Re: [RFC 09/14] mac80211: add power save support structure to mesh interface

[Marco Porsch <marco.porsch@etit.tu-chemnitz.de>]

On 11/17/2012 01:26 AM, Johannes Berg wrote:
> On Fri, 2012-11-16 at 22:48 -0800, Marco Porsch wrote:
>
>>   #ifdef CONFIG_MAC80211_MESH
>> diff --git a/net/mac80211/iface.c b/net/mac80211/iface.c
>> index bc3e3e1..8396602 100644
>> --- a/net/mac80211/iface.c
>> +++ b/net/mac80211/iface.c
>> @@ -769,6 +769,9 @@ static void ieee80211_do_stop(struct ieee80211_sub_if_data *sdata,
>>   		/* free all potentially still buffered bcast frames */
>>   		local->total_ps_buffered -= skb_queue_len(&sdata->u.ap.ps.bc_buf);
>>   		skb_queue_purge(&sdata->u.ap.ps.bc_buf);
>> +	} else if (sdata->vif.type == NL80211_IFTYPE_MESH_POINT) {
>> +		local->total_ps_buffered -= skb_queue_len(&sdata->u.mesh.ps.bc_buf);
>> +		skb_queue_purge(&sdata->u.mesh.ps.bc_buf);
>>   	} else if (sdata->vif.type == NL80211_IFTYPE_STATION) {
>>   		ieee80211_mgd_stop(sdata);
>>   	}
>
> This will not apply, and is in the wrong place anyway.

Moved it to mesh_stop. After disabling the beacon, these buffered 
broadcast frames are not going to be transmitted anywhere.

>
>> +	/* reset in case we have been in a mesh with PS peers before */
>
> should amend the comment to say "and leak all the frames that might be
> on the queue now.
>
>> +	} else if (sta->sdata->vif.type == NL80211_IFTYPE_MESH_POINT) {
>> +		ps = &sta->sdata->u.mesh.ps;
>> +		/* TIM map only for AID <= IEEE80211_MAX_AID */
>
> AID? Seems more like PLID to me:
>
>> +		id = le16_to_cpu(sta->plid) % IEEE80211_MAX_AID;
>
>> -	/* This is only necessary for stations on BSS interfaces */
>> -	if (!sta->sdata->bss)
>> +	/* This is only necessary for stations on BSS/MBSS interfaces */
>> +	if (!sta->sdata->bss &&
>> +	    !ieee80211_vif_is_mesh(&sta->sdata->vif))
>>   		return false;
>
> Instead of cluttering all the code paths like this, what if we changed
>
> struct ieee80211_if_ap *sdata->bss
>
> to
>
> struct ps_data *sdata->ps
>
>
> If we know the interface type is AP/AP_VLAN, we can still get the AP
> sdata from that pointer:
> 	container_of(sdata->ps, struct ieee80211_sub_if_sdata, u.ap.ps);
>
> This is only in a few places, and might keep all the RX/TX code simpler?

When counting, I found twice as many lines that would need the 
workaround, as the ones that would benefit. Also I don't really have a 
clue on the AP/VLAN code.

>
>> +++ b/net/mac80211/tx.c
>> @@ -329,6 +329,8 @@ static void purge_old_ps_buffers(struct ieee80211_local *local)
>>
>>   		if (sdata->vif.type == NL80211_IFTYPE_AP)
>>   			ps = &sdata->u.ap.ps;
>> +		else if (sdata->vif.type == NL80211_IFTYPE_MESH_POINT)
>> +			ps = &sdata->u.mesh.ps;
>
> This could just be "ps = sdata->ps" etc.

--Marco