Return-path: Received: from smtp.nokia.com ([192.100.122.230]:34765 "EHLO mgw-mx03.nokia.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750865AbYK2GyN (ORCPT ); Sat, 29 Nov 2008 01:54:13 -0500 To: "Johannes Berg" Cc: linux-wireless@vger.kernel.org Subject: Re: [RFC v3 2/4] mac80211: implement dynamic power save References: <1227901784-5325-1-git-send-email-kalle.valo@nokia.com> <1227901784-5325-3-git-send-email-kalle.valo@nokia.com> <1227910460.3543.3.camel@johannes.berg> From: Kalle Valo Date: Sat, 29 Nov 2008 08:53:34 +0200 In-Reply-To: <1227910460.3543.3.camel@johannes.berg> (ext Johannes Berg's message of "Fri\, 28 Nov 2008 23\:14\:20 +0100") Message-ID: <87vdu710r5.fsf@nokia.com> (sfid-20081129_075416_513020_6EF9A11E) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Sender: linux-wireless-owner@vger.kernel.org List-ID: Johannes Berg writes: > On Fri, 2008-11-28 at 21:49 +0200, Kalle Valo wrote: > >> + /* always wake queues to avoid having them stopped forever */ >> + netif_tx_wake_all_queues(local->mdev); > > I think we really need to think about keeping track of _who_ wanted the > queues stopped now... Makes sense. > We have a number of users of stopping queues now: > > * driver, obviously > * scanning code > * PS code > * ..? > > So when the driver wants a queue stopped, but the PS code wakes them up, > the driver will end up having to drop a packet, which isn't really good. > We really need to keep track, for each queue, who wants it > stopped/started and then only start the real queue if all the code > consents. Frankly I'm not that familiar with mac80211 queues, but it's a good time to learn now :) I'll look into this and will make a proposal. -- Kalle Valo