pkt_sched: fq: fix pacing for small frames

For performance reasons, sch_fq tried hard to not setup timers for every
sent packet, using a quantum based heuristic : A delay is setup only if
the flow exhausted its credit.

Problem is that application limited flows can refill their credit
for every queued packet, and they can evade pacing.

This problem can also be triggered when TCP flows use small MSS values,
as TSO auto sizing builds packets that are smaller than the default fq
quantum (3028 bytes)

This patch adds a 40 ms delay to guard flow credit refill.

Fixes: afe4fd062416 ("pkt_sched: fq: Fair Queue packet scheduler")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Maciej Żenczykowski <maze@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/include/uapi/linux/pkt_sched.h b/include/uapi/linux/pkt_sched.h
index 885001b62c83b64ea3993bb10667464d50c7cfba..a806687ad98fc1258ae0b36c75164ecf5d59eec4 100644 (file)
--- a/include/uapi/linux/pkt_sched.h
+++ b/include/uapi/linux/pkt_sched.h
@@ -768,6 +768,9 @@ enum {
        TCA_FQ_FLOW_MAX_RATE,   /* per flow max rate */
 
        TCA_FQ_BUCKETS_LOG,     /* log2(number of buckets) */
+
+       TCA_FQ_FLOW_REFILL_DELAY,       /* flow credit refill delay in usec */
+
        __TCA_FQ_MAX
 };
 

diff --git a/net/sched/sch_fq.c b/net/sched/sch_fq.c
index 26906ab51c52ccd23647641466427a1ca7073ca0..95d843961907bf4b852743ea81015e7954db342b 100644 (file)
--- a/net/sched/sch_fq.c
+++ b/net/sched/sch_fq.c
@@ -88,6 +88,7 @@ struct fq_sched_data {
        struct fq_flow  internal;       /* for non classified or high prio packets */
        u32             quantum;
        u32             initial_quantum;
+       u32             flow_refill_delay;
        u32             flow_max_rate;  /* optional max rate per flow */
        u32             flow_plimit;    /* max packets per flow */
        struct rb_root  *fq_root;
@@ -114,6 +115,7 @@ static struct fq_flow detached, throttled;
 static void fq_flow_set_detached(struct fq_flow *f)
 {
        f->next = &detached;
+       f->age = jiffies;
 }
 
 static bool fq_flow_is_detached(const struct fq_flow *f)
@@ -366,17 +368,20 @@ static int fq_enqueue(struct sk_buff *skb, struct Qdisc *sch)
        }
 
        f->qlen++;
-       flow_queue_add(f, skb);
        if (skb_is_retransmit(skb))
                q->stat_tcp_retrans++;
        sch->qstats.backlog += qdisc_pkt_len(skb);
        if (fq_flow_is_detached(f)) {
                fq_flow_add_tail(&q->new_flows, f);
-               if (q->quantum > f->credit)
-                       f->credit = q->quantum;
+               if (time_after(jiffies, f->age + q->flow_refill_delay))
+                       f->credit = max_t(u32, f->credit, q->quantum);
                q->inactive_flows--;
                qdisc_unthrottled(sch);
        }
+
+       /* Note: this overwrites f->age */
+       flow_queue_add(f, skb);
+
        if (unlikely(f == &q->internal)) {
                q->stat_internal_packets++;
                qdisc_unthrottled(sch);
@@ -454,7 +459,6 @@ begin:
                        fq_flow_add_tail(&q->old_flows, f);
                } else {
                        fq_flow_set_detached(f);
-                       f->age = jiffies;
                        q->inactive_flows++;
                }
                goto begin;
@@ -608,6 +612,7 @@ static const struct nla_policy fq_policy[TCA_FQ_MAX + 1] = {
        [TCA_FQ_FLOW_DEFAULT_RATE]      = { .type = NLA_U32 },
        [TCA_FQ_FLOW_MAX_RATE]          = { .type = NLA_U32 },
        [TCA_FQ_BUCKETS_LOG]            = { .type = NLA_U32 },
+       [TCA_FQ_FLOW_REFILL_DELAY]      = { .type = NLA_U32 },
 };
 
 static int fq_change(struct Qdisc *sch, struct nlattr *opt)
@@ -664,6 +669,12 @@ static int fq_change(struct Qdisc *sch, struct nlattr *opt)
                        err = -EINVAL;
        }
 
+       if (tb[TCA_FQ_FLOW_REFILL_DELAY]) {
+               u32 usecs_delay = nla_get_u32(tb[TCA_FQ_FLOW_REFILL_DELAY]) ;
+
+               q->flow_refill_delay = usecs_to_jiffies(usecs_delay);
+       }
+
        if (!err)
                err = fq_resize(q, fq_log);
 
@@ -699,6 +710,7 @@ static int fq_init(struct Qdisc *sch, struct nlattr *opt)
        q->flow_plimit          = 100;
        q->quantum              = 2 * psched_mtu(qdisc_dev(sch));
        q->initial_quantum      = 10 * psched_mtu(qdisc_dev(sch));
+       q->flow_refill_delay    = msecs_to_jiffies(40);
        q->flow_max_rate        = ~0U;
        q->rate_enable          = 1;
        q->new_flows.first      = NULL;
@@ -733,6 +745,8 @@ static int fq_dump(struct Qdisc *sch, struct sk_buff *skb)
            nla_put_u32(skb, TCA_FQ_INITIAL_QUANTUM, q->initial_quantum) ||
            nla_put_u32(skb, TCA_FQ_RATE_ENABLE, q->rate_enable) ||
            nla_put_u32(skb, TCA_FQ_FLOW_MAX_RATE, q->flow_max_rate) ||
+           nla_put_u32(skb, TCA_FQ_FLOW_REFILL_DELAY,
+                       jiffies_to_usecs(q->flow_refill_delay)) ||
            nla_put_u32(skb, TCA_FQ_BUCKETS_LOG, q->fq_trees_log))
                goto nla_put_failure;