Return-path: Received: from mga01.intel.com ([192.55.52.88]:59443 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750920AbXLSQ3z convert rfc822-to-8bit (ORCPT ); Wed, 19 Dec 2007 11:29:55 -0500 MIME-Version: 1.0 Subject: RE: [PATCH 5/8] mac80211: A-MPDU Rx handling aggregation reordering Date: Wed, 19 Dec 2007 18:29:37 +0200 Message-ID: (sfid-20071219_163000_905159_C2A2EC80) In-Reply-To: <1198079220.4747.10.camel@johannes.berg> References: <11979070692599-git-send-email-ron.rindjunsky@intel.com> <11979070782762-git-send-email-ron.rindjunsky@intel.com> <1197915515.4885.86.camel@johannes.berg> (sfid-20071218_134457_890406_A2932F58) <1197986248.4885.149.camel@johannes.berg> <1879838866982C46A9CB3D56BA49ADEB0431391B@hasmsx411.ger.corp.intel.com> <1198079220.4747.10.camel@johannes.berg> From: "Winkler, Tomas" To: "Johannes Berg" , "Rindjunsky, Ron" Cc: , , , "Zhu, Yi" Content-Type: text/plain; charset="us-ascii" Sender: linux-wireless-owner@vger.kernel.org List-ID: >-----Original Message----- >From: Johannes Berg [mailto:johannes@sipsolutions.net] >Sent: Wednesday, December 19, 2007 5:47 PM >To: Rindjunsky, Ron >Cc: linville@tuxdriver.com; linux-wireless@vger.kernel.org; >flamingice@sourmilk.net; Winkler, Tomas; Zhu, Yi >Subject: RE: [PATCH 5/8] mac80211: A-MPDU Rx handling aggregation >reordering > > >> > I'm just not sure why you'd want to. As far as I can the frames >> already >> > passed __ieee80211_rx(), no? Maybe only as part of the aggregation? >> >> No, not passed it yet, they are still queued for reordering, but as >> aggregation is being tear down I am expected to drop the frames >> that are still not in order. > >But haven't they passed through __ieee80211_rx() as the aggregate frame? AMsDU aggregation is different from AMpDU aggregation. For AMPDU the disaggregating is happening let say in PHY level. First one is many packets in packed into one. Second one is packets in a row without ACK between them. (You answer with Block Ack to them). They appear as regular frames to mac80211 but they are not in order. Only ordered sequence might be passed to the network stack. So we might drop the frames are result of non order. So what Ron was concerning about is whether we it's not waste of time to pass packets through rx handler before we know the sequence is completed. Tomas >johannes --------------------------------------------------------------------- Intel Israel (74) Limited This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.