From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: [bug?] r8169: hangs under heavy load Date: Sat, 26 Nov 2011 05:07:03 +0100 Message-ID: <1322280423.10212.3.camel@edumazet-laptop> References: <4E7F4D45.7040902@wolke7.net> <1316978632.4122.103.camel@deadeye> <4E82C0B5.1050902@wolke7.net> <1317254171.4068.12.camel@deadeye> <20111124044137.GA18262@elie.hsd1.il.comcast.net> <4ECFE7A7.5070300@wolke7.net> <20111125201936.GA26692@elie.hsd1.il.comcast.net> <1322253122.2550.3.camel@edumazet-laptop> <20111125222211.GA13719@electric-eye.fr.zoreil.com> <1322262357.2550.12.camel@edumazet-laptop> <20111126004447.GA14744@electric-eye.fr.zoreil.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Jonathan Nieder , netdev@vger.kernel.org, nic_swsd@realtek.com, linux-kernel@vger.kernel.org, Armin Kazmi , Gerd To: Francois Romieu Return-path: Received: from mail-ww0-f44.google.com ([74.125.82.44]:47832 "EHLO mail-ww0-f44.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750920Ab1KZEHH (ORCPT ); Fri, 25 Nov 2011 23:07:07 -0500 In-Reply-To: <20111126004447.GA14744@electric-eye.fr.zoreil.com> Sender: netdev-owner@vger.kernel.org List-ID: Le samedi 26 novembre 2011 =C3=A0 01:44 +0100, Francois Romieu a =C3=A9= crit : > Eric Dumazet : > > Le vendredi 25 novembre 2011 ?? 23:22 +0100, Francois Romieu a ??cr= it : > [...] > > > This path is not used. > >=20 > > OK, then we receive a RxFIFOOver indication while napi handler is > > running (quite possible if machine under network load) >=20 > Welcome to the post 92fc43b4159b518f5baae57301f26d770b0834c9 world. := o/ >=20 > The hack below could help (the XID matches RTL_GIGA_MAC_VER_22): >=20 > diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ether= net/realtek/r8169.c > index 6f06aa1..4fcfb3f 100644 > --- a/drivers/net/ethernet/realtek/r8169.c > +++ b/drivers/net/ethernet/realtek/r8169.c > @@ -5818,7 +5818,6 @@ static irqreturn_t rtl8169_interrupt(int irq, v= oid *dev_instance) > switch (tp->mac_version) { > /* Work around for rx fifo overflow */ > case RTL_GIGA_MAC_VER_11: > - case RTL_GIGA_MAC_VER_22: > case RTL_GIGA_MAC_VER_26: > netif_stop_queue(dev); > rtl8169_tx_timeout(dev); > @@ -5835,6 +5834,7 @@ static irqreturn_t rtl8169_interrupt(int irq, v= oid *dev_instance) > case RTL_GIGA_MAC_VER_31: > /* Experimental science. Pktgen proof. */ > case RTL_GIGA_MAC_VER_12: > + case RTL_GIGA_MAC_VER_22: > case RTL_GIGA_MAC_VER_25: > if (status =3D=3D RxFIFOOver) > goto done; >=20 >=20 > Traffic may stop flowing though. The patch below should be better: >=20 Yes, this is a workaround, and following _is_ the way to go... > diff --git a/drivers/net/ethernet/realtek/r8169.c b/drivers/net/ether= net/realtek/r8169.c > index 6f06aa1..54be9e1 100644 > --- a/drivers/net/ethernet/realtek/r8169.c > +++ b/drivers/net/ethernet/realtek/r8169.c > @@ -5820,8 +5820,7 @@ static irqreturn_t rtl8169_interrupt(int irq, v= oid *dev_instance) > case RTL_GIGA_MAC_VER_11: > case RTL_GIGA_MAC_VER_22: > case RTL_GIGA_MAC_VER_26: > - netif_stop_queue(dev); > - rtl8169_tx_timeout(dev); > + rtl8169_schedule_work(dev, rtl8169_reset_task); > goto done; > /* Testers needed. */ > case RTL_GIGA_MAC_VER_17: >=20 Please Jonathan, Gerd, try this fix !