]> git.hungrycats.org Git - linux/commitdiff
net_sched: make qdisc_tree_decrease_qlen() work for non mq
authorEric Dumazet <edumazet@google.com>
Tue, 15 Dec 2015 17:43:12 +0000 (09:43 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 23 Jan 2016 04:54:15 +0000 (20:54 -0800)
[ Upstream commit 225734de70cd0a9e0b978f3583a4a87939271d5e ]

Stas Nichiporovich reported a regression in his HFSC qdisc setup
on a non multi queue device.

It turns out I mistakenly added a TCQ_F_NOPARENT flag on all qdisc
allocated in qdisc_create() for non multi queue devices, which was
rather buggy. I was clearly mislead by the TCQ_F_ONETXQUEUE that is
also set here for no good reason, since it only matters for the root
qdisc.

Fixes: 4eaf3b84f288 ("net_sched: fix qdisc_tree_decrease_qlen() races")
Reported-by: Stas Nichiporovich <stasn77@gmail.com>
Tested-by: Stas Nichiporovich <stasn77@gmail.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
net/sched/sch_api.c

index d4b6f3682c14c7826e502770cf462d642594a59c..68c599a5e1d1d76ec1f9b06a57b71f23444849b0 100644 (file)
@@ -950,7 +950,7 @@ qdisc_create(struct net_device *dev, struct netdev_queue *dev_queue,
                }
                lockdep_set_class(qdisc_lock(sch), &qdisc_tx_lock);
                if (!netif_is_multiqueue(dev))
-                       sch->flags |= TCQ_F_ONETXQUEUE | TCQ_F_NOPARENT;
+                       sch->flags |= TCQ_F_ONETXQUEUE;
        }
 
        sch->handle = handle;