[<prev] [next>] [thread-next>] [day] [month] [year] [list]
Message-Id: <20230104174744.22280-1-rrameshbabu@nvidia.com>
Date: Wed, 4 Jan 2023 09:47:45 -0800
From: Rahul Rameshbabu <rrameshbabu@...dia.com>
To: netdev@...r.kernel.org
Cc: Tariq Toukan <tariqt@...dia.com>, Gal Pressman <gal@...dia.com>,
Saeed Mahameed <saeedm@...dia.com>,
Jamal Hadi Salim <jhs@...atatu.com>,
Cong Wang <xiyou.wangcong@...il.com>,
Jiri Pirko <jiri@...nulli.us>,
Rahul Rameshbabu <rrameshbabu@...dia.com>,
Eric Dumazet <edumazet@...gle.com>,
Maxim Mikityanskiy <maxtram95@...il.com>
Subject: [PATCH net] sch_htb: Fix prematurely activating netdev during htb_destroy_class_offload
When the netdev qdisc is updated correctly with the new qdisc before
destroying the old qdisc, the netdev should not be activated till cleanup
is completed. When htb_destroy_class_offload called htb_graft_helper, the
netdev may be activated before cleanup is completed. The new netdev qdisc
may be used prematurely by queues before cleanup is done. Call
dev_graft_qdisc in place of htb_graft_helper when destroying the htb to
prevent premature netdev activation.
Fixes: d03b195b5aa0 ("sch_htb: Hierarchical QoS hardware offload")
Signed-off-by: Rahul Rameshbabu <rrameshbabu@...dia.com>
Acked-by: Saeed Mahameed <saeedm@...dia.com>
Cc: Eric Dumazet <edumazet@...gle.com>
Cc: Maxim Mikityanskiy <maxtram95@...il.com>
---
net/sched/sch_htb.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c
index 2238edece1a4..f62334ef016a 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -1557,14 +1557,16 @@ static int htb_destroy_class_offload(struct Qdisc *sch, struct htb_class *cl,
WARN_ON(!q);
dev_queue = htb_offload_get_queue(cl);
- old = htb_graft_helper(dev_queue, NULL);
- if (destroying)
+ if (destroying) {
+ old = dev_graft_qdisc(dev_queue, NULL);
/* Before HTB is destroyed, the kernel grafts noop_qdisc to
* all queues.
*/
WARN_ON(!(old->flags & TCQ_F_BUILTIN));
- else
+ } else {
+ old = htb_graft_helper(dev_queue, NULL);
WARN_ON(old != q);
+ }
if (cl->parent) {
_bstats_update(&cl->parent->bstats_bias,
--
2.36.2
Powered by blists - more mailing lists