[PATCH 1/2] supplicant: Use high-priority queue for management packets.
Mon Mar 17 03:34:20 PDT 2014
On 26 February 2014 07:45, Janusz Dziedzic <janusz.dziedzic at tieto.com> wrote:
> On 25 February 2014 19:39, Ben Greear <greearb at candelatech.com> wrote:
>> On 02/25/2014 09:16 AM, Jouni Malinen wrote:
>>> On Fri, Jan 31, 2014 at 02:55:17PM +0100, Johannes Berg wrote:
>>>> The obvious other alternative would be to just do this in the kernel,
>>>> assigning high priority to any packets that have
>>>> tx->sdata->control_port_protocol == tx->skb->protocol (speaking in
>>>> mac80211 language) - a number of drivers already do some special things
>>>> via IEEE80211_TX_CTRL_PORT_CTRL_PROTO.
>>>> I'm not entirely sure that's a better solution overall, but in terms of
>>>> driver compatibility it would certainly be less risky.
>>> With commit 1bf4bbb4024dcdab5e57634dd8ae1072d42a53ac ('mac80211: send
>>> control port protocol frames to the VO queue') now in
>>> wireless-testing.git and tagged for stable, I'm planning on dropping
>>> these wpa_supplicant/hostapd changes that use setsockopt() with the
>>> special SO_PRIORITY values.
> Regarding this mac80211 patch I am just curious what in such case
> AP (UAPSD) <--> STA (AC: VO - default we have in mac80211)
> STA go to PS
> AP send EAP (VO) packet - we will not set PVB
> STA will not wake-up ...
> I am not sure about real word here - if vendors/mac80211 allow STA go
> to PS during EAP exchange ...
Did we introduce a possible bug in case of AP UAPSD?
More information about the Hostap