survey_start(struct ath10k *ar);
> +void ath10k_survey_stop(struct ath10k *ar);

Please use ath10k_mac_ prefix for all functions you are adding to mac.c.
Yeah, I know not all the existing one even have that, should fix it at
some point.

-- 
https://wireless.wiki.kernel.org/en/developers/documentation/submittingpatches

Reply via email to