From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cong Wang Subject: Re: [Bonding-devel] [v3 Patch 2/3] bridge: make bridge support netpoll Date: Wed, 14 Apr 2010 16:16:09 +0800 Message-ID: <4BC579C9.8060006@redhat.com> References: <20100408062234.4499.17042.sendpatchset@localhost.localdomain> <20100408062246.4499.5670.sendpatchset@localhost.localdomain> <20100408083710.2b61ee44@nehalam> <4BC2F7E2.7020309@redhat.com> <1271068737.16881.18.camel@edumazet-laptop> <20100412083842.26d71bda@nehalam> <4BC43214.6030009@redhat.com> <8304.1271177567@death.nxdomain.ibm.com> <20100413103320.11a2a4f7@nehalam> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: Jay Vosburgh , Eric Dumazet , Neil Horman , netdev@vger.kernel.org, Andy Gospodarek , bridge@lists.linux-foundation.org, linux-kernel@vger.kernel.org, bonding-devel@lists.sourceforge.net, Jeff Moyer , Matt Mackall , David Miller To: Stephen Hemminger Return-path: In-Reply-To: <20100413103320.11a2a4f7@nehalam> Sender: linux-kernel-owner@vger.kernel.org List-Id: netdev.vger.kernel.org Stephen Hemminger wrote: > On Tue, 13 Apr 2010 09:52:47 -0700 > Jay Vosburgh wrote: >=20 >> Cong Wang wrote: >> >>> Stephen Hemminger wrote: >>>> On Mon, 12 Apr 2010 12:38:57 +0200 >>>> Eric Dumazet wrote: >>>> >>>>> Le lundi 12 avril 2010 =C3=A0 18:37 +0800, Cong Wang a =C3=A9crit= : >>>>>> Stephen Hemminger wrote: >>>>>>> There is no protection on dev->priv_flags for SMP access. >>>>>>> It would better bit value in dev->state if you are using it as = control flag. >>>>>>> >>>>>>> Then you could use=20 >>>>>>> if (unlikely(test_and_clear_bit(__IN_NETPOLL, &skb->dev->sta= te))) >>>>>>> netpoll_send_skb(...) >>>>>>> >>>>>>> >>>>>> Hmm, I think we can't use ->state here, it is not for this kind = of purpose, >>>>>> according to its comments. >>>>>> >>>>>> Also, I find other usages of IFF_XXX flags of ->priv_flags are a= lso using >>>>>> &, | to set or clear the flags. So there must be some other thin= gs preventing >>>>>> the race... >>>>> Yes, its RTNL that protects priv_flags changes, hopefully... >>>>> >>>> The patch was not protecting priv_flags with RTNL. >>>> For example.. >>>> >>>> >>>> @@ -308,7 +312,9 @@ static void netpoll_send_skb(struct netp >>>> tries > 0; --tries) { >>>> if (__netif_tx_trylock(txq)) { >>>> if (!netif_tx_queue_stopped(txq)) { >>>> + dev->priv_flags |=3D IFF_IN_NETPOLL; >>>> status =3D ops->ndo_start_xmit(skb, dev); >>>> + dev->priv_flags &=3D ~IFF_IN_NETPOLL; >>>> if (status =3D=3D NETDEV_TX_OK) >>>> txq_trans_update(txq); >>> Hmm, but I checked the bonding case (IFF_BONDING), it doesn't >>> hold rtnl_lock. Strange. >> I looked, and there are a couple of cases in bonding that don't >> have RTNL for adjusting priv_flags (in bond_ab_arp_probe when no sla= ves >> are up, and a couple of cases in 802.3ad). I think the solution the= re >> is to move bonding away from priv_flags for some of this (e.g., conv= ert >> bonding to use a frame hook like bridge and macvlan, and greatly >> simplify skb_bond_should_drop), but that's a separate topic. >> >> The majority of the cases, however, do hold RTNL. Bonding >> generally doesn't have to acquire RTNL itself, since whatever called >> into bonding is holding it already. For example, the slave add and >> remove paths (bond_enslave, bond_release) are called either via sysf= s or >> ioctl, both of which acquire RTNL. All of the set and clear operati= ons >> for IFF_BONDING fall into this category; look at bonding_store_slave= s >> for an example. >> >> Bonding does acquire RTNL itself when performing failovers, >> e.g., bond_mii_monitor holds RTNL prior to calling bond_miimon_commi= t, >> which will change priv_flags. >> >=20 > All this was related to netpoll. And netpoll processing often needs t= o occur > in hard IRQ context. Therefor netpoll stuff and RTNL (which is a mute= x), > really don't mix well. Keep RTNL for what it was meant for network > reconfiguration. Don't turn it into a network special BKL. >=20 Hmm, I think for my patch, holding RTNL lock is not necessary, because there're no other call pathes to change IFF_IN_NETPOLL bit, which is unlike bonding or bridge cases where sysfs/ioctl is provided to change it. The only chance to change IFF_IN_NETPOLL is in netpoll_send_skb() which can't be called simultaneously because there are other locks protecting it. Or am I still missing something? Thanks.