[RFC 1/2] Revert "wcn36xx: Pad TIM PVM if needed"

Eugene Krasnikov k.eugene.e at gmail.com
Mon Mar 10 04:57:49 EDT 2014


Pontus added beacon_length6 because firmware needs an offset. Of
course this can be the reason why workaround for mesh network does not
work any more.

Pontus, did you test mesh network when you added this patch?:)

2014-03-07 5:02 GMT+00:00 Yeoh Chun-Yeow <yeohchunyeow at gmail.com>:
>> This would need a better commit log, but let me at least add some
>> context for others' benefit: I found the patch in question broke
>> mesh beaconing, i.e.  no beacons at all are sent after the change.
>>
>> But, mesh is doing some weird things with the TIM offset to
>> avoid having the firmware update it, so it could be mesh-specific
>> breakage and perhaps a workaround for mesh vif is needed.
>
> Yes, we just simply ignore the TIM.
>
> I take a look on the changes on wcn36xx_smd_send_beacon and found out
> that there is also a new member (beacon_length6) added in the
> wcn36xx_hal_send_beacon_req_msg.
> wcn36xx relies on ieee80211_beacon_get_tim. For mesh, the tim_off is
> always 0 and msg_body.beacon_length may be wrong with pad and pvm_len
> wrongly calculated.
>
> Perhaps, we can just set pad = 0 if it is mesh interface for the work
> around such as follow:
> https://github.com/chunyeow/wcn36xx/commit/54dd5deae8a85c02473d21d6c722bb8a4a0e7139
>
> ---
> Chun-Yeow
>
> _______________________________________________
> wcn36xx mailing list
> wcn36xx at lists.infradead.org
> http://lists.infradead.org/mailman/listinfo/wcn36xx



-- 
Best regards,
Eugene



More information about the wcn36xx mailing list