From a9312ae89324438b0edc554eb36c3ec6bf927d04 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Sun, 17 Aug 2008 21:51:03 -0700 Subject: pkt_sched: Add 'deactivated' state. This new state lets dev_deactivate() mark a qdisc as having been deactivated. dev_queue_xmit() and ing_filter() check for this bit and do not try to process the qdisc if the bit is set. dev_deactivate() polls the qdisc after setting the bit, waiting for both __QDISC_STATE_RUNNING and __QDISC_STATE_SCHED to clear. This isn't perfect yet, but subsequent changesets will make it so. This part is just one piece of the puzzle. Signed-off-by: David S. Miller --- include/net/sch_generic.h | 1 + 1 file changed, 1 insertion(+) (limited to 'include/net/sch_generic.h') diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index a7abfda3e447..757ab087adbf 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -27,6 +27,7 @@ enum qdisc_state_t { __QDISC_STATE_RUNNING, __QDISC_STATE_SCHED, + __QDISC_STATE_DEACTIVATED, }; struct qdisc_size_table { -- cgit v1.2.3 From 1e0d5a5747772182d1bb2525d8153da640fdfb58 Mon Sep 17 00:00:00 2001 From: "David S. Miller" Date: Sun, 17 Aug 2008 22:31:26 -0700 Subject: pkt_sched: No longer destroy qdiscs from RCU. We can now kill them synchronously with all of the previous dev_deactivate() cures. This makes netdev destruction and shutdown saner as the qdiscs hold references to the device. Signed-off-by: David S. Miller --- include/net/sch_generic.h | 1 - net/sched/sch_generic.c | 27 +++++++++------------------ 2 files changed, 9 insertions(+), 19 deletions(-) (limited to 'include/net/sch_generic.h') diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index 757ab087adbf..84d25f2e6188 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -61,7 +61,6 @@ struct Qdisc struct gnet_stats_basic bstats; struct gnet_stats_queue qstats; struct gnet_stats_rate_est rate_est; - struct rcu_head q_rcu; int (*reshape_fail)(struct sk_buff *skb, struct Qdisc *q); diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c index 30b76aec723b..6f96b7bc0809 100644 --- a/net/sched/sch_generic.c +++ b/net/sched/sch_generic.c @@ -518,14 +518,19 @@ void qdisc_reset(struct Qdisc *qdisc) } EXPORT_SYMBOL(qdisc_reset); -/* this is the rcu callback function to clean up a qdisc when there - * are no further references to it */ +/* Under qdisc_lock(qdisc) and BH! */ -static void __qdisc_destroy(struct rcu_head *head) +void qdisc_destroy(struct Qdisc *qdisc) { - struct Qdisc *qdisc = container_of(head, struct Qdisc, q_rcu); const struct Qdisc_ops *ops = qdisc->ops; + if (qdisc->flags & TCQ_F_BUILTIN || + !atomic_dec_and_test(&qdisc->refcnt)) + return; + + if (qdisc->parent) + list_del(&qdisc->list); + #ifdef CONFIG_NET_SCHED qdisc_put_stab(qdisc->stab); #endif @@ -542,20 +547,6 @@ static void __qdisc_destroy(struct rcu_head *head) kfree((char *) qdisc - qdisc->padded); } - -/* Under qdisc_lock(qdisc) and BH! */ - -void qdisc_destroy(struct Qdisc *qdisc) -{ - if (qdisc->flags & TCQ_F_BUILTIN || - !atomic_dec_and_test(&qdisc->refcnt)) - return; - - if (qdisc->parent) - list_del(&qdisc->list); - - call_rcu(&qdisc->q_rcu, __qdisc_destroy); -} EXPORT_SYMBOL(qdisc_destroy); static bool dev_all_qdisc_sleeping_noop(struct net_device *dev) -- cgit v1.2.3 From 2540e0511ea17e25831be543cdf9381e6209950d Mon Sep 17 00:00:00 2001 From: Jarek Poplawski Date: Thu, 21 Aug 2008 05:11:14 -0700 Subject: pkt_sched: Fix qdisc_watchdog() vs. dev_deactivate() race dev_deactivate() can skip rescheduling of a qdisc by qdisc_watchdog() or other timer calling netif_schedule() after dev_queue_deactivate(). We prevent this checking aliveness before scheduling the timer. Since during deactivation the root qdisc is available only as qdisc_sleeping additional accessor qdisc_root_sleeping() is created. With feedback from Herbert Xu Signed-off-by: Jarek Poplawski Signed-off-by: David S. Miller --- include/net/sch_generic.h | 5 +++++ net/sched/sch_api.c | 4 ++++ net/sched/sch_cbq.c | 4 ++++ 3 files changed, 13 insertions(+) (limited to 'include/net/sch_generic.h') diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index 84d25f2e6188..b1d2cfea89c5 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -193,6 +193,11 @@ static inline struct Qdisc *qdisc_root(struct Qdisc *qdisc) return qdisc->dev_queue->qdisc; } +static inline struct Qdisc *qdisc_root_sleeping(struct Qdisc *qdisc) +{ + return qdisc->dev_queue->qdisc_sleeping; +} + /* The qdisc root lock is a mechanism by which to top level * of a qdisc tree can be locked from any qdisc node in the * forest. This allows changing the configuration of some diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c index ef0efeca6352..45f442d7de47 100644 --- a/net/sched/sch_api.c +++ b/net/sched/sch_api.c @@ -444,6 +444,10 @@ void qdisc_watchdog_schedule(struct qdisc_watchdog *wd, psched_time_t expires) { ktime_t time; + if (test_bit(__QDISC_STATE_DEACTIVATED, + &qdisc_root_sleeping(wd->qdisc)->state)) + return; + wd->qdisc->flags |= TCQ_F_THROTTLED; time = ktime_set(0, 0); time = ktime_add_ns(time, PSCHED_US2NS(expires)); diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c index 47ef492c4ff4..8fa90d68ec6d 100644 --- a/net/sched/sch_cbq.c +++ b/net/sched/sch_cbq.c @@ -521,6 +521,10 @@ static void cbq_ovl_delay(struct cbq_class *cl) struct cbq_sched_data *q = qdisc_priv(cl->qdisc); psched_tdiff_t delay = cl->undertime - q->now; + if (test_bit(__QDISC_STATE_DEACTIVATED, + &qdisc_root_sleeping(cl->qdisc)->state)) + return; + if (!cl->delayed) { psched_time_t sched = q->now; ktime_t expires; -- cgit v1.2.3 From f6f9b93f1624206c802ac9162c9302edaf59bfd9 Mon Sep 17 00:00:00 2001 From: Jarek Poplawski Date: Wed, 27 Aug 2008 02:25:17 -0700 Subject: pkt_sched: Fix gen_estimator locks While passing a qdisc root lock to gen_new_estimator() and gen_replace_estimator() dev could be deactivated or even before grafting proper root qdisc as qdisc_sleeping (e.g. qdisc_create), so using qdisc_root_lock() is not enough. This patch adds qdisc_root_sleeping_lock() for this, plus additional checks, where necessary. Signed-off-by: Jarek Poplawski Signed-off-by: David S. Miller --- include/net/sch_generic.h | 8 ++++++++ net/sched/sch_api.c | 14 +++++++++++--- net/sched/sch_cbq.c | 4 ++-- net/sched/sch_hfsc.c | 4 ++-- net/sched/sch_htb.c | 4 ++-- 5 files changed, 25 insertions(+), 9 deletions(-) (limited to 'include/net/sch_generic.h') diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index b1d2cfea89c5..ef8a7e2e12e7 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -217,6 +217,14 @@ static inline spinlock_t *qdisc_root_lock(struct Qdisc *qdisc) return qdisc_lock(root); } +static inline spinlock_t *qdisc_root_sleeping_lock(struct Qdisc *qdisc) +{ + struct Qdisc *root = qdisc_root_sleeping(qdisc); + + ASSERT_RTNL(); + return qdisc_lock(root); +} + static inline struct net_device *qdisc_dev(struct Qdisc *qdisc) { return qdisc->dev_queue->dev; diff --git a/net/sched/sch_api.c b/net/sched/sch_api.c index ad9cda1b8c0a..506b709510b6 100644 --- a/net/sched/sch_api.c +++ b/net/sched/sch_api.c @@ -830,9 +830,16 @@ qdisc_create(struct net_device *dev, struct netdev_queue *dev_queue, sch->stab = stab; } if (tca[TCA_RATE]) { + spinlock_t *root_lock; + + if ((sch->parent != TC_H_ROOT) && + !(sch->flags & TCQ_F_INGRESS)) + root_lock = qdisc_root_sleeping_lock(sch); + else + root_lock = qdisc_lock(sch); + err = gen_new_estimator(&sch->bstats, &sch->rate_est, - qdisc_root_lock(sch), - tca[TCA_RATE]); + root_lock, tca[TCA_RATE]); if (err) { /* * Any broken qdiscs that would require @@ -884,7 +891,8 @@ static int qdisc_change(struct Qdisc *sch, struct nlattr **tca) if (tca[TCA_RATE]) gen_replace_estimator(&sch->bstats, &sch->rate_est, - qdisc_root_lock(sch), tca[TCA_RATE]); + qdisc_root_sleeping_lock(sch), + tca[TCA_RATE]); return 0; } diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c index 8fa90d68ec6d..9b720adedead 100644 --- a/net/sched/sch_cbq.c +++ b/net/sched/sch_cbq.c @@ -1839,7 +1839,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t if (tca[TCA_RATE]) gen_replace_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), + qdisc_root_sleeping_lock(sch), tca[TCA_RATE]); return 0; } @@ -1930,7 +1930,7 @@ cbq_change_class(struct Qdisc *sch, u32 classid, u32 parentid, struct nlattr **t if (tca[TCA_RATE]) gen_new_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), tca[TCA_RATE]); + qdisc_root_sleeping_lock(sch), tca[TCA_RATE]); *arg = (unsigned long)cl; return 0; diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c index c2b8d9cce3d2..c1e77da8cd09 100644 --- a/net/sched/sch_hfsc.c +++ b/net/sched/sch_hfsc.c @@ -1045,7 +1045,7 @@ hfsc_change_class(struct Qdisc *sch, u32 classid, u32 parentid, if (tca[TCA_RATE]) gen_replace_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), + qdisc_root_sleeping_lock(sch), tca[TCA_RATE]); return 0; } @@ -1104,7 +1104,7 @@ hfsc_change_class(struct Qdisc *sch, u32 classid, u32 parentid, if (tca[TCA_RATE]) gen_new_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), tca[TCA_RATE]); + qdisc_root_sleeping_lock(sch), tca[TCA_RATE]); *arg = (unsigned long)cl; return 0; } diff --git a/net/sched/sch_htb.c b/net/sched/sch_htb.c index 0df0df202ed0..97d4761cc31e 100644 --- a/net/sched/sch_htb.c +++ b/net/sched/sch_htb.c @@ -1372,7 +1372,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, goto failure; gen_new_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), + qdisc_root_sleeping_lock(sch), tca[TCA_RATE] ? : &est.nla); cl->refcnt = 1; cl->children = 0; @@ -1427,7 +1427,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid, } else { if (tca[TCA_RATE]) gen_replace_estimator(&cl->bstats, &cl->rate_est, - qdisc_root_lock(sch), + qdisc_root_sleeping_lock(sch), tca[TCA_RATE]); sch_tree_lock(sch); } -- cgit v1.2.3 From fe439dd09d3e4da6a44d35df7371b9c6a2661b99 Mon Sep 17 00:00:00 2001 From: Jarek Poplawski Date: Wed, 27 Aug 2008 02:27:10 -0700 Subject: pkt_sched: Fix sch_tree_lock() Use new qdisc_root_sleeping_lock() instead of qdisc_root_lock() as sch_tree_lock() because this lock could be used while dev is deactivated, but we never need to use this with noop_qdisc as a root. Signed-off-by: Jarek Poplawski Signed-off-by: David S. Miller --- include/net/sch_generic.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'include/net/sch_generic.h') diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h index ef8a7e2e12e7..e5569625d2a5 100644 --- a/include/net/sch_generic.h +++ b/include/net/sch_generic.h @@ -232,12 +232,12 @@ static inline struct net_device *qdisc_dev(struct Qdisc *qdisc) static inline void sch_tree_lock(struct Qdisc *q) { - spin_lock_bh(qdisc_root_lock(q)); + spin_lock_bh(qdisc_root_sleeping_lock(q)); } static inline void sch_tree_unlock(struct Qdisc *q) { - spin_unlock_bh(qdisc_root_lock(q)); + spin_unlock_bh(qdisc_root_sleeping_lock(q)); } #define tcf_tree_lock(tp) sch_tree_lock((tp)->q) -- cgit v1.2.3