From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Dumazet Subject: Re: BUG: unable to handle kernel paging request at 000041ed00000001 Date: Mon, 14 Jun 2010 10:31:35 +0200 Message-ID: <1276504295.2478.35.camel@edumazet-laptop> References: <9D7251E7-0EFD-4645-BC30-A96191D1046E@res.lt> <1276185609.2448.12.camel@edumazet-laptop> <80C864F3-B075-4E3A-B72E-6FCD945A8058@res.lt> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: QUOTED-PRINTABLE Cc: netdev@vger.kernel.org To: Arturas Return-path: Received: from mail-wy0-f174.google.com ([74.125.82.174]:59248 "EHLO mail-wy0-f174.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752988Ab0FNIbk (ORCPT ); Mon, 14 Jun 2010 04:31:40 -0400 Received: by wyb40 with SMTP id 40so3550775wyb.19 for ; Mon, 14 Jun 2010 01:31:38 -0700 (PDT) In-Reply-To: <80C864F3-B075-4E3A-B72E-6FCD945A8058@res.lt> Sender: netdev-owner@vger.kernel.org List-ID: Le lundi 14 juin 2010 =C3=A0 10:05 +0300, Arturas a =C3=A9crit : > Hi, >=20 > your patch fixes hangs and i get a warning (see bellow if needed) whe= n it triggers.=20 > As I understand it is workaround and a real fix should be different?=20 > What about bonding to be multiqueue aware? >=20 But your problem is about bridge, not bonding (see trace). And 2.6.34 wont accept such changes, its already released. > I also have another issue with NMI. On older machine with 5500 xeons = i=20 > have almost no overhead with nmi_watchdog enabled, but on this it is = about twice. > without nmi enabled cpu peak average is 30%, and with nmi enabled i h= ave 53%. > When traffic is not passing all cpus are idling at 100%. > Maybe overhead could be a little bit smaller? :-) >=20 I am a bit lost here, NMI have litle to do with network stack ;) > [ 8064.562106] WARNING: at net/core/dev.c:1964 dev_queue_xmit+0x504/0= x520() > [ 8064.562108] Hardware name: S5520UR > [ 8064.562108] br0 > [ 8064.562109] Modules linked in: ipt_REDIRECT xt_tcpudp ipt_set ipta= ble_filter iptable_nat nf_nat ipt_addrtype xt_dscp xt_string xt_owner x= t_multiport xt_mark xt_iprange xt_hashlimit xt_conntrack xt_connmark xt= _DSCP xt_NFQUEUE xt_MARK xt_CONNMARK ip_tables x_tables ip_set_ipmap ip= _set cls_u32 sch_htb ipmi_watchdog ipmi_devintf ipmi_si ipmi_msghandler= nf_conntrack_ipv4 nf_defrag_ipv4 nf_conntrack bonding ipv6 ixgbe igb m= dio > [ 8064.562125] Pid: 8643, comm: lighttpd Not tainted 2.6.34-gentoo #6 > [ 8064.562126] Call Trace: > [ 8064.562133] [] ? warn_slowpath_common+0x73/0xb0 > [ 8064.562135] [] ? warn_slowpath_fmt+0x40/0x50 > [ 8064.562137] [] ? dev_queue_xmit+0x504/0x520 > [ 8064.562141] [] ? ip_queue_xmit+0x182/0x3e0 > [ 8064.562145] [] ? tcp_init_tso_segs+0x2e/0x50 > [ 8064.562147] [] ? tcp_write_xmit+0x75/0xa00 > [ 8064.562151] [] ? lock_timer_base+0x33/0x70 > [ 8064.562153] [] ? tcp_transmit_skb+0x3ac/0x820 > [ 8064.562155] [] ? tcp_sendmsg+0x866/0xbf0 > [ 8064.562156] [] ? tcp_write_xmit+0x1ec/0xa00 > [ 8064.562161] [] ? lock_sock_nested+0x3d/0xe0 > [ 8064.562163] [] ? sock_aio_write+0x0/0x150 > [ 8064.562166] [] ? __tcp_push_pending_frames+0x19= /0x80 > [ 8064.562167] [] ? do_tcp_setsockopt+0x53a/0x690 > [ 8064.562171] [] ? do_sync_readv_writev+0xa9/0xf0 > [ 8064.562173] [] ? lock_timer_base+0x33/0x70 > [ 8064.562174] [] ? do_sync_read+0xbf/0x100 > [ 8064.562176] [] ? do_readv_writev+0x172/0x220 > [ 8064.562179] [] ? d_kill+0x5f/0x80 > [ 8064.562181] [] ? dput+0xb8/0x180 > [ 8064.562183] [] ? sockfd_lookup_light+0x22/0x80 > [ 8064.562185] [] ? sys_setsockopt+0x6d/0xd0 > [ 8064.562188] [] ? system_call_fastpath+0x16/0x1b >=20 Could you please test another patch ? Before calling sk_tx_queue_set(sk, queue_index); we should check if dst dev is current device. --- net/core/dev.c.orig 2010-06-10 17:52:17.000000000 +0200 +++ net/core/dev.c 2010-06-14 10:25:25.000000000 +0200 @@ -1958,12 +1958,10 @@ static inline u16 dev_cap_txqueue(struct net_device *dev, u16 queue_in= dex) { if (unlikely(queue_index >=3D dev->real_num_tx_queues)) { - if (net_ratelimit()) { - WARN(1, "%s selects TX queue %d, but " - "real number of TX queues is %d\n", - dev->name, queue_index, - dev->real_num_tx_queues); - } + WARN_ONCE("%s selects TX queue %d, but " + "real number of TX queues is %d\n", + dev->name, queue_index, + dev->real_num_tx_queues); return 0; } return queue_index; @@ -1977,6 +1975,7 @@ =20 if (sk_tx_queue_recorded(sk)) { queue_index =3D sk_tx_queue_get(sk); + queue_index =3D dev_cap_txqueue(dev, queue_index); } else { const struct net_device_ops *ops =3D dev->netdev_ops; =20 @@ -1991,7 +1990,7 @@ if (sk) { struct dst_entry *dst =3D rcu_dereference_bh(sk->sk_dst_cache); =20 - if (dst && skb_dst(skb) =3D=3D dst) + if (dst && skb_dst(skb) =3D=3D dst && dst->dev =3D=3D dev) sk_tx_queue_set(sk, queue_index); } }