| Message ID | 20110114110342.4d95ad5b@nehalam |
|---|---|
| State | Accepted, archived |
| Delegated to: | David Miller |
| Headers | show |
Le vendredi 14 janvier 2011 à 11:03 -0800, Stephen Hemminger a écrit : > From Eric Dumazet <eric.dumazet@gmail.com> > > In commit 44b8288308ac9d (net_sched: pfifo_head_drop problem), we fixed > a problem with pfifo_head drops that incorrectly decreased > sch->bstats.bytes and sch->bstats.packets > > Several qdiscs (CHOKe, SFQ, pfifo_head, ...) are able to drop a > previously enqueued packet, and bstats cannot be changed, so > bstats/rates are not accurate (over estimated) > > This patch changes the qdisc_bstats updates to be done at dequeue() time > instead of enqueue() time. bstats counters no longer account for dropped > frames, and rates are more correct, since enqueue() bursts dont have > effect on dequeue() rate. > > Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> > Acked-by: Stephen Hemminger <shemminger@vyatta.com> > > CC: Patrick McHardy <kaber@trash.net> > CC: Jarek Poplawski <jarkao2@gmail.com> > CC: jamal <hadi@cyberus.ca> > --- > sch_fifo now changed to use __qdisc_queue_drop_head which > keeps correct statistics and is actually clearer. > > Thanks for doing this Stephen, this version seems fine. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Fri, Jan 14, 2011 at 11:03:42AM -0800, Stephen Hemminger wrote: > From Eric Dumazet <eric.dumazet@gmail.com> > > In commit 44b8288308ac9d (net_sched: pfifo_head_drop problem), we fixed > a problem with pfifo_head drops that incorrectly decreased > sch->bstats.bytes and sch->bstats.packets > > Several qdiscs (CHOKe, SFQ, pfifo_head, ...) are able to drop a > previously enqueued packet, and bstats cannot be changed, so > bstats/rates are not accurate (over estimated) > > This patch changes the qdisc_bstats updates to be done at dequeue() time > instead of enqueue() time. bstats counters no longer account for dropped > frames, and rates are more correct, since enqueue() bursts dont have > effect on dequeue() rate. > > Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> > Acked-by: Stephen Hemminger <shemminger@vyatta.com> > > CC: Patrick McHardy <kaber@trash.net> > CC: Jarek Poplawski <jarkao2@gmail.com> > CC: jamal <hadi@cyberus.ca> ... > --- a/net/sched/sch_drr.c 2011-01-14 09:19:00.830857886 -0800 > +++ b/net/sched/sch_drr.c 2011-01-14 09:28:20.398631228 -0800 > @@ -376,7 +376,6 @@ static int drr_enqueue(struct sk_buff *s > } > > bstats_update(&cl->bstats, skb); Why leave leaf classes with different stats? Jarek P. > - qdisc_bstats_update(sch, skb); > > sch->q.qlen++; > return err; ... -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jan 17, 2011 at 6:35 AM, Jarek Poplawski <jarkao2@gmail.com> wrote: > On Fri, Jan 14, 2011 at 11:03:42AM -0800, Stephen Hemminger wrote: >> From Eric Dumazet <eric.dumazet@gmail.com> >> >> In commit 44b8288308ac9d (net_sched: pfifo_head_drop problem), we fixed >> a problem with pfifo_head drops that incorrectly decreased >> sch->bstats.bytes and sch->bstats.packets >> >> Several qdiscs (CHOKe, SFQ, pfifo_head, ...) are able to drop a >> previously enqueued packet, and bstats cannot be changed, so >> bstats/rates are not accurate (over estimated) >> >> This patch changes the qdisc_bstats updates to be done at dequeue() time >> instead of enqueue() time. bstats counters no longer account for dropped >> frames, and rates are more correct, since enqueue() bursts dont have >> effect on dequeue() rate. >> >> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> >> Acked-by: Stephen Hemminger <shemminger@vyatta.com> >> >> CC: Patrick McHardy <kaber@trash.net> >> CC: Jarek Poplawski <jarkao2@gmail.com> >> CC: jamal <hadi@cyberus.ca> > ... >> --- a/net/sched/sch_drr.c 2011-01-14 09:19:00.830857886 -0800 >> +++ b/net/sched/sch_drr.c 2011-01-14 09:28:20.398631228 -0800 >> @@ -376,7 +376,6 @@ static int drr_enqueue(struct sk_buff *s >> } >> >> bstats_update(&cl->bstats, skb); > > Why leave leaf classes with different stats? > HTB also has the same problem, and I have fixed in my own product, but the patch was rejected. http://patchwork.ozlabs.org/patch/6227/ > Jarek P. > >> - qdisc_bstats_update(sch, skb); >> >> sch->q.qlen++; >> return err; > ...
Le dimanche 16 janvier 2011 à 23:35 +0100, Jarek Poplawski a écrit : > Why leave leaf classes with different stats? > I wanted to address this point in a followup patch, once general qdisc stats changes are accepted. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Le lundi 17 janvier 2011 à 08:09 +0800, Changli Gao a écrit : > On Mon, Jan 17, 2011 at 6:35 AM, Jarek Poplawski <jarkao2@gmail.com> wrote: > > On Fri, Jan 14, 2011 at 11:03:42AM -0800, Stephen Hemminger wrote: > >> From Eric Dumazet <eric.dumazet@gmail.com> > >> > >> In commit 44b8288308ac9d (net_sched: pfifo_head_drop problem), we fixed > >> a problem with pfifo_head drops that incorrectly decreased > >> sch->bstats.bytes and sch->bstats.packets > >> > >> Several qdiscs (CHOKe, SFQ, pfifo_head, ...) are able to drop a > >> previously enqueued packet, and bstats cannot be changed, so > >> bstats/rates are not accurate (over estimated) > >> > >> This patch changes the qdisc_bstats updates to be done at dequeue() time > >> instead of enqueue() time. bstats counters no longer account for dropped > >> frames, and rates are more correct, since enqueue() bursts dont have > >> effect on dequeue() rate. > >> > >> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> > >> Acked-by: Stephen Hemminger <shemminger@vyatta.com> > >> > >> CC: Patrick McHardy <kaber@trash.net> > >> CC: Jarek Poplawski <jarkao2@gmail.com> > >> CC: jamal <hadi@cyberus.ca> > > ... > >> --- a/net/sched/sch_drr.c 2011-01-14 09:19:00.830857886 -0800 > >> +++ b/net/sched/sch_drr.c 2011-01-14 09:28:20.398631228 -0800 > >> @@ -376,7 +376,6 @@ static int drr_enqueue(struct sk_buff *s > >> } > >> > >> bstats_update(&cl->bstats, skb); > > > > Why leave leaf classes with different stats? > > > > HTB also has the same problem, and I have fixed in my own product, but > the patch was rejected. > > http://patchwork.ozlabs.org/patch/6227/ > Hmm, considering qdisc stats are not used in kernel (only updated and reported to tc users) it seems to me counting arrival instead of departure rates is mostly useless for the user, if drops are ignored. (I am not speaking of direct drops, when we try to enqueue() this skb, but later ones, when another skb is enqueued and we drop a previously enqueued skb) User really wants to see the effective departure rate, to check its qdisc parameters in respect with kernel ones (HZ=100/1000, HIGH res timers off/on, ...) Arrival rates are of litle use. However, it might be good to have a second "bstats" only for dropped packets/bytes, or extend bstats in a compatible way (maybe adding fields to the end of structure) -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
* Eric Dumazet | 2011-01-17 08:17:49 [+0100]: >Hmm, considering qdisc stats are not used in kernel (only updated and >reported to tc users) it seems to me counting arrival instead of >departure rates is mostly useless for the user, if drops are ignored. > >(I am not speaking of direct drops, when we try to enqueue() this skb, >but later ones, when another skb is enqueued and we drop a previously >enqueued skb) > >User really wants to see the effective departure rate, to check its >qdisc parameters in respect with kernel ones (HZ=100/1000, HIGH res >timers off/on, ...) > >Arrival rates are of litle use. However, it might be good to have a >second "bstats" only for dropped packets/bytes, or extend bstats in a >compatible way (maybe adding fields to the end of structure) Sure, qdiscs like CHOKe, SFQ, pfifo_head are only analyzable with this kind of additional information. E.g. pfifo_head currently provides no statistic that the queue length is possible underestimated and tunning is required. Hagen -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
From: Stephen Hemminger <shemminger@vyatta.com> Date: Fri, 14 Jan 2011 11:03:42 -0800 > From Eric Dumazet <eric.dumazet@gmail.com> > > In commit 44b8288308ac9d (net_sched: pfifo_head_drop problem), we fixed > a problem with pfifo_head drops that incorrectly decreased > sch->bstats.bytes and sch->bstats.packets > > Several qdiscs (CHOKe, SFQ, pfifo_head, ...) are able to drop a > previously enqueued packet, and bstats cannot be changed, so > bstats/rates are not accurate (over estimated) > > This patch changes the qdisc_bstats updates to be done at dequeue() time > instead of enqueue() time. bstats counters no longer account for dropped > frames, and rates are more correct, since enqueue() bursts dont have > effect on dequeue() rate. > > Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com> > Acked-by: Stephen Hemminger <shemminger@vyatta.com> Applied to net-2.6, thanks everyone. -- To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
--- a/include/net/sch_generic.h 2011-01-14 09:19:00.730849868 -0800 +++ b/include/net/sch_generic.h 2011-01-14 09:47:59.058551676 -0800 @@ -445,7 +445,6 @@ static inline int __qdisc_enqueue_tail(s { __skb_queue_tail(list, skb); sch->qstats.backlog += qdisc_pkt_len(skb); - qdisc_bstats_update(sch, skb); return NET_XMIT_SUCCESS; } @@ -460,8 +459,10 @@ static inline struct sk_buff *__qdisc_de { struct sk_buff *skb = __skb_dequeue(list); - if (likely(skb != NULL)) + if (likely(skb != NULL)) { sch->qstats.backlog -= qdisc_pkt_len(skb); + qdisc_bstats_update(sch, skb); + } return skb; } @@ -474,10 +475,11 @@ static inline struct sk_buff *qdisc_dequ static inline unsigned int __qdisc_queue_drop_head(struct Qdisc *sch, struct sk_buff_head *list) { - struct sk_buff *skb = __qdisc_dequeue_head(sch, list); + struct sk_buff *skb = __skb_dequeue(list); if (likely(skb != NULL)) { unsigned int len = qdisc_pkt_len(skb); + sch->qstats.backlog -= len; kfree_skb(skb); return len; } --- a/net/sched/sch_cbq.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_cbq.c 2011-01-14 09:28:20.398631228 -0800 @@ -390,7 +390,6 @@ cbq_enqueue(struct sk_buff *skb, struct ret = qdisc_enqueue(skb, cl->q); if (ret == NET_XMIT_SUCCESS) { sch->q.qlen++; - qdisc_bstats_update(sch, skb); cbq_mark_toplevel(q, cl); if (!cl->next_alive) cbq_activate_class(cl); @@ -649,7 +648,6 @@ static int cbq_reshape_fail(struct sk_bu ret = qdisc_enqueue(skb, cl->q); if (ret == NET_XMIT_SUCCESS) { sch->q.qlen++; - qdisc_bstats_update(sch, skb); if (!cl->next_alive) cbq_activate_class(cl); return 0; @@ -971,6 +969,7 @@ cbq_dequeue(struct Qdisc *sch) skb = cbq_dequeue_1(sch); if (skb) { + qdisc_bstats_update(sch, skb); sch->q.qlen--; sch->flags &= ~TCQ_F_THROTTLED; return skb; --- a/net/sched/sch_drr.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_drr.c 2011-01-14 09:28:20.398631228 -0800 @@ -376,7 +376,6 @@ static int drr_enqueue(struct sk_buff *s } bstats_update(&cl->bstats, skb); - qdisc_bstats_update(sch, skb); sch->q.qlen++; return err; @@ -403,6 +402,7 @@ static struct sk_buff *drr_dequeue(struc skb = qdisc_dequeue_peeked(cl->qdisc); if (cl->qdisc->q.qlen == 0) list_del(&cl->alist); + qdisc_bstats_update(sch, skb); sch->q.qlen--; return skb; } --- a/net/sched/sch_dsmark.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_dsmark.c 2011-01-14 09:28:20.398631228 -0800 @@ -260,7 +260,6 @@ static int dsmark_enqueue(struct sk_buff return err; } - qdisc_bstats_update(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; @@ -283,6 +282,7 @@ static struct sk_buff *dsmark_dequeue(st if (skb == NULL) return NULL; + qdisc_bstats_update(sch, skb); sch->q.qlen--; index = skb->tc_index & (p->indices - 1); --- a/net/sched/sch_fifo.c 2011-01-09 09:34:32.690685246 -0800 +++ b/net/sched/sch_fifo.c 2011-01-14 10:43:39.534246186 -0800 @@ -46,17 +46,14 @@ static int pfifo_enqueue(struct sk_buff static int pfifo_tail_enqueue(struct sk_buff *skb, struct Qdisc* sch) { - struct sk_buff *skb_head; struct fifo_sched_data *q = qdisc_priv(sch); if (likely(skb_queue_len(&sch->q) < q->limit)) return qdisc_enqueue_tail(skb, sch); /* queue full, remove one skb to fulfill the limit */ - skb_head = qdisc_dequeue_head(sch); + __qdisc_queue_drop_head(sch, &sch->q); sch->qstats.drops++; - kfree_skb(skb_head); - qdisc_enqueue_tail(skb, sch); return NET_XMIT_CN; --- a/net/sched/sch_hfsc.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_hfsc.c 2011-01-14 09:28:20.428633918 -0800 @@ -1600,7 +1600,6 @@ hfsc_enqueue(struct sk_buff *skb, struct set_active(cl, qdisc_pkt_len(skb)); bstats_update(&cl->bstats, skb); - qdisc_bstats_update(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; @@ -1666,6 +1665,7 @@ hfsc_dequeue(struct Qdisc *sch) } sch->flags &= ~TCQ_F_THROTTLED; + qdisc_bstats_update(sch, skb); sch->q.qlen--; return skb; --- a/net/sched/sch_htb.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_htb.c 2011-01-14 09:28:20.438634799 -0800 @@ -574,7 +574,6 @@ static int htb_enqueue(struct sk_buff *s } sch->q.qlen++; - qdisc_bstats_update(sch, skb); return NET_XMIT_SUCCESS; } @@ -842,7 +841,7 @@ next: static struct sk_buff *htb_dequeue(struct Qdisc *sch) { - struct sk_buff *skb = NULL; + struct sk_buff *skb; struct htb_sched *q = qdisc_priv(sch); int level; psched_time_t next_event; @@ -851,6 +850,8 @@ static struct sk_buff *htb_dequeue(struc /* try to dequeue direct packets as high prio (!) to minimize cpu work */ skb = __skb_dequeue(&q->direct_queue); if (skb != NULL) { +ok: + qdisc_bstats_update(sch, skb); sch->flags &= ~TCQ_F_THROTTLED; sch->q.qlen--; return skb; @@ -884,11 +885,8 @@ static struct sk_buff *htb_dequeue(struc int prio = ffz(m); m |= 1 << prio; skb = htb_dequeue_tree(q, prio, level); - if (likely(skb != NULL)) { - sch->q.qlen--; - sch->flags &= ~TCQ_F_THROTTLED; - goto fin; - } + if (likely(skb != NULL)) + goto ok; } } sch->qstats.overlimits++; --- a/net/sched/sch_multiq.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_multiq.c 2011-01-14 09:28:20.438634799 -0800 @@ -83,7 +83,6 @@ multiq_enqueue(struct sk_buff *skb, stru ret = qdisc_enqueue(skb, qdisc); if (ret == NET_XMIT_SUCCESS) { - qdisc_bstats_update(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; } @@ -112,6 +111,7 @@ static struct sk_buff *multiq_dequeue(st qdisc = q->queues[q->curband]; skb = qdisc->dequeue(qdisc); if (skb) { + qdisc_bstats_update(sch, skb); sch->q.qlen--; return skb; } --- a/net/sched/sch_netem.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_netem.c 2011-01-14 09:28:20.438634799 -0800 @@ -240,7 +240,6 @@ static int netem_enqueue(struct sk_buff if (likely(ret == NET_XMIT_SUCCESS)) { sch->q.qlen++; - qdisc_bstats_update(sch, skb); } else if (net_xmit_drop_count(ret)) { sch->qstats.drops++; } @@ -289,6 +288,7 @@ static struct sk_buff *netem_dequeue(str skb->tstamp.tv64 = 0; #endif pr_debug("netem_dequeue: return skb=%p\n", skb); + qdisc_bstats_update(sch, skb); sch->q.qlen--; return skb; } @@ -476,7 +476,6 @@ static int tfifo_enqueue(struct sk_buff __skb_queue_after(list, skb, nskb); sch->qstats.backlog += qdisc_pkt_len(nskb); - qdisc_bstats_update(sch, nskb); return NET_XMIT_SUCCESS; } --- a/net/sched/sch_prio.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_prio.c 2011-01-14 09:28:20.438634799 -0800 @@ -84,7 +84,6 @@ prio_enqueue(struct sk_buff *skb, struct ret = qdisc_enqueue(skb, qdisc); if (ret == NET_XMIT_SUCCESS) { - qdisc_bstats_update(sch, skb); sch->q.qlen++; return NET_XMIT_SUCCESS; } @@ -116,6 +115,7 @@ static struct sk_buff *prio_dequeue(stru struct Qdisc *qdisc = q->queues[prio]; struct sk_buff *skb = qdisc->dequeue(qdisc); if (skb) { + qdisc_bstats_update(sch, skb); sch->q.qlen--; return skb; } --- a/net/sched/sch_red.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_red.c 2011-01-14 09:28:20.438634799 -0800 @@ -94,7 +94,6 @@ static int red_enqueue(struct sk_buff *s ret = qdisc_enqueue(skb, child); if (likely(ret == NET_XMIT_SUCCESS)) { - qdisc_bstats_update(sch, skb); sch->q.qlen++; } else if (net_xmit_drop_count(ret)) { q->stats.pdrop++; @@ -114,11 +113,13 @@ static struct sk_buff * red_dequeue(stru struct Qdisc *child = q->qdisc; skb = child->dequeue(child); - if (skb) + if (skb) { + qdisc_bstats_update(sch, skb); sch->q.qlen--; - else if (!red_is_idling(&q->parms)) - red_start_of_idle_period(&q->parms); - + } else { + if (!red_is_idling(&q->parms)) + red_start_of_idle_period(&q->parms); + } return skb; } --- a/net/sched/sch_sfq.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_sfq.c 2011-01-14 09:28:20.438634799 -0800 @@ -402,10 +402,8 @@ sfq_enqueue(struct sk_buff *skb, struct q->tail = slot; slot->allot = q->scaled_quantum; } - if (++sch->q.qlen <= q->limit) { - qdisc_bstats_update(sch, skb); + if (++sch->q.qlen <= q->limit) return NET_XMIT_SUCCESS; - } sfq_drop(sch); return NET_XMIT_CN; @@ -445,6 +443,7 @@ next_slot: } skb = slot_dequeue_head(slot); sfq_dec(q, a); + qdisc_bstats_update(sch, skb); sch->q.qlen--; sch->qstats.backlog -= qdisc_pkt_len(skb); --- a/net/sched/sch_tbf.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_tbf.c 2011-01-14 09:28:20.438634799 -0800 @@ -134,7 +134,6 @@ static int tbf_enqueue(struct sk_buff *s } sch->q.qlen++; - qdisc_bstats_update(sch, skb); return NET_XMIT_SUCCESS; } @@ -187,6 +186,7 @@ static struct sk_buff *tbf_dequeue(struc q->ptokens = ptoks; sch->q.qlen--; sch->flags &= ~TCQ_F_THROTTLED; + qdisc_bstats_update(sch, skb); return skb; } --- a/net/sched/sch_teql.c 2011-01-14 09:19:00.830857886 -0800 +++ b/net/sched/sch_teql.c 2011-01-14 09:28:20.438634799 -0800 @@ -83,7 +83,6 @@ teql_enqueue(struct sk_buff *skb, struct if (q->q.qlen < dev->tx_queue_len) { __skb_queue_tail(&q->q, skb); - qdisc_bstats_update(sch, skb); return NET_XMIT_SUCCESS; } @@ -107,6 +106,8 @@ teql_dequeue(struct Qdisc* sch) dat->m->slaves = sch; netif_wake_queue(m); } + } else { + qdisc_bstats_update(sch, skb); } sch->q.qlen = dat->q.qlen + dat_queue->qdisc->q.qlen; return skb;