On Fri, May 8, 2009 at 6:18 PM, Marcel Holtmann <marcel@xxxxxxxxxxxx> wrote: > Hi Johannes, > >> >> > > Copying Wey-Yi, but I think she's aware of this? It's definitely a >> >> > > driver issue. I _suspect_, but am not sure, that the driver kills the >> >> > > aggregation session when we disassoc or something and we only tell it >> >> > > about the STA removal after that, when it already killed the STA. Thus >> >> > > it would be a consequence of iwlwifi not implementing a proper >> >> > > sta_notify() callback. >> >> > >> >> > I don't see it with nl80211 driver, but that might be pure luck. >> >> >> >> Hmm, ok, that points to something else maybe... Now I wish I had my >> >> ftrace-mac80211 patch ready. Well, the patch is almost ready but we >> >> can't use ftrace in modules yet... >> >> >> >> Maybe there's some timing issue. I still think it's iwlwifi getting >> >> confused about what mac80211 is telling it as described, but it seems >> >> it's a little more subtle than what I said. >> > >> > something similar now happened during normal operation: >> >> Just curious, have you applied johill's latest idle fix patch? I've >> been running iperf overnight with it and my woes are gone. I tested RX >> though, I'm leaving TX running now but haven't yet seen any warn. > > this is still running wireless-testing.git from yesterday with the debug > addBA patch from Johannes. Ok yeah wl from yesterday didn't have johill's patch applied yet. >From the looks of it the intel hardware doesn't want to stop the aggregation session which is API-breakage -- which is why you get the WARN_ON. This is happening when you are disassociating. I have a feeling johill's other patch will fix your issue but regardless I believe this just highlights more we should flush our queues (both RX and TX) before scanning. Luis -- 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
