| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Thu, 16 Dec 2010 11:18:35 +0100
From: Eric Dumazet <eric.dumazet@...il.com>
To: Jarek Poplawski <jarkao2@...il.com>
Cc: David Miller <davem@...emloft.net>,
netdev <netdev@...r.kernel.org>,
Patrick McHardy <kaber@...sh.net>
Subject: [PATCH v2 net-next-2.6] net_sched: sch_sfq: add backlog info in
sfq_dump_class_stats()
Le jeudi 16 décembre 2010 à 08:16 +0000, Jarek Poplawski a écrit :
> I don't think you can walk this list without the qdisc lock.
>
I assumed it was already the case, but I did not check
Me confused...
If I use following patch, I get a recursive lock :
net/sched/sch_sfq.c | 17 ++++++++++++++---
1 file changed, 14 insertions(+), 3 deletions(-)
diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index 3cf478d..a2cde03 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -547,9 +547,20 @@ static int sfq_dump_class_stats(struct Qdisc *sch, unsigned long cl,
struct gnet_dump *d)
{
struct sfq_sched_data *q = qdisc_priv(sch);
- sfq_index idx = q->ht[cl-1];
- struct gnet_stats_queue qs = { .qlen = q->qs[idx].qlen };
- struct tc_sfq_xstats xstats = { .allot = q->allot[idx] };
+ sfq_index idx;
+ struct gnet_stats_queue qs = { 0 };
+ struct tc_sfq_xstats xstats = { 0 };
+ struct sk_buff_head *list;
+ struct sk_buff *skb;
+
+ spin_lock_bh(qdisc_root_sleeping_lock(sch));
+ idx = q->ht[cl - 1];
+ list = &q->qs[idx];
+ xstats.allot = q->allot[idx];
+ qs.qlen = list->qlen;
+ skb_queue_walk(list, skb)
+ qs.backlog += qdisc_pkt_len(skb);
+ spin_unlock_bh(qdisc_root_sleeping_lock(sch));
if (gnet_stats_copy_queue(d, &qs) < 0)
return -1;
Dec 16 10:49:34 edumdev kernel: [ 616.452080] sch->qstats.backlog=185420
Dec 16 10:49:34 edumdev kernel: [ 616.452146]
Dec 16 10:49:34 edumdev kernel: [ 616.452147] =============================================
Dec 16 10:49:34 edumdev kernel: [ 616.452265] [ INFO: possible recursive locking detected ]
Dec 16 10:49:34 edumdev kernel: [ 616.452329] 2.6.37-rc1-01820-g4be8976-dirty #456
Dec 16 10:49:34 edumdev kernel: [ 616.452425] ---------------------------------------------
Dec 16 10:49:34 edumdev kernel: [ 616.452489] tc/8747 is trying to acquire lock:
Dec 16 10:49:34 edumdev kernel: [ 616.452550] (&qdisc_tx_lock){+.-...}, at: [<ffffffffa01331d5>] sfq_dump_class_stats+0x65/0x160 [sch_sfq]
Dec 16 10:49:34 edumdev kernel: [ 616.452753]
Dec 16 10:49:34 edumdev kernel: [ 616.452754] but task is already holding lock:
Dec 16 10:49:34 edumdev kernel: [ 616.452867] (&qdisc_tx_lock){+.-...}, at: [<ffffffff8145474a>] gnet_stats_start_copy_compat+0x4a/0xc0
Dec 16 10:49:34 edumdev kernel: [ 616.453068]
Dec 16 10:49:34 edumdev kernel: [ 616.453069] other info that might help us debug this:
Dec 16 10:49:34 edumdev kernel: [ 616.453184] 2 locks held by tc/8747:
Dec 16 10:49:34 edumdev kernel: [ 616.453243] #0: (rtnl_mutex){+.+.+.}, at: [<ffffffff8149dbc2>] netlink_dump+0x52/0x1e0
Dec 16 10:49:34 edumdev kernel: [ 616.453510] #1: (&qdisc_tx_lock){+.-...}, at: [<ffffffff8145474a>] gnet_stats_start_copy_compat+0x4a/0xc0
Dec 16 10:49:34 edumdev kernel: [ 616.453745]
Dec 16 10:49:35 edumdev kernel: [ 616.453746] stack backtrace:
Dec 16 10:49:35 edumdev kernel: [ 616.453857] Pid: 8747, comm: tc Tainted: G W 2.6.37-rc1-01820-g4be8976-dirty #456
Dec 16 10:49:35 edumdev kernel: [ 616.453943] Call Trace:
Dec 16 10:49:35 edumdev kernel: [ 616.454004] [<ffffffff8107bd1e>] validate_chain+0x10be/0x1330
Dec 16 10:49:35 edumdev kernel: [ 616.454072] [<ffffffff8107b144>] ? validate_chain+0x4e4/0x1330
Dec 16 10:49:35 edumdev kernel: [ 616.454142] [<ffffffff810cd6cc>] ? get_page_from_freelist+0x2bc/0x730
Dec 16 10:49:35 edumdev kernel: [ 616.454211] [<ffffffff81079200>] ? trace_hardirqs_on_caller+0x110/0x190
Dec 16 10:49:35 edumdev kernel: [ 616.454281] [<ffffffff8107c3e9>] __lock_acquire+0x459/0xbe0
Dec 16 10:49:35 edumdev kernel: [ 616.454379] [<ffffffff8107cc10>] lock_acquire+0xa0/0x140
Dec 16 10:49:35 edumdev kernel: [ 616.454446] [<ffffffffa01331d5>] ? sfq_dump_class_stats+0x65/0x160 [sch_sfq]
Dec 16 10:49:35 edumdev kernel: [ 616.454520] [<ffffffff815aee46>] _raw_spin_lock_bh+0x36/0x50
Dec 16 10:49:35 edumdev kernel: [ 616.454586] [<ffffffffa01331d5>] ? sfq_dump_class_stats+0x65/0x160 [sch_sfq]
Dec 16 10:49:35 edumdev kernel: [ 616.454657] [<ffffffffa01331d5>] sfq_dump_class_stats+0x65/0x160 [sch_sfq]
Dec 16 10:49:35 edumdev kernel: [ 616.454727] [<ffffffff81202ef4>] ? nla_put+0x34/0x40
Dec 16 10:49:35 edumdev kernel: [ 616.454794] [<ffffffff8145478f>] ? gnet_stats_start_copy_compat+0x8f/0xc0
Dec 16 10:49:35 edumdev kernel: [ 616.454865] [<ffffffff8147a2f1>] tc_fill_tclass+0x1b1/0x250
Dec 16 10:49:35 edumdev kernel: [ 616.454932] [<ffffffff8147a3ce>] qdisc_class_dump+0x3e/0x40
Dec 16 10:49:35 edumdev kernel: [ 616.454999] [<ffffffff81483a68>] ? cbq_walk+0x78/0xc0
Dec 16 10:49:35 edumdev kernel: [ 616.455064] [<ffffffffa013228c>] sfq_walk+0x5c/0x90 [sch_sfq]
Dec 16 10:49:35 edumdev kernel: [ 616.455131] [<ffffffff81479f3a>] tc_dump_tclass_qdisc+0xba/0x110
Dec 16 10:49:35 edumdev kernel: [ 616.455199] [<ffffffff8147a390>] ? qdisc_class_dump+0x0/0x40
Dec 16 10:49:35 edumdev kernel: [ 616.455266] [<ffffffff8147a00f>] tc_dump_tclass_root+0x7f/0xa0
Dec 16 10:49:35 edumdev kernel: [ 616.455332] [<ffffffff8147a0bc>] tc_dump_tclass+0x8c/0x110
Dec 16 10:49:35 edumdev kernel: [ 616.455426] [<ffffffff8149dbdd>] netlink_dump+0x6d/0x1e0
Dec 16 10:49:35 edumdev kernel: [ 616.455494] [<ffffffff814a0c0c>] netlink_dump_start+0x19c/0x210
Dec 16 10:49:35 edumdev kernel: [ 616.455562] [<ffffffff8147a030>] ? tc_dump_tclass+0x0/0x110
Dec 16 10:49:35 edumdev kernel: [ 616.455628] [<ffffffff8147a030>] ? tc_dump_tclass+0x0/0x110
Dec 16 10:49:35 edumdev kernel: [ 616.455694] [<ffffffff8146bfa9>] rtnetlink_rcv_msg+0xb9/0x260
Dec 16 10:49:35 edumdev kernel: [ 616.455763] [<ffffffff8146bef0>] ? rtnetlink_rcv_msg+0x0/0x260
Dec 16 10:49:35 edumdev kernel: [ 616.455832] [<ffffffff8149ef29>] netlink_rcv_skb+0x99/0xc0
Dec 16 10:49:35 edumdev kernel: [ 616.455898] [<ffffffff8146bed5>] rtnetlink_rcv+0x25/0x40
Dec 16 10:49:35 edumdev kernel: [ 616.455963] [<ffffffff8149ea95>] ? netlink_unicast+0xf5/0x2d0
Dec 16 10:49:35 edumdev kernel: [ 616.456030] [<ffffffff8149ec42>] netlink_unicast+0x2a2/0x2d0
Dec 16 10:49:35 edumdev kernel: [ 616.456098] [<ffffffff810e8cb3>] ? might_fault+0x53/0xb0
Dec 16 10:49:35 edumdev kernel: [ 616.456163] [<ffffffff81451fed>] ? memcpy_fromiovec+0x6d/0x90
Dec 16 10:49:35 edumdev kernel: [ 616.456231] [<ffffffff8149fbdd>] netlink_sendmsg+0x24d/0x390
Dec 16 10:49:35 edumdev kernel: [ 616.456299] [<ffffffff814463d0>] sock_sendmsg+0xc0/0xf0
Dec 16 10:49:36 edumdev kernel: [ 616.456390] [<ffffffff810e8cb3>] ? might_fault+0x53/0xb0
Dec 16 10:49:36 edumdev kernel: [ 616.456457] [<ffffffff814767ce>] ? verify_compat_iovec+0x6e/0x110
Dec 16 10:49:36 edumdev kernel: [ 616.456526] [<ffffffff81447164>] sys_sendmsg+0x194/0x320
Dec 16 10:49:36 edumdev kernel: [ 616.456593] [<ffffffff815b2e02>] ? do_page_fault+0x102/0x4e0
Dec 16 10:49:36 edumdev kernel: [ 616.456661] [<ffffffff8107cd4d>] ? lock_release_non_nested+0x9d/0x2e0
Dec 16 10:49:36 edumdev kernel: [ 616.456729] [<ffffffff810e8cb3>] ? might_fault+0x53/0xb0
Dec 16 10:49:36 edumdev kernel: [ 616.456796] [<ffffffff810e8cb3>] ? might_fault+0x53/0xb0
Dec 16 10:49:36 edumdev kernel: [ 616.456863] [<ffffffff81476154>] compat_sys_sendmsg+0x14/0x20
Dec 16 10:49:36 edumdev kernel: [ 616.456929] [<ffffffff814770fe>] compat_sys_socketcall+0x1be/0x210
Dec 16 10:49:36 edumdev kernel: [ 616.457000] [<ffffffff8102f1d0>] sysenter_dispatch+0x7/0x33
Dec 16 10:49:36 edumdev kernel: [ 616.457067] [<ffffffff815ae9a9>] ? trace_hardirqs_on_thunk+0x3a/0x3f
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Powered by blists - more mailing lists