From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp86.iad3a.emailsrvr.com (smtp86.iad3a.emailsrvr.com [173.203.187.86]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by lists.bufferbloat.net (Postfix) with ESMTPS id 47A273CB36 for ; Fri, 2 Feb 2018 11:55:17 -0500 (EST) Received: from smtp35.relay.iad3a.emailsrvr.com (localhost [127.0.0.1]) by smtp35.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id F23845C5F; Fri, 2 Feb 2018 11:55:16 -0500 (EST) X-SMTPDoctor-Processed: csmtpprox beta Received: from smtp35.relay.iad3a.emailsrvr.com (localhost [127.0.0.1]) by smtp35.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id EDFB15C2F; Fri, 2 Feb 2018 11:55:16 -0500 (EST) Received: from app62.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by smtp35.relay.iad3a.emailsrvr.com (SMTP Server) with ESMTP id CA4F55A6B; Fri, 2 Feb 2018 11:55:16 -0500 (EST) X-Sender-Id: dpreed@deepplum.com Received: from app62.wa-webapps.iad3a (relay-webapps.rsapps.net [172.27.255.140]) by 0.0.0.0:25 (trex/5.7.12); Fri, 02 Feb 2018 11:55:16 -0500 Received: from deepplum.com (localhost.localdomain [127.0.0.1]) by app62.wa-webapps.iad3a (Postfix) with ESMTP id B5F2C4130B; Fri, 2 Feb 2018 11:55:16 -0500 (EST) Received: by apps.rackspace.com (Authenticated sender: dpreed@deepplum.com, from: dpreed@deepplum.com) with HTTP; Fri, 2 Feb 2018 11:55:16 -0500 (EST) X-Auth-ID: dpreed@deepplum.com Date: Fri, 2 Feb 2018 11:55:16 -0500 (EST) From: "dpreed@deepplum.com" To: make-wifi-fast@lists.bufferbloat.net, linux-wireless@vger.kernel.org MIME-Version: 1.0 Content-Type: text/plain;charset=UTF-8 Content-Transfer-Encoding: quoted-printable Importance: Normal X-Priority: 3 (Normal) X-Type: plain In-Reply-To: <20180202151105.30043-1-toke@toke.dk> References: <20180202151105.30043-1-toke@toke.dk> Message-ID: <1517590516.742814797@apps.rackspace.com> X-Mailer: webmail/12.11.0-RC X-Mailman-Approved-At: Sat, 03 Feb 2018 04:53:53 -0500 Subject: Re: [Make-wifi-fast] [PATCH] mac80211: Adjust TSQ pacing shift X-BeenThere: make-wifi-fast@lists.bufferbloat.net X-Mailman-Version: 2.1.20 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 02 Feb 2018 16:55:17 -0000 I'm curious about the "WiFi Aware" initiative by the WiFi Alliance.=0A=0ADo= es LEDE and/or Linux support this protocol? I know gSupplicant is potential= ly the way such things are supposed to work, at least according to its supp= orters.=0A=0AThe general NAN (Neighborhood-Aware-Networking) concept makes = a lot of sense at one level, but as an Internet guy, it troubles me that th= ey decided to split from the Internet and go a balkanized direction. To me,= the neighborhood is interesting only as part of a larger Internet.=0A=0AIt= also troubles me that WiFi Aware is a "certification program" rather than = a real standard.=0A=0A-----Original Message-----=0AFrom: "Toke H=C3=B8iland= -J=C3=B8rgensen" =0ASent: Friday, February 2, 2018 10:11am=0A= To: make-wifi-fast@lists.bufferbloat.net, linux-wireless@vger.kernel.org=0A= Cc: "Toke H=C3=B8iland-J=C3=B8rgensen" =0ASubject: [Make-wifi= -fast] [PATCH] mac80211: Adjust TSQ pacing shift=0A=0ASince we now have the= convenient helper to do so, actually adjust the=0ATSQ pacing shift for pac= kets going out over a WiFi interface. This=0Asignificantly improves through= put for locally-originated TCP=0Aconnections. The default pacing shift of 1= 0 corresponds to ~1ms of=0Aqueued packet data. Adjusting this to a shift of= 8 (i.e. ~4ms) improves=0A1-hop throughput for ath9k by a factor of 3, wher= eas increasing it more=0Ahas diminishing returns.=0A=0AAchieved throughput = for different values of sk_pacing_shift (average of=0A5 iterations of 10-se= c netperf runs to a host on the other side of the=0AWiFi hop):=0A=0Ask_paci= ng_shift 10: 43.21 Mbps (pre-patch)=0Ask_pacing_shift 9: 78.17 Mbps=0Ask= _pacing_shift 8: 123.94 Mbps=0Ask_pacing_shift 7: 128.31 Mbps=0A=0ALatenc= y for competing flows increases from ~3 ms to ~10 ms with this=0Achange. Th= is is about the same magnitude of queueing latency induced by=0Aflows that = are not originated on the WiFi device itself (and so are not=0Alimited by T= SQ).=0A=0ASigned-off-by: Toke H=C3=B8iland-J=C3=B8rgensen =0A= ---=0A net/mac80211/tx.c | 8 ++++++++=0A 1 file changed, 8 insertions(+)=0A= =0Adiff --git a/net/mac80211/tx.c b/net/mac80211/tx.c=0Aindex 25904af38839.= .69722504e3e1 100644=0A--- a/net/mac80211/tx.c=0A+++ b/net/mac80211/tx.c=0A= @@ -3574,6 +3574,14 @@ void __ieee80211_subif_start_xmit(struct sk_buff *sk= b,=0A =09if (!IS_ERR_OR_NULL(sta)) {=0A =09=09struct ieee80211_fast_tx *fas= t_tx;=0A =0A+=09=09/* We need a bit of data queued to build aggregates prop= erly, so=0A+=09=09 * instruct the TCP stack to allow more than a single ms = of data=0A+=09=09 * to be queued in the stack. The value is a bit-shift of = 1=0A+=09=09 * second, so 8 is ~4ms of queued data. Only affects local TCP= =0A+=09=09 * sockets.=0A+=09=09 */=0A+=09=09sk_pacing_shift_update(skb->sk,= 8);=0A+=0A =09=09fast_tx =3D rcu_dereference(sta->fast_tx);=0A =0A =09=09i= f (fast_tx &&=0A-- =0A2.16.0=0A=0A_________________________________________= ______=0AMake-wifi-fast mailing list=0AMake-wifi-fast@lists.bufferbloat.net= =0Ahttps://lists.bufferbloat.net/listinfo/make-wifi-fast