From b379135c40163ae79ba7a54e6928b53983e74ee8 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Sat, 1 Sep 2012 03:19:57 +0000 Subject: fq_codel: dont reinit flow state When fq_codel builds a new flow, it should not reset codel state. Codel algo needs to get previous values (lastcount, drop_next) to get proper behavior. Signed-off-by: Dave Taht Signed-off-by: Eric Dumazet Acked-by: Dave Taht Signed-off-by: David S. Miller --- net/sched/sch_fq_codel.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/sched') diff --git a/net/sched/sch_fq_codel.c b/net/sched/sch_fq_codel.c index 9fc1c62ec80e..4e606fcb2534 100644 --- a/net/sched/sch_fq_codel.c +++ b/net/sched/sch_fq_codel.c @@ -191,7 +191,6 @@ static int fq_codel_enqueue(struct sk_buff *skb, struct Qdisc *sch) if (list_empty(&flow->flowchain)) { list_add_tail(&flow->flowchain, &q->new_flows); - codel_vars_init(&flow->cvars); q->new_flow_count++; flow->deficit = q->quantum; flow->dropped = 0; @@ -418,6 +417,7 @@ static int fq_codel_init(struct Qdisc *sch, struct nlattr *opt) struct fq_codel_flow *flow = q->flows + i; INIT_LIST_HEAD(&flow->flowchain); + codel_vars_init(&flow->cvars); } } if (sch->limit >= 1) -- cgit v1.2.3 From bdfc87f7d1e253e0a61e2fc6a75ea9d76f7fc03a Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 11 Sep 2012 13:11:12 +0000 Subject: net-sched: sch_cbq: avoid infinite loop Its possible to setup a bad cbq configuration leading to an infinite loop in cbq_classify() DEV_OUT=eth0 ICMP="match ip protocol 1 0xff" U32="protocol ip u32" DST="match ip dst" tc qdisc add dev $DEV_OUT root handle 1: cbq avpkt 1000 \ bandwidth 100mbit tc class add dev $DEV_OUT parent 1: classid 1:1 cbq \ rate 512kbit allot 1500 prio 5 bounded isolated tc filter add dev $DEV_OUT parent 1: prio 3 $U32 \ $ICMP $DST 192.168.3.234 flowid 1: Reported-by: Denys Fedoryschenko Tested-by: Denys Fedoryschenko Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/sched/sch_cbq.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'net/sched') diff --git a/net/sched/sch_cbq.c b/net/sched/sch_cbq.c index 6aabd77d1cfd..564b9fc8efd3 100644 --- a/net/sched/sch_cbq.c +++ b/net/sched/sch_cbq.c @@ -250,10 +250,11 @@ cbq_classify(struct sk_buff *skb, struct Qdisc *sch, int *qerr) else if ((cl = defmap[res.classid & TC_PRIO_MAX]) == NULL) cl = defmap[TC_PRIO_BESTEFFORT]; - if (cl == NULL || cl->level >= head->level) + if (cl == NULL) goto fallback; } - + if (cl->level >= head->level) + goto fallback; #ifdef CONFIG_NET_CLS_ACT switch (result) { case TC_ACT_QUEUED: -- cgit v1.2.3 From e29fe837bfa3ca0a9f4ef1d4a90e6e0a74b6b8a0 Mon Sep 17 00:00:00 2001 From: David Ward Date: Thu, 13 Sep 2012 05:22:32 +0000 Subject: net_sched: gred: correct comment about qavg calculation in RIO mode Signed-off-by: David Ward Acked-by: Jamal Hadi Salim Signed-off-by: David S. Miller --- net/sched/sch_gred.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/sched') diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c index e901583e4ea5..fca73cdf44d9 100644 --- a/net/sched/sch_gred.c +++ b/net/sched/sch_gred.c @@ -176,7 +176,7 @@ static int gred_enqueue(struct sk_buff *skb, struct Qdisc *sch) skb->tc_index = (skb->tc_index & ~GRED_VQ_MASK) | dp; } - /* sum up all the qaves of prios <= to ours to get the new qave */ + /* sum up all the qaves of prios < ours to get the new qave */ if (!gred_wred_mode(t) && gred_rio_mode(t)) { int i; -- cgit v1.2.3 From c22e464022f935b0cbd8724b1d99d800d49518a9 Mon Sep 17 00:00:00 2001 From: David Ward Date: Thu, 13 Sep 2012 05:22:33 +0000 Subject: net_sched: gred: eliminate redundant DP prio comparisons Each pair of DPs only needs to be compared once when searching for a non-unique prio value. Signed-off-by: David Ward Acked-by: Jamal Hadi Salim Signed-off-by: David S. Miller --- net/sched/sch_gred.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'net/sched') diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c index fca73cdf44d9..e19d4ebfea1c 100644 --- a/net/sched/sch_gred.c +++ b/net/sched/sch_gred.c @@ -102,9 +102,8 @@ static inline int gred_wred_mode_check(struct Qdisc *sch) if (q == NULL) continue; - for (n = 0; n < table->DPs; n++) - if (table->tab[n] && table->tab[n] != q && - table->tab[n]->prio == q->prio) + for (n = i + 1; n < table->DPs; n++) + if (table->tab[n] && table->tab[n]->prio == q->prio) return 1; } -- cgit v1.2.3 From 1fe37b106b039d9358fd1211c39b1fa199e547a8 Mon Sep 17 00:00:00 2001 From: David Ward Date: Thu, 13 Sep 2012 05:22:34 +0000 Subject: net_sched: gred: fix qave reporting via netlink q->vars.qavg is a Wlog scaled value, but q->backlog is not. In order to pass q->vars.qavg as the backlog value, we need to un-scale it. Additionally, the qave value returned via netlink should not be Wlog scaled, so we need to un-scale the result of red_calc_qavg(). This caused artificially high values for "Average Queue" to be shown by 'tc -s -d qdisc', but did not affect the actual operation of GRED. Signed-off-by: David Ward Acked-by: Jamal Hadi Salim Signed-off-by: David S. Miller --- net/sched/sch_gred.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'net/sched') diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c index e19d4ebfea1c..b2570b59d85e 100644 --- a/net/sched/sch_gred.c +++ b/net/sched/sch_gred.c @@ -534,6 +534,7 @@ static int gred_dump(struct Qdisc *sch, struct sk_buff *skb) for (i = 0; i < MAX_DPs; i++) { struct gred_sched_data *q = table->tab[i]; struct tc_gred_qopt opt; + unsigned long qavg; memset(&opt, 0, sizeof(opt)); @@ -565,7 +566,9 @@ static int gred_dump(struct Qdisc *sch, struct sk_buff *skb) if (gred_wred_mode(table)) gred_load_wred_set(table, q); - opt.qave = red_calc_qavg(&q->parms, &q->vars, q->vars.qavg); + qavg = red_calc_qavg(&q->parms, &q->vars, + q->vars.qavg >> q->parms.Wlog); + opt.qave = qavg >> q->parms.Wlog; append_opt: if (nla_append(skb, sizeof(opt), &opt) < 0) -- cgit v1.2.3 From ba1bf474eae07a128fae6252bf7d68eaef0eacf8 Mon Sep 17 00:00:00 2001 From: David Ward Date: Thu, 13 Sep 2012 05:22:35 +0000 Subject: net_sched: gred: actually perform idling in WRED mode gred_dequeue() and gred_drop() do not seem to get called when the queue is empty, meaning that we never start idling while in WRED mode. And since qidlestart is not stored by gred_store_wred_set(), we would never stop idling while in WRED mode if we ever started. This messes up the average queue size calculation that influences packet marking/dropping behavior. Now, we start WRED mode idling as we are removing the last packet from the queue. Also we now actually stop WRED mode idling when we are enqueuing a packet. Cc: Bruce Osler Signed-off-by: David Ward Acked-by: Jamal Hadi Salim Signed-off-by: David S. Miller --- net/sched/sch_gred.c | 26 +++++++++++++++----------- 1 file changed, 15 insertions(+), 11 deletions(-) (limited to 'net/sched') diff --git a/net/sched/sch_gred.c b/net/sched/sch_gred.c index b2570b59d85e..d42234c0f13b 100644 --- a/net/sched/sch_gred.c +++ b/net/sched/sch_gred.c @@ -136,6 +136,7 @@ static inline void gred_store_wred_set(struct gred_sched *table, struct gred_sched_data *q) { table->wred_set.qavg = q->vars.qavg; + table->wred_set.qidlestart = q->vars.qidlestart; } static inline int gred_use_ecn(struct gred_sched *t) @@ -259,16 +260,18 @@ static struct sk_buff *gred_dequeue(struct Qdisc *sch) } else { q->backlog -= qdisc_pkt_len(skb); - if (!q->backlog && !gred_wred_mode(t)) - red_start_of_idle_period(&q->vars); + if (gred_wred_mode(t)) { + if (!sch->qstats.backlog) + red_start_of_idle_period(&t->wred_set); + } else { + if (!q->backlog) + red_start_of_idle_period(&q->vars); + } } return skb; } - if (gred_wred_mode(t) && !red_is_idling(&t->wred_set)) - red_start_of_idle_period(&t->wred_set); - return NULL; } @@ -290,19 +293,20 @@ static unsigned int gred_drop(struct Qdisc *sch) q->backlog -= len; q->stats.other++; - if (!q->backlog && !gred_wred_mode(t)) - red_start_of_idle_period(&q->vars); + if (gred_wred_mode(t)) { + if (!sch->qstats.backlog) + red_start_of_idle_period(&t->wred_set); + } else { + if (!q->backlog) + red_start_of_idle_period(&q->vars); + } } qdisc_drop(skb, sch); return len; } - if (gred_wred_mode(t) && !red_is_idling(&t->wred_set)) - red_start_of_idle_period(&t->wred_set); - return 0; - } static void gred_reset(struct Qdisc *sch) -- cgit v1.2.3