pkt_sched: Add and use qdisc_root() and qdisc_root_lock().

When code wants to lock the qdisc tree state, the logic
operation it's doing is locking the top-level qdisc that
sits of the root of the netdev_queue.

Add qdisc_root_lock() to represent this and convert the
easiest cases.

In order for this to work out in all cases, we have to
hook up the noop_qdisc to a dummy netdev_queue.

Signed-off-by: David S. Miller <davem@davemloft.net>

7 files changed, 55 insertions, 26 deletions

diff --git a/include/net/sch_generic.h b/include/net/sch_generic.h
index bc2a09da21b1..92417825d387 100644
--- a/include/net/sch_generic.h
+++ b/include/net/sch_generic.h
@@ -161,6 +161,18 @@ struct tcf_proto
         struct tcf_proto_ops    *ops;
 };
 
+static inline struct Qdisc *qdisc_root(struct Qdisc *qdisc)
+{
+        return qdisc->dev_queue->qdisc;
+}
+
+static inline spinlock_t *qdisc_root_lock(struct Qdisc *qdisc)
+{
+        struct Qdisc *root = qdisc_root(qdisc);
+
+        return &root->dev_queue->lock;
+}
+
 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 830ccc544a15..19c244a00839 100644
--- a/net/sched/sch_api.c
+++ b/net/sched/sch_api.c
@@ -633,7 +633,7 @@ qdisc_create(struct net_device *dev, struct netdev_queue *dev_queue,
         if (!ops->init || (err = ops->init(sch, tca[TCA_OPTIONS])) == 0) {
                 if (tca[TCA_RATE]) {
                         err = gen_new_estimator(&sch->bstats, &sch->rate_est,
-                                                &sch->dev_queue->lock,
+                                                qdisc_root_lock(sch),
                                                 tca[TCA_RATE]);
                         if (err) {
                                 /*
@@ -675,7 +675,7 @@ static int qdisc_change(struct Qdisc *sch, struct nlattr **tca)
         }
         if (tca[TCA_RATE])
                 gen_replace_estimator(&sch->bstats, &sch->rate_est,
-                                      &sch->dev_queue->lock, tca[TCA_RATE]);
+                                      qdisc_root_lock(sch), tca[TCA_RATE]);
         return 0;
 }
 
@@ -967,7 +967,7 @@ static int tc_fill_qdisc(struct sk_buff *skb, struct Qdisc *q, u32 clid,
         q->qstats.qlen = q->q.qlen;
 
         if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS,
-                                         TCA_XSTATS, &q->dev_queue->lock, &d) < 0)
+                                         TCA_XSTATS, qdisc_root_lock(q), &d) < 0)
                 goto nla_put_failure;
 
         if (q->ops->dump_stats && q->ops->dump_stats(q, &d) < 0)
@@ -1216,7 +1216,7 @@ static int tc_fill_tclass(struct sk_buff *skb, struct Qdisc *q,
                 goto nla_put_failure;
 
         if (gnet_stats_start_copy_compat(skb, TCA_STATS2, TCA_STATS,
-                                         TCA_XSTATS, &q->dev_queue->lock, &d) < 0)
+                                         TCA_XSTATS, qdisc_root_lock(q), &d) < 0)
                 goto nla_put_failure;
 
         if (cl_ops->dump_stats && cl_ops->dump_stats(q, cl, &d) < 0)
diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c
index 4efc836cbf38..37ae653db683 100644
--- a/net/sched/sch_cbq.c
+++ b/net/sched/sch_cbq.c
@@ -1744,12 +1744,13 @@ static void cbq_put(struct Qdisc *sch, unsigned long arg)
 
         if (--cl->refcnt == 0) {
 #ifdef CONFIG_NET_CLS_ACT
+                spinlock_t *root_lock = qdisc_root_lock(sch);
                 struct cbq_sched_data *q = qdisc_priv(sch);
 
-                spin_lock_bh(&sch->dev_queue->lock);
+                spin_lock_bh(root_lock);
                 if (q->rx_class == cl)
                         q->rx_class = NULL;
-                spin_unlock_bh(&sch->dev_queue->lock);
+                spin_unlock_bh(root_lock);
 #endif
 
                 cbq_destroy_class(sch, cl);
@@ -1828,7 +1829,7 @@ 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,
-                                              &sch->dev_queue->lock,
+                                              qdisc_root_lock(sch),
                                               tca[TCA_RATE]);
                 return 0;
         }
@@ -1919,7 +1920,7 @@ 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,
-                                  &sch->dev_queue->lock, tca[TCA_RATE]);
+                                  qdisc_root_lock(sch), tca[TCA_RATE]);
 
         *arg = (unsigned long)cl;
         return 0;
diff --git a/net/sched/sch_generic.c b/net/sched/sch_generic.c
index ac208c2b2d10..739a8711ab30 100644
--- a/net/sched/sch_generic.c
+++ b/net/sched/sch_generic.c
@@ -151,14 +151,17 @@ static inline int qdisc_restart(struct netdev_queue *txq,
 {
         int ret = NETDEV_TX_BUSY;
         struct net_device *dev;
+        spinlock_t *root_lock;
         struct sk_buff *skb;
 
         /* Dequeue packet */
         if (unlikely((skb = dequeue_skb(q)) == NULL))
                 return 0;
 
-        /* And release queue */
-        spin_unlock(&txq->lock);
+        root_lock = qdisc_root_lock(q);
+
+        /* And release qdisc */
+        spin_unlock(root_lock);
 
         dev = txq->dev;
 
@@ -167,7 +170,7 @@ static inline int qdisc_restart(struct netdev_queue *txq,
                 ret = dev_hard_start_xmit(skb, dev, txq);
         HARD_TX_UNLOCK(dev, txq);
 
-        spin_lock(&txq->lock);
+        spin_lock(root_lock);
 
         switch (ret) {
         case NETDEV_TX_OK:
@@ -345,12 +348,18 @@ struct Qdisc_ops noop_qdisc_ops __read_mostly = {
         .owner          =       THIS_MODULE,
 };
 
+static struct netdev_queue noop_netdev_queue = {
+        .lock           =       __SPIN_LOCK_UNLOCKED(noop_netdev_queue.lock),
+        .qdisc          =       &noop_qdisc,
+};
+
 struct Qdisc noop_qdisc = {
         .enqueue        =       noop_enqueue,
         .dequeue        =       noop_dequeue,
         .flags          =       TCQ_F_BUILTIN,
         .ops            =       &noop_qdisc_ops,
         .list           =       LIST_HEAD_INIT(noop_qdisc.list),
+        .dev_queue      =       &noop_netdev_queue,
 };
 EXPORT_SYMBOL(noop_qdisc);
 
@@ -666,19 +675,21 @@ static bool some_qdisc_is_running(struct net_device *dev, int lock)
 
         for (i = 0; i < dev->num_tx_queues; i++) {
                 struct netdev_queue *dev_queue;
+                spinlock_t *root_lock;
                 struct Qdisc *q;
                 int val;
 
                 dev_queue = netdev_get_tx_queue(dev, i);
                 q = dev_queue->qdisc;
+                root_lock = qdisc_root_lock(q);
 
                 if (lock)
-                        spin_lock_bh(&dev_queue->lock);
+                        spin_lock_bh(root_lock);
 
                 val = test_bit(__QDISC_STATE_RUNNING, &q->state);
 
                 if (lock)
-                        spin_unlock_bh(&dev_queue->lock);
+                        spin_unlock_bh(root_lock);
 
                 if (val)
                         return true;
diff --git a/net/sched/sch_hfsc.c b/net/sched/sch_hfsc.c
index 997d520ca580..5090708ba384 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,
-                                              &sch->dev_queue->lock,
+                                              qdisc_root_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,
-                                  &sch->dev_queue->lock, tca[TCA_RATE]);
+                                  qdisc_root_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 c8ca54cc26b0..ee48457eaa4a 100644
--- a/net/sched/sch_htb.c
+++ b/net/sched/sch_htb.c
@@ -1039,11 +1039,12 @@ static int htb_init(struct Qdisc *sch, struct nlattr *opt)
 
 static int htb_dump(struct Qdisc *sch, struct sk_buff *skb)
 {
+        spinlock_t *root_lock = qdisc_root_lock(sch);
         struct htb_sched *q = qdisc_priv(sch);
         struct nlattr *nest;
         struct tc_htb_glob gopt;
 
-        spin_lock_bh(&sch->dev_queue->lock);
+        spin_lock_bh(root_lock);
 
         gopt.direct_pkts = q->direct_pkts;
         gopt.version = HTB_VER;
@@ -1057,11 +1058,11 @@ static int htb_dump(struct Qdisc *sch, struct sk_buff *skb)
         NLA_PUT(skb, TCA_HTB_INIT, sizeof(gopt), &gopt);
         nla_nest_end(skb, nest);
 
-        spin_unlock_bh(&sch->dev_queue->lock);
+        spin_unlock_bh(root_lock);
         return skb->len;
 
 nla_put_failure:
-        spin_unlock_bh(&sch->dev_queue->lock);
+        spin_unlock_bh(root_lock);
         nla_nest_cancel(skb, nest);
         return -1;
 }
@@ -1070,10 +1071,11 @@ static int htb_dump_class(struct Qdisc *sch, unsigned long arg,
                           struct sk_buff *skb, struct tcmsg *tcm)
 {
         struct htb_class *cl = (struct htb_class *)arg;
+        spinlock_t *root_lock = qdisc_root_lock(sch);
         struct nlattr *nest;
         struct tc_htb_opt opt;
 
-        spin_lock_bh(&sch->dev_queue->lock);
+        spin_lock_bh(root_lock);
         tcm->tcm_parent = cl->parent ? cl->parent->common.classid : TC_H_ROOT;
         tcm->tcm_handle = cl->common.classid;
         if (!cl->level && cl->un.leaf.q)
@@ -1095,11 +1097,11 @@ static int htb_dump_class(struct Qdisc *sch, unsigned long arg,
         NLA_PUT(skb, TCA_HTB_PARMS, sizeof(opt), &opt);
 
         nla_nest_end(skb, nest);
-        spin_unlock_bh(&sch->dev_queue->lock);
+        spin_unlock_bh(root_lock);
         return skb->len;
 
 nla_put_failure:
-        spin_unlock_bh(&sch->dev_queue->lock);
+        spin_unlock_bh(root_lock);
         nla_nest_cancel(skb, nest);
         return -1;
 }
@@ -1365,7 +1367,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
                         goto failure;
 
                 gen_new_estimator(&cl->bstats, &cl->rate_est,
-                                  &sch->dev_queue->lock,
+                                  qdisc_root_lock(sch),
                                   tca[TCA_RATE] ? : &est.nla);
                 cl->refcnt = 1;
                 cl->children = 0;
@@ -1420,7 +1422,7 @@ static int htb_change_class(struct Qdisc *sch, u32 classid,
         } else {
                 if (tca[TCA_RATE])
                         gen_replace_estimator(&cl->bstats, &cl->rate_est,
-                                              &sch->dev_queue->lock,
+                                              qdisc_root_lock(sch),
                                               tca[TCA_RATE]);
                 sch_tree_lock(sch);
         }
diff --git a/net/sched/sch_netem.c b/net/sched/sch_netem.c
index bc585f2089ff..c5ea40c9eb21 100644
--- a/net/sched/sch_netem.c
+++ b/net/sched/sch_netem.c
@@ -180,7 +180,7 @@ static int netem_enqueue(struct sk_buff *skb, struct Qdisc *sch)
          * skb will be queued.
          */
         if (count > 1 && (skb2 = skb_clone(skb, GFP_ATOMIC)) != NULL) {
-                struct Qdisc *rootq = sch->dev_queue->qdisc;
+                struct Qdisc *rootq = qdisc_root(sch);
                 u32 dupsave = q->duplicate; /* prevent duplicating a dup... */
                 q->duplicate = 0;
 
@@ -319,6 +319,7 @@ static int get_dist_table(struct Qdisc *sch, const struct nlattr *attr)
         struct netem_sched_data *q = qdisc_priv(sch);
         unsigned long n = nla_len(attr)/sizeof(__s16);
         const __s16 *data = nla_data(attr);
+        spinlock_t *root_lock;
         struct disttable *d;
         int i;
 
@@ -333,9 +334,11 @@ static int get_dist_table(struct Qdisc *sch, const struct nlattr *attr)
         for (i = 0; i < n; i++)
                 d->table[i] = data[i];
 
-        spin_lock_bh(&sch->dev_queue->lock);
+        root_lock = qdisc_root_lock(sch);
+
+        spin_lock_bh(root_lock);
         d = xchg(&q->delay_dist, d);
-        spin_unlock_bh(&sch->dev_queue->lock);
+        spin_unlock_bh(root_lock);
 
         kfree(d);
         return 0;