diff mbox series

wcn36xx: Set operational channel to 0 on disassociation

Message ID 20200730110947.57979-1-bryan.odonoghue@linaro.org
State New
Headers show
Series wcn36xx: Set operational channel to 0 on disassociation | expand

Commit Message

Bryan O'Donoghue July 30, 2020, 11:09 a.m. UTC
After a disassociated event we need to make sure we reset the operational
channel so that subsequent channel scans will not exclude the channel we
were using.

Doing a software scan on a wcn3680 showed that if we disassociated from an
AP we would never see it re-appear in a scan - unless we shifted the AP to
a different channel.

Setting the operational channel to zero on disassociation ensures that this
situation will not arise in the wild.

Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>

---
 drivers/net/wireless/ath/wcn36xx/main.c | 1 +
 1 file changed, 1 insertion(+)

-- 
2.27.0


_______________________________________________
wcn36xx mailing list
wcn36xx@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/wcn36xx

Comments

Bryan O'Donoghue July 30, 2020, 7:11 p.m. UTC | #1
On 30/07/2020 12:09, Bryan O'Donoghue wrote:
> After a disassociated event we need to make sure we reset the operational

> channel so that subsequent channel scans will not exclude the channel we

> were using.

> 

> Doing a software scan on a wcn3680 showed that if we disassociated from an

> AP we would never see it re-appear in a scan - unless we shifted the AP to

> a different channel.

> 

> Setting the operational channel to zero on disassociation ensures that this

> situation will not arise in the wild.

> 

> Signed-off-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>

> ---

>   drivers/net/wireless/ath/wcn36xx/main.c | 1 +

>   1 file changed, 1 insertion(+)

> 

> diff --git a/drivers/net/wireless/ath/wcn36xx/main.c b/drivers/net/wireless/ath/wcn36xx/main.c

> index fb8978a3c11e..4681d085b683 100644

> --- a/drivers/net/wireless/ath/wcn36xx/main.c

> +++ b/drivers/net/wireless/ath/wcn36xx/main.c

> @@ -908,6 +908,7 @@ static void wcn36xx_bss_info_changed(struct ieee80211_hw *hw,

>   						bss_conf->bssid,

>   						vif->addr,

>   						WCN36XX_HAL_LINK_IDLE_STATE);

> +			wcn36xx_smd_switch_channel(wcn, vif, 0);

>   		}

>   	}

>   

> 


Hmm.

This one failed for me too.. eventually.

wcn36xx: ERROR hal_switch_channel=0 response failed err=6.

Looks like channel 0 as a safety channel will not always be accepted by 
the firmware.

I think the right fix is to set the working channel to any _valid_ 
channel in our set other than the current channel.

That way the software scan will still run but, the firmware should also 
still be happy to switch to the channel we have set.

Anyway - ignore this patch for now

---
bod
diff mbox series

Patch

diff --git a/drivers/net/wireless/ath/wcn36xx/main.c b/drivers/net/wireless/ath/wcn36xx/main.c
index fb8978a3c11e..4681d085b683 100644
--- a/drivers/net/wireless/ath/wcn36xx/main.c
+++ b/drivers/net/wireless/ath/wcn36xx/main.c
@@ -908,6 +908,7 @@  static void wcn36xx_bss_info_changed(struct ieee80211_hw *hw,
 						bss_conf->bssid,
 						vif->addr,
 						WCN36XX_HAL_LINK_IDLE_STATE);
+			wcn36xx_smd_switch_channel(wcn, vif, 0);
 		}
 	}