[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <70895dd0-cf65-b9bc-ac4f-4098be5a5f97@yandex-team.ru>
Date: Sat, 19 Aug 2017 15:42:03 +0300
From: Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
To: netdev@...r.kernel.org, "David S. Miller" <davem@...emloft.net>
Cc: Cong Wang <xiyou.wangcong@...il.com>,
Jiri Pirko <jiri@...nulli.us>, stable@...r.kernel.org
Subject: Re: [PATCH] net_sched: fix order of queue length updates in
qdisc_replace()
I've seen this as loops in hfsc rb-trees and hungs in hfsc_dequeue() -> rb_first().
Simple debug catched insering already inserted rb-tree nodes and similar mess.
Most likely this could explain warning in hfsc_dequeue()
https://bugzilla.kernel.org/show_bug.cgi?id=109581
On 19.08.2017 15:37, Konstantin Khlebnikov wrote:
> This important to call qdisc_tree_reduce_backlog() after changing queue
> length. Parent qdisc should deactivate class in ->qlen_notify() called from
> qdisc_tree_reduce_backlog() but this happens only if qdisc->q.qlen in zero.
>
> Missed class deactivations leads to crashes/warnings at picking packets
> from empty qdisc and corrupting state at reactivating this class in future.
>
> Signed-off-by: Konstantin Khlebnikov <khlebnikov@...dex-team.ru>
> Fixes: 86a7996cc8a0 ("net_sched: introduce qdisc_replace() helper")hfsc_dequeue
> Cc: Stable <stable@...r.kernel.org>
> ---
> include/net/sch_generic.h | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
> index 1c123e2b2415..67f815e5d525 100644
> --- a/include/net/sch_generic.h
> +++ b/include/net/sch_generic.h
> @@ -806,8 +806,11 @@ static inline struct Qdisc *qdisc_replace(struct Qdisc *sch, struct Qdisc *new,
> old = *pold;
> *pold = new;
> if (old != NULL) {
> - qdisc_tree_reduce_backlog(old, old->q.qlen, old->qstats.backlog);
> + unsigned int qlen = old->q.qlen;
> + unsigned int backlog = old->qstats.backlog;
> +
> qdisc_reset(old);
> + qdisc_tree_reduce_backlog(old, qlen, backlog);
> }
> sch_tree_unlock(sch);
>
>
Powered by blists - more mailing lists