On Mon, 2014-07-21 at 14:54 +0000, Sachin Kulkarni wrote: > From: Sachin Kulkarni<sachin.kulkarni@xxxxxxxxxx> > > During a hw scan ieee80211_iface_work ignores work items for all vifs. However > after the scan complete work is requeued only for STA, ADHOC and MESH iftypes. > This occasionally results in event processing getting delayed for iftype AP when > it coexists with a STA, and when the scan durations are very long. > > Signed-off-by: Sachin Kulkarni<sachin.kulkarni@xxxxxxxxxx> > Cc: linux-wireless@xxxxxxxxxxxxxxx > Cc: johannes@xxxxxxxxxxxxxxxx > --- > v3: Fix white space after comma > > We have a test scenario where there are 2 VIFs: one STA and one AP. The STA is > not connected to anyone while the AP has a station connected to it. The hardware > we are using currently has scan durations of ~8 sec and a new scan request is > generated every 10 s. This results in a window of only 2 s when work items can > be processed by ieee80211_iface_work. We came across scenarios where the AP is > trying to tear down a AMPDU session, and the IEEE80211_SDATA_QUEUE_AGG_STOP > event gets queued. However the work item for this gets ignored by the ieee80211_iface_work > since a hw scan is in progress most of the time. When the scan completes however > the work item for the AP does not get requeued, and by the time the > IEEE80211_SDATA_QUEUE_AGG_STOP gets processed in some other work items context, > the TX frames just keep getting queued up, since the tid has not been destroyed. > This results in disconnections since things like ARP timeouts etc start > happening. > --- > net/mac80211/mlme.c | 5 +++-- > 1 files changed, 3 insertions(+), 2 deletions(-) > > diff --git a/net/mac80211/mlme.c b/net/mac80211/mlme.c > index 931330b..49af420 100644 > --- a/net/mac80211/mlme.c > +++ b/net/mac80211/mlme.c > @@ -3647,9 +3647,10 @@ static void ieee80211_restart_sta_timer(struct ieee80211_sub_if_data *sdata) > if (!(flags & IEEE80211_HW_CONNECTION_MONITOR)) > ieee80211_queue_work(&sdata->local->hw, > &sdata->u.mgd.monitor_work); > - /* and do all the other regular work too */ > - ieee80211_queue_work(&sdata->local->hw, &sdata->work); > } > + > + /* and do all the other regular work too */ > + ieee80211_queue_work(&sdata->local->hw, &sdata->work); Ok actually, I'm not going to apply this. Sorry, I should have paid more attention before, but I don't think it's reasonable to do this in a function called ieee80211_restart_sta_timer() in mlme.c, that's clearly not intended to have side effects for other interface types. johannes -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
